INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandardmargins, and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.
University Microfilms International A Bell & Howell Information Company 3 00 North Z eeb Road. Ann Arbor. Ml 48106-1346 USA 313/761-4700 800/521-0600
Order Number 9427740
The relationship between women’s self-efficacy expectations and career decision status at the college level in Korea
Lee, Ji-Yeon, Ph.D.
The Ohio State University, 1994
UMI 300 N. Zeeb Rd. Ann Arbor, MI 48106
THE RELATIONSHIP BETWEEN WOMEN’S SELF-EFFICACY
EXPECTATIONS AND CAREER DECISION STATUS
AT THE COLLEGE LEVEL IN KOREA
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
By
Ji-Yeon Lee, B.A.,M.Ed.
1994
Committee:
Dewey A. Adams Approved by
J. Robert Warmbrod (2. (U hsrtU L/ Samuel H, Osipow .dviser Vocational Education, Comprehensive Program To my husband, Yong Bok, and my family
ii ACKNOWLEDGEMENTS
As finishing this dissertation, I would like to express my sincere appreciation to my advisor, Dr. Dewey A. Adams for his support throughout my Ph.D. program. Also I owe a debt to the other members of my dissertation committee, Dr. Samuel H. Osipow and Dr. J. Robert Warmbrod for excellent guidance and assistance. Without them, I could not haVe completed this dissertation. They worked closely and patiently with me from the beginning to the implementation of this study.
I also express my appreciation to everyone who has helped, supported and assisted this study, especially to the participants of this study.
I wish to express special thanks to my parents and my other family members for their invaluable love and understanding during all of my study at the Ohio State University. I also thanks my husband, Yong Bok, and my son, Benjamin, for their incredible understanding, compassion, and contribution.
Finally, sincere thanks go to my Lord. VITA
September 7, 1964 ...... Bom-Seoul, Korea
1983 - 1987 ...... B.S. in Korean Literature, Sungkunkwan Univ., Seoul, Korea 1987 - 1989 ...... ffiMj Korea
1988 - 1990 ...... M.ED. in Korean Language Education, Sungkunkwan Univ., Seoul, Korea
FIELD OF STUDY
Major Field: Vocational Education Dr. Dewey A. Adams
Studies in Vocational Psychology Dr. Samuel H. Osipow
Studies in Research and Statistics Dr. J. Robert Warmbrod TABLE OF CONTENTS
DEDICATION ...... ii
ACKNOWLEDGEMENTS ...... iii
VITA ...... iv
LIST OF TABLES ...... vii
LIST OF FIGURES ...... xiii
CHAPTER PAGE
I. INTRODUCTION ...... 1
Need for the Study ...... 3 Problem Statement ...... 6 Purpose and Research Questions ...... 8 Research Hypotheses ...... 11 Definition of Term s...... 12 Basic Assumptions of Self-Efficacy Theory...... 14 Limitations of the Study ...... 15
II. REVIEW OF LITERATURE ...... 16
Theories of Career Development and Career Indecision ...... 17 The State of Knowledge about Career Indecision and the Career Decision Scale (CDS) ...... 20 Social Learning Theory ...... 26 Bandura’s Self-Efficacy Theory ...... 32 Hackett and Betz’s Self-Efficacy Theory...... 34 Task-Specific Career Self-Efficacy Research ...... 35 The Career Development of Women ...... 38 Explanation of Figure-1...... 45
v III. METHODOLOGY 49
Population and Sample ...... 49 Research Design ...... 51 Instrumentation ...... 51 Variables ...... 55 Data Collection ...... 58 Data Analysis ...... 58
IV. FINDINGS AND CONCLUSIONS ...... 61
Descriptive Data for Sample ...... 73 Data Analysis ...... 105 The First Question and Results ...... 106 The Second Question and Results ...... 133 The Third Question and Results ...... 158 The Fourth Question and Results ...... 190 The Fifth Question and Results ...... 218
V. SUMMARY, DISCUSSION AND RECOMMENDATIONS ...... 244
Limitations of the study...... 246 Sampling...... 247 Instrument ...... 247 Variables ...... 247 Data analysis...... 248 Summary of findings ...... 249 Discussion of findings ...... 266 Recommendations ...... 271
APPENDICES ...... 274 A. Career Decision Scale (CD S)...... 275 B. Task-Specific Occupational Self-Efficacy Scale (TSOSS) ...... 277 C. Selective Background Characteristics Questionnaire ...... 282
REFERENCES 287 LIST OF TABLES
TABLE PAGE
1. Categorization of Extraneous Variables Based on the Social Learning Theory ...... 31
2. Nature of School...... 63
3. Total Number of Sample by University ...... 63
4. Age by University ...... 74
5. Grade Level by University ...... 76
6. Detailed Information of College Area by University ...... 77
7. College Area by University ...... 78
8. Selected High-School Curriculum by University...... 80
9. Person Who Most Influenced Career Decision byUniversity ...... 82
10. Major Decision by University ...... 84
11. Additional Enrichment Educational Activities by University ...... 86
12. If, "YES",How Many? ...... 86
13. Educational Plan by Rank Order ...... 87
14. Job Plan by Rank O rd er ...... 88
15. Attitude Regarding Career and F am ily...... 89
16. Socio Economic Status(SES)-Father ...... 91 17. Socio Economic Status(SES)-Mother ...... 92
18. Educational Background for Father ...... 94
19. Educational Background for Mother ...... 95
20. Location of Education by University...... 96
21. Home Town by University ...... 97
22. Number of Different Type of Career Experience by University...... 98
23. Grade Point Average by University...... 99
24. CDS and TSOSS by University ...... 101
25. Career Indecision Score (Item 3 -18) ...... 102
26. TSOSS Total Score for Total Respondents . . 104
27. Testing Hypothesis for the Combined C a s e...... 108
28. Rc and Rc2 for the Total Case ...... 110
29. Mean and Standard Deviation for DV and IV Sets ...... I ll
30. Summary of Canonical Correlation Analysis for Total C a se...... 112
31. Testing Hypothesis for Case A University...... 113
32. Rc and Rc2 for the Case A University...... 115
33. Mean and Standard Deviation for DV and IV Sets ...... 116
34. Summary of Canonical Correlation Analysis for Case A University.... 117
35. Testing Hypothesis for Case B University...... 118
36. Rc and Sq. Rc for Case B University ...... 121
37. Mean and Standard Deviation for DV and IV Sets for Case B University 121
38. Summary of Canonical Correlation Analysis for Case B University.... 122
viii 39. Test Hypothesis for Case C University...... 123
40. Rc and Sq Rc for Case C University...... 125
41. Mean and Standard Deviation for DV and IV Sets for Case C University 126
42. Summary of Canonical Correlation Analysis for Case C University.... 127
43. Test Hypothesis for Case D University...... 128
44. Rc and Sq of Rc for Case D University ...... 131
45. Mean and Standard Deviation for DV and IV Sets for Case D University 131
46. Summary of Canonical Correlation Analysis for Case D University.... 132
47. Test Hypothesis for Total C ase...... 135
48. Rc and Sq of Rc for Total Case ...... 136
49. Mean and Standard Deviation for DV and IV Sets for Total Case .... 137
50. Summary of Canonical Correlation Analysis for Total C a se...... 138
51. Testing Hypothesis for Case A University...... 139
52. Rc and Sq of Rc for Case A University ...... 141
53. Mean and Standard Deviation for DV and IV Sets for Case A University 142
54. Summary of Canonical Correlation Analysis for Case A University.... 143
55. Testing Hypothesis for Case B University...... 144
56. Rc and Sq of Rc for Case B University...... 146
57. Mean and Standard Deviation for DV and IV Sets for Case B University 147
58. Summary of Canonical Correlation Analysis for Case B University.... 148
59. Test Hypothesis for Case C University...... 149
60. Rc and Sq of Rc for Case C University...... 152 61. Mean and Standard Deviation for DV and IV Sets for Case C University 153
62. Summary of Canonical Correlation Analysis for Case C University.... 154
63. Testing Hypothesis for Case D University...... 155
64. Mean and Standard Deviation for DV and IV Sets for Case D University 156
65. Summary of Canonical Correlation Analysis for Case D University.... 157
66. Testing Hypothesis for Total Case ...... 160
67. Rc and Sq of Rc for Total Case ...... 163
68. Mean and Standard Deviation for DV and IV Sets for Total Case .... 164
69. Summary of Canonical Correlation Analysis for Total C a se...... 165
70. Testing Hypothesis for Case A University...... 166
71. Rc and Sq of Rc for Case A University...... 169
72. Mean and Standard Deviation for DV and IV Sets for Case A University 170
73. Summary of Canonical Correlation Analysis for Case A University.... 171
74. Testing Hypothesis for Case B University...... 172
75. Rc and Sq of Rc for Case B University •. . 175
76. Mean and Standard Deviation for DV and IV Sets for Case B University 176
77. Summary of Canonical Correlation Analysis for Case B University.... 177
78. Testing Hypothesis for Case C University...... 178
79. Rc and Sq of Rc for Case C University...... 181
80. Mean and Standard Deviation for DV and IV Sets for Case C University 182
81. Summary of Canonical Correlation Analysis for Case C University.... 183
82. Testing Hypothesis for Case D University...... 184
x 83. Rc and Sq of Rc for Case D University ...... 187
84. Mean and Standard Deviation for DV and IV Sets for Case D University 188
85. Summary of Canonical Correlation Analysis for Case D University.... 189
86. Classification Results for Total Cases ...... 194
87. Mean and Standard Deviation for Discriminating Variables for Total Case...... 196
88. Summary Data for Discriminant Analysis for Total Cases...... 197
89. Classification Results Case A University ...... 199
90. Mean and Standard Deviation for Discriminating Variables for Case A University ...... 201
91. Summary Data for Discriminant Analysis for Case A University...... 202
92. Classification Results for Case B University ...... 204
93. Mean and Standard Deviation for Discriminating Variables for Case B University ...... 206
94. Summary Data for Discriminant Analysis for Case B University...... 207
95. Classification Results for Case C University ...... 209
96. Mean and Standard Deviation for Discriminating Variables for Case C University ...... 211
97. Summary Data for Discriminant Analysis for Case C University...... 212
98. Classification Results for Case D University ...... 214
99. Mean and Standard Deviation for Discriminating Variables for Case D University ...... 216
100. Summary Data for Discriminant Analysis for Case D University...... 217
101. Classification Results for Total C ases...... 219 02. Mean and Standard Deviation for Discriminating Variables for Total Case ...... 221
03. Summary Data for Discriminant Analysis for Total C a se...... s 223
04. Summary Data for Discriminant Analysis for Case A University...... 226
05. Mean and Standard Deviation for Discriminating Variables for Case A University ...... 227
06. Classification Results for Case B University...... 230
07. Summary Data for Discriminant Analysis for Case B University...... 232
08. Mean and Standard Deviation for Discriminating Variables for Case B University ...... 234
09. Summary Data for Discriminant Analysis for Case C University...... 237
10. Mean and Standard Deviation for Discriminating Variables for Case C University ...... 238
11. Summary Data for Discriminant Analysis for Case D University...... 241
12. Mean and Standard Deviation for Discriminating Variables for Case D University ...... 242
13. Relationship between 4 Sub-Scales of CDS and TSOSS ...... 250
14. Decision Regarding Ho ...... 252
15. Relationship between the Career Status and the 4 Sub-scales of CDS and TSOSS ...... 253
16. Summarized Relationship between the 4 Sub-Scales of TSOSS and Selected Variables ...... 256
17. Summarized Discriminating Variables on the Career Status ...... 260
18. Summarized Discriminating Variables on the Career Status ...... 265
xii LIST OF FIGURE
FIGURE PAGE
1. Inspired by Samuel H. Osipow (1973) Fundamental Idea of Career Education ...... 44 CHAPTER I
INTRODUCTION
It would be as absurd for one to undertake to educate the young with no knowledge of ...psychology, as for one to attempt to produce a sonata while ignorant of the phenomena of sound.
Louisa Parsons Hopkins, Educational Psychology, 1886, p.3. -
The last two decades have shown a dramatic increase in interest and research
regarding women’s career development. This interest in the study of women’s careers
directly parallels the increase in women’s labor force participation. For example, in
1984, 63% of all women aged 18 to 64 were working outside of the home; by 1995
the percentage should increase to 76% - 82% (U.S. Department of Labor, 1992).
Similarly, over ninety percent of Korean college/university female students had a positive attitude regarding participation in the world of work (Korean Education
Development Institute, 1993).
This trend toward the increase of women in the workforce must be taken into account in vocational education at the higher education level in order to meet the needs of female college students. Vocational education programs also need to be founded on a theoretical understanding of women’s career development. In fact, career development is especially important to individuals who are making the
1 2 decision to enter into continued education or the world of work. Any of these decisions is likely to present an ongoing task that is comprised of many parts and not a single, one-time decision. Despite great concern about women’s positive attitudes towards the world of work, career indecision is a common experience for most college students (Gordon, 1984). Moreover, female students tend to have more career indecision than male students due to perhaps socialization.
The concept of self-efficacy in Bandura’s (1977) writings on Social Learning
Theory has special application to women’s career indecision. In fact, career choice and aspiration are a natural extension of the concept of self-efficacy, and it seems clear that self-efficacy is an important variable among the many that contribute to how career decisions and aspirations unfold in the process of career development.
Making an optimizing career choice depends upon having a critical and rational perspective on one’s own wishes and life plan. However, if one’s own desire is destroyed by barriers in reality, and if one’s choice does not correspond to his/her original wishes, it becomes a special concern of educators to understand student’s process of career development.
It goes without saying that an effective program of vocational education for women must operate by accompanying career counseling with general pedagogical and psychological advice, based on a theoretical framework highlighting harmony between one’s career development and personal life plans. 3 Therefore, this study is designed to investigate the utility of Bandura’s self-
efficacy theory to the understanding and treatment of career indecision for females
in Korea.
Need for the Study
There are several important reasons for proposing this study.
First, there has been a dramatic increases in women’s labor force
participation. This has directly paralleled the interest and research into women’s
career development.
Second, research is needed on women’s career development from different
cultural backgrounds in providing empirical data to test the validity of westem-society
oriented theories of career development. According to vocational psychologists,
current theories of career development cannot be adequately applied to women
(Holland, 1966; Osipow, 1983; Brown and Brooks, 1990) because most studies are based on empirical data generated from studies examining men’s careers. They also
suggest that separate theories are needed because the results of the unique socialization due to race and sex have never been adequately considered in the present theories, which were developed to explain career development and occupational choice in western society.
Third, one common experience for young adults in their career development is career indecision (Gordon, 1984). However educators are often insufficiently concerned with factors that influence about one-half of students’ decisions on careers which they did not originally want to practice. In fact, "career indecision" prevents 4 effective vocational education by lowering both individual motivation and satisfaction,
even though career indecision can be viewed as a healthy response to the ambiguity
of the college environment and experience. Thus, special attention needs to be paid
to understanding the career indecision of students during the counseling process.
Despite many studies investigating potential correlates of indecision, there is a lack
of sufficient clarity in understanding in career indecision, so that career development
research and practice have no effective guide (Lucas & Epperson, 1990). Instead
studies suggest that career indecision is a complex and multidimensional construct.
Thus, further research is needed to better understand the complex phenomenon of
career indecision.
Fourth, one potentially valuable approach to refining our understanding of
career indecision is the new concept of "self-efficacy"that has come out of the work
of Bandura (1978). He states that self-efficacy reflects one’s belief in the ability to
perform a behavior that will lead to a desirable outcome. Later Hackett and Betz
(1981) applied self-efficacy theory to career development, theorizing that self-efficacy
expectations determine the range of perceived career and academic options, as well as people’s perseverance and ultimate success in their careers. Hackett and Betz
(1981) also pointed out that self-efficacy theory could be useful in helping one to understand the career development of women.
However, despite growing interest in self-efficacy theory and its promise (e.g.,
Hackett and Betz, 1981; Lent, Brown and Larkin 1984; 1986; Branch and
Lichtenberg, 1987; Bores-Rangel, Church, Szendre, and Reeves, 1990; Matsui and 5 Tsukamoto’s, 1991; Osipow and Rooney, 1989; Rooney and Osipow, 1992), research
on the relationship of women’s career indecision to self-efficacy expectations, and
such selected extraneous variables as age, grade level, college area, socio economic
status (SES), parent’s educational background, additional enrichment educational
activities, perceived work experience, educational and job plan after completing
college, location where most education was completed, home town, grade point
average, and the attitude regarding whether considerations of planning to have a
family had influenced their career plans in any way.
Fifth, the implications of this study’s findings for the counseling of career
undecided individuals might be extremely important by suggesting increasing self-
efficacy expectations related to those tasks and behaviors, since theoretically, low expectations of self-efficacy would lead to avoidance of those tasks and behaviors and consequently, continued indecision.
In summary, this study is designed to investigate the utility of Bandura’s self- efficacy theory to the understanding and treatment of career indecision in female students in Korea. This will be accomplished through the administration of instruments translated into Korean including the Task Specific Occupational Self- efficacy Scale(TSOSS) developed by Osipow and Rooney (1989), The Career
Decision Scale (CDS) developed by Osipow and others (1978), and as well as one short background characteristics questionnaire developed by the researcher. 6 Problem Statement
Vocational guidance counselors, college advisors, and career counselors have long observed that persons may be undecided about their vocational or career choices. While much research in the field of career indecision has identified the demographic variables associated with career indecision in college students, relatively little research has been found to suggest the way to treat the processes associated with career indecision. Consequently most of the research in career indecision has failed to predict career indecision, and has found difficulty in identifying background variables associated with career indecision (Gordon, 1984; Osipow, 1980).
One component of indecision in need of further elaboration is that summarized by Osipow et al. (1976) as a lack of structure and confidence. in combination with choice anxiety, leading to the avoidance of a vocational choice.
Another component identified by Holland and Holland (1977) is a lack of confidence in decision-making skills. The consistent appearance of this component in factor- analytic studies (e.g.,Kazin, 1976; Osipow et al., 1976; Slaney et al., 1981) suggests its importance as an explanatory variable, but further refinements in its conceptualization, including the development of assessment methods having more direct implications for intervention, would increase its usefulness. One potentially valuable approach to the refinement of the component of lack of structure and confidence, particularly in terms of applications to intervention strategies involves the concept of self-efficacy expectations. These expectations are a person’s beliefs concerning his/her ability to successfully perform a given task or behavior. This 7 approach was postulated by Bandura(T9771 to be the major mediator of behavior and behavior change through four methods of performance: accomplishments, modeling, emotional arousal and verbal persuasion (Bandura. 1978a). According to Bandura, it is at the task level that self-efficacy has an effect on behavior. One leading instrument to assess self-efficacy expectations at the task-specific level is the Task-
Specific Occupational Self-Efficacy Scale (TSOSS) developed by Osipow and Rooney
(1989).
Despite growing concerns for self-efficacy theory and its promise (e.g.,Hackett and Betz, 1981; Lent, Brown, and Larkin, 1984; 1986; Branch and Lichtenberg, 1987;
Bores-Rangel, Church, Szendre, and Reeves, 1990; Matsui and Tsukamoto’s, 1991,
Osipow and Rooney, 1989; Rooney and Osipow, 1992) there is a weak research effort on different cultural background in the relationship of women’s career indecision, self-efficacy expectations at the task level, and extraneous variables such as personal, educational, environmental and attitude toward family and career.
In summary, there is no clear research to support a hypothesis of relationship between career indecision, self-efficacy expectations and other selected background characteristics. 8 Purpose and Research Questions
This study was designed to investigate the utility of Bandura’s self-efficacy theory to the understanding and treatment of career indecision by female students in Korea. Focuses are upon investigation of the relationship between career indecision and self-efficacy expectations as measured at the task-specific level.
The major purposes of this study are to 1) investigate the relationship between women’s career indecision, self-efficacy expectations, and other selected background characteristics such as age, grade level, college area, socio economic status (SES), parent’s educational background, additional enrichment educational activities, perceived work experience, attitude regarding whether considerations of planning to have a family had influenced their career plans in any way, educational and job plan after completing college, location where most education was completed, home town, and grade point average; and 2) to describe the differentiation between decided and undecided students on the suggested variables.
More specifically, this study will be conducted to answer the following research questions:
1. Are the four sub-scales of CDS related to the four sub-scales of
TSOSS ?
2. What is the relationship between four the sub-scales of CDS and age,
grade level, college area, socio economic status (SES), parent’s
educational background, additional enrichment educational activities,
perceived work experience, attitude regarding whether considerations of planning to have a family had influenced their career plans in any
way, educational and job plan after completing college, location where
most education was completed, home town, and grade point average
?
3. What is the relationship between four sub-scales of TSOSS and age,
grade level, college area, socio economic status (SES), parent’s
educational background, additional enrichment educational activities,
perceived work experience, attitude regarding whether considerations
of planning to have a family had influenced their career plans in any
way, educational and job plan after completing college, location where
most education was completed, home town, and grade point average
?
4. Are the four sub-scales of CDS and the four sub-scales of TSOSS
related to the students’ career decision status regarding decided or
undecided ? If the groups differ, describe how the groups are similar
or different. What variables are important in describing how the
groups differ?
5. Are there any discriminant variables to distinguish both groups of
decided and undecided on the basis of such variables such as age,
.grade level, college area, additional enrichment educational activities,
educational and job plans after completing college, attitude regarding
whether considerations of planning to have a family had influenced 10 their career plans in any way, socio economic status (SES), parent’s educational background, location where most education are completed, home town, perceived work experience, and grade point average ? If the groups differ, describe how the groups are similar or different.
What variables are important in describing how the groups differ?
r 11 Research Hypotheses
Based on each research questions, research hypotheses have been constructed.
1. There is a relationship between the four sub-scales of CDS and the task specific four sub-scales of the TSOSS.
2. There is a relationship between the four sub-scales of CDS and the selected variables age, grade level, college area, additional enrichment educational activities, educational and job plans after completing college, attitude regarding whether considerations of planning to have a family had influenced their career plans in any way, socio economic status (SES), parent’s educational background, location where most education was completed, home town, perceived work experience, and grade point average.
3. There is a relationship between the four sub-scales of the TSOSS and the selected variables age, grade level, college area, additional enrichment educational activities, educational and job plans after completing college, attitude regarding whether considerations of planning to have a family had influenced their career plans in any way, socio economic status (SES), parent’s educational background, location where most education was completed, home town, previous work experience, and grade point average.
4. There is a differentiation between undecided group and decided group on the 4 sub-scales of CDS and 4 sub-scales of TSOSS.
5. There is a differentiation between the undecided group and the decided group on the selected variables age, grade level, college area, additional enrichment 12 educational activities, educational and job plans after completing college, attitude
regarding whether considerations of planning to have a family had influenced their
career plans in any way, socio economic status (SES), parent’s educational
background, location where most of education was completed, home town, previous
work experience, and grade point average.
Definition of Terms
Self-Efficacy expectations is the belief in one’s ability to successfully
perform a given behavior (Bandura, 1977). The assumption is that self-efficacy
expectation determines whether one will attempt a behavior and how long one will persist in a given behavior.
Career indecision is the inability to reach closure for educational or vocational decisions (Osipow, 1980).
The Career Decision Scale (CDS') was developed by Osipow, Carney,
Winer, Yanico, and Koschier (1976) and was designed to standardize the identification of sources of career indecision, thus serving as a diagnostic aid by employing a series of structured interventions. In a graduate seminar led by Dr.
Osipow, 16 statements that clients often make to describe their state of career indecision were identified. These statements became the basic CDS items. The validity of CDS has been supported by much research. The rationale of CDS is that a finite number of relatively discrete circumstances are responsible for the problems people have in reaching career decisions (Osipow, 1987). 13 Four sub-scales, of CDS : These were constructed on the basis of previous factor analysis (Shimizu et al., 1988;Schulenberg et al., 1988). The strategy ot construct the factor scales was to take the mean of those items that were most highly correlated with the given factor in the Shimizu et al.(1988) exploratory factor analysis (cf. Gorsuch, 1983), which resulted in a range of
1 - 4 for each scale. This strategy was justified by the confirmatory factor analyses
(Schulenberg et al., 1988), in which items considered to be not associated with a given factor had zero loadings on that factor (cf. Kim & Mueller, 1978) and in which the loadings of the salient items on each factor were of similar magnitude. The four sub-scales of CDS are described as follows: (a) feelings of confusion, discouragement, and lack of experience or information about the making or career decisions (items,
7, 8 and 1 l);(b) uncertainty about how to proceed in making decisions and the need for additional support for initial decisions (items 12 16 and 18);(c) classical approach- approach conflict in which several possible careers are attractive (items 4, 15, and
17); and (d) external and internal barriers to career choice and lack of interest in making a decision (items 3,5,6, and 9).
The Task Specific Scale of Occupational Self-Efficacy Expectations
(TSOSS) was developed by Osipow and Rooney (1989) based on Bandura’s theoretical framework and the work characteristics of the occupations in the
Dictionary of Occupational Titles (DOT, 1977). The original scale consisted of 230 items that are based on the skills associated with the 66 work groups reported in Part 14 A of the volume entitled Selected Characteristics of Occupations Defined in the
Dictionary of Occupational Titles (U.S. Employment Service, 1981).
Four sub-scales of TSOSS : The 230 item of the TSOSS yields four distinctive factors (Osipow & Rooney, 1990): (a) verbal, interpersonal, (b) quantitative, scientific, business, (c) physical, outdoor, (d) aesthetic, artistic. Based on this factor analysis, the 230 item scale was reduced to 60 items, with 15 items representing each factor. The 60 items of TSOSS were used in this research.
Basic Assumptions of Self-Efficacy Theory
1. Efficacy expectations differ in magnitude and responses to tasks
ordered by level of difficulty. They may be different from those
randomly ordered tasks because of individuals’ overall estimates of
their efficacy limits.
2. Efficacy expectations differ in generality, suggesting that specificity of
tasks may be another dimension which can be varied to observe
differences in generalized efficacy versus circumscribed mastery.
3. Efficacy expectations differ in strength, suggesting that a useful scale
will provide subjects a range of responses which permits them to
estimate the intensity or potency of their efficacy by task (Bandura,
1978b, p. 142). 15 Limitation of The Study
This proposed research is relational in nature. Thus a relational study cannot
establish cause and effect relationships between variables (Miller, 1986). This
proposed research will seek only to explain and predict relationships among
characteristics.
Summary
There is no clear research to support hypothesis of relationship between
career indecision, self-efficacy expectations and other selected background characteristics for females. In addition, there is no research addressing this
relationship from different cultural background. Therefore, this study is designed to investigate the utility of Bandura’s self-efficacy theory to the understanding and treatment of career indecision for female students in Korea by using two translated
Korean version of instruments, the Task Specific Occupational Self-efficacy
Scale(TSOSS) developed by Osipow and Rooney (1989), The Career Decision Scale
(CDS) developed by Osipow and others (1978), and one background characteristics questionnaire developed by the researcher.
The remainder of this study is organized into four chapters. There chapters are: review of literature, research method, findings and conclusions and summary, discussion and recommendations. CHAPTER II
REVIEW OF LITERATURE
Theories are not only essential in guiding the directions and facilitating the understanding of research, but the lack of theory may lead to arbitrary interpretations of data.
(Osipow and Walsh, 1983, p. 80)
Introduction
Career development has a long history beginning with the work of a visionary
social reformer, Frank Parsons, who wrote the first recognizable career counseling
book (Parsons, 1909). Since Parson’s first career counseling book, many books and journals on career choice and developmental research theory and practice have been
written to further understanding with respect to people’s abilities, interests and
available work opportunities..
While there are volumes of work in career development, this chapter is
focused on describing and summarizing the literature in the area of career indecision
and the research on self-efficacy expectations.
The literature review is organized in the following manner: (1) theories of
career development and career indecision, (2) the state of knowledge about career
16 17 indecision and the Career Decision Scale (CDS), (3) social learning theory,
(4) Bandura’s self-efficacy theory, (5) Hackett and Betz’s self-efficacy theory,
(6) task-specific career self-efficacy research, (7) the career development of women, and (8) a brief explanation of extraneous variables in the figure-1.
Theories of Career Development and Career Indecision
For convenience, theories of career development can be simply explained by two models: the search model and the compromise model (Spokane & Richardson,
1992). The search model asserts that individuals search for a career option that represents a reasonable fit with their personal qualities. A reasonable fit seems likely to be reflected in satisfaction or persistence with respect to work. Stated another way, the search model can be described as "person-environment" matching, like Holland’s theory (1985a) or the Theory of Work Adjustment (Dawis and
Lofquist, 1984). The Theory of Work Adjustment by Dawis and Lofquist(1984) is consistent with the underlying assumptions of the search model. Because correspondence or difference between work personality such as celerity, pace, rhythm and endurance and work environment could be used to express the success of adjustment to work. In other words, if the work meets the needs of the individual and the individual meets the demands of the work environment, then work stability and satisfaction with work can be expected (Dawis and Lofquist, 1984).
Research results supporting the search model can be found in many studies
(e.g.,Borgen, 1986; Rounds & Tracey, 1990). If occupational interests scores are not highly stable, then it is likely that individuals will change or leave jobs that are 18 incongruent with their personalities as opposed to changing themselves in any
significant way. Nurses dissatisfied with their careers leaving nursing in search of
more congruent careers are a good example of the search model (Prestholdt, Lane,
and Matthews, 1987). According to assumptions about the search model, individuals
will not need to change themselves much to accomplish this transition (Spokane &
Richardson, 1992). In the search model, if correspondence between personal characteristics and work environment is not maintained as the Theory of Work
Adjustment would predict, there will be dissatisfaction resulting in discomfort about
the choice culminating in career change.
Holland’s theory also suggests that incongruence leads to maladjustment,
dissatisfaction, and instability. Stated another way, a poor match between personality
and environment tends to put an individual into the state of career indecision. In the
search model personality and biological factors seem likely to be stressed because of
its unchangeable factors.
The second model in the theories of career development is the compromise
model. The compromise model asserts that individuals continuously change over time by repeatedly having to compromise their aspirations because of barriers imposed by realities of the world of work. These compromises must be integrated into a developing self or ego identity (Gottfredson, 1981; Super, 1957; Vondracek,
Lemer, & Schulenberg, 1986). Indeed, that the human experience of change and development over time is more crucial than the portion that remains stable is the central point in the compromise model (Spokane & Richardson, 1992). 19 Support for this compromise position comes from many studies of career development. One example is Super’s theory which has examined how individual life stages can progress by ages. In addition the Social Learning Theory of career decision making (Krumboitz, 1979) seems to present the compromise model. The compromise model assumes that the career decision-making process includes such factors as genetic predisposition, environmental conditions and events, learning experiences and cognitive, emotional and performance responses skills. Those skills serve as reinforcements that play on the different combinations of interactions of the factors, and produce the multitude of different career choices that different individuals make (Mitchell & Krumboitz, 1990). Indeed considerable research by
Spokane now suggests that people vary widely in their ability to persist in a task and in their emotional, cognitive, and behavioral reactions to compromise situations
(Spokane, 1991, 1992). In the compromise model, environmental conditions and events are likely to be considered important aspects in developmental approaches because the environment influences how the individual responds to each stage and compromise as context. For instance, job opportunities, the number and nature of training opportunities, social politics, the labor market, the educational system, and neighborhood and community influences are all important factors that influence individuals to compromise their original aspirations.
In conclusion, two models: the search and the compromise in the theories of career development indicate that there are two kinds of career development processes, that is, personal change versus job change. Although the career 20 development process varies from one person to another, some individuals will simply switch to a different and more congruent job in the face of the reality of barriers, whereas other individuals will integrate the experience to result in a change in the self.
If the process of compromise or search regarding job change or personal change fails, it should lead to the state of career indecision.
The State of Knowledge about Career Indecision and the Career Decision Scale (CDS)
A great deal of research on career indecision has identified the antecedent variables that contribute to career indecision (e.g., Ashby, Wall, & Osipow, 1966).
Other research has attempted to describe the characteristics of vocationally decided and undecided students, and to identify appropriate interventions (Hartman, 1973;
Gordon, 1984; Lunneborg, 1975, Holland & Holland, 1977; Slaney, 1988).
Crites (1969) and Slaney (1988) organized the early research on undecided college students by describing how decided and undecided college students differed on personal and demographic traits.
According to Goodstein (1965), who proposed two types of career indecision, the first type of career indecision correlates to the literal conception of being undecided about a career, a condition that presumably could be alleviated by providing individuals with relevant information which would then lead to a satisfactory decision. Goodstein’s second type of career indecision represents a more intractable state, that of indecisiveness due to the anxiety involved in making decisions. Thus, indecisive persons presumably have a more difficult time making 21 decisions not only about their career but about most other things in their lives. This differentiation is intuitively appealing but poses problems from a measurement perspective. However, a recent study by Fuqua et al. (1988) represented a useful effort to demonstrate how anxiety correlates with various types of indecision.
Career indecision may merely be a normal, developmental state that can be remedied by obtaining relevant career information or through standard career interventions. Indecisiveness, on the other hand, may be more trait-like and respond only to intensive, fundamental and long lasting treatment (Slaney, 1988; Salomone,
1982). However, research findings are conflicting and confusing, and provide no empirical basis to permit the reliable differentiation of decided and undecided students on the basis of personal, social, or academic characteristics (Harman, 1973;
Gordon, 1984; Lunneborg, 1975; Holland & Holland, 1977; Slaney, 1988). Ashby,
Wall and Osipow (1966), comparing three groups of entering college freshmen, decided, undecided and tentative, studied differences on a variety of personality, background, and college performance characteristics. In their study, significant differences were found on only two variables: high school grades and dependence scores on the Bemreuter Personality Inventory. This study recommended that undecided students are capable enough, but that they may need extra support and encouragement.
In addition to the distinction between general indecisiveness and career undecided persons, much research has indicated that several types of career indecision may exist (Hartman, Fuqua, and Jenkins, 1986; Vondracek, Hostetler, 22 Schulenberg, & Shimizu, 1990). An explanation of career indecision as a unidimensional problem is not likely to be satisfactory. Instead, research has indicated that career indecision seems more likely to be a complex and multidimensional construct (e.g., Appel, Haak, and Witzke, 1970; Shimizu,
Vondracek, Schulenberg, and Hostetler, 1988; Vondracek, Hostetler, Schulenberg &
Shimizu, 1990).
The debate over the unidimensional versus multidimensional conceptualization of career indecision has important diagnostic and therapeutic implications. One of the leading instruments used to examine different dimensions of career indecision by many researchers is the Career Decision Scale (CDS) developed by Osipow, Carney,
Winer, Yanico, and Koschier (1976). There is a growing body of literature aimed at identifying the dimensions of career indecision as measured by the CDS. In an original factor approach study by Osipow and others (Osipow, 1980, 1987), four factors were found, but because the scale is short and because the number of items on some factors are as few as two, it was concluded that it would not be appropriate to rely on the factor scores to interpret the scores of respondents (CDS Manual,
1986). However many studies (Schulenberg, Shimizu, Vondracek & Hostetler, 1988;
Vondracek, Hostetler, Schulenberg, & Shimize, 1990 and so on) have continued to debate the suitability of different methods of factor analysis and have reported results suggesting a stable factor structure.
Recent studies have clarified the factor structure of the CDS, thereby permitting the development of four linearly independent scales to measure 23 dimensions of career indecision (Shimizu, Vondracek, Schulenberg, & Hostetler,
1988; Vondracek, Hostetler, Schulenberg & Shimizu, 1990). To determine the validity of the multidimensional model, Vondracek et al (1990) examined whether the CDS total score and the four sub-scales were related to the student’s career decision status, grade level, and gender. Their results showed the presence of significant differences related to gender and to career decision status on specific indecision scales. It was concluded that the four factor based sub-scales (diffusion, support, approach, and external barriers) proposed in their study possess relatively good discriminate validity.
On the other hand, Martin, F.,Sabourin, S.,Laplante, B.,and Jean-Claude
Coallier (1991) reported that one can not be sure about the number of sub-scales nor about the existence of different dimensions of career indecision. Martin and others’ research suggests the insight of a multidimensional approach is needed if we are to have stronger evidence to support the generality of the four factor model. However, there is need to consider the language and cross-cultural question because Martin et al (1991) used translated versions of CDS.
The debate not with standing, the four factors on the CDS appear to be as follows: (a) diffusion, which represents feelings of confusion, discouragement, and lack of experience or information about the making of career decisions (items 7,8, and 11); (b) support, which represents uncertainty about how to proceed in making decisions, and the need for additional support for initial decisions (items, 12,16 and
18); (c) approach-approach. which represents a classical approach-approach conflict 24 in which several possible careers are attractive (Items, 4, 15, and 17); and (d) external barriers, which represent both external barriers to career choice and lack of interest in making a decision (items, 3,5,6,and 9).
Using a different instrument: the Career Decision Readiness Inventory
(CDRI), early research by Appel, Haak, and Wtzke (1970), also examined factors associated with indecision about collegiate major and career choice. The authors described the six factors they found as follows: factor 1. Situation-specific choice anxiety, which reflects the personal discomfort and concern felt by those who have not arrived at a decision about collegiate major and career choice; factor 2. Data- seeking orientation, which suggests the need for an increased reality basis for judgment regarding a particular objective; factor 3. Concern with self-identitv. which indicates that there are some undecided students for whom the task of selecting a major or carer is not a relevant one; factor 4. Generalized indecisiveness. which reflects a general inability or at least difficulty in choosing among alternative courses of action; factor 5. Multiplicity of interests, which suggests that some undecided students have an abundance of alternatives which they consider as possible areas of study and subsequent career objectives; factor 6. Humanitarian orientation, which indicates that some students find their choice of major and subsequent career objectives confounded by the need to take into account not only the intrinsic appeal of a particular alternative, but the societal value of this choice.
Another study supporting a multidimensional perspective of career indecision from the developmental aspect was done by Gordon and Kline (1989). This 25 research indicated that entering freshmen are at many stages of ego-identity development and their perceived need for the type and degree of advising varies not only with being decided or undecided, but is a function of their status on the ego- identity developmental continuum, as well.
The results of the studies seeking to determine the subtypes of career indecision (e.g.,Appel, Haak, and Witzke, 1970; Shimizu, Vondracek, Schulenberg, and Hostetler, 1988; Vondracek, Hostetler, Schulenberg & Shimizu, 1990) provide evidence of the heterogeneity of the group classed as "undecided students." It might be suggested that there are multiple bases for the inability or unwillingness of college students to make an educational or vocational commitment.
In summary, in view of the results found to date the following may be concluded about career indecision: first, large numbers of undecided students exist in college (Astin, 1977; Lunneborg, 1975; Crites, 1981); second. both decided and undecided students are heterogeneous populations so that they cannot be described easily or accurately enough to predict future indecision; third, viewing vocational indecision as comprising multiple subtypes might be more useful than viewing it as a single type (Holland and Holland, 1977; Jones and Chenery, 1980; Osipow, Carney,
Winer, Yanico, and Koschier, 1976).
Nevertheless, Crites (1981) citing the complexity of the college and vocational environment and the danger of identity foreclosure as a result of making a premature vocational decision, observed that being undecided may be the best decision that a college student can make. This observation emphasizes that it is important to note 26 that a state of career indecision is not a necessarily negative consequence of people’s career development process. On the contrary, career indecision may lead a person into the right path with a better sense of direction, and is a healthy and appropriate response in many circumstances.
Social Learning Theory
The Social Learning Theory of career decision making (Krumboitz, 1979;
Krumboitz, Mitchell, and Jones, 1976) is an outgrowth of the general Social Learning
Theory of behavior, which is most often associated with the work of Bandura (1977) emphasizing the role of self-efficacy. The concept of self-efficacy, seen to be an important mechanism directing human behavior in the belief in one’s ability to perform a task leading to a desirable outcome, has its roots in the Social Learning
Theory even though self-efficacy was not highlighted. Thus this section describes 1)
Social Learning Theory, 2) Bandura’s Self-Efficacy Theory, and 3) Betz and Hackett’s
Self-Efficacy Expectations Theory.
Social Learning Theory asserts that the individual personalities and behavioral repertories of persons arise primarily from their unique learning experiences, rather than from innate developmental or psychic processes (Mitchell & Krumboitz, 1990).
These unique learning experiences consist of contact with the cognitive analysis of positively and negatively reinforcing events, and result in individual behavioral and cognitive skills and preferences that allow people to function effectively in one’s environment. There are three major types of learning experiences: instrumental learning experiences, associative learning experiences, and vicarious experiences. 27 Instrumental experiences occur when the individual is positively reinforced or punished for the exercise of some behavior and its associated cognitive skills. For example, the individual is more likely to take actions that lead to enrollment in a given course, or employment in a given occupation, if that individual has recently expressed a preference for that course or occupation.
Associative learning experiences occur when individuals associate some previously affectively neutral event or stimulus with an emotionally laden event or stimulus. For example, a person might associate the hospital setting with the death of a beloved relative and thus become extremely reluctant to engage in any activities that are associated with hospitals, such as entering the field of medicine as a career.
Instrumental and associative learning experiences occur through direct experience with reinforcing or punishing events (Mitchell and Krumboitz, 1990).
A third way in which persons learn a large part of their behavioral and cognitive skills and preferences has to do with vicarious experience.
According to Osipow (1990), the outcome emphasized by the Social Learning
Theory is the skills acquired in decision making, and their use in the career choice realm throughout life, rather than what an individual chooses for life work. In fact the Social Learning Theory of career decision making is designed to address the question of why people enter particular educational programs or occupations, why they may change educational programs or occupations at selected points in their lives, and why they may express various preferences for different occupational activities at selected points in their lives. 28 In addressing these questions, the theory examines the impact on the career
decision-making process of such factors as genetic predisposition, environmental
conditions and events, learning experiences, and cognitive, emotional, and
performance responses skills. In this theory, four factors play a part in all career
decisions that are made, but the different combinations of interactions of factors
produce the multitude of different career choices that different individuals make.
From this assumption, these four factors seems likely to lead to a state of career
indecision too. For instance, genetic endowment and special abilities are inherited
qualities that may set limits on educational and occupational preferences, skills and
related decisions. Environmental conditions and events also affect the career
indecision of an individual, either in planned or unplanned process. The number and
nature of job opportunities, the number and nature of training opportunities, social
, policies and procedures for selecting trainees, workers and students, the rate of
return for various occupations, labor laws, family training experiences, the
educational system, or neighborhood and community influences are examples of
environmental conditions and events.
Mitchell and Krumboitz (1990) pointed out that the application of Social
Learning Theory to career decision making delineates four outcomes of learning
experiences that determine the career decision making behavior of individuals: self
observation generalizations, world-view generalizations, task approach skills and
actions (entry behaviors). It is hypothesized that an individual will reject certain
educational and occupational choices when that individual (1) experiences negative 29 consequences for engaging in activities associated with certain occupations, (2) observes a valued model experiencing negative consequences for these activities, or
(3) has been positively reinforced by a valued person who does not advocate, or actively discourages, engaging in these activities. The following guidelines are suggested by Mitchell and Krumboitz (1990) for determining problematic beliefs and generalizations.
1. Examine the assumptions and presuppositions of the expressed belief.
2. Look for inconsistencies between words and actions.
3. Test simplistic answers for inadequacies.
4. Confront attempts to build an illogical consistency.
5. Identify barriers to the goal
6. Challenge the validity of key beliefs.
From this perspective, the Career Decision Scale (Osipow, Carney, Winer,
Yanico, and Koschier, 1976), an assessment instrument used with this approach, was designed to identify the specific reasons people might be undecided about a career choice (for example, lack of information or a need to satisfy others). Such techniques have been used to identify the content and the process by which certain career beliefs and generalizations have arisen.
The Social Learning Theory is especially attractive because of its special application to educational organizations. Social Learning Theory emphasizes specific career aspirations, the development of the means to implement those aspirations, and the development of skills and attitudes (such as self-efficacy). Consequently, it so 30 seems to provide ways to directly address the problems of career indecision in educational settings. Students tend to be indecisive when they feel that they have a high anxiety level, a negative image of an occupation or curriculum, and a lack of information or a lack of skills.
Thus, this research suggests eleven such selected background characteristics as age, grade level, college area, area of home town, location where most of education was completed, academic achievement, socio economic status, parent’s educational background, additional enrichment educational activities, previous work experiences, and attitude regarding career and family as extraneous variables which can pose a major threat to the validity of this study.
Extraneous variables used in this research can be categorized into four areas based of Social Learning Theory in career decision making as shown in table 1. Table 1
Categorization of Extraneous Variables Based on the Social Learning Theory
Extraneous variables Based on Social Learning Theory Genetic Age Predisposition Sex (Female) Envir. Socio Economic Status(SES) Conditions Parents ed. background Location where most ed. was completed Home town area
Events Additional enriching ed. activities Learning Previous work experience Experiences College Area Cognitive Self-efficacy expectations Emotional/ Academic achievement (GPA) Performance Attitude regarding career and family Responses Educational/job Plan Skills 32 Bandura’s Self-Efficacv Theory
Many books (for example, Bandura, 1969,1973, 1977) have been devoted to discussing empirical investigations that support the basic tenets of Social Learning
Theory and the related concept of self-efficacy. In Bandura’s writing on Social
Learning Theory (1977), the concept of self-efficacy, one’s beliefs that a given task or behavior can be successfully performed, plays an important role. Bandura (1982, p. 123) explains self-efficacy as "people avoid activities that they believe exceed their coping capabilities, but they undertake and perform assuredly those that they judge themselves capable of managing." More specifically, according to Bandura(1977,
1986), self-efficacy expectations vary in three dimensions: (1) level- that is, the degree of difficulty of the task that an individual feels capable of performing;(2) strength - that is, the confidence the person has in his or her estimates; (3) generality - that is, the range of situations in which the person feels efficacious.
In Bandura’s Self-Efficacy Theory, outcome expectations interact with self- efficacy expectations. Outcome expectations refer to the person’s beliefs about the consequences of performance, while self-efficacy expectations refer to beliefs about the ability to perform the behavior. According to Bandura, self-efficacy expectations are not readily distinguishable from outcome expectations, since people perceive outcomes to be contingent on their performance. Brooks (1990) gives as examples in this matter persons who judge themselves as efficacious in repairing automobiles and so will envision themselves as successfully completing a tuneup, while those who are less confident might anticipate a jerky ride, with misplaced spark plugs. In this 33 situation, outcome expectations will not influence behavior independently of self-
efficacy. However, outcome expectations are readily distinguishable from efficacy
expectations and do have an independent influence on behavior. An example is a
woman who may believe she has the abilities to perform the role of chief executive
officer of a company, but she does not expect that she would be selected for the job
if she applied, because company decision makers would prefer a male in the position.
In this example, environmental contingencies are perceived as controlling or
influencing the outcome rather than the level or quality of one’s behavior. Thus,
self-efficacy and outcome expectations need to be differentiated.
Another variable that influences whether behavior will be initiated in
Bandura’s Self-Efficacy Theory is incentives. For example, a man may feel efficacious about his abilities to perform the duties of a kindergarten teacher, but he
has no incentive to do so because he does not value the outcomes (for example, low
salary and status). Bandura remarks that "values determine behavior in that prized incentives can motivate activities required to secure them and disvalued incentives do not" (1977b, p. 139).
In Bandura’s Self-Efficacy Theory, four sources are thought to be the powerful influences on self-efficacy expectations: 1. performance accomplishments, 2. vicarious experiences (for example, observing others), 3. verbal persuasion or encouragement from others, and 4. physiological or emotional arousal (that is, anxieties). 34 Hackett and Betz’s Self-Efficacv Theory
In developing the application of Self-Efficacy Theory to the career domain,
Hackett and Betz (1981) point out that career development theories have failed to
specify the mechanisms through which societal beliefs and expectations affect
women’s vocational behavior. Hackett and Betz (1981) propose that differential sex
role socialization prevents women from gaining equal access to information from
which self-efficacy expectations are acquired. This socialization results in women
having lowered overall self-efficacy and lower career self-efficacy for traditionally
male occupations. Thus they postulate that women, compared to men, possess lower
and weaker career-related efficacy expectations, and that these differences help
explain women’s vocational behavior.
In their study college students were asked to rate their capabilities to
complete the educational requirements and perform job tasks for 20 occupations.
The Occupational Self-Efficacy Scale (OSES) was used for measuring occupational
self-efficacy as the first systematically derived method. OSES method is based on
Bandura’s (1978b) concepts of strength and level (magnitude). In this instrument,
level is assessed by a simple yes/no response to questions asking whether a subject
could complete the educational or job requirements of various careers. On the
other hand, strength is measured through use of a 10-point scale of confidence (for example completely unsure to completely sure) regarding the yes responses.
However, there are no measures of generality even though these are suggested by hackett and Betz (1981). 35 Other studies have applied the concept of self-efficacy to occupations (e.g.,
Wheeler, 1983; Lent, Brown & Larkin, 1984, 1986; Branch & Lichtenberg, 1987;
Lapan, Boggs & Morrill, 1989). These studies are based on respondents’ opinions of their capabilities to complete what they believe are the educational and work requirements for broadly defined occupational titles. Rooney and Osipow (1992) indicated that most of the published research to date does not provide the subjects with an objective listing of specific courses or job tasks for the various occupations.
They point out that the forgoing career self-efficacy research has been operationalized as a general entity, not in terms of specific tasks. Nevertheless, the work of Betz and Hackett (1981) describes a potentially useful method for operationalizing the generality construct and provides a basis for comparing the results of a task-specific career self-efficacy measure with a general measure to assess construct validity (Rooney and Osipow, 1992).
Task-Specific Career Self-Efficacy Research
The limitations of career self-efficacy research are very important in the perspective of developing instruments for self-efficacy based on the original theoretical framework of Bandura. According to Bandura, there are clear distinctions between efficacy expectations and outcome expectation: "An outcome expectation is defined as a person’s estimate that a given behavior will lead to certain outcomes. An efficacy expectation is the conviction that one can successfully execute the behavior required to produce the outcomes" (Bandura, 1978a, p. 240). Applying this distinction to career or occupational self-efficacy, specific job tasks may be 36 thought of as the behaviors important in assessing efficacy expectations (Rooney and
Osipow, 1992).
Theoretically, the appropriate level for measurement of self-efficacy is the task-specific level whereas measurement at the general level may be thought of as outcome rather than efficacy measurement.
Two studies, Bores-Rangel et al. (1990) and Matsuie and Tsukamato (1991) have operationalized career self-efficacy at the task level for a limited range of occupations.
Bores-Rangel, Church, Szendre and Reeves (1990) constructed a measure of self-efficacy using items from the United States Employment Service (USES) Interest
Inventory that combines both general and task-specific items related to 69 selected occupational activities. This work of Bores-Rangel et al conceptualize generality as efficacy across a wide range of occupations. In addition, they confirmed that self- efficacy more strongly relates to the extent of consideration of occupational activities for people with weaker overall self-efficacy (less generality of self-efficacy across a range of occupations) than for people with greater or stronger overall self-efficacy.
Matsui & Tsukamato (1991) operationalized career self-efficacy at the task level for 60 work activities related to 30 selected occupations and related it to
Holland occupational codes and model environments. However, they operationalized only strength estimates for efficacy even though Bandura’s Theory of Self-Efficacy postulates that both level and strength measurements are appropriate. One results of Matsui & Tsukamato is gender differences in self-efficacy. In their study, males 37 are higher in self-efficacy for the realistic domain and for eight occupations, and
women are higher in self-efficacy for the artistic domain, for tine work with concrete
objects, and for four occupations. But there are no gender differences for both
investigative and enterprising domains.
Two studies lend support to the assumption that an appropriate level for
measurement of occupational self-efficacy is the job task rather than the more
general job or occupation. In addition, such a scale consisting of task level may be
more powerful in assessing people’s strength and level of self-efficacy as an aid in
understanding and expanding their range of career choices.
Osipow and Rooney (1989) studied the importance of measuring job task self-
efficacy. They developed a 230 item prototype Task-Specific Occupational Self-
efficacy Scale (TSOSS) and compared it with a general measure of self-efficacy, the
OSES, for a college student population (Rooney & Osipow, 1990). One of the
gender differences in self-efficacy is that males are higher in their efficacy estimates
than females on tasks requiring strenuous physical activity; operating equipment,
tools, vehicles and machinery; judging, estimating and calculating speed, distance,
values and dimensions; understanding blueprints and drawings, panel boards and
gauges; interpreting and reporting scientific and technical data; and using logic. On
the other hand, female scores were higher (more efficacious) on tasks requiring
people skills, accurate speaking and grammar; performing clerical tasks; devising and performing dance routines; and knowing hair and skin care. In this study, the TSOSS demonstrated differential efficacy in journalism and psychology classes. In addition 38 the researchers suggested that males and females view occupations differently and
that these differences effect efficacy estimates. This study suggested that age and
college year differences require confirmation.
Based on this review of the literature of this field, one is persuaded that the
present study can be seen to extend the research concerning task specific
occupational self-efficacy in several ways. First, it extends the work of Matsui and
Tsukamato (1991), and Bores-Rangel, et al (1990) by examining the relationship of
self-efficacy to academic areas such as humanity, social sciences, natural
science/engineering and arts rather than to a range of occupations. Second, this study also extends the work of Rooney and Osipow by broadening the sample size and including background characteristics from Seoul, Korea. Finally, it attempts to relate aspects of Bandura’s Self-Efficacy Theory (1977) with aspects of the Social
Learning Theory of Career Decision making for women.
The Career Development of Women
The field of study concerning the career development of women is becoming a vital area of concern while the activity of women in the work force is increasing.
The United States Department of Labor statistics from 1940 to the present confirm the increasing number of women in the work force. Similarly, the Korean
Educational Development Institution reports that there are a remarkably increasing number of women in the work force as compared to the past (1992). In addition, women’s work patterns are often quite different from those of men. Thus, current career development theories do not adequately explain the choice of career for 39 females. This discovery influenced the proposed research focusing on women’s career choice and self-efficacy.
Attention to women’s roles in the work force has resulted in studies dealing with issues of women’s career development, from gender differences to employment opportunities (Fidell, 1970). Such studies have tried to find variables affecting the process of women’s career development (Betz N, 1988). A number of papers in a volume edited by Osipow(1975) deal with the factors involved in the career development of women, such as women’s interest development, parental factors, personal factors, the interaction of marriage and work for women, sex-role stereotyping, and barriers that women face in working with their careers.
Early studies dealing with the issue of women’s career development have tried to study how occupational stereotypes develop in children and college students in order to better understand sex-role perceptions of work roles.
The book The Career Psychology of Women written by Betz and Fitzgerald
(1987) reviewed extensively the research on the vocational psychology of women including women’s career choice and women’s career development by suggesting variables used to differentially describe women’s career development (i.e.,dependent variables). There was also a review of individual variables considered uniquely important to the understanding and prediction of that development. The dependent variables identified are "Homemaking versus career orientation", "variables describing career orientation" (i.e.,classifying them according to the extent to which they were traditional or non-traditional for women-pioneer, innovator, and nontraditional), and 40 "Career patterns" (i.e., whether a woman pursues her career without interruption
through marriage and childbearing or takes "time o ff). In addition the independent variables suggested are family background, abilities, interest, marital/family status,
sex-role attitudes and role conflict.
A great deal of the research in this area has examined various background factors such as parental attitudes and occupations (Tried & Angrist, 1971; Lemkau,
1979), role model influence (Tried & Angrist, 1971;Basow &Howe, 1980; Goldstein,
1979; Phelan, 1979; Stake & Granger, 1978), educational and academic success
(Greenfield, Greiner, & Wood, 1980), including course work in math and science
(Betz & Hackett, 1981; Peng & Jaffe, 1979) encouragement (Stake & Levitz, 1979), and work experience (Tried & Angrist, 1971). Other influential variables examined in research on women’s career development have been related to personality characteristics and attitudes such as emotional health (Lemkau, 1979), self-efficacy expectations (Betz & Hackett, 1981) and adult marital and parental status (Card,
Steel, & Abeles, 1980; Levition & Whitely, 1981; O’Leary, 1974; Wallston, Foster, &
Berger, 1978).
Fassinger (1985) tested the Betz and Fitzgerald model by using structural equation modeling. He found the most plausible model included such variables as ability, achievement orientation, feminist orientation which affected family orientation and career orientation which affected career choice. He suggested that understanding career decisions for females required a understanding of a number of variables besides personal characteristics, such as self-efficacy. Extensive research 41 by Fassinger (1990) also supported the notion that ability, agenetic personality
characteristics, and sex-role attitudes predict career choice and career orientation.
Farmer (1985) attempted to explain the relationship between background,
personal and environmental factors acting with an additional three motivation
dimensions: (a) aspiration — level of occupation chosen, (b) mastery — motivation
to achieve on short-range challenging tasks, and (c) career commitment - degree
of commitment to the long-range prospects of a career. Farmer developed a model
for long-range career commitment for males and for females using path analyses.
According to results of this study, the model for females included a higher number of factors (12) with more complex relationships between the factors than did the male model (7 factors). These findings tend to support the notion that there are more factors which influence the long-range career motivation of females as compared to males. Given this, it can be interpreted that the impact of any one variable on career decisions and indecision for women would be small (Temple and
Osipow, 1994).
Some findings indicate that males are significantly less undecided (Gordon and Osipow, 1976; Westbrook, Cutts, Madision, & Arcia, 1980), while other findings indicate that females are less undecided (Taylor, 1979). However other research suggests that there is no difference in level of decisiveness between males and females (Cellini, 1978; Limburg, 1980; Niece & Bradley, 1979; Osipow, Carney, and
Barak, 1976; Sutera, 1977).
Despite of lack of research addressing women’s career indecision, there have 42 been continuous efforts to explain career indecision using the concept of self-efficacy
by Osipow, Temple and Rooney (1993), Temple (1991), and Temple and Osipow
(1994). Temple (1991) investigated the relationship between task-specific
occupational self-efficacy measured by the Task-Specific Occupational Self-efficacy
Scale (Osipow & Rooney, 1989) and career indecision measured by the Career
Decision Scale (CDS: Osipow, Carney, Winer, Yanico, & Koschier, 1976), for both
males and females. Temple did not find any relationship between occupational self- efficacy and career indecision for females, but did find one for males. Given this,
Temple and Osipow (1994) concluded that any single factor, such as self-efficacy, would not contribute enough to the decision-making process of females. In order to test the explanation of the results give in the Temple (1991) study, a study by Temple and Osipow (1994) was designed to investigate the impact of sex-role orientation by testing whether females possessing an egalitarian sex-role orientation would display a decision-making process more similar to males than females with a less egalitarian sex-role orientation. The results of the study offer minimal support for the hypothesis that there would be a significant relationship between self-efficacy and career indecision depending upon sex-role orientation. But the strength of the findings is limited because of the small absolute variability of the sample on the sex- role measure.
In summary, despite the large quantity of research on factors related to women’s career choice, the lack of a unifying theory to describe the relationships among variables has made it difficult to determine the relative strength of the 43 variables and their influence on women’s career choice. Even if those variables summarized above all affect women’s career choice and affect their decision and indecision, the specific question of how women’s career indecision interacts with the selected variables within Table-1 has not been clearly addressed.
Figure-1 shows the process of the career indecision for females based on research and theory in the women’s career development mentioned above in order to answer the research questions developed in this study. 44
1 Career Indecision of Females varies as a function of Age Demographic Inf Marital/ Parental Status Person-Envi ronment
Intra-Personal Interpersonal Environmental Factors Factors Factors Self-Efficacy Expectations Vatued People SES Attitude Toward Family Members School Environment Family and Career Mass Media Location where com Career aspirations Advisor pleted by most of Ed. Additional Ed. Enrichment Major Field Activities Parents Educational Perceived Job Experience Background Academic Achievement (GPA) Hometown
Moderated by al Decision Status Personal Ed/Vocational Change Change By By Compromising Searching for Aspirations/ Satisfaction Barriers
-and results in harmony or conflict By
Time |situation| Short-term Reality of world of work Long-term Reality of self-awareness Sex-role Socialization Marital and parental Status I and leads to 10
11 Satisfaction Dissatisfaction Feeling comfortable Feeling Change/Instability
and leads to II 12
Figure -1
Inspired by Samuel H. Osipow (19731 Fundamental Idea of Career Education Explanation of Figure-1
Almost all college females experience career indecision. They tend to have different levels of commitment and career indecision.
Demographic information such as age, marital and parental status and ethnicity are more likely to offer plausible explanations for females’ status of career indecision more precisely, despite the widely endorsed statement of heterogeneity of both decided and undecided students.
Especially, women, unlike men, are more likely to have difficulties with their career decisions because of sex-role socialization.
Person-environment factors such as intra-personal, interpersonal, and external factors are apt to influence state of career indecision for females.
Intra-personal factors consist of underlying variables, such as the self- efficacy, individual’s attitude toward family and career concerns, academic achievement, previous work experience, and additional enriching educational activities. These variables are more likely to have a close relationship with career indecision. For instance, a low level of self-efficacy may be related to a high level of career indecision, other variables, such as attitude towards career and family are equally important since work and family are linked throughout life, especially for females.
Interpersonal factors also tend to affect females’ career indecision, 46 since individuals are not likely to make wise decisions independent of
family members, valued people, advisors or counselors. These
relationships may influence career indecision. Most individuals live
within the context of their relationships with their various family
members. Thus these relationships influence their career directions
and decisions in profound ways.
6. Environmental factors are also important in deciding on females’
career direction, but these are uncontrolled and unpredictable nature
such as cultural uniqueness and social perception. Underlying
variables regarding socio economic status, school environment such as
being at a single-sex university or a co-ed university, major field,
location where most of education was completed, and hometown are
likely to influence career indecision.
7. Influenced by the above three factors, an individual’s career indecision
is apt to be moderated as the result of personal change or
educational/vocational change. Personal change is apt to take place
by compromising on aspirations and barriers imposed in reality.
Educational/vocational change is apt to take place by searching out
alternatives for satisfaction based on interest and personality.
8. The process of search or compromise is often confusing, painful, and
exciting, all at the same time. Psychological factors such as anxiety are
a natural concomitant of career decision making. Psychological factors 47 can refer to the stress of choice and change, pressure to implement the
choice, or lack of structure (Osipow, 1987), which means that they
simply do not know how to proceed in making a career decision.
9. Personal change and vocational/educational change result in either
harmony or conflict based on time and situation. Situation can refer
to the reality of the world regarding the opportunity or popularity of
a job when an individual needs a decision. Time can refer to the
length of time one is undecided. Some students are undecided in the
short-term, and tend to change their decision in many ways, while
others are undecided in the long-term, finding enough information or
confidence for their decision.
10. The consequence of an individual’s career indecision can be expressed
as either satisfaction or dissatisfaction, but these are also influenced by
time and situation. Some students tend to allow career indecision to
postpone their educational/vocational decision, searching after more
satisfaction, and they are more comfortable in being career undecided
because sometimes it can be a more healthy and appropriate decision,
and leads to better, if later, satisfaction. On the other hand, other
students tend to be dissatisfied with being undecided because of their
lack of self-confidence, anxiety, or pressure for a decision.
11. Dissatisfaction/satisfaction tends to result in different degrees of career
indecision, and provides multiple subtypes of career indecision, such as 48 the four sub-scales of CDS: (a) diffusion, which represents feelings of
confusion, discouragement, and lack of experience or information
about the making of career decisions, (b) support, which represents
uncertainty about how to proceed in making decisions, and the need
for additional support for initial decisions, (c) approach-approach.
which represents a classical approach-approach conflict in which
several possible careers are attractive, and (d) external barriers, which
represents both external barriers to career choice and lack of interest
in making a decision.
12. In a life-long process, an individual tends to develop and mature, and
have more satisfaction and less stress on their vocational educational
decision by experiencing choice and compromising between aspiration
and reality.
13. Career indecision is not a one-time event. Even an individual in a
period of being decided, after completing an educational and training
program of their choice, may feel unhappy with their choice or change
(by the process of compromise or search), and return to the first step
of being undecided. It is a continuous process in the life-span, but
different levels of decidedness and commitment exist. CHAPTER III
METHODOLOGY
Methods and procedures used in conducting the research are described in this chapter. The chapter is organized in the following manner: (a) population and sample, (b) research design, (c) instrumentation, (d) variables, (e) data collection, and (f) data analysis.
Population and Sample
The target population of this study is four-year university female students at the undergraduate level in Seoul, the capital of the Republic of Korea. The accessible population is only female students enrolled in autumn semester, 1993, at two women’s universities and two co-ed universities.
A simple random sample is normally recommended as a way to ensure that every person has an equal chance of being selected. However, to randomly select university women from all over Seoul is cumbersome. Thus, a cluster sampling technique was used, where colleges are the sampling units. And then, a stratified random sampling technique was used to select the same number of students from each of four colleges. These colleges were : Humane, Social and Science, Natural
49 50 Science and Engineering, and Arts.
To assure representativeness of female students at the university in Seoul, two important cultural characteristics were considered. First, since Korea has different levels of universities in terms of quality of students and institution, determined by an entrance examination, sampling units were divided into two levels, the competitive admission institution and the non-competitive admission institution. Second, due to differences of educational environment between co-ed institutions and single-sex institutions, sampling units were also divided into two, a co-ed institution and a single-sex institution.
According to Fraenkel and Wallen (1990), a replication study is recommended when a random selection is difficult, in order to decrease the likelihood that the results obtained are not simply a one-time occurrence. Therefore, this study was designed to replicate four times with the same procedures to avoid a conclusion that may be a one-time occurrences. Therefore this study was repeated four times to enhance representativeness of the target population of all female students at the university level in Seoul, Korea. Research Design
Descriptive and relational research method was used in this study to obtain the data necessary to test the hypotheses. Data were collected with the use of two instruments and a questionnaire designed to obtain the information to answer the research questions proposed in this study.
Campbell and Stanley (1963) suggest that researchers seldom conduct research on samples representatively drawn from the entire United States. They indicate that we will learn how far we can generalize results only piece by piece through trial and error of generalization efforts. Attempts to learn whether results are consistent across similar studies in different settings are necessary to gain an understanding of the external validity of research findings. For this reason, four replications were conducted using similar procedures. Two studies (Cases A and C) were conducted in a co-ed university in Seoul, while the others two studies (Case B and D) were conducted in single-sex universities in Seoul.
Instrumentation
Data were collected by administering two types of instruments: the TSOSS
(Task Specific Occupational Self-Efficacy Expectations Scale, 1989) and the CDS
(Career Decision Scale, 1978). A third type of short questionnaire was used to gather demographic data. This short questionnaire developed by the researcher is designed to collect selected background characteristics on extraneous variables. The survey instruments are the Career Decision Scale (CDS) developed by Osipow,
Carney, Winer, Yanico, and Koschier (1976), and The Task Specific Occupational 52 Self-Efficacy Scale (TSOSS) developed by Osipow and Rooney (1989). The CDS was used to measure the different types of career indecision on the dependent variable, and the TSOSS was used to measure the different tasks of self-efficacy expectations on the independent variable. All data were collected during August and September
1993 in Soul, Korea.
Career Decision Scale (CDS)
The Career Decision Scale developed by Osipow, Carney, Winer, Yanico, and
Koschier (1976) was designed to devise a method to standardize the identification of sources of career indecision and to serve as a diagnostic aid in using a series of structured interventions. In a graduate seminar led by Dr. Osipow, 16 statements that clients often make to describe their state of career indecision were identified.
These statements became the basic CDS items. According to the CDS manuel
(19861. for reliability. two studies have reported test-retest correlations of individual items and Indecision Scale scores. Osipow, Carney, and Barak (1976) reported two retest correlations of .90 and .82 for the Indecision Scale for two separate samples of college students. Item correlations for the Certainty and Indecision Scales ranged from .34 and .82 with the majority of correlations falling in the .60 to .80 range. The study by Slangy, Palko-Nonemaker, and Alexdander (1981) examined test-retest reliability over a six week period for the Certainty and Indecision Scale items. Their results showed item correlations ranging from . 19 to .70 with total Career Decision
Scale scores yielding a correlation of . 70. The validity of CDS has been supported by much research such as group comparisons and correlations with instruments 53 measuring the construct of indecision, treatment studies, relationships with other personality variables of interest, and relationships with demographic variables. The rationale of CDS is that a finite number of relatively discrete circumstances are responsible for the problems people have in reaching career decisions (Osipow,
1987). "The CDS is intended as a rapid and reliable instrument for surveying high school and college students about their status in the decision-making process'1. The
CDS "provides an estimate of career indecision and its antecedents as well as an outcome measure for determining the effects of interventions relevant to career choice or career development" (CDS manual, 1986).
The Task Specific Scale of Occupational Self-Efficacv Expectations
The TSOSS was developed by Osipow and Rooney (1989) based on Bandura’s theoretical framework and the work characteristics of the occupations in the
Dictionary of Occupational Titles (DOT, 1977). The original scale consists of 230 items that are based on the skills associated with the 66 work groups reported in Part
A of the volume entitled Selected Characteristics of Occupations Defined in the
Dictionary of Occupational Titles (U.S. Employment Service, 1981). The 230 items of TSOSS yielded four distinctive factors (Osipow & Rooney, 1990; Osipow, Temple
& Rooney, 1993): (1) verbal, interpersonal, (2) quantitative, scientific, business, (3) physical strength and agility, and (4) aesthetic skills. Based on this factor analysis, the 230 item scale was reduced to 60 items, with 15 items representing each factor.
Cronbach’s alpha describing internal consistency for the four resulting scales are all above .90, as is test-retest reliability over a 2 week period. The 60 item TSOSS will 54 be used in this research.
The Translation Process of the CDS and TSOSS
Since the CDS and TSOSS were used for the first time in Korea, several
recommendations were considered in translation. The translation of both
instruments into the Korean language required: 1) receiving permission to translate
and adapt the two instruments from authors and publisher, 2) translation into the
Korean language by the researcher, 3) review by the Korean bilingual experts
consisting of one Ph.D student in the Linguistics Department, one professor in the
Architecture Department, and one professional counselor, 4) back-translation by one
expert who did not see the original English version, 5) final review of the back-
translation by Dr. Osipow and the publisher, Psychological Assessment Resources
(PAR) Inc., in order to compare the English version of the CDS and TSOSS to the
translated instruments, and 6) pilot testing on 24 students at both the undergraduate and graduate level at The Ohio State University to assess the reliability of the
translated instruments during the first week of August, 1993. The reliability value of the four sub-scales of CDS and TSOSS were as following.
1 sub-scale of the CDS : .86 2 sub-scale of the CDS : .32 3 sub-scale of the CDS : .62 4 sub-scale of the CDS : .67
1 sub-scale of the TSOSS : .88 2 sub-scale of the TSOSS : .86 3 sub-scale of the TSOSS : .84 4 sub-scale of the TSOSS : .82 55 The Short Form of The Selected Background Characteristics Questionnaire
The short form demographic questionnaire consists of 19 questions to measure extraneous variables. Items include such variables as age, grade level, area of home town, location which most education was completed, marital status, major field, grade point average, socio economic status (SES), parent’s educational background, additional enrichment of educational activities, previous work experience, educational and job plan after completion of college, and attitude regarding career and family.
For content validity and clarity of this questionnaire, three experts (committee members for this research) reviewed the short form of the selected characteristics questionnaire. This instrument was also pilot tested for content validity.
Variables
Dependent Variables
The Career Decision Scale was used to assess different type of career indecision by the four sub-scales of CDS, and to determine the career decision status on the dependent variables. The CDS consists of 19 items with items 1 and 2 indicating certainty of career choice, items 3 -1 8 representing a measure of career indecision, and item 19 being open-ended and not scored.
Career Decision Status: Item 1 of the CDS was used to divide all respondents into two groups, either decided or undecided. Item states: "I have decided on a career and feel comfortable with it. I also know how to go about implementing my choice". Those who responded "exactlylike me" or "verymuch like me" were placed in the decided group, and those who responded only "slightly like me" or "not at all 56 like me" were placed in the undecided group.
Four Sub-Scales of CDS: Four sub-scales of CDS were identified on the basis
of previous factor analyses of the study (Vondracek, Hostetler, Schulenberg, and
Shimizu, 1990). The four factor-based scales are described as follows: (a) diffusion.
which represents feelings of confusion, discouragement, and lack of experience or
information about the making of career decisions (items 7,8, and 11); (b) support.
which represents uncertainty about how to proceed in making decisions, and the need
for additional support for initial decisions (items, 12, 16 and 18); (c) approach-
approach. which represents a classical approach-approach conflict in which several possible careers are attractive (items, 4,15, and 17); and (d) external barriers, which represents both external barriers to career choice and lack of interest in making a decision (items, 3,5,6, and 9). According to Vondracek, Hostetler, Schulenberg, and
Shimizu (1990), correlations among the factor-based scales ranged from .36 to .58.
Factor scale correlations with the CDS total score ranged from .65. to .81. The
magnitude of these correlations is not surprising because each of the factor scales
shares its items with the total CDS scale (minus the three deleted items of 10, 13, and 14). In their study, deletion of the three CDS items appears to have had a negligible effect on the CDS total score, as correlations between factor scales and the
CDS total score, with and without the three deleted items, were nearly identical.
Further support for dropping the three items is found in the fact that the CDS total score with and without the three items correlated .98 at both times of measurement. 57 Independent Variables
The Task Specific Occupational Self-Efficacy Expectations Scale (TSOSS),
consisting of 60 items, was used to assess the students’ different self-efficacy
expectations on the four sub-scales of TSOSS on the independent variables.
Subjects indicated their confidence in their ability to perform each activity on
a five-point scale from A, labeled "No Confidence," to E, labeled "Absolute
Certainty." By assigning values of 1 for A through 5 for E, the 60 items were scored.
Four Factor-based TSOSS Scales: Four factor-based TSOSS scales were
identified on the basis of previous factor analyses of the measure (Osipow & Rooney,
1990; Osipow, Temple & Rooney, 1993). Originally the 230 items TSOSS yielded
four distinctive factors. Items were reduced to 60 with 15 items representing each
factor. The four factor-based scales are described as follows: fa) Verbal and
Interpersonal . (b) Quantitative. Scientific and Business. (cl Physical Strength and
Agility, and (dl Aesthetic Skills. Cronbach’s Alpha describing internal consistency for the four resulting scales are all above .90, as is test-retest reliability over a 2 week period.
Extraneous Variables
Extraneous variables can pose a major threat to the validity of the study, and thus should be incorporated into the study as suggested by Franenkel and Wallen
(1990). By incorporating these variables into the study variance in the dependent variable that might be accounted for by them can be controlled (Blustein, 1987;
Fraenkel and Wallen, 1990). Twelve extraneous variables were identified based on 58 the research in career indecision, women’s career development process, and Social
Learning Theory (Krumboltz, 1979), and incorporated into the study by administration of the short form of questionnaire. Extraneous variables include such variables as age, grade level, area of home town, location where most education was completed, marital status, college area, grade point average (GPA), social economic status (SES), parent’s educational background, additional enrichment of educational activities, previous work experience, educational and job plan after completion of college, and attitude regarding career and family.
Data Collection
Data were collected during the months of August and September, 1993. An introductory letter was mailed to inform each student’s administrator to provide information to the two women’s universities in Seoul that had agreed to participate.
After the researcher arrived in Seoul in the middle of August, additional two co-ed universities agreed to participate this research. A telephone call was made to discuss this study and data collection procedures more specifically. A second communication with universities agreed to participate was a personal visit with a designated contact person. The personal visit was made by the researcher to deliver two instruments, the CDS and the TSOSS, one short form selected background characteristics questionnaire, and to give instructions for each of three instruments.
Data were collected in one hour during the usual lecture times.
Data Analysis
The data were analyzed using the Statistical Package for the Social Science 59 Pc+ (SPSSx-user’s guide, 1985) at the Ohio State University. Descriptive statistics
were used to determine and describe the background characteristics of respondents.
Frequencies, means and standard deviations were used to describe the characteristics
of the university women.
In order to answer research questions 1, 2 and 3, canonical correlation
analysis was performed. Discriminant analysis was also used for research
questions 4 and 5.
Canonical correlation is an appropriate analysis for investigating the
relationship between two sets of variables. The general question addressed by
canonical correlation is to what extent can one set of two or more dependent
(criterion) variables be explained (or predicted) by another set of two or more
independent (predictor) variables ? To determine the proportion of the variance in
the dependent variable set explained by the independent variable set, total
redundancy (Rd) for the dependent variable set was used. In addition structure
coefficients (canonical loadings) were used to interpret the meaningful set of both
independent variable sets and the dependent variable sets.
Discriminant analysis is an appropriate technique for studying simultaneously
the differences between two or more groups with respect to several variables. To assess and describe the discriminating power of the discriminate function, eigenvalue and canonical correlation coefficients were used. In addition, to describe how the groups differ, standardized discriminant function coefficients, and structure coefficients loadings were used. 60 For testing Hypotheses 1,2,3,4 and 5, the Wilks lambda was used as the test statistic at an alpha level of .05.
Summary
In summary, data were collected by administering three types of translated instruments in Seoul, Korea to explain the relationship between career indecision, self-efficacy expectations, and extraneous variables for females in Seoul, Korea.
These instruments included the TSOSS, CDS and the short form of selected background characteristics questionnaire. Those were pilot tested for reliability and content validity. Data were collected on August and September of 1993 in Seoul,
Korea. Data analyses included use of descriptive statistics, canonical correlation, discriminant analysis and Wilks lambda. CHAPTER IV
FINDINGS AND CONCLUSIONS
This chapter describes the findings with respect to (a) population and sample,
(b) instrumentation, (c) variables, (d) data collection, (e) descriptive data for sample, and (f) data analysis. Research questions are answered and the results of hypothesis testing are shown.
Population and Sample
The target population of this study was four-year university female students at the undergraduate level in Seoul, the capital of the republic of Korea. There are
34 four-year colleges in Seoul including 13 women’s universities and 21 co-ed
Universities (KEDI, 1992).
The accessible population was female students enrolled in the Autumn semester, 1993 at two women’s universities and two co-ed universities.
A simple random sample is normally recommended as a way to ensure that every person has an equal chance of being selected. However, to randomly select university women from all over Seoul is cumbersome. According to Fraenkl and
Wallen (1990), a replication study is recommended when random selection is 62 difficult in order to decrease the likelihood that the results obtained were not simply a one-time occurrence. Thus this study was repeated four times under the same
procedures for collecting data from female students in Seoul, Korea. In this process,
two very important Korean educational characteristics were considered. First, since
Korea has different levels of universities in terms of the quality of students and the institution based upon The Entrance Examination For University, data were collected at both levels of institution, a competitive and a non-competitive. Second, due to
the differences in educational environment at co-ed institutions and single-sex institutions, this study was also replicated at both a co-ed and a single-sex institution.
Therefore, this study was repeated four times with female students from two women’s universities and two co-ed universities in Seoul. The total subjects were
1,275 female undergraduate students enrolled in 1993 Autumn semester at four universities in Seoul, the capital of Korea.
As shown in Table 2 and 3, the sample consisted of 1,275 female students and included representatives from competitive and non-competitive admissions universities and from co-ed and single-sex universities. Table 2
Nature of School
Institution Co-Ed Single-Sex
Non-competitive Case 1 (A) Case 2 (B)
Competitive Case 3 (C) Case 4 (D)
Table 3
Total Number of Samole bv Universitv (n = 12751
Universities Total Number Percent Cum Perc
Case A 234 18.4 18.4 Case B 507 39.8 58.2 Case C 267 20.9 79.1 Case D 267 20.9 100.0
University
i______i______i______i______i______i 0 120 240 360 480 600
Valid cases 1275 Missing cases 0 64 Instrumentations
Career Decision Scale (CDS)
The CDS consists of nineteen items. Items 1 and 2 indicate certainty of
career choice, items 3 through 18 measure career indecision, and item 19 is open-
ended and is not scored. The items are scored on a Likert scale from "not at all like
me" (1) to "exactly like me" (4). Due to the unique characteristics of the Korean
educational system, two CDS items (12, and 18) were changed: item 12, from "Iknow
what I’d like to major in" to "lam happy with my major"; and item 18, from "major
in" to "career in" with the permission of the publisher of the CDS.
The Task Specific Occupational Self-Efficacv Scales (TSQSS1
The TSOSS was developed by Osipow and Rooney (1989) based on Bandura’s
theoretical framework, and the work characteristics of the occupations in the
Dictionary of Occupational Titles (DOT. 1977). The original scale consists of 230
items based on the skills associated with the 66 work groups reported in Part A of
the volume entitled Selected Characteristics of Occupations Defined in the
Dictionary of Occupational Titles (U.S. Employment Service, 1981). The 230 items
of TSOSS yielded four distinctive factors (Osipow & Rooney, 1990; Osipow, Temple
& Rooney, 1993) : (1) verbal, interpersonal, (2) quantitative, scientific, business, (3) physical strength and agility, and (4) aesthetic skills. Based on this factor analysis,
the 230 item scale was reduced to 60 items and 15 items represented each factor.
Cronbach’s Alpha describing internal consistency for the four resulting scales are all above .90, as is test-retest reliability over a 2 week period. The 60 items of the 65 TSOSS was used in this proposed research.
The Translation Process of the CDS and TSOSS
Since CDS and TSOSS were used for the first time in Korea, the translation work into Korean of both instruments, the CDS and TSOSS required several steps
: 1) receiving permission to translate and adapt both instruments from the authors and publisher, 2) the translating of them into the Korean language by the researcher,
3) reviewing of the translations by Korean bilingual experts including one Ph.D student in Linguistics, one professor in the Architecture Department, and one professional counselor, 4) back-translating of the instruments by one expert who did not see the original English version, (5) final review of the back-translation by Dr.
Osipow and the publisher, Psychological Assessment Resources (PAR) Inc.,in order to compare the English version of the CDS and TSOSS to the translated versions of both instruments, and 6) a pilot testing of 24 students at both undergraduate and graduate levels at The Ohio State University to assess the reliability of the translated instruments during the first week of August, 1993. The reliability value of the translated four sub-scales of the CDS and TSOSS were as following.
1 sub-scale of the CDS : .86 2 sub-scale of the CDS : .32 3 sub-scale of the CDS : .62 4 sub-scale of the CDS : .67
1 sub-scale of the TSOSS : .88 2 sub-scale of the TSOSS : .86 3 sub-scale of the TSOSS : .84 4 sub-scale of the TSOSS : .82 66 The Short Form of Background Characteristics Questionnaire
The short form of the background characteristics questionnaire consists of 19 questions measuring demographic, educational, and environmental variables as well as attitude toward family and career. These items are related to information about such variables as age, grade level, college area, selected curriculum in high-school, person who has the most influence on career decision, person who had the most influence on major decision before entering the university, experience of additional enrichment educational activities, educational plans and career plans after completion of college, attitude regarding career and family, social economic status
(SES), parent’s educational background, location where most education was completed, home town, previous work experience, and grade point average(GPA).
Three experts (the committee members overseeing this research) reviewed the short from of the background characteristics questionnaire. It was also pilot tested for content validity with twenty-two Korean female students at The Ohio State
University. After the pilot test, some items in this questionnaire were changed or omitted before the instruments were administered in Seoul, Korea.
First, question 2, asking marital status, was omitted since over ninety percent of the total female students in the accessible population are single, according to the population report of school administrators.
Second, question 3, graduate level was omitted since it is difficult to have the same number of graduate female students in the each of four colleges of the four universities. 67 Third, question 4, asking about major and colleges, was divided into two questions as a multiple choice and an open-ended question: Q-4 In what college are you studying ? 1. College of Humanities, 2. College of Social/Science, 3. College of
Engineering Sciences, or 4. College of Fine Arts ; Q-5 What major you are studying?
Fourth, questions 5 and 6, item 3 was divided into two: 3. advisor and 4. counselor.
Fifth, questions 12, 13, and 17, asking about the social economic status of students, were changed to report more accurately based on Korean cultural uniqueness according to Research on Social Economic Status through occupational analysis (1990, Hong, Doo Seung).
Lastly, question 19, asking grade point average, was changed to reflect only four categories : 1. A+ - A, 2. B+ - B, 3. C+ - C, and 4. Below C, since each of the four universities have different academic grade point standards.
Finally, the short form of the selected background characteristics questionnaire, consisting of 18 questions was used in this research. Some questions were open-ended (e.g., "Q- Please specify the occupation in which you expect to make a career after you have completed your education ?"). Others were multiple choice (e.g.,"Q-2 What is your grade level at your university ?" [Please circle the number of your answer]). Multiple-choice responses were: 1 Freshman, 2
Sophomore, 3 Junior, and 4 Senior. In addition, two questions on this questionnaire were used to establish the college of enrollment. In a multiple-choice format, female students were asked to respond to the following question, "In what college are you 68 studying ?" by choose either College of Humanities, College of Social/Science,
College of Engineering Sciences, or College of Fine Arts. The answer to this
question was cross-checked with another open-ended question which asked the
students to list the specific major they were now taking. The current list of majors provided by each of universities was used to classify the students. This study is
subject to the limitations associated with self- reported data, such as question 18 asking grade point average since obtaining the grade point average of each student from the academic records of the four universities was too cumbersome. Variably
Dependent Variables
The Career Decision Scale(CDS) was used to assess the dependent variables,
which are the four factor-based scales and the career decision status regarding
undecided or decided. More specific information about dependent variables used
in this research follows.
1. Career Decision Status: Item 1 of the CDS, which was not scored with the
CDS total score, was used to divide all respondents into two groups as decided or undecided. Item 1 states: "I have decided on a career and feel comfortable with it.
I also know how to go about implementing my choice". Those who responded
"exactly like me" or "very much like me" were placed in the decided group, and those who responded only "slightly like me" or "not at all like me "were placed in the undecided group.
2. Four factor-based scales: Four sub-scales of CDS were identified on the basis of previous factor analyses of the study (Vondreacek, Hostetler, Schulenberg, and Shimizu, 1990). The four factor-based scales can be described as follows: (a) diffusion, which represents feelings of confusion, discouragement, and lack of experience or information about the making of career decisions (Items 7,8, and 11);
(b) support, which represents uncertainty about how to proceed in making decisions, and the need for additional support for initial decisions (Items, 12, 16 and 18); (c) approach-approach. which represents a classical approach-approach conflict in which several possible careers are attractive (Items, 4, 15, and 17); and (d) external 70 barriers, which represents both external barriers to career choice and lack of interest
in making a decision (Items 3, 5, 6, and 9).
Independent Variables
The Task Specific Occupational Self-Efficacy Expectations Scale (TSOSS),
consisting of 60 items was used to assess the students’ different self-efficacy
expectations. Subjects indicated their confidence in their ability to perform each
activity on a five-point scale from A, labeled "No Confidence," to E, labeled
"Absolute Certainty." By assigning values of 1 through 5 fora through E respectively,
the 60 items are scored. Possible scores range between 60 to 300. More specific
information about independent variables used in this research follows.
1. Four sub-scales of TSOSS : Four sub-scales of TSOSS were identified on
the basis of previous factor analyses of the study (Osipow & Rooney, 1990; Osipow,
Temple & Rooney, 1993). Originally the 230 items TSOSS yielded four distinctive
factors. The number of items was then reduced to 60, with 15 items representing
each factor. The four factor-based scales are described as follows: (a) Verbal and
Interpersonal . (bl Quantitative. Scientific and Business, (cl Physical strength and
Agility, and fdl Aesthetic skills. Cronbach’s Alphas, describing internal consistency for the four resulting scales, are all above .90, as is test-retest reliability over two week period. Osipow, Temple, and Rooney (1993) found the test-retest reliability of the short form over a two-week period to be .90 and coefficient alphas for the 4 scales were all above .90, based on 34 subjects. 71 Extraneous Variables
Extraneous variables can pose a major threat to the validity of the study, and thus they should be incorporated into the study as suggested by Franenkel and
Wallen (1990). By incorporating them into the study, variance in the dependent variable that might be attributed to them can be controlled (Blustein, 1987; Fraenkel and Wallen, 1990). Several extraneous variables are identified and incorporated into the study based on the characteristics of women’s career development processes
(Betz, 1981) and Social Learning Theory (Krumboltz, 1979). These variables are age, grade level, college area, experience of additional enrichment educational activities, educational and job plans after completing of college, attitude regarding career and family, socio economic status (SES), parent’s educational background, location where most education was completed, home town, previous work experience, and grade point average (GPA). 72 Data Collection
Data were collected from the end of August through September 18, 1993, in
Seoul, Korea. An introductory letter written by the researcher and dissertation chair
person, Dr. Dewey Adams, was mailed to the student administrators of the Korean
universities who had agreed to participate. Data collection procedures, the
instruments to be used, and the research purpose were included in this letter. Later,
when the researcher arrived in Seoul, three more identical letters were handed to the
other three university administrators to inform them of the nature Of this study.
Thus, all four universities consisting of two women’s universities and two co-ed
universities agreed to participate this research.
For Case C and B universities, all subjects completed the three instruments,
CDS, TSOSS and the short form of the selected background characteristics of
questionnaire, under the supervision of the researcher in students’ lecture time. All
subjects at case A and D universities completed the same three above instruments
under the supervision of the professors in their lecture time.
After distributing the instruments, the experimenter or professors orally
reviewed the instructions. After the subjects completed all three measures, a
debriefing was given by the experimenter or professors and questions were answered.
It took most subjects approximately 40 minutes to complete the questionnaires. The
scales were numbered to aid in matching each subject’s data, but no name or identification number were required, thereby maintaining anonymity. * Similar procedures were used to collect the data from all four universities. 73 Descriptive Data for Sample
In order to fulfill the first objective in describing sample, descriptive analysis was performed by using the cross tables consisting of the suggested variables by the each of four universities.
Age
As shown in the Table 4, the average age of all respondents at the four universities was twenty one years (20.515) old and ranged from seventeen to twenty- eight years. 74
Table 4
Age by University
University
Age A B C D All Cases
M 20.780 20.303 20.869 20.350 20.515 SD 1.723 1.746 1.752 1.413 1.723
Valid cases 1240 Missing cases 35
Age of Total Respondents
17 oh 8 18 131 19 254 20 hhhhhhbhmhhh 243 21 h b s h num BHBnHgrnansKKi 259 22 bhhhhhbhbhbbbhhhb 210
25 ™ T ’"" 26 M 4 2 7 * 5 28 — 4 i 0 80 160 240 320 400 75 Grade Level bv University
As shown in the Table 5, the highest percentage of female students of the case
A university (33%), and the case D university (28%) were at the Junior level.
However, a higher percentage of female students at the case B university (50%) were
Freshmen. At the case C university, the highest percentage of female students (32%) were Sophomores.
Thus, the percentages of respondents at each grade level(n = 1273) in the four universities were (a) Freshman, 34%; (b) Sophomore, 20%; (c) Junior, 25% and (d)
Senior, 22%. 76 Table 5
Grade Level By University
U n iv ersity T o tal GRADE A B C D cases (n=233) (n=507) (n=266) (n=276) (n=1273) No. 54 254 52 67 427 Freshman % 23.2 50.1 19.5 25.1 33.5 No. 63 49 85 56 253 Sophomore % 27.0 9.7 32.0 21.0 19.9
No. 77 92 70 75 314 Junior % 33.0 18.1 26.3 28.1 24.7 No. 39 112 59 69 279 Senior % 16.7 22.1 22.2 25.8 21.9
V alid cases 1273 M issing cases 2
Grade Level of Total Respondent
Fresh 427 Sopho. H H M H B H H nH B H H H B nB H H I 253 Junior 314 Senior mmmmmmmmmmmmmKmmtamaamamm 279 i______i______i______i___ 0 100 200 300 400 500 77 College Area bv University
Table 6 shows the detailed information of major fields in each college by universities. As shown in the Table 7, a higher percentage of female students(47%) from case A university were in the College of Humanities. However a higher percentage of female students (40%) of case B university were in the College of
Engineering Science. Both case C and D universities had higher percentages of female students (35% and 34%) in the Fine Arts College. Overall, the percentages of all respondents (n=1263) in each major field at the four Universities were (a)
Engineering Sciences, 30%; (b) Humanities, 27%; (c) Fine Arts, 26% and (d) Social
Sciences 20%.
Table 6
Detailed Information of College Area By University
C 0 L L G E Univ Humanities Social/Sci Eng/Science Arts
Korean Lit Account C.I.S. Drawing A French Lit
Eng Lit Account Physc Math Drawing B Korean Lit Psychology Statistics
Eng Lit Dance C Korean Lit Social Work Math Drawing
Eng Lit Law Pharmacy Drawing D Math Dance, Music 78
Table 7
College Area by University
Universities C ollege T o tal Area A B C D Cases (n=230 ) (n=506) (n=263) (n=264) (n=1263)
No 107 98 84 57 346 Humanities % 46.5 19.4 31.9 21.6 27.4
S o cial No 18 89 44 50 201 Sciences % 7.8 17.6 16.7 18.9 15.9 Engineering No 74 198 43 68 383 Sciences % 32.2 39.1 16.3 25.8 30.3
Fine No 31 121 92 89 333 A rts % 13.5 23.9 35.0 33.7 26.4
Valid cases 1263 Missing cases 12
College Area of Total Respondents
Engi/Sci 383 Fine Arts wamnmnmmmmmmmmmmammmaBmnmmmmmn 333
i______i______i______i______i______i 0 80 160 240 320 400 79 Selected High-school Curriculum bv University
As shown in the Table 8, at four universities, more than 50% of the female students had selected a general humanities curriculum in their high School.
Interestingly, a very small percentage of females had selected vocational high- school curriculum before entering the university. The percentages who had selected a vocational high-school curriculum for the four universities were (a) case
D, 17%; (b) case C, 11%; (c) case B, 1% and (d) case A, 0.4%.
Thus, the percentages of all respondents (n= 1,267) broken down by high school curriculum choice were (a) General High School Humanities Department, 60%;
(b) General High School Engineering Department, 32%; (c) Arts and Physical
Department, 7% and (d) Vocational High School Curriculum, 1%. 80 Table 8
Selected High School Curriculum by University
Universities S elected T otal High School Currie A B C D Cases (n= 231) (n=506) (n=266) (n=264) (n=1267) General No. 151 289 182 139 761 Human % 65.4 57.1 68.4 52.7 60.1
General No. 73 210 44 79 406 Engin % 31.6 41.5 16.5 29.9 32.0
V ocational No. 6 10 16 High School % 2.6 3.8 1.3 Arts/Physic No. 1 7 30 46 84 High School % .4 1.4 11.3 17.4 6.6
Valid cases 1267 Missing cases 8
High School Curriculum Choice of Total Respondents
Gen 761 Gen Eng 406 Voc B 16 Arts/Phyh 84
I______I______l______I______i 0 160 320 480 640 800
Career Decision bv University
As shown in the Table 9, at all four universities, a higher percentage of females were influenced in their career decision by both parents as compared to other influences. However, in case B, C, and D universities, mothers influenced the females career decision more than the fathers. The mass media was second 81 greatest influence on females’ career decision making at case A and case D universities. Overall, the percentages for the total respondents (n = 1,145) in terms of influence on their career decision making were (a) parents, 32%; (b) others,
14%; (c) mass media, 13%; (d) mother, 11%; (e) counselor, 10%; (f) father, 8%;
(g) friends, 8% and (h) advisor, 6%. Interestingly, advisors tended to have a less influence on females career decision making at the four universities. However, the "others" category included brother, sister, and relatives. 82 Table 9
Person Who Most Influenced Career Decision by University
Universities T o tal Career Decision A B C D Cases by (n=208) (n=458) (n=232) (n=247) (n=1145) No 22 39 16 15 92 F ath er % 10.6 8.5 6.9 6.1 8.0
No 19 48 28 28 123 Mother % 9.1 10.5 12.1 11.3 10.7 No 55 135 69 104 363 P are n ts % 26.4 29.5 29.7 42.1 31.7
No 8 29 20 9 66 A dvisor % 3.8 6.3 8.6 3.6 5.8
No 25 49 25 16 115 Counselor % 12.0 10.7 10.8 6.5 10.0 No 23 26 22 15 86 F riends % 11.1 5.7 9.5 6.1 7.5 No 30 65 18 31 144 Mass Media% 14.4 14.2 7.8 12.6 12.6
No 26 67 34 29 156 Others % 12.5 14.6 14.7 11.7 13.6
V alid cases 1145 Missing cases 130
Career Decision of Total Respondents
Mother SSSSSSSSLskb 123 P are n ts BBBHHHHBBggmnHHBnSBBHHBHHHHUHHBBBBHIBn 363 A dvisor h h h MBB 68 Counselorg^nMBHBH 115 F riends b h MMMMH 86 MaSSMediag^ggggggg^gggmmm 144 Others BBBBBHBHBBBBBHBn -1-56 i :____ i______i______i______i______i 0 80 160 240 320 400 Maior Decision By University
As shown in the Table 10, the first person who influenced females’ major decision before they entered the university was a classroom advisor or teacher in high school, for the case A and C universities respondents; while parents were the first influence on decision about a major before entering the university in case B and D respondents. Mothers had more influence on females’ major decisions at case B, C and D universities as compared to fathers. For case B and D respondents, those in the "others" category were the third most important influence on females’ major decisions. "Others" includes self, brother/sister and relatives.
Overall, the percentage breakdown of influences for the total number of respondents (n= 1063) were (a) parents, 20%; (b) classroom advisor or teacher in high school, 19%; (c) mother, 15%; (d) others, 14%; (e) father, 13%; (f) counselor,
10%; (g) mass media, 7% and (h) friends, 3%. Table 10
Major Decision by University
Universities Major Decision T otal By A B C D Cases (n=185) (n=439) (n=218) (n=221) (n=1063)
No. 33 50 27 29 139 Father % 17.8 11.4 12.4 13.1 13.1 No. 14 69 35 37 155 Mother % 7.6 15.7 16.1 16.7 14.6 No. 29 81 42 55 207 Parents % 15.7 18.5 19.3 24.9 19.5
No. 40 77 45 37 199 Advisor % 21.6 17.5 20.6 16.7 18.7
No. 25 47 26 11 109 Counselor% 13.5 10.7 11.9 5.0 10.3
No. 5 12 8 7 32 Friends % 2.7 2.7 3.7 3.2 3.0 No. 16 30 11 14 71 Mass Media% 8.6 6.8 5.0 6.3 6.7 No. 23 73 24 31 151 Others % 12.4 16.6 11.0 14.0 14.2
Valid cases 1063 Missing cases 212
Major Decision of Total Respondents
F ath er 139 Both Dad and Mom 207 Advisor nUUBBHmBHBBBnHBBH 199 Counselor b BBBBBBBBBBBB 109 F riends ggnBB 32 Mass Media mnaiHBHBB 71 O thers 151 ______1______1______1______I______I 0 80 160 240 320 400 85 Additional Enrichment Educational Activities bv University
As shown in the Table 11, in all four universities, over 70% of the females
responded that they had experienced enriching educational activities in addition to
the school curriculum. As shown in the table 11, the average number of the enriching educational experiences for all universities was 3 (2.50) among females who
responded "Yes" on this variable. 8 10 4 30 400 320 240 160 80 0 00-JO^U1£*U)bJM 347 cases issing M 928 cases alid V SD=1.44 d Actvii s itie tiv c A Ed. M=2.50 M=2.50 Value e % No Yes issing M Enrichment Enrichment D C B A Say Additional Enrichment Educational Activities bv University bv Activities Educational Enrichment Additional SD=1.42 172 m m m n M=2.50 M=2.50 87 No. No. No. 6920.9 26.9 % 1.3 % 94 f"e" Hw ay frTtl Respondents Total for many? How "Yes". If ver tes itie rs e iv n U f "E"Hw many? "YES".How If. (n=234) 1878.7 71.8 168 SD=1.31 al 12 Table A al 11 Table 63 M=2.22 3 (n=507) 9 9 239 191 106 399 B s itie rs e iv n U 4. 1.5 .7 .4 2 M=2.75 SD=1.61 (n=267) 27.7 71.5 C 74 2 SD=1.46 M=2.50 M=2.50 CASES TOTAL (n=267) 95 . 89.5 D 9.0 24 4 (n=1275) tal o T Cases 20.9 78.2 267 997 11 .9 86 87 Educational Plan after Graduation from the University
As shown in the Table 13, at all four universities, the top ranked educational plan for after graduation was self-study; the second ranked educational plan was to pursue graduate school. Taking courses from private vocational institutions for career preparation was ranked third for universities A, B, and C.
Table 13
Educational Plan by Rank Order
Rank Order By Universities
Ed Plan ABCD TOTAL CASE
Pursue Graduate 38%(2) 44% (2) 39% (2) 44% (2) 42% (2) School
Take course in 10%(3) 9% (3) 9% (3) 6% (4) 8% (4) private Voc Inst.
Self-Study 50%(1) 46%(1) 46%(1) 47% (1) 47% (1)
No Ed Plan 8% (4) 6% (4) 8% (4) 6% (5) 6% (5)
Other 5% (5) 9% (3) 6% (5) 14% (3) 9% (3)
Valid cases 1275 Missing cases 0 88 Job Plan after Graduation
Similarly, respondents at all four universities reported that they wanted to have full time employment after graduation as shown in Table 14. Self-employment appeared as the second in the rank order of job plans from the four universities.
These result reflect the sexual-discriminatory policy of Korean businesses toward females. Marriage does not appear as an important career plan after graduation.
Table 14
Job Plan by Rank Order
Rank Order by Universities
Career Plan A B CD TOTAL CASE
Full Time 48% (1) 47% (1) 47% (1) 57% (1) 49% (1) Employment
Part Time 11% (3) 13% (3) 16% (3) 15% (3) 14% (3) Employment
Self-Employment 38% (2) 40% (2) 35% (2) 39% (2) 38% (2)
Do not plan 3% (5) 1 % (6) 3% (6) 1% (6) 2% (6) to Work
Marriage 7% (4) 6% (4) 10% (4) 8% (5) 7% (4)
Other 3% (5) 5% (5) 5% (5) 5% (4) 5% (5)
Valid Cases 1275 Missing cases 0 Attitude Regarding Career and Family bv University
As shown in the Table 15, at all four universities, over 80% of the females
responded that they had a positive attitude toward having both a career and family
at the same time after graduation.
Overall, the percentages of total respondents (n = 1,256) broken down by
attitude regarding career and family were (a) yes, 83%; (b) do not know, 11% and
(c) no, 6%.
Table 15
Attitude Regarding Career and Family
Universities T o tal A ttitu d e A B C D Cases (n=230) (n=507) (n=257) (n=262) (1256)
No 186 409 221 224 1040 Yes % 80.9 80.7 86.0 85.5 82.8
No 14 39 17 11 81 No % 6.1 7.7 6.6 4.2 6.4 Do not No 30 59 19 27 135 Know % 13.0 11.6 7.4 10.3 10.7
Valid cases 1256 Missing cases 19
Attitude of Total Respondents
Yes < ■ ! 1040 No Don' t Know I______I______I______I______I______I 0 240 480 720 960 1200 90 Socio Economic Status (SES) bv University
The occupations reported for fathers and mothers were assigned a social
economic status (SES) index level developed by Hong, Doo Seung (1983). Values
from 1 (low status) through 8 (high status) were assigned.
SES-Father
As shown in the Table 16, mean SES scores for fathers for each of the
universities were (a) case A, 6;(B) case B, 7;(c) case C, 6 and (d) case D, 7.
Interestingly, both single-sex universities, case B and D universities tended to have
higher SES-fathers than the two co-ed Universities.
SES-Mother
As shown in the Table 17, mean scores of SES for mothers for each of the
universities were (a) case A, 2.76;(B) case B, 2.62;(c) case C, 2.95 and (d) case D,
2.28. There is a reason that the SES means for mothers were lower than those for
fathers. Over fifty percent of the females from each university responded that the
occupation of their mother was unemployed, which means housewife, and scored the lowest level SES.
From observing both parents SES in the descriptive data, the researcher
decided to look at only one side of SES. that is. the father’s occupation for further analysis. 91 Table 16
Socio Economic StatuslSESVFather
Universities T o tal A B C D Case
(n=216) (n=489) (n=241) (n=254) (n=1200) SES M=6.2 M=6.70 M=6.16 M=7.05 M=6.58 For F ath er SD=1.80 SD=1.39 SD=1.62 SD=1.28 SD=1.53 No 9 7 7 3 • 26 low est 1 % 4.2 1.4 2.9 1.2 2.2 No 4 1 2 1 8 2 % 1.9 .2 .8 .4 .7 No 12 18 15 3 48 3 % 5.6 3.7 6.2 1.2 4.0
No 3 2 6 4 15 4 % 1.4 .4 2.5 1.6 1.3
No 16 22 19 3 60 5 % 7.4 4.5 7.9 1.2 5.0 No 65 143 81 . 62 351 6 % 30.1 29.2 33.6 24.4 29.3 No 50 130 64 51 295 7 % 23.1 26.6 26.6 20.1 24.6
NO 57 166 47 127 397 highest % 26.4 33.9 19.5 50.0 33.1
Valid cases 1200 Missing cases 75
Socio Economic StatusfSESl-Father for Total Respondents low 1 b 26 2 8 3 ——. 48 4 15 5 — 60 6 7 m u in i in am in mu .... m u .. 295 high8 ■ii...... 397
______I______I______I______1 0 80 160 240 320 400 92 Table 17
Socio Economic StatusfSESl-Mother
Universities
AB C D T o tal Cases SES fo r (n=213) (n=489) (n=242) (n=248) (1192) Mother M=2.76 M=2.62 M=2.95 M=2.28 M=2.64 SD=2.40 SD=2.40 SD=2.49 SD=2.30 SD=2.41
Lowest 1 No. 129 324 142 185 780 % 60.6 66.3 58.7 74.6 65.4
2 No. 4 4 1 2 11 % 1.9 .8 .4 .8 .9 3 No. 12 17 11 4 44 % 5.6 3.5 4.5 1.6 3.7
4 No. 2 2 6 2 12 % .9 .4 2.5 .8 1.0 5 No. 13 15 10 5 43 % 6.1 3.1 4.1 2.0 3.6
6 No. 34 78 47 25 184 % 16.0 16.0 19.4 10.1 15.4 7 No. 14 43 19 19 95 % 6.6 8.8 7.9 7.7 8.0 8 No. 5 6 6 6 23 Highest % 2.3 1.2 2.5 2.4 1.9
Number of M issing O bservations: 83
Socio Economic Status-Mother for Total Respondents
Low 1 — — ——— —— ' . - .... 780 2 m n 3 mmm 44 4 B 12 6 184 8 r 1™ .
0 160 320 480 640 800 93 Parents’Educational Background bv University
As shown in the Table 18 and 19, the single-sex universities had 59% and 76% respectively of females whose fathers had more than a 4 year college education, while 43% of the case A university and 40% of the case D university females had fathers with more than a 4 year college education. However, the educational level of mothers showed a different pattern. The majority of the mother’s educational levels showed less than a 2 year vocational/technical education, as is shown in the following break down in those Universities ;(a) case A, 78%;(b) case B, 71% and (c) case C, 78%. However for case D university, about 50% respondents reported that their mothers had 4 year college/university education. 94 Table 18
Educational Background for Father
Universities F a th e r 's TOTAL Ed Background A B C D CASES (n=219) (n=493) (n=251) (n=257) (n=1217)
Less than No 36 27 35 7 105 High-School % 16.4 5.5 13.9 2.8 8.6
No 87 165 107 50 409 High School % 39.7 33.5 42.6 19.7 33.6
2 y rs No. 2 10 6 3 21 Voc Edc % .9 2.0 2.4 1.2 1.7
4 y r No 72 223 77 137 509 Coll or Unv % 32.9 45.2 30.7 53.9 41.8
Advanced No. 21 67 23 57 168 Degree % 9.6 13.6 9.2 22.4 13.8
Do Not No. 1 1 3 5 Know % .5 .2 1.2 .4
V alid cases 1217 M issing cases 58
Educational Background for Father of Total Respondents
Less than Hi sch 105 High School BBBBBSBnBBKBBBBBBBBHBBBBBnBBHBi 409 2 y rs Voc Educ 21 4 y r C oll o r Univ 509 Advanced Degree 168 Do not Know 5 I______I______I______I______1______I 0 120 240 360 480 600 95 Table 19
Educational Background for Mother
Universities M other' s TOTAL Ed Background A B C D CASES (n=224) (n=494) (n=251) (n=257) (n=1226) Less than No 65 88 80 20 253 High-school % 29.0 17.8 31.9 7.8 20.6
High School No 109 262 119 99 589 % 48.7 53.0 47.4 38.5 48.0 2 y r No 1 13 5 3 22 Voc Edc % .4 2.6 2.0 1.2 1.8
4 yr No 43 122 37 128 330 Coll or Univ% 19.2 24.7 14.7 49.8 26.9
Advanced No 5 6 5 6 22 Degree % 2.2 1.2 2.0 2.3 1.8 Do not No 1 3 5 1 10 Know % .4 .6 2.0 .4 .8
Valid cases 1226 Missing cases 49
Educational Background for Mother of total respondents
Less -than High sch 253 High School HHnnHBHBnunB 2 yrs Voc Educ 22 4 yr Coll or Univ Advanced Degree gg| 22 Do not Know g 10 I______I______I______I______I______I 0 120 240 360 480 600 96 Location where most education was completed bv University
As shown in the Table 20, at all four universities, the majority of females had
completed most of their education in a large city;(a) Case A, 69 %;(b) Case B,
78%;(c) 67% and (d) Case D, 85%. Female students at single-sex universities tended
to complete most of their education in large cities.
Table 20
Location of education by University
UNIVERSITIES T o tal Location A B C D cases (n=224) (n=500) (n=258 (n=257) (n=1239) No 21 18 23 4 66 Farm % 9.4 3.6 8.9 1.6 5.3
Sm all/ No 48 94 61 35 238 Mid C ity % 21.4 18.8 23.6 13.6 19.2 No 155 388 174 218 935 Urban % 69.2 77.6 67.4 84.8 75.5
Valid cases 1239 Missing cases 36
Location Completed Most of Education of Total Respondents
On farm —tii 66 Small/m id c ity flBBEBHHBBHEi 238 Large c ity BBBnaBBBBBHMBBMBBHBBMHWHMBmmMiBa B H B r a w M i 935
0 200 400 600 800 1000 Hometown bv University
As shown in the Table 21, at all four universities, the majority of females
reported their hometown to be large cities. A greater percentage of female students
at single-sex universities tended to come from large city hometown than the
percentage from co-ed universities.
Table 21
Home town by University
UNIVERSITIES T otal Home town A B C D cases (n=226) (n=501) (n=258) (n=257) (n=1242)
No 54 55 44 15 168 Farm % 23.9 11.0 17.1 5.8 13.5 Sm all/ No 33 99 65 43 240 Mid C ity % 14.6 19.8 25.2 16.7 19.3
Large No 139 347 149 199 834 C ity % 61.5 69.3 57.8 77.4 67.1
Valid cases 1242 Missing cases 33
Hometown of Total Respondents
On farm 168 Small to mid city HBBEBmHSHI 240 Large city ®^4 I______I______I______I______1______1 0 200 400 600 800 1000 98 Previous Work Experience bv University
As shown in the Table 22, 80% of females at the four universities had job experience in their school year at least one time.
Table 22
Number of Different Type of Career Experience bv University
Universities No. of T o tal Job exp A B C D Cases (n=234) (n=507) (n=267) (n=267) (n=1275)
* .00 40 125 43 41 249 17.1 24.7 16.1 15.4 19.5
1.00 103 240 136 158 637 44.0 47.3 50.9 59.2 50.0
2.00 61 102 56 55 274 26.1 20.1 21.0 20.6 21.5 3.00 24 32 26 13 95 10.3 6.3 9.7 4.9 7.5
* 4.00 6 8 6 20 2.6 1.6 2.2 1.6
Valid cases 1275 Missing cases *0 : No experience of any job *4 : 4 types of different job experience, those are Employment full time, Employed part time, Only During Summer or Winter Vocation, and tu to r .
Number of Different Job Experience for Total Case
.00 249 1.00 m 637 2.00 m 274 3.00 95 4.00 20
160 320 480 640 800 Grade Point Average CGPA)
As shown in the Table 23, over 80% of females of each four universities reported that their grade point average was over a "B"or higher. It should be noted that some bias in findings may result due to self-reported information.
Table 23
Grade Point Average by University
Universities T o tal GPA A B C D Cases (n=224) (n=493) (n=252) (n=240) (n=1209)
A NO 54 42 44 30 170 % 24.1 8.5 17.5 12.5 14.1
B No 143 396 179 186 904 % 63.8 80.3 71.0 77.5 74.8 C No 25 54 27 . 24 130 % 11.2 11.0 10.7 10.0 10.8 D No 2 1 2 5 % .9 .2 .8 .4
Valid cases 1209 Missing cases 66
Grade Point Average of Total Respondents
Below C 5
0 200 400 600 800 1000 100 Career Indecision Score (item 3 through 18) and TSOSS Score
As shown in the Table 24, at all four universities career indecision score
showed was higher (M=37.97) as compared to American college females. The
Career Decision Scale manual shows that female college student mean in the United
States score was 26.88 (Appendix P). On the other hand, respondents in case B
university showed higher TSOSS total score as compared to other universities.
The four sub-scales of CDS and TSOSS
As shown in the Table 25, respondents in case A, B and D universities had
higher score on second sub-scale of CDS, which represents uncertainty about how to proceed in making a decision and the need for additional support for initial decisions. Respondents in case C university had higher score on fourth sub-scale of
CDS, which represents both external barriers to career choice and lack of interest in
making a decision.
On the other hand, at all four universities respondents showed higher mean scores on first and fourth sub-scales of TSOSS comparedto other sub-scales of
TSOSS. First sub-scales of TSOSS is verbal/interpersonal related skills. Fourth sub scales of TSOSS is agility skills. Table 24
CDS and TSOSS by University
Universities Total AB C D 4 sub-scales of CDS 1 sub-scale M 6.9 7.2 7.0 7.3 7.13 2 sub-scale SD 2.6 2.6 2.7 2.6 2.62
M 8.1 8.4 8.3 8.2 8.29 3 sub-scale SD 2.0 1.9 2.0 2.0 1.96
M 7.3 8.0 7.5 7.6 7.67 4 sub-scale SD 2.2 2.2 2.5 2.3 2.30
M 8.0 7.7 8.5 7.7 7.92 SD 2.7 2.6 2.6 2.3 2.56
Career Indecision M 37.0 38.4 38.1 38.0 37.97 Score SD 8.4 8.0 8.7 7.8 8.19 4 sub-scales of TSOSS TSOSS Total M 205.3 213.4 208.7 203.3 208.8 SD 26.8 30.3 31.8 29.8 30.2 1 sub-scale M 54.9 55.9 56.1 53.8 55.32 SD 7.9 8.4 8.4 8.2 8.30 2 sub-scale M 49.5 52.5 50.2 51.1 51.18 SD 9.6 9.3 9.7 10.1 9.66 3 sub-scale M 49.7 51.1 51.0 48.6 50.29 SD 9.0 10.6 10.4 10.5 10.32 4 sub-scale M 51.2 53.6 51.4 50.0 51.94 SD 10.0 10.8 10.8 9.9 10.53 102
Table 25
Career Indecision Score fltem 3-181
V alid Cum Value Label Value Frequency P ercent P ercen t Percent
Minimum score 16.00 1 .1 .1 .1 17.00 1 .1 .1 .2 18.00 2 .2 .2 .3 19.00 6 .5 .5 .8 20.00 9 .7 .7 1.5 21.00 5 .4 .4 2.0 22.00 14 1.1 1.1 3.1 23.00 15 1.2 1.2 4.3 24.00 16 1.3 1.3 5.6 25.00 19 1.5 1.5 7.2 26.00 23 1.8 1.9 9.1 27.00 27 2.1 2.2 11.3 28.00 28 2.2 2.3 13.5 29.00 31 2.4 2.5 16.1 30.00 32 2.5 • 2.6 18.7 31.00 43 3.4 3.5 22.2 32.00 49 3.8 4.0 26.2 33.00 45 3.5 3.7 29.9 34.00 58 4.5 4.7 34.6 35.00 50 3.9 4.1 38.7 36.00 57 4.5 4.6 43.3 37.00 53 4.2 4.3 47.6 38.00 53 4.2 4.3 52.0 39.00 48 3.8 3.9 55.9 40.00 53 4.2 4.3 60.2 41.00 49 3.8 4.0 64.2 42.00 64 5.0 5.2 69.4 43.00 57 4.5 4.6 74.1 44.00 40 3.1 3.3 77.3 45.00 32 2.5 2.6 79.9 46.00 49 3.8 4.0 83.9 47.00 44 3.5 3.6 87.5 48.00 30 2.4 2.4 90.0 49.00 31 2.4 2.5 92.5 50.00 18 1.4 1.5 94.0 51.00 20 1.6 1.6 95.6 52.00 14 1.1 1.1 96.7 53.00 12 .9 1.0 97.7 54.00 9 .7 .7 98.5 55.00 3 .2 .2 98.7 56.00 6 .5 .5 99.2 57.00 7 .5 .6 99.8 58.00 2 .2 .2 99.9 Maximum score 60.00 1 .1 .1 100.0 • 49 3.8 M issing Total 1275 100.0 100.0
V alid cases 1225 M issing cases 49 103 (Continued Table 25) Career Indecision Score for total Respondents
16. 00 17. 00 18. 00 19. 00 2 0 . 00 21. 00 22. 00 23. 00 24. 00 25. 00 19 26. 00 23 27. 00 27 28. 00 ■ 28 29. 00 31 30. 00 32 31. 00 43 32. 00 49 33. 00 45 34. 00 58 35. 00 36. 00 571 37. 00 38. 00 39. 00 40. 00 41. 00 42. 00 64 43. 00 57 44. 00 40 45. 00 32 46. 00 49 47. 00 44 48. 00 30 49. 00 i 31 50. 00 18 51. 00 a 20 52. 00 53. 00 54. 00 55. 00 56. 00 57. 00 58. 00 60. 00
15 30 45 60 75 104 Table 26
TSOSS Total Score for Total Respondents
Total TSOSS (Item 1 - 60)
V alid Cum Value Label Value Frequency Percent P ercent P ercent 90-99 1.00 1 .1 .1 .1 120-129 2.00 1 .1 .1 .2 130-139 3.00 6 .5 .5 .7 140-149 4.00 13 1.0 1.1 1.8 150-159 5.00 25 2.0 2.1 3.9 160-169 6.00 61 4.8 5.2 9.2 170-179 7.00 76 6.0 6.5 15.7 180-189 8.00 130 10.2 11.1 26.8 190-199 9.00 153 12.0 13.1 39.9 200-209 10.00 157 12.3 13.4 53.3 210-219 11.00 145 11.4 12.4 65.8 220-229 12.00 121 9.5 10.4 76.1 230-239 13.00 91 7.1 7.8 83.9 240-249 14.00 71 5.6 6.1 90.0 250-259 15.00 51 4.0 4.4 94.3 260-269 16.00 32 2.5 2.7 97.1 270-279 17.00 15 1.2 1.3 98.4 280-289 18.00 17 1.3 1.5 99.8 290-299 19.00 2 .2 .2 100.0 • 107 8.4 M issing T otal 1275 100.0 100.0
V a lid cases 1275 M issing cases 107
90-99 120-129 130-139 140-149 150-159 160-169 61 170-179 76 180-189 6 ' "0 190-199 153 200-209 . 157 210-219 145 220-229 121 230-239 91 240-249 71 250-259 51 260-269 32 270-279 280-289 290-299
0 40 80 120 160 200 Data Analysis
All completed questionnaires were coded and the data were entered into a
personal computer by the researcher. The Statistical Package for the Social Sciences
was used to analyze the data. This study was designed to investigate the utility of
Bandura’s self-efficacy theory to the understanding and treatment of career indecision in female students in Korea. The major purposes of this study were to 1) investigate
the relationship between women’s career indecision and self-efficacy expectations and other selected background characteristics, and 2) to differentiate between decided and undecided students on the suggested variables.
More specifically, this study was conducted to answer the following research questions. The First Question and Results
Question: Are the four sub-scales of CDS related to the 4 sub-scales of TSOSS? Explain what information is being carried bv the canonical variates if each of the pairs of independent and dependent variable canonical variates (canonical roots) are meaningful?
To answer this question canonical correlation was used. The general question addressed by canonical correlation is to what extent can one set of two or more dependent (criterion) variables be explained (or predicted) by another set of two or more independent (predictor) variables. To answer this question by canonical correlation, the following variables were used:
INDEPENDENT VARIABLE SET (The Four Sub-scales of TSOSS)
VERBAL Verbal and Interpersonal (Total score of the'TSOSS items 1, 5, 9, 13, 17, 21, 25, 29, 33, 37, 41,45,49,53, and 57.)
SCIENCE Quantitative, Scientific and Business (Total score of the TSOSS items 2, 6,10, 14, 18, 22, 26, 30, 34, 38, 42, 46, 50, 54, and 58)
STRENGTH Physical strength and Agility (Total score of the TSOSS items 3, 7,11, 15,19, 23, 27, 31, 35, 39, 43, 47, 51, 55, and 59)
AESTHE Aesthetic skills (Total score of the TSOSS items 4, 8,12, 16, 20, 24, 28, 32, 36, 40, 44, 48, 52, 56, and 60) 107 DEPENDENT VARIABLESET (Four Sub-scales of CDS)
CONFUS Total Score of the CDS item 7, 8, and 11. (Diffusion, which represents feelings of confusion, discouragement and lack of experience or information about the making of career decisions.)
UNCERT Total Score of the CDS item 12,16, and 18. (Support, which represents uncertainty about how to proceed in making decisions, and the need for additional support for initial decisions.)
CLASIC Total Score of the CDS item 4, 15, and 17. (Approach, which represents a classical approach-approach conflict in which several possible careers are attractive.)
EXTERNL Total Score of the CDS item 3,5,6 and 9 (External Barrier, which represents both external barriers to career choice and lack of interest in making a decision.)
Statistical Hypothesis: There are no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho: All squared canonical correlation coefficients (Rc2) are equal to zero.
Total Case (n=1135)
For testing the hypothesis, Wilks’ lambda for each squared canonical correlation coefficient (Rc2) was used to test the statistical hypotheses. The Wilks’ lambda statistic is transformed to an F distribution which is used as the test statistic.
The information for testing the hypothesis for all the respondents follows.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .89; F = 8.80; p < .001
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .99;F = 1.85;p =.54 Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .996; F = .998;p =.407 108
Table 27
Testing Hypothesis for The Combined Cases
Roots Wilks L. F Significance Decision
1 TO 4 .88476 8.80113 <.001 Reject 2 TO 4 .98537 1.85298 .054 Fail to Reject 3 TO 4 .99647 .99872 .407 Fail to Reject 4 TO 4 1.00000 .00336 .954 Fail to Reject
Therefore, only canonical root 1 is statistically significant at an alpha level of .05. These significance tests can be used for deciding which canonical functions to interpret. In addition, statistics for determining the meaningfulness of canonical roots are based on a value of squared canonical correlation coefficients (Rc2) greater than .10, and the redundancy index. The summary Table 29 presents information about the roots and the appropriate statistics. Therefore, primarily canonical root
1 is statistically significant. However the value of Rc2 demonstrates that canonical root 1 explains only 10 % (Rc2 = .10) of the total variance in the dependent canonical variate (a linear combination of the dependent variable set) shared with the independent canonical variate (a linear combination of the independent variable set).
Moreover, the value of the redundancy index demonstrates that canonical root 1 explains only2% o fthe total variance in the dependent variable set accounted for by the independent variable set. Therefore, even if it is interpreted as meaningful, canonical root 1 practically adds little to the explanation of the variance. The reason that the hypothesis proved statistically significant may be due to the large 109 sample size in this case (n = 1135).
In order to explain the nature(direction) of the relationship between the independent variable set and the dependent variable set, structure coefficients
(canonical loadings) were used. Structure coefficients (loadings) are product-moment correlations between a variable in the dependent or independent variable set and the canonical variate. Structure coefficients identify what variables the variates are carrying information about. The rule of thumb is that structure coefficients of .30 or greater are meaningful.
As shown in the Table 29, dependent variate 1 carries information about
CONFUS (s=. 81) in the dependent variable set. CONFUS represents the first sub scale of CDS, that is "diffusion" which represents feelings of confusion, discouragement, and lack of experience or information about the making of career decisions. Females with higher values on CONFUS then have higher scores on dependent variate 1; females with lower values on CONFUS have lower scores on dependent variate 1.
However, independent variate 1 carries information about all four sub-scales of TSOSS in the independent variable set, since the values of the structure coefficient for the four variables are greater than .30, but the direction is negative. In other words, females with higher values on each of the 4 sub-scales of TSOSS, have lower scores on independent variate 1; females with lower values on each of the 4 sub scales of TSOSS have higher scores on independent variate 1.
Therefore, it can be interpreted that females who had higher confusion (first 110 sub-scales of CDS1 in making a career decision typically had lower scores on the four sub-scales of TSOSS. which are verbal/interpersonal, quantitative scientific/business, physical strength and agility, and aesthetics related skills: females who had lower confusion in making a career decision typically had higher scores on the four sub scales of the TSOSS.
However, even though canonical root 1 is statically significant and meaningful, it should be noted that only 2% (Redundancy index = .021) of variance in the dependent variable set is explained by the first independent canonical variate (linear combination of the independent variable set).
Table 28
Rc and Rc2 for the Total Case
Root No. Canon Cor. Sq. Cor. significant
1 .320 .102 <.001 2 .106 .011 .054 3 .059 .004 .407 4 .002 .000 .954 I ll Table 29
Mean, and Standard Deviations for DV and IV sets ______fn = 11351______
Variables Mean Std Dev N Label
CONFUS 7.13 2.62 1250 Dv’s Set UNCERT 8.29 1.96 1268 CLASIC 7.67 2.30 1249 EXTRNL 7.93 2.56 1242 VERBAL 55.32 8.30 1237 Iv’s Set SCIENCE 51.18 9.66 1216 STRENGTH 50.29 10.32 1217 AESTHE 51.94 10.53 1226 112
Table 30
Summary of Canonical Correlation Analysis for Total Case (n=1135)
Canonical Canonical Canonical Canonical V ariables V ariate 1 Variate 2 Variate 3 V ariate 4 b s b s b s b s
For IV Set
VERBAL -.872 -.991 .147 -.023 .818 .106 .716 .073 SCIENCE -.156 -.712 .593 .313 -1.075 -.550 -.368 -.304 STRENGTH -.015 -.495 -.040 -.231 .581 .221 -1.058 -.808 AESTHE -.032 -.529 -1.037 -.779 -.628 -.308 .144 -.132
PV .051 .002 .000 .000
Rc2(1)= .102 Rc2(2)=.011 Rc2(3)=.004 Rc2(4)=.000
For DV Set
CONFUS 1.165 .812 -.131 -.360 -.520 -.010 .003 .459 UNCERT -.113 -.222 -.859 -.853 -.218 -.285 -.478 -.376 CLASIC -.594 -.116 -.143 -.438 -.340 -.143 .932 .880 EXTRNL -.140 .285 -.353 - .446 1.172 .754 -.003 .389
PV .200 .312 .168 .003 Rd .021 .003 .000 .000 Rd .021 .024 .025 .025
Rc2 = Proportion of variance in the dependent canonical variate (linear combination of dependent variable set) shared with the independent canonical variate (linear combination of the independent variable set).
PV = Proportion of variance in the dependent variable set explained by the jth dependent canonical v a ria te .
Rd = Proportion of variance in the dependent variable set explained by the jth independent canonical variate.
Rd = Total proportion of the variance in the dependent variable set explained by the independent variable set. b = Standardized canonical coefficients (weights) s = Structure coefficients 113 Replication 1: Case A University (n= 204)
Statistical Hypothesis: There are no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho: All squared canonical correlation coefficients (Rc2) are equal to zero.
For testing the Hypothesis, the following information demonstrates the roots and the appropriate statistics for Case A University.
Ho : Rc(l) 2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .83; F = 2.30; p = .003
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .95; F = 1.07;p =.383
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .97; F =1.40; p =.232
Table 31
Testing Hypothesis for Case A University
Roots Wilks L. F Significance Decision
1 TO 4 .83302 2.30880 .003 Reject 2 TO 4 .95272 1.07121 .383 Fail to Reject 3 TO 4 .97226 1.40238 .232 Fail to Reject 4 TO 4 _ .9.9146_ _ 1.71347 .192 -Fail.to. Reject...... -
As shown in the Table 31, canonical root 1 is significant at an alpha level of
.05. These significance tests can be used for deciding which canonical functions to 114 interpret. Therefore, the most practical meaningful strategy is to interpret canonical root 1 only. Rc2 for canonical root 1 is .13, which supports the meaningful interpretation of root 1. Structure coefficients (loadings), which identify what variables the variates are carrying information about, were used to describe the nature (direction) of the relationship. As shown in the summary Table 34 for canonical root 1, the dependent variate primarily carries information about
CONFUS(s=.86), and EXTRNL (s=.45) in the dependent variable set. In other words, females with higher values on Confusion and External barriers on making career decision have higher scores of dependent variate 1; females with lower values on confusion and external barriers on making career decisions have lower scores on dependent variate 1. However, independent variate 1 carries information about
VERBAL (s=-.91), SCIENCE (s=-.71), STRENGTH (s=-.30),and AESTHE (s=-
.57) in the independent variable set. In other words, females with higher values on those 4 sub-scales of TSOSS have higher scores on independent variate 1; females with lower values on those 4 sub-scales of TSOSS have lower scores on independent variate 1.
Therefore, by looking at both sides of variate 1, the nature of this relationship can be interpreted as demonstrating that females who had more confusion and greater external barriers (first and fourth sub-scales of the CDS') in making career decision typically had lower scores on the 4 sub-scales of the TSOSS: females who had less confusion and external barriers in making their career decision typically had higher scores on the four sub-scales of the TSOSS. However, even though 115 canonical root 1 is statistically significant and meaningful, only (redundancy index=.032) 3% of the variance in the dependent variable set is explained by the first independent canonical variate (linear combination of the independent variable set).
Therefore root 1 practically adds little to the explanation of the variance.
Table 32
Rc and Rc2 for the Case A University
Root No. Canon Cor. Sq. Cor
1 .354 .126 2 .142 .020 3 .139 .019 4 .092 .009 116 Table 33
Means and Standard Deviation for DV and IV sets (n=229)
Means and Standard Deviation for DV set
Variable Mean Std Dev N Label
CONFUS 6.88 2.61 230 UNCERT 8.09 2.03 234 CLASIC 7.34 2.24 229 EXTRNL 8.04 2.65 230
Means and Standard Deviation for IV set (n=208)
Variable Mean Std Dev N Label
VERBAL 54.92 7.88 224 SCIENCE 49.46 9.59 219 STRENGTH 49.69 8.96 219 AESTHE 51.16 9.95 221 117
Table 34
Summary of Canonical Correlation Analysis for Case A University (n=204)
Canonical Canonical Canonical Canonical V ariables V ariate 1 V ariate 2 V ariate 3 V ariate 4 b s b s b s b s
For IV Set
VERBAL -.720 - .914 .638 .333 .461 -,.066 .724 .221 SCIENCE -.335 - .712 -.991 .471 -.572 -, .520 - .065 .038 STRENGTH .264 - .286 .631 .540 -.942 -..735 - .143 .293 AESTHE -.314 -•.570 -.102 .193 .448 .090 -1 .009 .794
PV .055 .003 .004 .002
Rc2<1)=. 126 Rc2(2)=..020 Rc2(3)=.019 Rc2(4)=.,009
For DV Set
CONFUS 1.079 .861 -.677 -.230 -.138 .367 -.129 .267 UNCERT .303 .249 .583 .447 .216 .441 -.772 .737 CLASIC -.575 .076 -.340 -.117 .995 .967 .194 .212 EXTRNL .087 .446 1.039 .524 -.023 .309 .646 .657
PV .252 .135 .340 .273 Rd .031 .003 .007 .002 13 .031 .034 .041 .043
b : Standized canonical coefficients (weights) s : Structure coefficients PV : Proportion of variance explained Rd : Redundancy Rd : Total redundancy 118 Replication 2: Case B University (n=476)
Statistical Hypothesis: There are no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients (Rc2) are equal to zero.
For testing the hypothesis, the following information represents the roots and the appropriate statistics for Case B University.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .88; F = 3.93; p < .000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .98; F = .965; p =.468
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .99; F =.969;p =.333
Table 35
Testing Hypotheses for Case B University
Roots Wilks L. F Significance Decision
1 TO 4 .87514 3.92953 <.001 Reject 2 TO 4 .98146 .96474 .468 Fail to Reject 3 TO 4 .99168 .96922 .423 Fail to Reject 4 TO 4 .99798 .93829 .333 Fail to Reject
As shown in the Table 35, canonical root 1 is significant at an alpha level of
.05. These significance tests can be used for deciding which canonical functions to 119 interpret. Therefore, the most practical meaningful strategy is to interpret canonical root 1. Rc2 for canonical root 1 i s . 108 which supports meaningful interpretation of root 1. Structure coefficients (loadings) were used, which have more than .30, to identify what variables the variates are carrying information about.
As shown in the Table 38, for canonical root 1, the dependent variate primarily carries most information about CONFUS (s=.75),and UNCERT (s=.47) in the dependent variable set, and independent variate carries information about
VERBAL(s=-.92), AESTHE (s=-.76), SCIENCE (s=-.56),and STRENGTH (s=-
.38), in the independent variable set. In other words, females with higher values on
CONFUS and UNCERT have higher scores on dependent variate 1; females with lower values on CONFUS and UNCERT have lower scores on dependent variate 1.
In addition females with higher values on 4 sub-scales of TSOSS have lower scores on independent variate 1; females with lower values on 4 sub-scales of TSOSS have higher values on independent variate 1. Therefore, the nature (direction) of the relationship between independent variable and dependent variable set is that females who had higher confusion and uncertainty (first and second sub-scales of CDS’) in making a career decision typically had the lower scores on the four sub-scales of the
TSOSS: females who had lower confusion and uncertainty (first and second sub-scales of the CDS') in making a decision typically had the higher scores on four sub-scales of the TSOSS.
However even though canonical root 1 is statistically significant and 120 meaningful only (Rd = .002) 2% of the variance in the dependent variable set is explained by the first independent canonical variate (linear combination of the independent variable set). Even if it is interpreted as meaningful, canonical root 1 practically adds little to the explanation of the variance. 121
Table 36
Rc and Sq. Rc for Case B University
Root No. Canon Cor. Sq. Cor
1 .329 .108 2 .102 .010 3 .079 .006 4 .045 .002
Table 37
Means and Standard Deviation for DV and IV sets for Case B Universitv
Means and Standard Deviation for DV set (n=498)
Variable Mean Std Dev N Label
CONFUS 7.20 2.62 503 UNCERT 8.44 1.90 505 CLASIC 7.97 2.18 503 EXTRNL 7.72 2.56 502
Means and Standard Deviation for IV set (n=476)
Variable Mean Std Dev N Label
VERBAL 55.91 8.38 501 SCIENCE 52.48 9.32 495 STRENGTH 51.08 10.63 494 AESTHE 53.57 10.77 496 Table 38
Summary of Canonical Correlation Analysis for Case B University (n=469)
Canonical Canonical Canonical Canonical V ariables V ariate 1 V ariate 2 V ariate 3 V ariate 4 b s b s b s b s
For IV Set
VERBAL -.755 -.923 -.382 -.157 -.889 -.016 -.651 -.352 SCIENCE -.012 -.564 -.548 -.316 1.157 .743 .223 -.175 STRENGTH .183 -.382 .801 .663 .280 .406 -.851 -.500 AESTHE -.480 -.763 .466 .507 .079 .162 1.048 .367
PV .051 .002 .001 .000
Rc2(1) =.108 Rc2(2)= .010 Rc2(3)=.006 Rc2(4)= .002
For DV Set
CONFUS 1.106 .749 .500 .511 -.073 -.306 .370 -.289 UNCERT -.361 .473 .553 .600 -.512 -.368 .583 .530 CLASIC -.365 .123 .628 .754 .687 .202 -.559 -.613 EXTRNL -.410 .108 -.338 .181 -.956 -.680 -.648 -.702
PV .203 .310 .183 .310 Rd .022 .003 .001 .000 EcT .022 .025 .026 .026 b : Standized canonical coefficients (weights) s : Structure coefficients PV : Proportion of variance explained Rd : Redundancy E3" : Total redundancy 123
Replication 3: Case C University (n=233)
Statistical Hypothesis: There are no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho: AI1 squared canonical correlation coefficients (Rc2) are equal to zero.
For testing the hypothesis, the following information represents the roots and the appropriate statistics for Case C University.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .81; F = 2.93; p < .000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .95; F =1.28; p =.247
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .98; F =1.12; p =.345
Table 39
Test Hypothesis for Case C University
Root No. Wilks L. F Significance Decision
1 TO 4 .80941 2.93114 <.001 Reject 2 TO 4 .94854 1.27618 .247 Fail to Reject 3 TO 4 .97954 1.12214 .345 Fail to Reject 4 TO 4 .99789 .45818 .499 Fail to Reject
As shown in the Table 39, canonical root 1 is significant at an alpha level of
.05 These significance tests can be used for deciding which canonical functions to interpret. Therefore, the most practical meaningful strategy is to interpret canonical 124 root 1 only. Rc2 for canonical root 1 is .147, which supports meaningful interpretation of root 1. Structure coefficients (loadings) were used, which have more than .30, to identify what variables the variates are carrying information about.
As shown in the summary Table 42, for root 1, dependent variate carries most information about CONFUS (s=.90), and EXTRNL (s=.69) in the dependent variable set, and independent variate carries information about VERBAL(s=-.79),
SCIENCE(s=-.65), and STRENGTH (s=-.46), in the independent variable set. In other words, females with higher values on CONFUS and EXTRNL have higher scores on dependent variate 1; females with lower values on CONFUS and EXTRNL have lower scores on dependent variate 1. In addition females with higher values on
VERBAL, SCIENCE and STRENGTH related self-efficacy skills have lower scores on independent variate 1; females with lower values on VERBAL, SCIENCE and
STRENGTH related self-efficacy skills have higher values on independent variate 1.
Therefore, the nature (direction) of the relationship between independent variable and dependent variable set for canonical root 1 is that females who had higher confusion and greater external barriers (first and fourth sub-scales of CDSi in making a career decision ranked lowest scores on verbal/interpersonal, quantitative scientific/business, and physical strength related skills (first, second, and third sub scales on TSOSS'): females who had lower confusion and smaller external barriers in making a career decision had higher scores on first, second and third sub-scales on the TSOSS. 125 However, even though canonical root 1 is statistically significant and meaningful, only (Rd=.049)5% of the variance in the dependent variable set is explained by the first independent canonical variate (linear combination of the independent variable set). Even though interpreted as meaningful, root 1 practically adds little to the explanation of the variance.
Table 40
Rc and Sq Rc for Case C University
Root No. Canon Cor. Sq. Cor
1 .383 .147 2 .178 .032 3 .136 .018 4 .046 .000 Table 41
Mean and Standard Deviation for DV and IV sets for Case C University
Means and Standard Deviation for DV set (n=251)
Variable Mean Std Dev N Label
CONFUS 7.02 2.67 256 UNCERT 8.25 1.98 264 CLASIC 7.49 2.47 256 EXTRNL 8.46 2.60 254
Means and Standard Deviation for IV set (n=233)
Variable Mean Std Dev N Label
VERBAL 56.08 8.43 252 SCIENCE 50.23 9.66 245 STRENGTH 51.02 10.41 246 AESTHE 51.44 10.84 248 127
Table 42
Summary of Canonical Correlation Analysis for Case C University (n=469)
Canonical Canonical Canonical Canonical V ariables V ariate 1 Variate 2 Variate 3 V ariate 4 b s b s b s b s
For IV Set
VERBAL -1.023 -.791 .719 -.369 .989 .386 .245 .291 SCIENCE -.129 -.648 -.600 -.596 -1.206 .323 .487 .348 STRENGTH -.281 -.462 -.334 -.658 -.041 .143 -1.136 .578 AESTHE .773 -.033 -.831 -.827 .460 .510 .442 .234
PV .046 .013 .002 .000
Rc2(1)=.147 Rc2(2)=.032 Rc2(3)=. 018 Rc2(4)=.002
For DV Set
CONFUS .933 .903 .630 .190 .232 .380 .766 -.066 UNCERT .005 -.166 -.618 -.643 .419 .457 .696 .592 CLASIC -.478 .117 .279 .070 .903 .875 -.529 -.465 EXTRNL .310 .690 -1.019 -.455 -.356 .197 -.745 -.527
PV .333 .165 .290 .212 Rd .049 .005 .005 .000 Rd .049 .054 .059 .059
b: Standardized canonical coefficients (weights) s: Structure coefficients PV: Proportion of variance explained Rd: redundancy Rd: Total redundancy 128 Replication 4: Case D University (n=2Q4)
Statistical Hypothesis: There is no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients (Rc2) are equal to zero.
For testing the Hypothesis, the following information presents about the roots and the appropriate statistics for Case D University.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .78; F = 3.77; p < .000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .98;F = .608;p =.791
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .99; F =.296;p =.998
Table 43
Test Hypothesis for Case D University
Roots Wilks L. F Significance Decision
1 TO 4 .77947 3.76742 <.001 Reject 2 TO 4 .97692 .60761 .791 Fail to Reject 3 TO 4 .99495 .29647 .880 Fail to Reject 4 TO 4 1.00000 .00001 .998 Fail to Reject
As shown in the Table 43 canonical root 1 is significant at an alpha level of
.05 This significance test can be used for deciding which canonical functions to interpret. Therefore, the most practical meaningful strategy is to interpret canonical root 1 only. Rc2 for canonical root 1 is .202, which support meaningful interpretation 129 of root 1. Structure coefficients (loadings) were used, which have more than .30, to identify what variables the variates are carrying information about.
As shown in the summary Table 46, for root 1, dependent variate carries most information about CONFUS (s=. 68) in the dependent variable set, and independent variate carries information about VERBAL (s=-.91), SCIENCE (s=-.76), and
STRENGTH (s=-.46) in the independent variable set. In other words, females with higher values on CONFUS have higher scores on dependent variate 1; females with lower values on CONFUS have lower scores on dependent variate 1. In addition, females with higher values on VERBAL, SCIENCE and STRENGTH related self- efficacy skills have lower scores on independent variate 1; females with lower values on VERBAL, SCIENCE and STRENGTH related self-efficacy skills have higher values on independent variate 1.
Therefore, the nature (direction) of the relationship between independent variable and dependent variable set is that females who had higher confusion and greater external barriers (first and fourth sub-scales of CDS') in making a career decision had lowest scores on verbal/interpersonal, quantitative scientific/business, and physical strength related skills 6first, second, and third sub-scales on TSOSS): females who had lower confusion and smaller external barriers in making a career decision had higher scores on first, second and third sub-scales on the TSOSS.
However, even though canonical root 1 is statistically significant and meaningful, only (Rd=.03) 3% of the variance in the dependent variable set is explained by the first independent canonical variate (linear combination of the 130 independent variable set). Even though interpreted as meaningful, root 1 practically adds little to explanation of variance. 131 Table 44
Rc and Sq of Rc for Case D University
Root No. Canon Cor. Sq. Cor
1 .450 .202 2 .135 .018 3 .071 .005 4 .000 .000
Table 45
Means and Standard Deviation for DV and IV sets for Case D Universitv
Means and Standard Deviation for DV set (n= 254)
Variables Mean Std Dev N Label
CONFUS 7.33 2.55 261 UNCERT 8.23 1.97 265 CLASIC 7.56 2.34 261 EXTRNL 7.68 2.35 256
Means and Standard Deviation for IV set (n= 251)
Variables Mean Std Dev N Label
VERBAL 53.81 8.20 260 SCIENCE 51.05 10.09 257 STRENGTH 48.61 10.51 258 AESTHE 50.00 9.85 261 132
Table 46
Summary of Canonical Correlation Analysis for Case D University
Canonical Canonical Canonical V ariables V ariate 1 V ariate 2 V ariate 3 b s b s b s
For IV Set
VERBAL -.871 -.913 -.376 , .066 -.892 -.028 SCIENCE -.351 -.758 .729 ,.439 .479 .417 STRENGTH .000 -.452 -.770 -, .297 .920 .765 AESTHE .365 -.169 .822 .579 , .211 .334
PV .083 .003 .001
Rc2( 1 )=.202 Rc2(2)=.018 Rc2(3)=.005 Rc2(4)=.000
For DV Set
CONFUS 1.143 .677 .310 .619 -.511 - .042 UNCERT -.022 -.025 .547 .573 .288 .244 CLASIC -.874 -.255 .653 .799 -.164 -.052 EXTRNL .009 .278 -.108 .252 1.119 .804
PV .150 .353 .178 Rd .030 .006 .000 Rd .030 .036 .036
b : Standized canonical coefficients (weights) s : Structure coefficients PV : Proportion of variance explained Rd : Redundancy RcT : Total redundancy 133 The Second Question and Results
Question: Are the four sub-scales of CDS related to age, grade level (year of school), college area, educational and Job plan after completion of college, attitude regarding family and career. SES. parent’s educational background, enrichment educational experience, and location where completed most of education, area of home town, previous work experience, and grade point average ?
INDEPENDENT VARIABLE SET
Grade level (Year of School), Major field, Educational plan, Job plan, Attitude regarding family and career, SES, Enrichment educational experience, and Location completed most of education.
DEPENDENT VARIABLE SET (Four Factors of CDS)
CONFUS Total Score of the CDS item 7, 8, and 11. (Diffusion, which represents feelings of confusion, discouragement, and lack of experience or information about the making of career decisions)
UNCERT Total Score of the CDS item 12, 16, and 18. (Support, which represents uncertainty about how to proceed in making decisions, and the need for additional support for initial decisions)
CLASIC Total Score of the CDS item 4, 15, and 17. (Approach, which represents a classical approach-approach conflict • in which several possible careers are attractive)
EXTRNL Total Score of the CDS item 3,5,6 and 9 (External barrier, represents both external barriers to career choice and lack of interest in making a decision)
Due to the many dummy variables, one strategy for deciding which of the varialbes in the independent variable set to include in the canonical correlation was recommended by Dr. Warmbrod. The strategy was to regress (multiple regression) each of the dependent variables on the backgrand characteristics in 134 order to decide which of the independent variables are important (significant).
This involves the regression of 4 sub-scales of CDS with the selected background characteristics. Then, include in the canonical correlation analysis the sets of dummy variables that were demonstrated to be important by the multiple regression analysis.
Based on those steps, important background variables were identified as follows for each of the four sub-scales of CDS.
CONFUSE: Age, Major field, Educational plan, Job plan, SES
UNCERT: Grade level, Major field, Enrichment educational experience, Job plan and Location most education was completed.
CLASIC: Grade level, Educational Plan, Job plan and Attitude regarding family and career.
EXTERNAL: Grade level and Educational plan.
Therefore, the above variables were included as the independent variable set in the canonical correlation, except age, due to its relationship with grade.
Statistical Hypothesis: There are no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients(Rc2) are equal to zero.
Total Case (n=1232)
Following is the information for testing the hypothesis for all combined respondents.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .80; F = 2.48; p < .000 135 Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .87; F = 2.00; p < .000
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .97; F = 1.21;p =.055
Table 47
Test Hypothesis for Total Case
Roots Wilks L. F Significance Decision
1 TO 4 .79882 2.47738 <.001 Reject 2 TO 4 .87699 1.99002 <.001 Reject 3 TO 4 .94456 1.33733 .055 Fail to Reject 4 TO 4 .97553 1.20684 .221 Fail to Reject
As shown in the Table 47, primarily canonical roots 1 and 2 are statistically significant at an alpha level of .05. These significance tests can be used for deciding which canonical functions to interpret. In addition, statistics for determining the meaningfulness of canonical roots are based on the value of squared canonical correlation coefficients (Rc2) greater than .10, and the redundancy index. As shown in the Table 50, the values of Rc2 for roots 1 and 2, which are statistically significant, represent that both canonical root 1 and 2 explain less than 10 % (Rc2(l) =.089,Rc2(2) = .072) of the total variance in the dependent canonical variate (linear combination of dependent variable set) shared with the independent canonical variate (linear combination of the independent variable set). Moreover, the values of redundancy index demonstrate that canonical root 1 and 2 explain only 2% and 4% of the total variance in the 136 dependent variable set explained by the-independent variable set. Therefore, even though they are statistically significant, canonical roots 1 and 2 practically add little to the explanation of the variance. The significance of the hypothesis may be explained by the large sample size in this case(n = 12321.
Table 48
Rc and Sq of Rc for Total Case
Root No. Canon Cor. Sq. Cor
1 .299 .089 2 .267 .072 3 .178 .032 4 .156 .024 Table 49 Mean and Standard Deviation for DV and IV Sets for Total Case (n=1232) Means and Standard Deviation for DV Set Variables Mean Std Dev N Label CONFUS 7.13 2.62 1250 UNCERT 8.29 1.96 1268 CLASIC 7.67 2.30 1249 EXTRNL 7.93 2.56 1242 Means and Standard Deviation for IV Set(n=1275) Grade Level FRESH .33 .47 1275 SOPH .20 .40 1275 JR .25 .43 1275 Major Field HUMTY .27 .44 1275 SOCSCI .16 .36 1275 SCIENG .30 .46 1275 Enrichment Ed. Activities EXTYES .78 .41 1275 If "Yes".How manv 9 EXTED2 1.83 1.67 1275 Ed. Plan EPLAN1 .42 .49 1275 Pursue Graduate Studies EPLAN2 .09 .28 1275 Study priv voc inst EPLAN3 .47 .50 1275 Self-study EPLAN4 .07 .25 1275 No Further Educ plan Job Plan JPLAN1 .49 .50 1275 Full time emp JPLAN2 .14 .34 1275 Part time emp JPLAN3 .38 .49 1275 Self employment JPLAN4 .02 .14 1275 dont plan to work JPLAN5 .07 .26 1275 Marriage Attitude Regarding Career and Familv ATTYES .82 .39 1275 ATTNO .06 .24 1275 Social Economic Status SESF1 .02 .14 1275 Lowest SES SESF2 .01 .08 1275 SESF3 .04 .19 1275 SESF4 .01 .11 1275 SESF5 .05 .21 1275 SESF6 .28 .45 1275 SESF7 .23 .42 1275 Highest SES Location comoleted most of Ed. LTOWN .19 .39 1275 LURBAN .73 .44 1275 138 Table 50 Summary of Canonical Correlation Analysis for Total Case (n=1232) Canonical Canonical Canonical . Canonical Variables Variate 1 Variate 2 Variate 3 Variate 4 b s b S b s b S
For XV set FRESH -.6 6 1 -.5 2 0 .793 .357 -.119 -.197 .082 .022 SOPH -.0 6 4 .128 .419 -.072 .080 -.0 0 7 .213 .139 JR -.1 8 6 .123 .619 .175 .166 .215 .160 .087 HUMTY .258 .025 .2,91 .097 .385 .228 .282 .096 SOCSCI .269 .190 .287 -.024 .343 .177 .225 .005 SCIENG .368 .177 .441 .241 .243 .005 .158 .049 EXTYES -.3 8 1 -.179 .090 .029 -.139 .080 .124 -.161 EXTED2 .191 .086 .051 -.029 .328 .282 -.442 -.389 EPLAN1 -.2 4 3 -.4 3 8 -.335 -.211 .383 .270 .091 .280 EPLAN2 .129 .135 .119 .221 .080 .071 -.358 -.308 EPLAN3 .104 .134 -.128 .043 .102 .031 -.377 -.345 EPLAN4 .368 .414 -.051 .006 -.107 -.2 3 3 -.147 -.021 JPLAN1 -.105 .159 .375 .175 .065 .273 -.042 .016 JPLAN2 .038 .056 .450 .276 -.1 7 8 -.235 .184 .156 JPLAN3 -.055 -.1 9 0 .419 .227 -.055 -.0 9 4 -.282 -.3 2 1 JPLAN4 -.210 -.147 -.090 -.270 -.220 -.200 -.099 -.0 8 8 JPLAN5 -.0 0 4 .079 .034 .005 -.4 6 6 -.477 .083 .058 ATTYES -.155 -.065 .100 .108 -.0 4 1 -.071 -.498 -.353 ATTNO -.167 -.148 .029 -.047 .113 .127 -.233 .003 SESF1 -.0 3 0 .015 -.073 -.067 .183 .147 -.299 -.339 SESF2 -.067 -.075 -.180 -.222 -.059 -.025 -.027 -.0 1 3 SESF3 .264 .224 .002 -.027 -.071 -.115 .088 .126 SESF4 .230 .198 -.003 .028 -.302 -.343 -.0 7 7 -.042 SESF5 -.1 6 8 -.132 -.089 -.090 -.046 -.097 -.177 -.2 0 6 SESF6 .050 .044 .028 .110 .075 .038 .157 .128 SESF7 -.002 -.047 -.066 -.045 .233 .257 .087 .109 LTOWN .122 .044 .226 .031 .140 .145 -.087 .004 LURBAN .137 -.075 .256 .135 .050 -.0 6 6 -.114 -.0 7 8 PV 003 .002 . 001 .000
Rc2(1) = .089 Rc2(2) = .072 Rc2(3) = . 032 Rc2( 4 ) =.024
For DV Set
CONFUS .089 .329 .765 .647 -.567 -.680 844 .107 UNCERT -.717 -.777 -.316 -.249 -.641 -.573 038 -.077 CLASIC -.310 -.098 .602 .633 .280 -.245 -. 903 -.728 EXTRNL .644 .596 -.735 -.062 -.477 -.662 -. 566 -.451
PV .269 .221 .322 .188 Rd .024 .016 .012 .005 RD .024 .040 .052 .057 139
Replication 1: Case A University (n=229)
Statistical Hypothesis: There are no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients (Rc2) are equal to zero.
Following is the information for testing the hypothesis for Case A
University.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .50;F = 1.33; p =.017
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .66; F = 1.09;p =.289
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .80; F = .886; p =.697
Table 51
Testing Hypothesis for Case A University
Roots Wilks L. F Significance Decision
1 TO 4 .50050 1.33435 .017 Reject 2 TO 4 .66048 1.08920 .289 Fail to Reject 3 TO 4 .80319 .88638 .697 Fail to Reject 4 TO 4 .92113 .68501 .868 Fail to Reject
As shown in the Table 51, primarily canonical root 1 is statistically significant at an alpha level of .05. The value of Rc2 for root 1, which is statistically significant, indicates that for canonical root 1 24% (Rc2(l) =.24) of the total variance in the dependent canonical variate (linear combination of 140 dependent variable set) is shared with the independent canonical variate (linear combination of the independent variable set).
In order to explain the information on independent and dependent canonical variates (canonical roots), that the researcher interprets as meaningful, structure coefficients were used, which have values greater than .30.
As shown in the Table 54, dependent variate 1 carries information about
CONFUS(s=-.927), CLASIC (s=-.363), and EXTRNAL (s=-.337)in the dependent variable set. In other words, females with higher values on CONFUS,
CLASIC and EXTRNAL have lower scores on dependent variate 1; females with lower values on CONFUS, CLASIC and EXTERNAL have higher scores on dependent variate 1. However, independent variate 1 carries information about
FRESH (s=-.30), EPLAN1 (s=.302), JPLAN l(s= -.308), JPLAN 2 (s=-.307), and
JPLAN 4 (s = .313)in the independent variable set. In other words, females who are freshman, NOT pursing graduate school as an educational plan after graduation, who are full time/part time employment jobs plan after graduation have lower values on independent variate 1; females with the opposite characteristics of those mentioned above have higher values on independent variate 1.
Therefore, the nature of the relationship between the 4 sub-scales of CDS and selected variables can be explained as follows : females who were freshman, who had no plans to go to graduate school, who wished to have full time/part time employment after graduation typically had higher confusion, conflict, and 141 who felt greater external barriers in making a career decision: females who were not freshman, who wished to pursue graduate study, and who did not have a job plan after graduation had lower confusion, conflict and felt less external barriers in making career decision.
As shown in Table 52, the magnitude of this relationship between above variable sets is indicating by the Rc value of .49. This indicates that 24% (Rc2(l)) of variance is shared by the dependent and independent canonical variate 1.
Table 52
Rc and Sq of Rc for Case A University
Root No. Canon Cor. Sq. Cor
(Rc) (Rc2) 1 .492 .242 2 .422 .178 3 .358 .128 4 .281 .079 Mean and Stand Deviation for DV and IV Sets for Case A University (n=229)
Variables . Mean Std Dev N Label CONFUS 6.88 2.61 230 4-CDS UNCERT 8.09 2.03 234 CLASIC 7.34 2.24 229 EXTRNL 8.04 2.65 230 Means and Standard Deviation for IV set (=234) Grade Level FRESH .23 .42 234 SOPH .27 .44 234 JR .33 .47 234 Maior Field HUMTY .46 .50 234 SOCSCI .08 .27 234 SCIENG .32 .47 234 Enrichment Ed. Activities EXTYES .72 .45 234 If "ves".How manv? EXTED2 1.63 1.67 234 Ed. Plan EPLAN1 .38 .49 234 Pursue Graduate Studies EPLAN2 .10 .30 234 Study priv voc inst EPLAN3 .50 .50 234 Self-study EPLAN4 .08 .27 234 No Further Educ plan Job Plan JPLAN 1 .48 .50 234 Full time emp JPLAN2 .11 .31 234 Part time emp JPLAN3 .38 .49 234 Self employment JPLAN4 .03 .17 234 dont plan to work JPLAN5 .07 .26 234 Marriage Attitude reeardine Career and Familv ATTYES .79 .40' 234 ATTNO .06 .24 234 Social Economic Status SESF1 .04 .19 234 SESF2 .02 .13 234 SESF3 .05 .22 234 SESF4 .01 .11 234 SESF5 .07 .25 234 SESF6 .28 .45 234 SESF7 .21 .41 234 Location comoleted most of Ed. LTOWN .21 .40 234 LURBAN .66 .47 234 143
Table 54
Summary of Canonical Correlation Analysis for Case A University (n=229)
Canonical Canonical Canonical Canonical Variables Variate 1 Variate 2 Variate 3 Variate 4 bs bs bs bs
For IV s e t
FRESH -.6 2 8 -.288 -.394 -.166 .059 -.1 4 0 -.2 2 9 -.257 SOPH -.324 .032 -.366 -.109 .125 .098 .171 .133 JR -.445 .053 -.345 .037 .192 .100 .002 .036 HUMTY -.322 -.193 -.417 -.210 -.099 -.0 2 4 .490 -.0 9 4 SOCSCI .020 .212 -.272 -.052 -.161 -.052 .566 .294 SCIENG -.329 -.187 -.358 .038 -.027 -.025 .644 .180 EXTYES .078 .142 -.340 -.2 4 3 .193 .503 -.1 4 3 .108 EXTED2 -.097 .067 .228 .069 .306 .434 .284 .191 EPLAN1 .267 .302 -.484 -.3 3 0 .077 .093 .268 .136 EPLAN2 -.052 -.057 -.015 .111 .258 .262 .230 .170 EPLAN3 -.109 -.086 -.358 -.121 .063 .049 .163 -.057 EPLAN4 -.228 -.1 4 0 -.140 .061 -.088 -.187 •.071 .046 JPLAN1 -.554 -.308 -.033 -.0 2 6 .455 .194 -.289 -.030 JPLAN2 -.666 -.307 -.128 -.102 -.101 -.265 -.082 .137 JPLAN3 ' -.361 .073 .049 .119 .553 .373 -.1 5 3 -.1 0 9 JPLAN4 .023 .313 -.065 -.1 7 3 -.124 -.1 6 0 -.1 4 6 -.145 JPLAN5 -.2 0 8 -.1 3 1 .172 .221 -.190 -.2 3 8 .210 .155 ATTYES -.3 1 4 -.227 .318 .428 -.059 -.009 -.2 2 9 -.085 ATTNO -.144 .071 -.190 -.3 5 1 .136 .110 -.2 6 1 -.1 5 9 SESF1 .102 .135 -.062 -.062 .298 .268 .027 -.0 0 8 SESF2 .184 .217 .098 .067 .006 -.112 -.0 4 0 -.0 7 9 SESF3 .096 .134 -.238 -.291 -.166 -.323 .188 .121 SESF4 -.084 -.082 .216 .227 .157 .048 .335 .354 SESF5 .056 .069 -.047 .122 -.163 -.1 4 4 -.3 3 9 -.3 1 8 SESF6 -.134 -.140 -.036 .074 -.050 .063 .033 .237 SESF7 -.234 -.136 -.349 -.326 .076 .023 -.2 0 8 -.2 3 1 LTOWN -.183 -.117 -.159 -.134 -.164 -.2 8 6 .565 .205 LURBAN -.120 -.020 .005 .163 .087 .349 .487 .044 PV .007 .006 .006 .002
Rcz(l) = .242 Rcz( 1) = . 178 Rcz(l) = .128 Rcz(l) = .079
For DV Set CONFUS -1.070 -.927 -.340 .264 -.494 -.080 -.2 2 7 -.2 5 3 UNCERT .337 .219 .249 .237 -.431 -.214 -.841 -.922 CLASIC -.027 -.363 -.009 .339 1.160 .753 -.278 -.431 EXTRNL .227 -.337 1.108 .933 -.173 .030 .379 .126 PV .289 .278 .155 .279 RD .070 .050 .020 .022 m .070 .120 .140 .162 144 Replication 2: Case B University (n=498)
Statistical Hypothesis: There are no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients (Rc2) are equal to zero.
Following is the information for testing the hypotheses for Case B
University.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .67; F = 1.73;p =.000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .82;F = 1.16;p =.152
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .90; F = .973; p =.530
Table 55
Testing Hypothesis for Case B University
Roots Wilks L. F Sig. of F D ecision 1 TO 4 .67303 1.73487 <.001 R eject 2 TO 4 .82216 1.16753 .152 Fail to Reject 3 TO 4 .90006 .97308 .530 Fail to Reject 4 TO 4 .96503 .67985 .878 Fail to Reject
As shown in Table 55, primarily canonical root 1 is statistically significant at an alpha level of .05. The values of Rc2 for root 1, which is statistically significant, indicates that for canonical root 1 18% (Rc2(l) =.181) of the total variance in the dependent canonical variate (linear combination of dependent 145 variable set) is shared with the independent canonical variate (linear combination of the independent variable set). The values of Rc (.42) for root 1 also represent the magnitude of this relationship.
In order to explain the each of the pairs of independent and dependent canonical variates (canonical roots), that the researcher interprets as meaningful, structure coefficients were used, which have values greater than .30.
As shown in the Tabke 58, dependent variate 1 carries information about
CONFUS(s=.514), UNCERT(s=-.693), and EXTRNAL (s=.724)in the dependent variable set. In other words, females with higher values on CONFUS and EXTRNAL, and with lower values on UNCERT have higher scores on dependent variate 1; females with lower values on UNCERT and CONFUS, and with higher values on UNCERT have lower scores on dependent variate 1.
However, independent variate 1 carries information about FRESH (s=-.424),
EPLAN1 (s=-.386), EPLAN4 (s=.362), and SES4 (s=310) in the independent variable set. In other words, females who are freshman, pursing graduate school as an educational plan after graduation, plan to work, and who have the middle level of Social Economic Status have lower values on independent variate 1; females with the opposite characteristics of those mentioned above variables have higher values on independent variate 1.
Therefore, by looking at both sets of relationship, it can be concluded that females who were freshman, had plans to go to graduate school, had plans to get a job after graduation, and did not place in the middle level on SES typically had 146 lower confusion . and felt lower external barriers in making a career decision, but they had higher scores on uncertainty: females who were not freshman, wished to pursue graduate study, who did not want to further their education, and who fell in the middle level on SES typically had higher confusion and external barriers. but they less uncertainty in making a career decision.
Table 56
Rc and Sq of Rc for Case B University
Root No. Canon Cor. Sq. Cor
1 .426 .181 2 .294 .087 3 .259 .067 4 .187 .035 147
Table 57
Mean and Stand Deviation for DV and IV Sets for Case B University (n=498)
Means and Standard Deviation for DV sets (n=498)
Variable Mean Std Dev N Label CONFUS 7.20 2.62 503 UNCERT 8.44 1.90 505 CLASIC 7.97 2.18 503 EXTRNL 7.72 2.56 502
Means and Standard D eviation fo r IV s e ts (n=507)
V a ria b le Mean Std Dev N Label Grade Level FRESH .50 .50 507 SOPH .10 .30 507 JR .18 .39 507 Maior Field HUMTY .19 .40 507 SOCSCI .18 .38 507 SCIENG .39 .49 507 Enrichment Ed. Activities EXTYES .79 .41 507 I f "ves". How manv EXTED2 1.90 1.64 507 Ed. Plan EPLAN1 .44 .50 507 Pursue Graduate Studies EPLAN2 .09 .28 507 Study priv voc inst EPLAN3 .46 .50 507 S e lf-stu d y EPLAN4 .06 .23 507 No Further Educ plan Job Plan JPLAN1 .47 .50 507 Full time emp JPLAN2 .13 .33 507 Part time emp JPLAN3 .40 .49 507 Self employment JPLAN4 .01 .12 507 dont plan to work JPLAN5 .06 .24 507 Marriage Attitude reaardina Career and Family ATTYES .81 .40 507 ATTNO .08 .27 507 Social Economic Status SESF1 .01 .12 507 Lowest SES SESF2 .00 .04 507 SESF3 .04 .19 507 SESF4 .00 .06 507 SESF5 .04 .20 507 SESF6 .28 .45 507 SESF7 .26 .44 507 Highest SES Location corrmleted most of ED. LTOWN .19 .39 507 LURBAN .77 .42 507 148 Table 58
Summary of Canonical Correlation Analysis for Case B University (n=498)
Canonical Canonical Canonical Canonical Variables Variate I Variate 2 Variate 3 Variate 4 bsbsb s b s
For IV set FRESH -.683 -.424 -.658 -.200 .138 .079 -.077 .094 SOPH .043 .164 -.527 -.125 .297 .134 -.1 3 4 -.0 8 0 JR -.475 -.099 -.447 -.127 .053 .049 -.272 -.2 5 8 HUMTY .407 .103 -.500 -.267 -.107 -.089 -.261 -.1 1 9 SOCSCI .118 .206 -.198 .209 -.179 -.1 2 9 -.272 -.020 SCIENG .485 .236 -.587 -.209 .069 .078 -.267 -.1 6 1 EXTYES -.275 -.040 -.035 .023 .424 .170 -.016 .104 EXTED2 .136 .122 -.101 .028 -.421 -.167 .128 .108 EPLAN1 -.153 -.386 .509 .253 -.123 .044 -.323 -.3 2 0 EPLAN2 .151 .199 -.118 -.244 -.176 -.141 -.170 -.132 EPLAN3 .207 .256 .296 -.012 -.362 -.307 .218 .404 EPLAN4 .275 .362 .313 .144 .012 .263 -.213 -.207 JPLAN1 -.111 .221 -.411 -.142 .173 .032 -.0 1 3 -.2 3 9 JPLAN2 -.214 -.124 -.412 -.232 .011 -.0 4 0 -.039 .004 JPLAN3 -.022 -.198 -.453 -.155 .063 -.016 .262 .359 JPLAN4 -.132 -.063 -.029 .081 .162 .103 .071 .009 JPLAN5 .050 .054 .151 .210 .451 .406 -.0 3 8 -.101 ATTYES -.050 .054 -.060 -.100 -.419 -.325 • -.045 .074 ATTNO -.207 -.210 .149 .165 -.205 .010 -.187 -.1 5 3 SESF1 -.061 .024 -.015 .011 -.154 -.153 .180 .251 SESF2 -.026 .014 .063 .081 .154 .061 .493 .545 SESF3 .248 .272 -.065 -.135 .107 .096 .099 .089 SESF4 .274 .310 .146 .116 .277 .332 -.024 -.0 2 4 SESF5 -.089 -.033 .013 -.050 -.211 -.165 -.119 -.0 5 5 SESF6 .011 .003 .004 -.022 -.111 -.027 -.2 4 8 -.1 3 7 SESF7 -.012 -.068 .184 .125 -.282 -.258 -.297 -.2 6 8 LTOWN .083 .005 .136 .007 -.041 -.2 6 5 .121 -.1 2 9 LURBAN .116 -.052 .103 -.033 .152 .250 .241 .093 PV .007 .002 .002 .001
Rc2(1) = .181 Rc2(2) =.087 Rc2(3)=.067 Rc2(4) = .035
For DV Set
CONFUS .177 .514 -.579 -.552 1.009 .643 -.459 .131 UNCERT -.593 -.693 .270 .175 .603 .461 .501 .526 CLASIC -.2 8 1 .003 -.841 -.771 -.582 -.163 .419 .615 EXTRNL .690 .724 .673 -.023 -.137 .165 .804 .670 PV .317 .233 .170 .280 RD .057 .020 .011 .009 RD .057 .077 .088 .097 149 Replication 3: Case C University (n=251)
Statistical Hypothesis: There is no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients (Rc2) are equal to zero.
Following is the information for testing the hypotheses for Case C
University.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .47;F = 1.63;p =.000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .63;F = 1.35;p =.027
Ho : Rc(3)2 = Rc(4)2 = 0 . Wilks lambda = .81;F = .944;p =.586
Table 59
Test Hypothesis for Case C University
Roots Wilks L. F Sig. of F Decision
1 TO 4 .46987 1.63150 .000 Reject 2 TO 4 .63152 1.35119 .027 Reject 3 TO 4 .80992 .94488 .586 Fail to reject 4 TO 4 .92677 .70169 .853 Fail to reject 150
As shown in the Table 59, primarily canonical root 1 and 2 are statistically significant at an alpha level of .05. The values of Rc2 for roots 1 and 2, which are statistically significant, demonstrate that for canonical roots 1 and 2 26% and 22%
(Rc2(1) = .2 5 6 ,R c 2 (2) = .220) of the total variance in the dependent canonical variate (linear combination of dependent variable set) is shared with the independent canonical variate (linear combination of the independent variable set). In addition the magnitude of relationship is indicating by the value of Rc
(Rc(l) = .50, and Rc(2) = .47).
In order to explain the each of the pairs of independent and dependent canonical variate (canonical roots), that researcher interpret as meaningful, structure coefficients were used, which have greater than .30.
As shown in the Table 62, dependent variate 1 carries information about
CONFUS (s= .855), UNCERT (s=-.560), CLASIC (s= .509) and EXTRNAL
(s=.473)in the dependent variable set. In other words, females with higher score on CONFUS CLASIC and EXTRNAL, and lower scores on UNCERT have higher scores on dependent variate 1; females with lower scores on CONFUS,
CLASIC and EXTERNAL, and higher scores on UNCERT have lower scores on dependent variate 1. However, independent variate 1 carries information about
SOCSCI (s=.389),EPLANl (s=-.400),EPLAN4 (s=.302), JPLAN2 (s=.367)and
JPLAN4 (s=-.312)in the independent variable set. In other words, females with enrolled in Social Science college, have no further education plan, have part time employment or any job plan after graduation, have higher scores on independent 151 variate 1; females with the opposite characteristics of those mentioned above variables have lower scores on independent variate 1.
Therefore, by looking both side of relationship for canonical root 1, the nature of the relationship can be concluded that females who were pursuing graduate school after completion of college, and had no job plan, and were not enrolled in the Social Science College typically had lower scores on confusion. conflict and external barriers, but higher scores on uncertainty: females who were enrolled in the Social science college, had no plan for further education, but had a plan for work after graduation, typically had higher scores on confusion, conflict and external barriers, but lower scores on uncertainty.
As shown in the summary Table 62, dependent variate 2 carries information about CONFUS (s=.375), UNCERT (s=.610), CLASIC (s=.571) and EXTRNAL (s=.745) in the dependent variable set. In other words, females with higher score on CONFUS, UNCERT, CLASIC and EXTRNAL have higher scores on dependent variate 1; females with lower score on CONFUS, UNCERT,
CLASIC and EXTRNAL have lower scores on dependent variate 1. However, independent variate 2 carries information about JR (s=-.299), HUMTY (s=-.366),
JPLAN1 (s=-.439), JPLAN4 (s=.477), and JPLAN5 (s=.382) in the independent variable set. In other words, females with Junior school year, enrolled in
Humanities College, having full-time employment or any job plan after graduation, and do not want to marriage after graduation have lower scores on 152 independent variate 1; females with versus above mentioned variables have higher scores on independent variate 1.
Therefore, by looking at both sides of relationship for canonical root 2, the nature of this relationship can be concluded that females who were enrolled in Humanities College and were juniors, who had full-time employment or any job plan after graduation rather than marriage typically had lower scores on overall four sub-scales on CDS: females who were not enrolled in the Humanities College and were not juniors, who had no plan to work and had marriage plans typically had higher scores on confusion, uncertainty, conflict, and external barriers.
Table 60
Rc and Sq of Rc for Case C University
Root No. Canon Cor. Sq. Cor
1 .506 .256 2 .469 .220 3 .355 .126 4 .271 .073 153 Table 61
Mean and Standard Deviation for DV and IV Sets for Case C University (n=251)
Means and Standard D eviation for DV set (n=251)
V ariab le Mean Std Dev N Label
CONFUS 7.02 2.67 256 4-CDS UNCERT 8.25 1.98 264 CLASIC 7.49 2.47 256 EXTRNL 8.46 2.60 254
Means and Standard Deviation for XV set (n=267)
Variable Mean Std Dev N Label Grade Level FRESH .19 .40 267 SOPH .32 .47 267 JR .26 .44 267 Maior Field HUMTY .31 .47 267 SOCSCX .16 .37 267 SCIENG .16 .37 267 Enrichment Ed. Activities EXTYES .72 .45 267 I f "ves". How manv ?
EXTED2 1.48 1.50 267 Ed Plan EPLAN1 .39 .49 267 Pursue Graduate Studies EPLAN2 .09 .29 267 Study p riv voc in s t EPLAN3 .46 .50 267 S e lf-stu d y EPLAN4 .08 .27 267 No Further Educ plan Job Plan JPLAN1 .47 .50 267 F u ll tim e emp JPLAN2 .16 .37 267 Part time emp JPLAN3 .35 . .48 267 S e lf employment JPLAN4 .03 .17 267 dont plan to work JPLAN5 .10 .30 267 M arriage Attitude Reaardina Career■ and Familv ATTYES .83 .38 267 ATTNO .06 .24 267 Social Economic Status SESF1 .03 .16 267 Lowest SES SESF2 .01 .09 267 SESF3 .06 .23 267 SESF4 .02 .15 267 SESF5 .07 .26 267 SESF6 .30 .46 267 SESF7 .24 .43 267 H ighest SES Location completed most of ED. LTOWN .23 .42 267 LURBAN .65 .48 267 154 Table 62
Summary of Canonical Correlation Analysis for Case C University (n=251)
Canonical Canonical Canonical Canonical Variables Variate X Variate 2 Variate 3 Variate 4 b s b s b s b s
For XV set
FRESH -.041 .016 .160 .257 .406 -.055 -.5 6 7 -.2 3 1 SOPH .044 -.114 .046 -.045 .535 .133 -.4 8 6 -.0 8 1 JR .208 .227 -.274 -.299 .523 .193 -.514 -.148 HUMTY .074 -.139 -.407 -.366 .387 .356 .044 .005 SOCSCI .390 .389 -.177 -.055 .067 -.095 .152 -.050 SCIENG .005 -.078 -.205 -.125 .014 -.111 .189 .181 EXTYES -.307 -.243 .190 .179 -.227 .120 .067 .116 EXTED2 .165 -.109 -.029 .113 .451 .318 .151 .117 EPLAN1 -.449 -.400 -.116 .036 .270 .081 .215 -.1 2 7 EPLAN2 -.054 -.029 .040 .083 .382 .404 .268 .098 EPLAN3 -.163 .034 .018 -.112 .181 .005 .295 .014 EPLAN4 .224 .302 .110 .186 .219 .096 .491 .414 JPLAN1 -.219 -.246 -.259 -.439 .062 .117 -.244 .114 JPLAN2 .363 .367 -.060 .039 -.085 -.028 -.405 -.3 3 9 JPLAN3 .140 .220 .100 .267 -.050 .116 -.304 -.108 JPLAN4 -.321 -.312 .483 .477 .026 -.008 -.195 -.057 JPLAN5 .092 .168 .401 .382 .314 .319 -.297 -.113 ATTYES -.196 -.221 .145 .064 .039 -.024 -.154 -.271 ATTNO .019 .185 .290 .259 .198 .114 .309 .269 SESF1 .021 -.031 .084 .086 .084 . 121 .124 .059 SESF2 -.087 -.138 -.060 -.138 -.198 -.1 5 6 -.3 7 8 -.2 5 4 SESF3 .288 .073 .034 .029 -.282 -.285 .166 .149 SESF4 .202 .079 .029 .167 -.1 7 4 -.0 4 7 -.1 7 2 -.2 3 6 SESF5 -.076 -.146 .030 .042 .090 .108 -.0 7 1 -.0 0 9 SESF6 .306 .104 .016 .055 -.049 .064 -.004 .021 SESF7 .165 .044 .170 .023 -.353 -.245 -.042 .006 LTOWN .372 .113 .088 -.002 .313 .137 -.0 1 9 .110 LURBAN .296 -.009 .155 .088 .227 .065 -.0 8 7 -.1 8 1 PV .001 .009 .004 .002
Rc2(l)=.256 Rc2(2)=.220 Rc2(3) = .126 R c2(4) = .073
For DV Set CONFUS .798 .855 -.194 .375 -.404 -.220 -1.064 -.2 8 4 UNCERT -.499 -.560 .583 .610 -.088 .065 -.665 -.5 5 6 CLASIC .236 .509 .260 .571 1.128 .643 .179 .034 EXTRNL -.173 .473 .764 .745 -.5 9 0 -.325 .942 .341 PV .382 .348 .143 .127 RD .098 .077 .018 .009 ID .098 .175 .193 .202 155
Replication 4: Case D University In=2511
Statistical Hypothesis: There are no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients (Rc2) are equal to zero.
Following is the information for testing the hypotheses for Case D
University.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .56;F = 1.18;p =.112
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .70;F = 1.03;p =.393
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .81;F = .938;p =.581
Table 63
Testing Hypothesis for Case D University
Roots Wilks L. F Sig. of F Decision
1 TO 4 .57554 1.17830 . 112 Fail to Reject 2 TO 4 .70127 1.03864 . 393 Fail to Reject 3 TO 4 .81323 .93822 .599 Fail to Reject 4 TO 4 .90748 .91754 .581 Fail to Reject
As shown in the Table 63, the statistical hypothesis was not rejected. Thus the conclusion is that the 4 sub-scales of CDS and selected variables have no relationship for this case. 156 Table 64
Mean and Standard Deviation for DV and IV Sets for Case D University (n=254)
Means and Standard Deviation for DV set (n=254)
Variable Mean Std Dev N Label CONFUS 7.33 2.55 261 UNCERT 8.23 1.97 265 CLASIC 7.56 2.34 261 EXTRNL 7.68 2.35 256
Means and Standard Deviation for DV set (n=267) Variable Mean Std Dev N Label Grade Level FRESH .25 .43 267 SOPH .21 .41 267 JR .28 .45 267 Manor F ie ld HUMTY .21 .41 267 SOCSCI .19 .39 267 SCIENG .25 .44 267 Enrichment Ed. Activities EXTYES .90 .31 267 I f "yes". How manv 9 EXTED2 2.20 1.82 267 Ed. Plan EPLAN1 .44 .50 267 Pursue G raduate S tu d ie s EPLAN2 .06 .24 267 Study p riv voc in s t EPLAN3 .47 .50 267 S e lf-stu d y EPLAN4 .06 .24 267 No Further Educ plan Job Plan JPLAN1 .57 .50 267 Full time emp JPLAN2 .15 .35 267 Part time emp JPLAN3 .39 .49 267 Self employment JPLAN4 .01 .11 267 dont plan to work JPLAN5 .07 .26 267 M arriage Attitude reoardina Career and Family ATTYES .84 .37 267 ATTNO .04 .20 267 Social Economici S tatu s SESF1 .01 .11 267 Lowest SES SESF2 .00 .06 267 SESF3 .01 .11 267 SESF4 .01 .12 267 SESF5 .01 .11 267 SESF6 .23 .42 267 SESF7 .19 .39 267 Highest SES Location completed most of: Ed. LTOWN .13 .34 267 LURBAN .82 .39 267 157
Table 65
Summary of Canonical Correlation Analysis for Case D University (n=254)
Canonical Canonical Canonical Canonical Variables Variate I Variate 2 Variate 3 Variate 4 b s b s b s b s
For IV set
FRESH -.617 -.450 .156 .114 .444 -.087 .111 .077 SOPH -.0 9 8 -.016 -.019 .081 .510 .154 -.0 0 6 -.245 JR -.058 .128 .031 -.139 .828 .431 .540 .166 HUMTY .055 -.162 -.256 -.107 -.296 -.0 4 1 -.2 7 8 -.0 9 1 SOCSCI .193 .140 -.129 .073 -.150 -.3 2 4 .171 .225 SCXENG .067 -.019 -.186 -.145 .090 .233 -.032 .169 EXTYES -.266 -.123 -.179 -.094 -.131 -.063 -.192 -.051 EXTED2 .354 .408 .019 -.047 .138 -.057 -.119 .063 EPLAN1 -.158 -.193 -.203 .037 -.139 -.057 -.199 -.272 EPLAN2 -.170 -.165 .096 .058 .248 .343 -.110 -.0 3 8 EPLAN3 -.148 -.108 -.455 -.283 -.222 -.1 0 3 .024 .154 EPLAN4 .254 .293 -.347 -.176 -.122 .072 -.040 .053 JPLAN1 .166 .104 -.181 -.367 .233 .050 .436 .411 JPLAN2 .171 -.039 -.241 -.090 .241 .220 -.092 -.211 JPLAN3 -.036 -.178 .321 .371 .184 .051 .371 .132 JPLAN4 .057 .096 -.083 -.059 .410 .373 .006 -.129 JPLAN5 -.093 -.029 .140 .066 .191 .120 -.117 -.1 7 3 ATTYES -.373 -.394 -.070 -.057 -.135 -.129 .639 .537 ATTNO .066 .178 .123 .098 -.276 -.138 .191 -.173 SESF1 .072 .161 .016 .019 .121 .094 .119 .038 SESF2 -.088 -.009 .019 -.064 -.135 -.112 .270 .202 SESF3 -.103 -.080 -.215 -.314 .076 .099 -.016 .110 SESF4 -.056 -.091 -.111 -.102 .034 .086 .140 .072 SESF5 -.344 -.347 -.333 -.368 -.119 -.079 -.144 -.157 SESF6 -.089 -.168 .504 .476 -.106 .043 -.109 -.022 SESF7 .203 .189 .017 .004 -.286 -.266 .138 .132 LTOWN .138 -.1 8 4 .382 .026 .100 -.101 .324 -.0 0 7 LURBAN .182 .155 .479 .108 .071 .025 .550 .172
PV .007 .005 .003 .003
Rc2( 1) = . 179 Rc2(2 ) = . 138 Rc2(3) = .104 Rc2(4) = .093
For DV se t CONFUS -.580 -.531 .585 -.161 .875 .818 -.453 -.1 5 6 UNCERT -.605 -.609 -.188 -.198 -.536 -.550 -.559 -.536 CLASIC -.429 -.559 -.585 -.574 -.217 .285 .925 .525 EXTRNL .569 .147 -.893 -.807 .111 .465 -.431 -.333
PV .247 .261 .317 .175 RD .044 .036 .033 .016 RD .044 .080 .113 .129 158 The Third Question and Results
Question: Are the four sub-scales of TSOSS related to age, grade level college area, educational and job plan after completion of college, attitude regarding family and career, socio economic status fSESl. parent’s educational background, additional enrichment educational activity, location where most education was completed, home town, previous work experience, and grade point average ('GPA1 ?
INDEPENDENT VARIABLE SET
College area, additional enrichment educational activities, educational plan, attitude regarding family and career, socio economic status (SES), mother’s educational background, grade point average (GPA), and location where most education was completed.
DEPENDENT VARIABLE SET (Four Factors of TSOSS)
VERBAL Verbal and Interpersonal (Total score of the TSOSS items 1, 5, 9, 13, 17,21, 25, 29, 33, 37, 41, 45, 49, 53, and 57.)
SCIENCE Quantitative, Scientific and Business (Total score of the TSOSS items 2, 6, 10, 14, 18, 22, 26, 30, 34, 38, 42, 46, 50, 54, and 58)
STRENGTH Physical strength and Agility (Total score of the TSOSS items 3, 7, 11, 15, 19, 23, 27, 31, 35, 39, 43, 47, 51, 55, and 59)
AESTHE Aesthetic skills (Total score of the TSOSS items 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44,48, 52, 56, and 60)
Using the same technical strategy as used in Question 2 to identify the significant variables to include in the independent variable sets for the canonical correlation, the following variables were identified as meaningful. 159 VERBAL: Age, additional enrichment educational activities, educational plan, attitude regarding career and family, socio economic status (SES), father’s educational background, grade point average (GPA) and mother’s educational background.
SCIENCE: College area, additional enrichment educational activities, educational plan, socio economic status (SES), hometown, location most education was completed and grade point average (GPA)
STRENGTH: College area and attitude regarding family and career
AESTHETIC: College area, additional enrichment educational activities, educational plan, mother’s educational background and Hometown.
Therefore, such variables as college area, additional enrichment educational activities, educational plan, attitude regarding family and career, SES, mother’s educational background, grade point average (GPA), hometown and location most education was completed were included as the independent variable set in the canonical correlation.
Statistical Hypothesis: There are no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients(Rc2) are equal to zero.
Total Case (n=1168)
Following is the information for testing the hypotheses for all the respondents.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .54; F = 6.24; p < .000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .73; F = 4.24; p < .000 160 Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .88; F = 2.45; p < .000
Table 66
Testing Hypothesis for Total Case
Roots Wilks L. F Sig. of F Decision
1 TO 4 .54278 6.24543 .000 Reject 2 TO 4 .73459 4.23969 .000 Reject 3 TO 4 .88907 2.45678 .000 Reject 4 TO 4 .96306 1.61532 .025 Reject
As shown in the Table 66, all canonical roots 1 through 4 are statistically significant at an alpha level of .05. These significance tests can be used for deciding which canonical functions to interpret as meaningful. However, practically speaking, canonical roots 1 and 2 are meaningful to interpret due to a greater than .lO ofRc2. (Rc2(l) = .261 and Rc2(2) = .174)
Structure coefficients which have values greater values than .30 were used to explain the independent and dependent canonical variates (canonical roots) that the researcher interpreted as meaningful. As can be seen in the Table 69, dependent variate 1 carries information about SCIENCE (s=-.65) and AESTHE
(s=.42) in the dependent variable set. In other words, females with higher scores on AESTHE and lower scores on SCIENCE have higher scores on dependent variate 1; females with lower scores on AESTHE and higher scores on SCIENCE 161 have lower scores on dependent variate 1. However, independent variate 1 carries information about SCIENG(s=-.759) only in the independent variable set.
In other words, females who enrolled in the Natural Science and Engineering
College have lower values on independent variate 1; females who not enrolled in the Natural Science and Engineering College have higher values on independent variate 1.
Therefore, the nature of the relationship between the 4 sub-scales of
TSOSS and the selected variable sets can be described as follows : females who were enrolled in the Natural Science and Engineering College typically had higher scores on science related skills, but lower scores on aesthetic skills on the TSOSS: females who were not enrolled in the Natural Science and Engineering College typically had lower scores on science related skills, but higher scores on aesthetic skills on the TSOSS. As can be seen in the Table 67, the magnitude of this relationship between the above variable sets is indicating by the value of Rc
(Rc(l) = .51). In addition the square of Rc (Rc2) for root 1 explains 26% of the variance shared by the dependent and independent canonical variates.
In addition, dependent variate 2 carries information about SCIENCE (s=-
.31), STRENGTH (s=-.35),and AESTHE (s=-.69) in the dependent variable set.
In other words, females with higher values on SCIENCE, STRENGTH and
AESTHE have lower scores on dependent variate 2; females with lower values on
SCIENCE, STRENGTH, and AESTHE have higher scores on dependent variate
2. However, independent variate 2 carries information about HUMTY(s= .78) 162 SCIENG(s=-.30), and MLH (s=.43) in the independent variable set. In other
words, females enrolled in the Humanities, not enrolled in the Natural Science and Engineering College, and females whose mothers had less than a high school education, have higher scores on independent variate 1; females with the opposite characteristics to those mentioned above have lower scores on independent variate 1.
Therefore, the nature of the relationship between the 4 sub-scales of
TSOSS and the selected variable sets can be described as follows: females who were enrolled in the Natural Science and Engineering College, not enrolled in the
Humanities College, and females whose mothers had a high school education typically had higher scores on science, physical strength, and aesthetic related skills on the TSOSS: females who were enrolled in the Humanities College and not enrolled in the Natural Science and Engineering College, and females whose mothers had less than a high school education typically had lower scores on science, physical strength, and aesthetic related skills on the TSOSS.
As can be seen the Table 67, the magnitude of this relationship between the above variable sets is indicating by the value of Rc (Rc(2) = .42). In addition the square of Rc (Rc2) for root 2 explains 20% of the variance shared by the dependent and independent canonical variate. Table 67
Rc and Sq of Rc for Total Case
Root No. Canon Cor. Sq. Cor (Rc) (Rc)
1 .511 .261 2 .417 .174 3 .277 .077 4 .192 .037 Table 68
Mean and Standard Deviation for DV and IV Sets for Total Case (n = 1168)
Means and Standard Deviation for DV set(n=1168)
V ariab le Mean Std Dev N Label
VERBAL 55.32 8.30 1237 SCIENCE 51.18 9.66 1216 STRENGTH 50.29 10.32 1217 AESTHE 51.94 10.53 1226
Means and Standard Deviation for IV i3et(n=1275) V ariab le Mean Std Dev N Label
Maior Field HUMTY .27 .44 1275 SOCSCI .16 .36 1275 SCIENG .30 .46 1275 Enrichment Ed. Activities EXTYES .78 .41 1275 I f "Yes", How manv ? EXTED2 1.83 1.67 1275 Ed Plan EPLAN1 .42 .49 1275 Pursue G raduate S tu d ies EPLAN2 .09 .28 1275 Study p riv voc in s t EPLAN3 .47 .50 1275 S e lf-stu d y EPLAN4 .07 .25 1275 No F u rth er Educ p lan Attitude Reaardina Career and Familv ATTYES .82 .39 1275 ATTNO .06 .24 1275 Social Economic Status SESF1 .02 .14 1275 Lowest SES SESF2 .01 .08 1275 SESF3 .04 .19 1275 SESF4 .01 .11 1275 SESF5 .05 .21 1275 SESF6 .28 .45 1275 SESF7 .23 .42 1275 H ighest SES M other' s Ed Background MLH .20 .40 1275 MHS .46 .50 1275 M2YR .02 .13 1275 MCOLL4 .26 .44 1275 MADVDEG .02 .13 1275 Location completed most of Ed. LTOWN .19 .39 1275 LURBAN .73 .44 1275 Hometown HTOWN .19 .39 1275 HURBAN .65 .48 1275 Grade Point Averaae GPAA .13 .34 1275 GPAB .71 .45 1275 GPAC .10 .30 1275 165 Table 69
Summary of Canonical Correlation Analysis for Total Case (n= 1168)
Canonical Canonical Canonical Canonical Variables Variate 1 Variate 2 Variate 3 Variate 4 b s b s b s b s
For IV set
HUMTY -.353 .242 .853 .779 .089 .079 -.345 -.2 1 7 SOCSCI -.6 2 7 -.239 .326 .071 .006 .063 -.125 .000 SCIENG -1.077 -.759 .129 -.298 .014 -.1 0 6 -.022, .132 EXTYES -.010 .054 -.082 -.136 .010 .283 -.2 8 1 -.0 6 9 EXTED2 .008 -.002 .006 -.066 .406 .452 .378 .185 EPLAN1 -.062 .008 .031 -.052 .304 .345 -.1 4 5 -.1 7 3 EPLAN2 .115 .143 -.060 -.052 .040 .016 .312 .277 EPLAN3 -.0 5 6 -.057 .058 .126 .113 .047 .026 -.0 0 6 EPLAN4 -.0 5 3 -.0 7 4 .112 .107 -.318 -.4 1 9 .022 .030 ATTYES .093 .063 .168 .146 .437 .370 .225 -.1 4 5 ATTNO .065 .044 .017 -.097 .210 -.016 ■.617 .483 SESF1 .003 -.0 2 5 .068 .151 .178 .150 .037 .101 SESF2 -.0 4 4 -.0 7 0 .105 .216 .049 .045 .125 .207 SESF3 -.0 3 8 -.0 7 1 -.024 .178 -.076 -.180 .145 .237 SESF4 .107 .092 .042 .077 -.2 8 6 -.326 .133 .167 SESF5 .038 -.009 -.002 .067 .056 .037 .020 .075 SESF6 -.0 1 8 -.032 -.108 -.148 -.062 -.054 -.206 -.138 SESF7 .027 -.027 .068 .077 .083 .118 -.209 -.189 MLH -.113 -.142 .433 .433 -.259 -.1 8 3 .330 .202 MHS -.078 .027 .313 -.043 -.169 .068 .366 .000 M2YR -.145 -.121 -.050 -.175 -.078 -.039 .057 .019 MCOLL4 -.127 .081 .194 -.230 -.129 .191 .192 -.0 7 6 MADVDEG .012 .094 .062 -.057 -.213 -.0 7 6 -.2 1 5 -.2 6 5 LTOWN .115 -.128 -.233 .024 -.0 4 1 -.0 3 8 -.6 7 9 -.1 3 9 LURBAN .220 .159 -.106 -.170 -.115 .136 -.478 -.039 HTOWN -.132 -.117 -.150 .000 .141 -.0 1 5 .047 -.0 3 4 HURBAN -.1 3 4 .101 -.314 -.269 .297 .246 -.0 7 1 -.110 GPAA -.115 -.152 .113 .069 .616 .381 .331 -.0 6 1 GPAB -.032 .028 .201 -.022 .388 -.124 .461 -.024 GPAC .093 .110 .202 .042 .270 -.0 7 3 .486 .221 PV .008 .008 .003 .001
Rc2 (1) =. 261 R c2(2) = . 174 Rc2(3) = .077 R c2(4) = . 037
For DV set
VERBAL .423 .041 1.013 .135 .848 .911 .188 .386 SCIENCE -1.137 -.651 -.557 -.312 .194 .638 -.1 4 1 .268 STRENGTH .074 .084 -.131 -.346 -.456 .167 1.115 .919 AESTHE .564 .419 -.939 -.685 .309 .581 -.461 .129 PV .152 .176 .401 .270 RD .040 .031 .030 .010 RD .040 .072 .102 .112 166 Replication 1; Case A University (n=208)
Statistical Hypothesis: There are no relationships between the dependent (criterion! and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients(Rc2) are equal to zero.
Following is the information for testing the hypothesis for Case A
University.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .32; F = 1.90; p < .000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0 . Wilks’ lambda = .53; F = 1.40; p = .014
Ho : Rc(3)2 =Rc(4)2 = 0 . Wilks lambda = .73; F = 1.04; p = .390
Table 70
Testing Hypothesis for Case A University
Roots Wilks L. F Sig. of F Decision
1 TO 4 .32281 1.90364 .000 Reject 2 TO 4 .53461 1.40353 .014 Reject 3 TO 4 .73470 1.04760 .390 Fail to Reject 4 TO 4 .89896 .73681 .824 Fail to Reject 167 As can be seen in the table 70, canonical root 1 and 2 are statistically significant at an alpha level of .05. These significance tests can be used for deciding which canonical functions to interpret. Structure coefficients which have the value greater than .30 were used in order to explain each of the pairs of independent and dependent canonical variate (canonical roots), that the researcher interprets as meaningful.
As shown in the summary Table 73, dependent variate 1 carries information primarily about SCIENCE (s=-.40) and AESTHE (s=,781) in the dependent variable set. In other words, females with higher values on science and lower values on aesthetic related skills have lower scores on dependent variate 1; females with lower values on science and higher values on aesthetic related skills have higher scores on dependent variate 1. However, independent variate 1 primarily carries information about SCIENG (s=-.55),and MLH (-.41) in the independent variable set. In other words, females enrolled in the Natural Science and Engineering, whose mothers had less than a high school education have lower scores on independent variate 1; females with opposite characteristics to those mentioned in the above variables have higher scores on independent variate 1.
Therefore, the natural relationship between the 4 sub-scales of TSOSS and the selected variable sets on canonical root 1 can be described as follows : females who were enrolled in the Natural Science and Engineering College, and females whose mothers had less than a high school education had higher scores on science, but lower on Aesthetic related skills on the TSOSS:females who were not 168 enrolled in the Natural Science and Engineering College, and femalse whose mothers had more than a high school education had lower scores on science, hut higher on aesthetic related skills on the TSOSS. As can be seen in the
Table 71, the magnitude of this relationship between the above variable sets is indicating by Rc (Rc(l) = .63). In addition, the value of Rc2 explains 40% of the total variance shared by the dependent and independent canonical variates.
In addition, dependent variate 2 carries information about VERBAL (s=-
.30), SCIENCE (s=.47),and AESTHE (s=.45)inthe dependent variable set. In other words, females with higher values on SCIENCE, AESTHE and lower values on VERBAL have higher scores on dependent variate 2; females with higher values on VERBAL, and lower values on SCIENCE and AESTHE have lower scores on dependent variate 2. However, independent variate 2 carries information about HUMTY (s=-.64) SCIENG (s=.53),MLH (s=-.37),and
HURBAN (s=.30)in the independent variable set. In other words, females enrolled in the Humanities and not enrolled in the Natural Science and
Engineering College, whose mothers had less than a high school education , and females whose home town is not in urban area have lower scores on independent variate 2; females with opposite characteristics to those variables mentioned variables have higher scores on independent variate 2.
Therefore, the relationship between the 4 sub-scales of TSOSS and the selected variable sets on canonical variate 2 can be interpreted as follows : females who were enrolled in the Natural Science and Engineering College and 169 not enrolled in the Humanities College, and females whose mothers had more than a high school education, and females from an urban area home town typically had higher scores on science and aesthetic, but lower scores on verbal related skills on TSOSS: females who were enrolled in the Humanities College and not enrolled in the Natural Science and Engineering College, and females whose mothers had less than high school education typically had lower scores on science and aesthetic, but higher scores on verbal related skills on the TSOSS.
As can be seen in the Table 71, the magnitude of this relationship between the variable sets is indicating by Rc. (Rc(2) =.52). In addition the value of square of
Rc explains 27 % of total variance shared by dependent and independent canonical variate.
Table 71
Rc and Sq of Rc for Case A University
Root No. Canon Cor. Sq. Cor (Rc) (Rc2)
1 .629 .396 2 .522 .272 3 .427 .183 4 .318 .101 170 Table 72
Mean and Standard Deviation for DV and IV Sets for Case A University (n=208)
Means and Standard Deviation for DV set
Variable Mean Std Dev N Label
VERBAL 54.92 7.88 224 SCIENCE 49.46 9.59 219 STRENGTH 49.69 8.96 219 AESTHE 51.16 9.95 221 Means and Standard Deviation for IV iset (n=234)
V a ria b le Mean Std Dev N Label
Maior Field HUMTY .46 .50 234 SOCSCI .08 . .27 234 SCIENG .32 .47 234 Enrichment Ed. Activities EXTYES .72 .45 234 I f r "ves" How manv ? EXTED2 1.63 1.67 234 Ed. Plan EPLAN1 .38 .49 234 Pursue Graduate Studies EPLAN2 .10 .30 234 Study priv voc inst EPLAN3 .50 .50 234 S e lf-stu d y EPLAN4 .08 .27 234 No F u rth er Educ p lan Attitude reaardina Career iand Family ATTYES .79 .40 234 ATTNO .06 .24 234 Social Economic Status SESF1 .04 .19 234 Lowest SES SESF2 .02 .13 234 SESF3 .05 .22 234 SESF4 .01 .11 234 SESF5 .07 .25 234 SESF6 .28 .45 234 SESF7 .21 .41 234 Highest SES Mother's Ed. Backaround MLH .28 .45 234 MHS .47 .50 234 M2YR .00 .07 234 MCOLL4 .18 .39 234 MADVDEG .02 .14 234 Location Completed most of Ed. LTOWN .21 .40 234 LURBAN .66 .47 234 Home Town HTOWN .14 .35 234 HURBAN .59 .49 234 Grade Point Averaae GPAA .23 .42 234 GPAB .61 .49 234 GPAC .11 .31 234 171 Table 73
Summary of Canonical Correlation Analysis for Case A University (n=208)
Canonical Canonical Canonical Canonical Variables Variate I Variate 2 Variate 3 Variate 4 b s b s b s b s
For IV set HUMTY -.761 .126 -.391 -.635 .126 .155 -.365 -.190 SOCSCI -.507 -.190 .041 .043 -.115 -.063 -.210 -.1 8 9 SCIENG -1.158 -.546 .285 .534 .088 -.046 -.1 5 1 .228 EXTYES .073 .150 .222 .168 .306 .558 -.2 0 8 -.0 2 9 EXTED2 .109 .139 -.140 -.018 .276 .508 .282 .092 EPLAN1 -.304 -.066 .154 .121 -.005 .119 -.383 -.267 EPLAN2 .065 .226 -.006 .046 .161 .163 .050 .079 EPLAN3 -.216 -.048 .002 -.117 .013 .077 -.253 -.1 2 5 EPLAN4 -.156 -.080 -.147 -.117 -.192 -.307 .322 .348 ATTYES .109 .009 -.130 -.0 9 8 .339 .194 -.2 5 9 -.292 ATTNO .067 .105 .034 .139 .330 .238 .130 .187 SESF1 -.002 -.107 -.052 -.126 .024 -.033 -.169 -.183 SESF2 -.091 -.169 -.102 -.270 .118 .113 -.079 -.108 SESF3 .087 -.072 -.126 -.277 .036 .007 -.092 .003 SESF4 .216 .056 .143 .160 -.309 -.357 -.120 -.0 3 9 SESF5 .046 -.037 .099 .084 -. 146 -.198 -.180 -.058 SESF6 -.051 -.011 .027 .074 -.009 . 106 .234 .188 SESF7 .197 .130 -.140 -.093 .072 .183 .184 .126 MLH -.426 -.4 1 4 -.828 -.373 .647 -.011 .836 .109 MHS -.252 .219 -.717 .009 .573 .094 .759 .081 M2YR -.295 -.258 -.142 -.023 .092 .009 -.0 4 5 -.1 7 3 MCOLL4 -.216 .137 -.347 .243 .444 -.011 .390 -.2 2 4 MADVDEG -.036 .025 -.221 .091 .113 -.001 .291 .078 LTOWN .249 -.133 .094 -.094 -.030 .149 -.665 .021 LURBAN .291 .200 .059 .196 -.218 -.093 -.304 -.167 HTOWN -.090 -.156 .323 .185 .365 .261 .465 .420 HURBAN -.037 .218 .298 .289 .312 .006 -.2 1 6 -.3 7 1 GPAA -.139 -.072 -.129 .027 -.108 .374 -.382 -.112 GPAB -.244 -.060 -.240 -.070 -.468 -.215 -.265 .075 GPAC .065 .098 -.286 -.093 -.407 -.138 -.2 1 7 .035 PV .013 .013 .008 .003
Rc2 (1)= .396 Rc2(2 )= .272 Rc2(3 ) = .183 Rc2(4)=. 101
For DV s e t
VERBAL .190 .164 -1.042 --.299 .689 .939 -.247 -.040 SCIENCE -.713 -.403 .845 .465 .384 .784 -.223 -.085 STRENGTH .009 .234 .090 .183 .077 .503 1.162 .812 AESTHE .870 .781 .627 .445 .030 .433 -.411 -.069
PV .214 .134 .484 .168 RD .085 .037 .088 .017 RD .085 .122 .210 .227 172 Replication 2: Case B Universitversitv (n=476)
Statistical Hypothesis: There are no relationships between the dependent (criterion! and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients (Rc2) are equal to zero.
Following is the information for testing the hypothesis for Case B
Universityersity.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .46; F = 3.18;p < .000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .70; F = 1.90; p < .000
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .86; F = 1.24; p = .115
Table 74
Testing Hypothesis for Case B University
Roots Wilks L. F Sig. of F Decision
1 TO 4 .45822 3.18082 .000 Reject 2 TO 4 .70307 1.90501 .000 Reject 3 TO 4 .86014 1.24069 .115 Fail to Reject 4 TO 4 .93529 1.14028 .288 Fail to Reject
As can be seen in the Table 74, canonical roots 1 and 2 are statistically significant at an alpha level of .05. These significance tests can be used for deciding which canonical functions to interpret. Structure coefficients were used, 173 which have the value more than .30 in order to explain each of the pairs of independent and dependent canonical variate (canonical roots), that the researcher interprets as meaningful,
As shown in the summary Table 77, dependent variate 1 carries primarily information about SCIENCE (s=-.54),and AESTHE (s=.58) related skills on the
TSOSS in the dependent variable set. In other words, females with lower values on science but higher values on aesthetic related skills have higher scores on dependent variate 1; females with lower values on aesthetic but higher values on science related skills have lower scores on dependent variate 1. However, independent variate 1 carries information about SCIENG (s=-.78) in the independent variable set. In other words, females enrolled in the Natural
Science and Engineering College have lower scores on independent variate 1; females not enrolled in the Natural Science and Engineering College have higher scores on independent variate 1.
Therefore, the nature of relationship between the 4 sub-scales of TSOSS and the selected variable sets on Canonical root 1 can be described as follows : females who were enrolled in the Natural Science and Engineering College typically had higher scores on science but lower scores on aesthetic related skills on the TSOSS: females who were not enrolled in Natural Science and
Engineering College typically had lower scores on science but higher aesthetic related skills on the TSOSS. As can be seen in the Table 74, the magnitude of this relationship between above variable sets is indicating by the value of Rc 174 (Rc(l) = .60). In addition, the square Rc explains 34% of the total variance shared
by the dependent and independent canonical variates.
In addition, dependent variate 2 carries information about VERBAL
(s=.34) and AESTHE (s=-.49) in the dependent variable set. In other words,
females with higher values on verbal but lower values on aesthetic related skills
have higher scores on dependent variate 2; females with lower values on verbal
but higher values on aesthetic related skills have lower scores on dependent
variate 2. However, independent variate 2 carries information HUMTY(s=.75)
in the independent variable set. In other words, females enrolled in the
Humanities College have higher scores on independent variate 2; females not
enrolled in Humanities College have lower scores on independent variate 2.
Therefore, the nature of relationship between the 4 sub-scales of TSOSS
and the selected variable sets on canonical variate 2 can be interpreted as follows
: females who were enrolled in the Humanities College typically had higher
scores on verbal but lower scores on aesthetic related skills on the TSOSS:
females who were not enrolled in the Humanities College typically had lower
scores on verbal but higher scores on aesthetic related skills on the TSOSS.
As can be seen in the Table 75, the magnitude of this relationship between above
variable sets is indicating by the value of Rc (Rc(2) = .43). In addition, the value
of the square Rc explains 18% of the total variance shared by the dependent and independent canonical variates (Rc2(2)=.18). 175 Table 75
Rc and Sq of Rc for Case B University
Roots Canon Cor. sq. cor (Rc) (Rc2)
1 .590 .348 2 .427 .183 3 .283 .080 4 .254 .055 176 Table 76
Mean and Standard Deviation for DV and IV Sets for Case B University (n=476)
Means and Standard Deviation for DV set (n=476) V ariab le Mean Std Dev N Label
VERBAL 55.91 8.38 501 SCIENCE 52.48 9.32 495 STRENGTH 51.08 10.63 494 AESTHE 53.57 10.77 496
Means and Standard Deviation for IV !set (n=507)
Variable Mean Std Dev N Label maior Field HUMTY .19 .40 507 SOCSCI .18 .38 507 SCIENG .39 .49 507 Enrichment Ed. Activities EXTYES .79 .41 507 I f , "ves". How manv ? EXTED2 1.90 1.64 507 Ed. Plan EPLAN1 .44 .50 507 Pursue Graduate Studies EPLAN2 .09 .28 507 Study p riv voc in s t EPLAN3 .46 .50 507 S e lf-stu d y EPLAN4 .06 .23 507 No Further Educ plan Attitude Reaardino Career and Familv ATTYES .81 .40 507 ATTNO .08 .27 507 Social Economic Status SESF1 .01 .12 507 Lowest SES SESF2 .00 .04 507 SESF3 .04 .19 507 SESF4 .00 .06 507 SESF5 .04 .20 507 SESF6 .28 .45 507 SESF7 .26 .44 507 Highest SES Mother's Ed Backaround MLH .17 .38 507 MHS .52 .50 507 M2YR .03 .16 507 MCOLL4 .24 .43 507 MADVDEG .01 .11 507 Location completed most of Ed. LTOWN .19 .39 507 LURBAN .77 .42 507 Home town HTOWN .20 .40 507 HURBAN .68 .47 507 Grade Point Averaae GPAA .08 .28 507 GPAB .78 .41 507 GPAC .11 .31 507 177 Table 77
Summary of Canonical Correlation Analysis for Case B University (n=476)
Canonical Canonical Canonical Canonical Variables Variate I Variate 2 Variate 3 Variate 4 b s b s b s b s
For IV set
HUMTY -.3 6 0 .189 .932 .754 .023 -.159 -.157 -.0 4 0 SOCSCI -.6 1 4 -.0 9 6 .381 .078 .030 .015 -.115 -.113 SCIENG -1.107 -.777 .309 -.248 .235 .120 -.075 .017 EXTYES -.093 .023 .048 .107 .019 .222 -.181 .006 EXTED2 .074 .055 -.006 .070 .346 .411 .225 .157 EPLAN1 .006 .090 .253 .151 .179 .237 -.1 1 8 -.1 5 0 EPLAN2 .025 .048 .040 .030 -.013 -.010 .314 .323 EPLAN3 .022 -.053 .142 .114 -.006 -.071 -.079 -.1 2 4 EPLAN4 -.0 6 9 -.1 1 9 .067 .094 -.242 -.262 .045 .094 ATTYES .193 .206 .216 .085 .219 .018 -.227 -.345 ATTNO .026 -.051 .212 .080 .406 .284 .234 .344 SESF1 .004 -.027 .126 .133 .296 .268 .026 -.010 SESF2 .011 .035 .147 .232 .205 .217 .171 .188 SESF3 -.205 -.206 -.087 .120 -.252 -.3 7 6 .494 .476 SESF4 .009 .003 -.076 -.021 -.189 -.232 .246 .252 SESF5 -.055 -.111 -.137 -.102 .059 .068 -.032 -.066 SESF6 -.0 9 4 -.110 -.209 -.253 -.013 .033 -.0 9 8 -.082 SESF7 -.038 -.017 .029 .115 -.164 -.214 -.2 5 8 -.335 MLH .078 -.207 .265 .190 -.096 -.256 .262 .036 MHS .233 .012 .233 -.067 .008 -.114 .491 -.048 M2YR .021 -.055 -.007 -.098 .128 .108 -.094 -.225 MCOLL4 .211 .198 .214 -.041 .218 .354 .493 .199 MADVDEG .107 .079 .068 .084 -.196 -.140 -.027 -.155 LTOWN -.075 -.078 -.474 -.092 -.205 -.177 -.2 9 1 -.1 4 0 LURBAN -.0 9 6 .120 -.330 -.030 -.278 .169 -.294 .022 HTOWN -.015 -.030 -.030 .011 .129 -.190 -.0 2 6 -.2 2 4 HURBAN -.072 .110 -.160 -.091 .408 .371 .159 .076 GPAA -.099 -.124 .137 -.042 -.088 .201 -.088 -.288 GPAB .001 .043 .346 -.062 -.335 -.164 .208 .182 GPAC .082 .065 .368 .156 -.259 -.024 .102 .060 PV .001 .006 .004 .003
Rc2( 1) = . 348 Rc2(2) = . 183 Rc2(3) = .080 Rc2(4) = .065
For DV s e t VERBAL .367 .098 1.210 .345 .567 .925 -.177 -.125 SCIENCE -.981 -.536 -.702 -.233 .314 .795 -.381 -.159 STRENGTH -.101 .063 .001 -.208 .173 .634 1.208 .742 AESTHE .771 .576 -.859 -.488 .177 .654 -.457 -.045 PV .158 .114 .579 .149 RD .055 .021 .047 .010 RD .055 .076 .122 .132 178 Replication 3; Case C Universitversitv (n=233)
Statistical Hypothesis; There is no relationships between the dependent (criterion! and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients (Rc2) are equal to zero.
Following is the information for testing the hypotheses for Case C Universityersity.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .37; F == 1.89; p < .000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .53;F = 1.60;p =.001
Ho : Rc(3)2 = Rc(4)2 = 0. Wilks lambda = .72; F = .1.30p = .079
Table 78
Testing Hypothesis for Case C University
Roots Wilks L. F Sig. of F Decision
1 TO 4 .36865 1.88732 .000 Reject 2 TO 4 .53526 1.60023 .001 Reject 3 TO 4 .71613 1.30428 .079 Fail to Reject 4 TO 4 .88778 .94574 .546 Fail to Reject
As can be seen in the Table 78, canonical roots 1 and 2 are statistically significant at an alpha level of .05. These significance tests can be used for deciding which canonical functions to interpret. Structure coefficients were used, 179 which have the value greater than .30 in order to explain each of the pairs of independent and dependent canonical variate (canonical roots), that the researcher interprets as meaningful,
As shown in the summary Table 81, dependent variate 1 carries information about STRENGTH (s=-.46),and AESTHE (s=-.68) related skills on the TSOSS in the dependent variable set. In other words, females with lower values on physical strength and aesthetic related skills on the TSOSS have higher scores on dependent variate 1; females with higher values on physical strength and aesthetic related skills on the TSOSS have lower scores on dependent variate 1.
However, independent variate 1 carries information about HUMTY (s=.72),
EXTYES (s=-.30), ATTNO (s=-.37),and MLH (s=-.39)in the independent variable set. In other words, females enrolled in the Humanities College, did not have experience of the enriching educational activities, had positive attitude regarding family and career, and females whose mothers had less than a high- school education have higher scores on independent variate 1; females with the opposite characteristics to those mentioned above variables have lower scores on independent variate 1. Therefore, the nature of relationship between the 4 sub scales of TSOSS and the selected variable sets on canonical root 1 can be described as follows : females who were enrolled in the Humanities College, and did not have additional enriching educational experiences, and who had a positive attitude regarding family and career, and females whose mothers had less than high-school education typically had lower scores on physical strength and aesthetic related skills on the TSOSS: females who were not enrolled in the Humanities
College, did not have additional enriching educational experiences, who had a negative attitude regarding family and career, and females whose mothers had more than high school education typically had higher scores on physical strength and aesthetic related skills on the TSOSS. As can be seen in the Table 79, the magnitude of this relationship is indicating by the value of Rc (Rc(l) = .56). In addition, the value of the square of Rc explains 31% of the total variance shared by the dependent and independent canonical variates (Rc2= .31)
In addition, dependent variate 2 carries information about VERBAL (s=-
.40), SCIENCE (s=-.79)and AESTHE (s=-.47) in the dependent variable set. In other words, females with higher values on verbal, science, and aesthetic related skills on the TSOSS have lower scores on dependent variate 2; females with lower values on verbal, science, and aesthetic related skills on the TSOSS have higher scores on dependent variate 2. However, independent variate 2 carries information about SCIENG (s=-.44),EXTED2 (s=-.33), ATTNO (s=.30)
MCOLL4 (s=-.45)and GPA"A" (s=-.35) in the independent variable set. In other words, females enrolled in the Natural Science and Engineering College, had lower number of enriching educational experience, had a positive attitude regarding career and family, 4 year of college level of mother’s educational background, and females who reported GPA of "A" had lower scores on independent variate 2; females with the opposite characteristics to those mentioned above variables have higher scores on independent variate 2. 181 Therefore, the nature relationship between the 4 sub-scales of TSOSS and the selected variable sets on canonical variate 2 can be interpreted as follows : females who were enrolled in the Natural Science and Engineering College, had a lower number of additional enriching: educational experience, had a positive attitude regarding career and family, whose mothers had 4 year of college education, and females who reported a GPA of "A” typically had higher scores on verbal, science, and aesthetic related skills on the TSOSS: females who were not enrolled in the Natural Science and Engineering College, had large number of additional enriching educational experience, had a negative attitude regarding career and family, whose mothers did not have 4 year of college education, and females who reported a GPA lower than "A" typically had lower scores on verbal, science and aesthetic related skills on the TSOSS. The magnitude of this relationship between the above variable sets is indicating by the value of Rc
(Rc(2)=50). In addition the value of square of the Rc explains 25% the total variance shared by the dependent and independent canonical variates (Rc2 =
.253) as can be seen in the Table 78.
Table 79
Rc and Sq of Rc for Case C University
Root No. Canon Cor. Sq. Cor (Rc) (Rc2)
.558 .311 2 .503 .253 3 .440 .193 4 .335 .112 182 Table 80
Mean and Standard Deviation for DV and IV Sets for Case C University (n=233)
Means and Standard Deviation for DV sets
V ariab le Mean Std Dev N Label VERBAL 56.08 8.43 252 SCIENCE 50.23 9.66 245 STRENGTH 51.02 10.41 246 AESTHE 51.44 10.84 248
Means and Standard Deviation for DV isets (n=267) V ariab le Mean Std Dev N Label
Maior Field HUMTY .31 .47 267 SOCSCI .16 .37 267 SCIENG .16 .37 267 Enrichment Ed. A c tiv itie s EXTYES .72 .45 267 "v e s", how manv ? EXTED2 1.48 1.50 267 Ed Plan EPLAN1 .39 .49 267 Pursue Graduate Studies EPLAN2 .09 .29 267 Study priv voc inst EPLAN3 .46 .50 267 Self-study EPLAN4 .08 .27 267 No Further Educ plan Attitude regarding Career and Family ATTYES .83 .38 267 ATTNO .06 .24 267 Social Economic Status SESF1 .03 .16 267 Lowest SES SESF2 .01 .09 267 SESF3 .06 .23 267 SESF4 .02 .15 267 SESF5 .07 .26 267 SESF6 .30 .46 267 - SESF7 .24 .43 267 H ighest SES Mother's Ed. Backaround MLH .30 .46 267 MHS .45 .50 267 M2YR .02 .14 267 MCOLL4 .14 .35 267 MADVDEG .02 .14 267 Location completed most of Ed. LTOWN .23 .42 267 LURBAN .65 .48 267 HomeTown HTOWN .24 .43 267 HURBAN .56 .50 267 Grade Point Averaae GPAA .16 .37 267 GPAB .67 .47 267 GPAC .10 .30 267 183 Table 81
Summary of Canonical Correlation Analysis for Case C University (n=233)
Canonical Canonical Canonical C anonical Variables Variate X V a riate 2 V a ria te 3 V a ria te 4 b s b b b s b s
For IV s e t
HUMTY .887 .719 .084 .271 -.042 .170 .282 .204 SOCSCI .319 -.011 -.098 -.0 2 8 -.1 7 1 -.0 9 0 .061 -.032 SCIENG .385 .076 -.332 -.444 -.395 -.329 -.0 3 1 .019 EXTYES -.267 -.290 -.041 -.170 .106 .197 .414 .274 EXTED2 .087 -.104 -.228 -.332 .109 .214 -.089 .113 EPLAN1 .104 -.049 -.173 -.2 6 1 .349 .298 -.420 -.208 EPLAN2 -.117 -.098 .123 .103 .215 .179 -.173 -.045 EPLAN3 .081 .207 -.020 .073 .209 .201 -.3 0 4 -.0 4 4 EPLAN4 .074 .011 .108 .133 -.2 1 6 -.349 -.110 .051 ATTYES .223 .274 .214 -.070 -.088 .070 -.346 -.204 ATTNO -.1 1 8 -.374 .349 .286 .134 -.052 -.1 4 0 -.049 SESF1 .067 .128 .083 .130 .225 .157 -.1 2 4 -.1 8 4 SESF2 .046 .082 -.003 -.013 -.241 -.244 .052 -.007 SESF3 -.031 .071 -.093 -.0 1 9 .151 .009 .267 .068 SESF4 .053 .031 .160 .208 -.086 -.084 .108 .054 SESF5 .055 .170 .190 .193 .477 .347 -.1 7 0 -.3 0 6 SESF6 -.032 -.133 .047 .095 .179 .012 .475 .346 SESF7 -.068 -.047 -.098 -.2 3 1 .162 -.023 .407 .125 MLH .280 .387 .326 .264 -.170 -.036 .329 .110 MHS .258 -.188 .238 -.008 -.125 .018 -.0 2 8 -.128 M2YR .016 -.218 .034 .030 -.423 -.319 -.208 -.173 MCOLL4 .245 -.045 -.093 -.4 5 4 .018 .188 .164 .087 MADVDEG .099 -.066 -.019 -.117 -.147 .000 .126 .092 LTOWN -.287 -.012 .071 -.110 .131 -.129 .651 .094 LURBAN -.395 -.153 .386 -.013 .182 .214 .397 .008 HTOWN .013 -.077 -.207 -.103 -.133 -.152 -.027 -.026 HURBAN .045 -.066 -.436 -.184 .190 .217 .229 .066 GPAA .088 .149 -.951 -.351 .114 .108 -.6 1 6 -.120 GPAB .117 .073 -.920 .009 .035 .069 -.821 .135 GPAC .056 -.188 -.592 .111 -.313 -.205 -.806 -.252 PV .014 .010 .007 .003
Rc2(1) = .311 Rc2(2 ) = .253 Rc2(3) = .193 Rc2(4)=.112
For DV s e t
VERBAL .761 .113 .422 -.393 1.249 .651 -.417 -.6 4 0 SCIENCE .223 .117 -1.130 -.794 -.8 4 4 -.002 -.1 2 6 -.597 STRENGTH -.417 -.464 .605 .022 -.433 -.021 -.885 -.885 AESTHE -1.017 -.683 -.534 -.477 .357 .495 .511 -.2 4 6 PV .177 .253 .167 .402 RD .055 .064 .032 .045 TW .055 .119 .151 .197 184 Replication 4: Case D Universitversitv (n=251)
Statistical Hypothesis; There are no relationships between the dependent (criterion) and the independent (predictor! variable sets.
Ho : All squared canonical correlation coefficients(Rc2) are equal to zero.
Following is the information for testing the hypothesis for Case D
Universityersity.
Ho : Rc(l)2 = Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .33; F = 2.26;p < .000
Ho : Rc(2)2 = Rc(3)2 = Rc(4)2 = 0. Wilks’ lambda = .58;F = 1.48;p =.004
Ho : Rc(3)2 =Rc(4)2 = 0 . Wilks lambda = .79; F = .936; p = .607
Table 82
Testing Hypothesis for Case D University
Roots Wilks L. F Sig. of F Decision
1 TO 4 .33706 2.26868 .000 Reject 2 TO 4 .58224 1.48753 .004 Reject 3 TO 4 .79760 .93635 .607 Fail to Reject 4 To 4 .91362 .77034 .787 Fail to Reject 185 As can be seen in the Table 82, canonical roots 1 and 2 are statistically
significant at an alpha level of .05. These significance tests can be used for
deciding which canonical functions to interpret. Structure coefficients were used,
which have a value greater than .30 in order to explain each of the pairs of
independent and dependent canonical variate (canonical roots) that the researcher
interprets as meaningful.
As shown in the summary Table 84, dependent variate 1 carries
information about SCIENCE (s=-.74) related skills on the TSOSS in the
dependent variable set. In other words, females with have lower values on
science related skills have higher scores on dependent variate 1; females with
higher values on Science related skills have lower scores on dependent variate 1.
However, independent variate 1 primarily carries information about HUMTY
(s=.30), SOSCI (s=~.44),and SCIENG (s=-.60) in the independent variable set.
In other words, females enrolled in the Humanities College, not enrolled in the
Social Science College and the Natural Science and Engineering College have
higher scores on independent variate 1; females not enrolled in the Humanities
College, enrolled in the Social Science and Natural Science and Engineering
College have lower scores on independent variate 1.
Therefore, the nature of relationship between the 4 sub-scales of TSOSS and the selected variable sets on canonical root 1 can be described as follows : females who were enrolled in the Humanities College and not enrolled in the
Social Science and the Natural Science Engineering College typically had lower 186 scores on science related skills on the TSOSS: females who were enrolled in the
Social Science and the Natural Science and Engineering Colleges, but not
enrolled in the Humanities College typically had higher scores on science related
skills on the TSOSS. As can be seen in the Table 83, the magnitude of this
relationship between above variable sets is indicating by the value of Rc
(Rc(l) = .65). In addition, the value of the square of Rc explains 42% of the total
variance shared by the dependent and independent canonical variates. (Rc2 = .42)
In addition, dependent variate 2 carries information about VERBAL
(s = .44),and AESTHE (s=-.53)in the dependent variable set. In other words,
females with higher values on verbal but lower values on aesthetic related skills
on the TSOSS have higher scores on dependent variate 2; females with lower
values on verbal but higher values on aesthetic related skills on the TSOSS have lower scores on dependent variate 2. However, independent variate 2 primarily carries information about HUMTH(s=.38), SOCSCI(s=.30), SCIENG (s=-.32) and M2YR (s=-.35) in the independent variable set. In other words, females enrolled in the Humanities, Social Science, not enrolled in Natural Science and
Engineering colleges, and females whose mothers had not 2 year vocational education have higher scores on independent variate 2; females with the opposite characteristics to those mentioned above variables have higher scores on independent variate 2.
Therefore, the nature of relationship between the 4 sub-scales of TSOSS and the selected variable sets on canonical variate 2 can be interpreted as follows 187 : females who were enrolled in the Natural Science and Engineering College, not enrolled in the Humanities and the Social Science Colleges, and females whose mothers had 2 year vocational education typically had lower scores on verbal, but higher scores on aesthetic related skills on the TSOSS : females who were enrolled in the Humanities and the Social Science Colleges, and not enrolled in the Natural Science and Engineering, and females whose mothers had not 2 year vocational/technical education typically had higher scores on verbal, but lower scores on aesthetic related skills on the TSOSS.
The magnitude of this relationship between above variable sets is indicating by the value of Rc (Rc(2) = .52). In addition the value of square Rc explains 27% the total variance shared by the dependent and independent canonical variates as can be seen in the Table 82.
Table 83
Rc and Sq of Rc for Case D University
Root No. Canon Cor. Sq. Cor (Rc) (Rc2)
1 .649 .421 2 .520 .270 3 .356 .127 4 .294 .086 188 Table 84
Mean and Standard Deviation for DV and IV Sets for Case D University (n=251)
Means and Standard Deviation for DV set (n=251)
V aria b le Mean Std Dev N Label
VERBAL 53.81 8.20 260 SCIENCE 51.05 10.09 257 STRENGTH 48.61 10.51 258 AESTHE 50.00 9.85 261 Means and Standard Deviation for IV !set (n=267) V ariab le Mean Std Dev N Label
Maior Field HUMTY .21 .41 267 SOCSCI .19 .39 267 SCIENG .25 .44 267 Enrichment Ed. Activities EXTYES .90 .31 267 I f r "v es". How manv ? EXTED2 2.20 1.82 267 Ed. Plan EPLAN1 .44 .50 267 Pursue Graduate Studies EPLAN2 .06 .24 267 Study priv voc inst EPLAN3 .47 .50 267 S e lf-stu d y EPLAN4 .06 .24 267 No Further Educ plan Attitude reaardina Career and Family ATTYES .84 .37 267 ATTNO .04 .20 267 Social Economic Status SESF1 .01 .11 267 Lowest SES SESF2 .00 .06 267 SESF3 .01 .11 267 SESF4 .01 .12 267 SESF5 .01 .11 267 SESF6 .23 .42 267 SESF7 .19 .39 267 Highest SES Mother's Ed. Backaround MLH .07 .26 267 MHS .37 .48 267 M2YR .01 .11 267 MCOLL4 .48 .50 267 MADVDEG .02 .15 267 Location completed most of Ed LTOWN .13 .34 267 LURBAN .82 .39 267 Home town HTOWN .16 .37 267 HURBAN .75 .44 267 Graduate Point Averaae GPAA .11 .32 267 GPAB .70 .46 267 GPAC .09 .29 267 189 Table 85
Summary of Canonical Correlation Analysis for Case D University (n=251)
Canonical Canonical Canonical Canonical Variables Variate I Variate 2 Variate 3 Variate 4 b s b s b s b s
For XV set
HUMTY -.131 .291 .451 .378 -.638 -.445 -.0 5 4 -.085 SOCSCI -.665 -.435 .377 .301 -.228 -.005 .164 .025 SCIENG -.855 -.602 -.124 -.321 -.134 .148 -.0 4 6 .046 EXTYES .009 .145 -.133 -.100 -.100 .009 .015 -.177 EXTED2 .008 -.046 .066 .145 .193 .248 -.3 5 1 -.316 EPLAN1 -.004 -.005 -.200 -.143 -.106 -.147 -.2 8 5 -.277 EPLAN2 .129 .077 -.169 -.113 .208 .150 -.0 4 6 .081 EPLAN3 -.095 -.118 -.106 -.071 .083 .125 .192 .231 EPLAN4 -.016 -.064 -.063 -.038 -.259 -.2 4 4 .107 .202 ATTYES .039 .041 .180 .272 .481 .329 -.4 2 4 -.4 0 1 ATTNO -.036 -.040 -.057 -.090 .272 .009 -.0 6 8 .144 SESF1 -.106 -.116 -.034 .000 .132 .137 -.232 -.162 SESF2 -.026 -.090 .208 .221 .111 .142 .107 .096 SESF3 .013 -.105 -.085 .045 .322 .321 -.0 8 6 .034 SESF4 .132 .190 .234 .188 .067 .168 .369 .481 SESF5 -.032 .024 • .094 .091 -.199 -.125 .008 .067 SESF6 -.053 -.034 -.174 -.145 -.156 -.135 -.2 5 4 -.2 0 4 SESF7 -.123 -.124 .118 .217 -.132 -.063 .045 .003 MLH -.333 -.095 .272 .234 .265 .211 .712 .354 MHS -.702 -.176 .383 .238 .056 -.183 . 541 -.1 8 9 M2YR -.264 -.196 -.202 -.349 .058 .004 .314 .221 MCOLL4 -.752 .172 .164 -.162 .181 .187 . 680 -.0 5 1 MADVDEG -.176 .201 -.109 -.239 -.1 9 6 -.2 0 8 .220 -.001 LTOWN .500 -.236 -.189 .078 -.190 -.221 -.3 9 1 -.049 LURBAN .739 .274 .090 -.073 .195 .239 -.218 -.056 HTOWN . 049 -.150 -.168 .147 .071 -.1 6 3 .010 -.0 4 8 HURBAN -.023 .192 -.468 -.253 -.101 .149 -.0 7 4 -.0 8 4 GPAA -.203 -.147 .447 .216 .063 .040 -.3 3 0 -.3 0 5 GPAB -.065 .064 .343 -.013 .039 .004 -.025 .078 GPAC .120 .108 .193 .035 .186 .098 -.121 .089 PV .016 .001 .004 .003
Rc2( 1) =. 421 Rc2(2 ) =.270 Rc2( 3) =. 127 Rc2( 4) =. 086
For DV se t VERBAL .446 -.028 1.150 .438 .334 .781 -.558 -.444 SCIENCE -1.201 -.747 -.371 -.030 .038 .588 -.223 -.307 STRENGTH .090 -.027 -.104 -.060 .641 .913 1.020 .403 AESTHE .429 .273 -.909 -.527 .203 .644 -.5 6 6 -.483
PV .159 .119 .551 .172 RD .067 .032 .070 .015 RD .067 .099 .169 .184 190
The Fourth Question and Results
Question; Are the four sub-scales of CDS and the four sub-scales of TSOSS related to the students’ career decision status regarding decided or undecided ? If the groups differ, describe how the groups are similar or different. What variables are important in describing how the groups different?
To answer questions 4 and 5, discriminant analysis was performed using
SPSS-PC. Discriminant analysis is the statistical technique for studying simultaneously the differences between two or more groups with respect to several variables. When comparing two groups, the undecided and decided, in terms of many variables, the interest lies not only in whether the groups differ significantly from one another; but, if they do differ significantly, in understanding the nature of the differences of both the undecided and decided groups.
In the discriminant analysis for this question, the variables are as follows.
The Dependent Variable (Group Variable)
Career decision status (Cl) containing two groups as undecided (1) or decided (2). Item 1 of the CDS, was used to divide all respondents into two groups as decided or undecided. The item 1 states: "I have decided on a career and feel comfortable with it. I also know how to go about implementing my choice." Those who responded "exactly like me" or "very much like me" were placed in the decided group, and those who responded only "slightly like me" or
"not at all like me" were placed in the undecided group. 191 The Independent Variables (Discriminating Variables)
CONFUS Total Score of the CDS items 7, 8, and 11. Diffusion, which represents feelings of confusion, discouragement, and lack of experience or information about the making of career decisions)
UNCERT Total Score of the CDS items 12, 16, and 18. (Support, which represents uncertainty about how to proceed in making decisions, and the need for additional support for initial decisions)
CLASSIC Total Score of the CDS items 4, 15, and 17. (Approach, which represents a classical approach-approach conflict in which several possible careers are attractive)
EXTRNL Total Score of the CDS items 3,5,6 and 9 (External barrier, which represents both external barriers to career choice and lack of interest in making a decision)
VERBAL Verbal and Interpersonal (Total score of the TSOSS items 1, 5, 9, 13, 17, 21, 25, 29, 33, 37, 41, 45, 49, 53, and 57.)
SCIENTIF Quantitative, Scientific and Business (Total score of the TSOSS items 2, 6, 10,14, 18, 22, 26, 30, 34, 38, 42, 46,50, 54, and 58)
STRENGTH Physical strength and agility (Total score of the TSOSS items 3, 7, 11, 15, 19, 23, 27, 31, 35,39, 43, 47, 51, 55, and 59)
AESTHE Aesthetic skills (Total score of the TSOSS items 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48, 52, 56, and 60)
Hypothesis : The group centroids (means) for the two groups are equal on the 4 sub-scales of CDS and the 4 sub-scales of TSOSS. 192 Total Responses(n= 1135)
For testing the hypothesis, Wilks’ lambda, which is a multivariate measure of group differences over several discriminating variables, was used. As Wilks’ lambda increases toward its maximum value of 1.0, it indicates progressively less discrimination. When Wilks’ lambda equals 1.0, it indicates no discrimination.
The group centroids are equal and there are no group differences.
As shown in the summary Table 88, the statistical hypothesis was rejected
(p<.001). Thus, the conclusion is that the groups, undecided and decided differ significantly on the discriminant function; that is it is unlikely that the groups have the same group centroids on the discriminant function. In other words, the independent variable set, which includes the 4 sub-scales of CDS and the 4 sub scale of TSOSS, discriminate between the undecided and decided groups.
To describe the power of the statistically significant discriminant function, eigenvalue and canonical correlation coefficient statistics were used to assess and describe the discriminating power of the discriminant function. The eigenvalue is the ratio (for the discriminant scores) of the between groups sum of squares to the within groups sum of squares. For the two-group situation here, there is only one discriminant function, hence only one eigenvalue. The larger the eighenvalue, the greater the power of the discriminate function. The eigenvalue for the function here was .3514. Therefore, the discriminating power of this function is not great. 193 The canonical correlation coefficient is a measure of the degree of association between the groups and discriminate scores. Values vary from 0 (no association) to 1.0 (maximum association). The canonical correlation coefficient squared is the proportion of variance explained by the groups. Since this is a two- group situation, the canonical correlation is the Pearson product-moment correlation coefficient between the discriminant score and the group variable.
Canonical correlation is a tool in judging the discriminating power of the discriminant function. A high coefficient indicates a strong relationship between the groups and the discriminant function. In this case, the coefficient of .5099 indicates that the discriminant function has some (but not great) power as can be seen in the summary Table 88. By examining both the relative percentages
(associated with the eigenvalues) and the canonical correlations for the discriminant functions, it can be determined that this discriminant function had substantively meaningful discriminating power but not greater in explaining the groups differences, undecided and decided. The classification of cases Table can be used as a means of predicting the group membership for cases. It can also be used as an index of the effectiveness of the discriminant function. For all cases,
74.36% of the cases were correctly classified indicating the accuracy of the discriminant function as an additional measure of group differences. More detailed information on the classification results is shown in the following Table
86. 194 Table 86
Classification Results for Total Cases
No. of Predicted Grouo Actual Group Cases Undecided Decided
Undecided 384 278 106 72.4% 27.6%
Decided 751 185 566 24.6% 75.4%
Percent of cases correctly classified: 74.36%
For this case, only the cases with complete information for all predictor variables are included in the classification results Table. Of the 384 females in the undecided group, 278 were correctly predicted to be members of the undecided group (72%), while 106 (28%) were assigned incorrectly to the decided group. Similarly, 566 out of 751 (75%) of the decided group females were identified correctly, and 185(25%) were misclassified. The overall percentage of cases classified correctly was 74%.
In identifying what variables are important in describing how the groups differ, standardized discriminant function coefficients (discriminant weights) and structure coefficients (discriminant loadings) were used.
Standardized discriminant function coefficients (discriminant weights) describe the relative importance of the discriminating variables. They are also used to determine which variables contribute most to determining scores on the 195 discriminant function. The larger the magnitude of the standardized coefficient
the greater is that discriminating variable’s contribution. Values higher than one-
half of the highest coefficient are considered important. As shown in the
summary Table 87, CONFS (b=.65) and UNCERT (b=-.49), which represent
uncertainty about how to proceed in making decisions and the need for additional
support for initial decisions, are considered as important contributors to
determining scores on the discriminant function. Thus, the most highly
distinguishing (discriminating) attributes of Group I (undecided) when compared
with Group II (decided) are that females in the undecided group tend to be
lacking in experience or information, having feelings of confusion, feeling
discouragement about the making of career decisions, wanting extra support and
needing additional support for initial decisions.
Structure coefficients are product-moment correlations between
discriminating variables and discriminant scores. A rule of thumb is to consider
all structure coefficients that are equal to or greater than .30 as meaningful. For
all cases combined, CONFUS(=.83), EXTRNL (=.62), UNCERT (=-.41),
CLASIC (=.39) and VERBAL (-.31) are variables that are closely related to the
function. By examining both values on two groups, the nature of the relationship between groups and discriminant variables can be concluded. The results indicate that undecided females were more likely to have confusion, uncertainty and external barriers, but less likely to have classic factor, which represents a classical approach conflict compared to decided females. In addition undecided females 196 tended to have lower scores on verbal/interpersonal related skills on the TSOSS compared to decided females.
Table 87
Mean and Standard Deviation for Discriminating Variables for Total Cases (n=1135)
GROUPCONFUSUNCERTCLASICEXTNAL
Undecided M 8.72135 7.62500 8.40625 9.12240 (N=384) SD 2.34781 2.03366 2.09227 2.38536
Decided M 6.28362 8.58988 7.31558 7.28895 (N=751) SD 2.34850 1.80062 2.30888 2.36370
GROUP VERBAL SCIENCE STRENGTH AESTHE
Undecided M 53.24740 49.41146 48.97917 50.65104 SD 8.59769 9.68418 10.00729 9.93283
Decided M 56.38216 52.09055 50.88948 52.55126 SD 7.99194 9.65759 10.35051 10.83167 197
Table 88
Summary Data for Discriminant Analysis for Total Cases 10= 11351
Discriminant Function I Variables b s Group Centroids
Undecided .82828 Decided -.42351 CONFUS .64513 .82929 UNCERT -.48382 .61773 CLASIC .11445 -.40943 EXTRNL .27791 .38931 VERBAL -.16665 -.30533 SCIENCE -.01075 -.22140 STRENGTH -.01951 -.14909 AESTHE .03813 -.14407
Eigenvalue Rc Wilks' Lambda p .3514 .5099 .7400 <.001 b = Standardized Canonical Discriminant Function Coefficients s = Structure Coefficient Rc= Canonical Correlation Coefficient
Group Centroids and Discriminating Variables f s value >.30)
- . 4Z D score 783 Decided Group Undecided Group
LOW CONFUS HIGH CONFUS LOW UNCERT HIGH UNCERT HIGH CLASIC LOW CLSIC LOW EXTRNL HIGH EXTRNL HIGH VERBAL LOW VERBAL 198 Replication 1: Case A University (n=204)
Hypothesis: The group centroids (means') for the two groups are equal on 4 sub-scales of CDS and 4 suh-scaies of TSOSS.
Ho: Group centroids for the two groups are equal.
As shown in the summary Table 91, the decision regarding the statistical
hypothesis was rejected by Wilks’ lambda = .73;Chi square = 62.2;p<.01. Thus,
the conclusion is that the undecided and decided groups differ significantly on the
discriminant function for case A university; that is, it is unlikely that the groups
have the same centroids (means) on the discriminant function. Therefore, the
test indicates that the discriminant function is statistically significant for case A
university.
To describe the power of the statistically significant discriminant function,
eigenvalue and canonical correlation coefficient statistics were used to assess and
describe the discriminating power of the discriminant function. For case A
university, the eigenvalue for this function was .37 and the canonical correlation
coefficient was .52. These two values are not large enough to be strong.
However, as can be seen in the Table 89, the classification results Table shows that for case A University,72% of the cases were correctly classified, which indicates the accuracy of the discriminant function on the independent variable set. 199 Table 89
Classification Results for Case A University
No. of Predicted Grouo Actual Group Cases Undecided Decided
Undecided 68 48 20 70.6% 29.4%
Decided 136 38 98 27. 9% 72.1%
Percent of "grouped" cases correctly classified: 71.57%
To determine what variables are important in describing how the groups differ, standardized coefficients (discriminant weights) were used. If the coefficient is positive (negative), the group with the higher group centroid has higher (lower) values on the discriminant variables than the group with the lower group centroid. For case A university, CONFS (b=-.74), which represents feelings of confusion, discouragement, and lack of experience or information about the making career decisions, and UNCERT (b=-.59), which represents uncertainty about how to proceed in making decisions, and the need for additional support for initial decisions, are also considered as important contributors. Therefore the most highly distinguishing (discriminating) attributes of Group I (undecided) when compared with Group II (decided) are that females who are in the undecided group tend to be more confused but less uncertain in making career decisions. The value of the structure coefficients indicate that CONFUS (=-.72),
EXTRNL (=-.47),UNCERT(=.41), VERBAL (.39), STRENGTH (.33) and
AESTHE (.30) are variables that are closely related to the discriminant function.
It can be concluded that undecided females were more likely to have confusion and external barriers, as compared to decided females, who were more likely to have uncertainty. In addition undecided females were less likely to demonstrate verbal/interpersonal, physical strength, and aesthetic related skills on the TSOSS as compared to decided females. 201
Table 90
0i=2041
Cl CONFUS UNCERTCLASIC EXTRNL
1* M 8.25000 7.33824 7.72059 9.01471 SD 2.30104 1.96697 1.74361 2.53037
2* M 6.06618 8.34559 7.19853 7.49265 SD 2.39815 1.88318 2.40315 2.50332
Cl VERBAL SCIENCE STRENGTH AESTHE
1* M 52.35294 47.25000 47.10294 48.55882 SD 7.95653 9.06753 8.28078 9.16089
2* M 56.26471 50.71324 50.90441 52.38235 SD 7.77959 9.95836 9.17475 10.16316
Cl : Career Decision Status 1* — > Undecided females( n=68) 2* — > Decided females (n=136) 202
Table 91
Summary Data for Discriminant Analysis for Case A University.
Discriminant Function 1 Variables b s Group Centroids
CONFUS -.74005 -.71950 Undecided -.85509 UNCERT .58507 .41090 Decided .42754 CLASIC .12572 -.18448 EXTRNL -.17102 -.47234 VERBAL .15854 .38907 SCIENCE .07898 .27917 STRENGTH .22803 .33345 AESTHE .03227 .30288
Eiaenvalue Rc Wilks' Lambda E .37 .52 .73 <.001 b = Standardized Canonical Discriminant Function Coefficient s = Structure Coefficient Rc= Canonical Correlation Coefficient
Groups Centroids and Discriminating Variables(s value > .301
- 0- -.86 D score .43 (Undecided Group) (Decided Group)
High CONFUS Low CONFUS Low UNCERT High UNCERT High EXTRNL Low EXTERNAL Low VERBAL High VERBAL Low STRENGTH High STRENG Low AESTHE High AESTHE 203 Replication 2; Case B University (n=469)
Hypothesis: The group centroids (means) for the two groups are equal on 4 sub scales of CDS and 4 sub-scales of TSOSS.
Ho: Group centroids for the two groups are equal.
As can be seen in the summary Table 94, the decision regarding the statistical hypothesis was rejected by Wilks’ lambda = .70; Chi square =162.2; pC .O l. Thus, the conclusion is that, undecided and decided groups differ significantly on the discriminant function on the case B university; that is, it is unlikely that the groups have the same centroids (means) on the discriminant function. Therefore, the test indicates that the discriminant function is statistically significant for case B university.
To describe the power of the statistically significant discriminant function eigenvalue and canonical correlation coefficient statistics were used to assess and describe the discriminating power of the discriminant function. For the case B university, the eigenvalue for this function was .42, and the canonical correlation coefficient was .54. These two values are not large enough to be strong.
However, as can be seen in the Table 92, the classification of results indicates that for case B university, 77% of the respondents were correctly classified, which indicates the accuracy of the discriminant function on independent variable set. 204
Table 92
Classification Results for Case B University
No. of Predicted Grouo
Actual Group Cases Undecided Decided
Undecided 150 113 37 75.3% 24.7%
Decided 319 73 246 22.9% 77.1%
Percent of cases correctly classified: 76.55%
To determine what variables are important in describing how the groups differ, standardized discriminant function coefficients (discriminate weights) and structure coefficients (discriminant loadings) were used. For case B university,
CONFS (b = .46), which represents feelings of confusion, discouragement, and lack of experience or information about the making of career decisions, UNCERT
(b=-.46), which represents uncertainty about how to proceed in making decisions, and the need for additional support for initial decisions, and EXTRNL (b=.46), which represents external barriers to career choice are considered as important contributor.
The values of structure coefficients represent that CONFUS (=.77),
EXTRNL(=.74), CLASIC(=.44), and UNCERT (= -.40) are variables that are closely related to the discriminant function. Therefore the nature of this relationship can be interpreted that undecided females were more likelv to have confusion, conflict.and feel external barriers to career choice compared to decided group. In addition decided females were more likely to be uncertain so that they need more support about how to proceed in making decisions compared to undecided females. Interestingly for case B university overall self-efficacv expectations failed to discriminant of either groups. 206
Table 93
Mean and Standard Deviation for Discriminating Variables for Case B University (n=469)
Cl CONFUS UNCERT CLASIC EXTRNL
1* M 8.90000 7.72667 8.82667 9.28667 SD 2.41847 2.13283 2.11973 2.37531
2* M 6.38871 8.73668 7.53292 6.95298 SD 2.32420 1.69524 2.10525 2.23487
Cl VERBAL SCIENCE STRENGTHAESTHE
1* M 54.56667 51.28667 50.89333 51.84000 SD 9.00665 9.05415 10.18035 10.29764
2* M 56.71473 52.93730 51.21944 54.29154 SD 7.98368 9.46705 10.75903 10.97480
Cl : Career Decision Status
1* — > Undecided females (n= 150) 2* — > Decided females (n=319) 207
Table 94
Summary Data for Discriminant Analysis for Case B University (n=469)
Variables b s Group Centroids
CONFUS .46108 .76834 Undecided .944 UNCERT -.46092 -.39414 Decided -.444 CLASIC .22199 .44176 EXTRNL .46230 .73719 VERBAL -.07571 -.18592 SCIENCE -.01060 -.12736 STRENGTH .09427 -.02221 AESTHE -.0723 2 -.16409
Eicrenvalue Rc Wilks' Lambda o .42 .54 .70 <.001 b = Standardized Canonical Discriminant Function Coefficients s = Structure Coefficient Rc= Canonical Correlation Coefficient
Group Centroids and Discriminatincr Variables( s value > .30)
n -.44 D score .94 (Decided Group) (Undecided Group)
Low CONFUS High CONFUS High UNCERT Low UNCERT Low CLASIC High CLASIC Low EXTRNL High EXTRNL 208 Replication 3. Case C University (n=222f
Hypothesis: The group centroids (means) for the two groups are equal on 4 sub-scales of CDS and 4 sub-scales of TSOSS.
Ho: Group centroids for the two groups are equal.
As can be seen in the Table 97, the decision regarding the statistical
hypothesis was rejected by Wilks’ lambda = .76; Chi square = 58.9;p < .01.
Thus, the decision is that, undecided and decided groups differ significantly on the
discriminant function for the case C university; that is, it is unlikely that the
groups have the same centroids (groups centroids) on the discriminant function.
Therefore, the test indicates that the discriminant function is statistically
significant for case C university.
To describe the power of the statistically significant discriminant function eigenvalue and canonical correlation coefficient statistics were used to assess and
describe the discriminating power of the discriminant function. For the case C
university, the eigenvalue for this function was .31, and the canonical correlation coefficient was .49. As can be seen in the Table 94, the classification of results indicates that for case C university, 74% of the cases were correctly classified, which indicates the accuracy of the discriminant function on independent variable sets. 209 Table 95
Classification Results for Case C University
No. of Predicted Group Actual Group Cases Undecided Decided
Undecided 70 50 20 71.4% 28.6%
Decided 152 37 115 24.3% 75.7%
Percent of "grouped" cases correctly classified: 74 %
To determine what variables are important in describing how the groups differ, standardized discriminant function coefficients (discriminate weights) and structure coefficients (discriminant loadings) were used. For case C university, the value of standardized canonical discriminant function coefficients represents
CONFS (b = .83), which represents feelings of confusion, discouragement, and lack of experience or information about the making of career decisions. CONFUS is the primary variable to be considered as an important contributor to discriminate among the undecided and decided groups. Thus, the most highly distinguishing
(discriminating) attributes of Group I (undecided) when compared with Group II
(decided) are that females who are in the undecided group tend to lack more experience or information, express more feelings of confusion and discouragement about the making of career decisions compared to the decided group.
The values of the structure coefficients indicate that CONFUS (s=. 882),
EXTRNL(s=.499), UNCERT(s=-.42), CLASIC (s=.397), and VERBAL (s=- 210 .389) are variables that are closely related to the discriminant function. It can be concluded that undecided females were more likely to have confusion, conflicts and external barriers as compared to decided females. However, the decided group females were more likely to need support which represents uncertainty about how to proceed in making decisions. The decided group females were also more likely to have higher scores on verbal/interpersonal related skills on the
TSOSS compared to the undecided females. 211 Table 96
Mean and Standard Deviations for Discriminating Variables for Case C University (n=222)
Cl CONFUS UNCERT CLASIC EXTRNL
1* M 8.72857 7.61429 8.27143 9.50000 SD 2.32143 1.92838 2.21275 2.33282
2* M 6.15789 8.56579 7.11842 8.00000 SD 2.47398 1.86523 2.50280 2.57600.
Cl VERBAL SCIENCE STRENGTH AESTHE
1* M 53.17143 48.12857 49.25714 50.18571 SD 8.32835 9.14286 9.85207 9.73012
2* M 57.01974 51.09868 51.55921 51.88158 SD 8.17966 10.02299 10.35063 11.33239
Cl : Career Decision Status
1 * — > Undecided females (n= 70) 2* — > Decided females (n=152) 212
Table 97
Summary Data for Discriminant Analysis for Case C University
(n = 222 )
Variables b s Group Centroids
CONFUS .83459 .88213 Undecided .822 UNCERT -.39058 -.42036 Decided -.379 CLASIC .14302 .39755 EXTRNL -.09372 .49928 VERBAL -.39296 -.38962 SCIENCE .20530 -.25358 STRENGTH .04417 -.188Q3 AESTHE .02467 -.13012
Eicrenvalue RC Wilks’ Lambda . JB .31 .49 .76 <.001 b = Standardized Canonical Discriminant Function Coefficients s = Structure Coefficient Rc= Canonical Correlation Coefficient
Group Centroids and Discriminating Variables( s value > .301
-.38 D score .82 (Decided Group) (Undecided Group)
Low CONFUS High CONFUS Low EXTRNL High EXTRNL High UNCERT Low UNCERT Low CLASSIC High CLASSIC High VERBAL Low VERBAL 213 Replication 4. Case D University (n=240)
Hypothesis: The group centroids (means) for the two groups are equal on 4 sub-scales of CDS and 4 sub-scales of TSOSS.
Ho: Group centroids for the two groups are equal.
As can be seen in the Table 100, the decision regarding the statistical hypothesis was rejected by Wilks’ lambda = .70;Chi square =82.4;p < .01. Thus, the conclusion is that, undecided and decided groups differ significantly on the discriminant function for the case D university; that is, it is unlikely that the groups have the same centroids on the discriminant function. Therefore, the test indicates that the discriminant function is statistically significant for case D university.
To describe the power of the statistically significant discriminant function eigenvalue and canonical correlation coefficient statistics were used to assess and describe the discriminating power of the discriminant function. For the case D university, the eigenvalue for this function was .42, and the canonical correlation coefficient was .54. As can be seen in the Table 97, the classification of results indicates that for case D university, 73% of the cases were correctly classified, which indicates the accuracy of discriminant function on independent variables set. 214 Table 98
Classification Results for Case D University
No. of Predicted Group Actual Group Cases Undecided Decided
Undecided 96 67 29 69.8% 30.2%
Decided 144 36 108 25.0% 75.0%
Percent of "grouped" cases correctly classified: 72.92%
In order to determine, what variables are important in describing how the groups differ, standardized discriminant function coefficients (discriminate
weights) and structure coefficients (discriminant loadings) were used. For case D university, CONFS (b = .661), which is first sub-scales of CDS, diffusion which represents feelings of confusion, discouragement and lack of experience or information about the making of career decisions, is the variable to be considered as an important contributor to discriminate the undecided and decided groups.
Thus, the most highly distinguishing (discriminating) attributes of Group I
(undecided) when compared with Group II (decided) are that females who are in
the undecided group tend lack more of experience or information, express more feelings of confusion and discouragement about the making of career decisions.
The values of structure coefficients of CONFUS(=.80), EXTRNL(=.54),
CLASIC(=.39), UNCERT (= -.33), and VERBAL (-.30) are variables that are closely related to the discriminant function. Therefore, it can be concluded that 215 undecided females were more likely to have confusion, external barriers, and conflicts, but they were less likely to have verbal/interpersonal related skills on
the TSOSS. However, decided group females were more likely to need additional
support and show uncertainty about their choice. 216 Table 99
Mean and Standard Deviation for Discriminating Variables for Case D University(n =240)
Cl CONFUS UNCERT CLASIC EXTRNL
1* M 8.77083 7.67708 8.33333 8.66667 SD 2.27795 2.00785 2.06559 2.29263
2* M 6.38889 8.52083 7.15278 7.09028 SD 2.21915 1.86603 2.41881 2.09866
Cl VERBAL SCIENCE STRENGTH AESTHE
1* M 51.87500 48.94792 47.11458 50.61458 SD 8.39831 10.97688 10.51002 9.89776
2* M 55.08333 52.56250 49.43750 49.56250 SD 7.94166 9.24622 10.44179 9.90565
Cl : Career Decision Status
1 * — > Undecided females (n= 96) 2* — > Decided females (n=144) 217 Table 100
Summary Data for Discriminant Analysis for Case D University
Variables Group Centroids
CONFUS .66108 .80412 Undecided .792 UNCERT -.46391 -.33206 Decided -.528 CLASIC .06946 .39129 EXTRNL .27404 .54797 VERBAL -.18753 -.2989 SCIENCE -.16572 -.27442 STRENGTH -.06466 -.16800 AESTHE .30627 .08044
Eicrenvalue Rc Wilks' Lambda P .42 .54 .70 <.001
b = Standardized Canonical Discriminant Function Coefficients s = Structure Coefficient Rc= Canonical Correlation Coefficient
Group Centroids and Discriminating Variables ( s value >.30)
-.52 D score ,79 (Undecided Group) (Decided Group)
Low CONFUS High CONFUS Low EXTRNL Low UNCERT Low CLASIC High CLASIC High UNCERT High EXTRNL High VERBAL Low VERBAL 218
The Fifth Question and Results
Question: Are there any discriminant variables to distinguish both decided and undecided groups on the suggested variables such as age, grade level college area, educational and job plan after completion of college, attitude regarding family and career, socio economic status(SES). parent’s educational background, additional enrichment educational activity, location where most education was completed, home town, previous work experience, and grade point average(GPA) ?
Hypothesis: The group centroids (means) for the two groups are equal on the selected variables.
Total Responsefn = 1223)
As shown in the summary Table 103, Wilks’ lambda, used to test the
hypothesis, is .91, which indicates progressively less discrimination. But the statistical
hypothesis was rejected at alpha .05. This means that the discriminant function is
statistically significant. Thus, the results indicate that undecided and decided groups
differ significantly on the discriminant function; that is, it is unlikely that the groups
have the same means (groups centroids) on the discriminant function.
In order to describe the power of this statistically significant
discriminant function, eigenvalue and canonical correlation coefficient statistics were
used. The eigenvalue for this function was .0983. The canonical correlation
coefficient was .2991. Both statistics indicates that the discriminant function is not
very powerful. However, as can be seen in the Table 101, the classification of results indicates that 68% of the cases were correctly classified. Thus, 68% can be used as an index of the effectiveness of the discriminant function on the independent variable sets, so that 68% of the cases can be classified by the discriminating function. However, note that only 21% of the undecided group was correctly classified and 92% of the decided group was correctly classified.
Table 101
Classification Results for Total Cases
No. of Predicted Group Actual Group Cases Undecided Decided
Undecided Group 406 84 322 20.7% 79.3%
Decided'Group 817 66 751 8.1% 91.9%
Percent of "grouped" cases correctly classified: 68.27;
To identify which variables are important in describing how the groups differ standardized discriminant function coefficients (discriminate weights) and structure coefficients (discriminant loadings) were used. As can be seen in the summary Table 103, the values of standardized discriminant function coefficients, ofHUMTY (b=.49)and SCIENG (b=.52) describe the relative importance of the discriminating variables that contribute most to determining scores on the discriminant function. In addition, the positive coefficient indicates that females on decided group tend to be enrolled in the Humanities and the Natural
Science/Engineering colleges. 220 The value of structure coefficients indicate that AGE (s=-.30),
EPLANl(s=-.28), and SESF4(s=.26) are variables that are closely related to the function. It can be concluded that undecided females are more likely to be younger in age, not pursuing graduate school as a further educational plan: decided females are more likely to be older in age and pursuing graduate school.
However, females in the undecided group are more likely to be at a middle level of social economic standing compared to decided group. 221
Table 102
Mean and Standard Deviation for Discriminating Variables for Total Cases (n=1223)
Number of Cases Cl U nw eighted Weighted Label U ndecided (1) 406 4 0 6.0 D ecided (2) 817 8 1 7.0 T o ta l 1223 1223.0
G roup m eans
Cl AGE FRESH SOPH JR 1 20.28818 .36453 .21675 .25862 2 20.63158 .31701 .18605 .23745 Cl HUMTY SOCSCI SCIENG EXTYES 1 .30788 .14039 .34483 .74631 2 .25581 .16769 .27540 .80539 Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1 1.74138 .35714 .10345 .50000 2 1.88739 .44920 .07834 .46389 Cl EPLAN4 JPLAN1 JPLAN2 JPLAN3 1 .08621 .50246 .16502 .41626 2 .04896 .49327 .12607 .36965 Cl JPLAN4 JPLAN5 ATTYES ATTNO 1 .02217 .06650 .81773 .05419 2 .01714 .07834 .82130 .06854 Cl SESF1 SESF2 SESF3 SESF4 1 .01970 .00000 .02956 .01970 2 .02203 .00979 .04284 .00367 Cl SESF5 SESF6 SESF7 FLH 1 .04187 .26108 .22906 .09606 2 .04651 .28519 .23501 .07344 Cl FHS F2YR FCOLL4 FADVDEG 1 .30049 .00739 .41379 .14532 2 .33660 .02203 .39535 .12485 Cl MLH MHS M2YR MCOLL4 1 .19951 .44335 .02463 .28818 2 .20073 .48103 .01469 .24113 Cl MADVDEG LTOWN LURBAN HTOWN 1 .01232 .18966 .75369 .18227 2 .01836 .18482 .72827 .19584 Cl HURBAN GPAA GPAB GPAC 1 .67980 .10591 .71429 .12808 2 .64259 .14321 .72093 .08935 222
(Continued Table 102)
Cl CAREERO CAREER1 CAREER2 CAREER3 1 .17241 .54187 .20443 .07143 2 .20073 .47980 .22032 .07956
Ge o u d Standard Deviations
Cl AGE FRESH SOPH JR 1 1.59574 .48189 .41254 .43842 2 1.77334 .46560 .38938 .42578 Cl HUMTY SOCSCI SCIENG EXTYES 1 .46219 .34782 .47590 .43566 2 .43659 .37382 .44699 .39615 Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1 1.64343 .47975 .30492 .50062 2 1.68020 .49772 .26886 .49900
Cl EPLAN4 JPLAN1 JPLAN2 JPLAN3 1 .28102 .50061 .37166 .49355 2 .21592 .50026 .33213 .48300 Cl JPLAN4 JPLAN5 ATTYES ATTNO 1 .14741 .24947 .38654 .22667 ■2 .12986 .26886 .38334 .25283 Cl SESF1 SESF2 SESF3 SESF4 1 .13915 .00000 .16957 .13915 2 .14688 .09853 .20262 .06052 Cl SESF5 SESF6 SESF7 FLH 1 .20054 .43977 .42075 .29504 2 .21072 .45178 .42426 .26102 Cl FHS F2YR FCOLL4 FADVDEG 1 .45904 .08575 .49312 .35286 2 .47284 .14688 .48922 .33075 C l MLH MHS M2YR MCOLL4 1 .40012 .49739 .15519 .45347 2 .40080 .49995 .12037 .42803 C l MADVDEG LTOWN LURBAN HTOWN 1 .11042 .39251 .43139 .38654 2 .13433 .38839 .44512 .39709 C l HURBAN GPAA GPAB GPAC 1 .46713 .30810 .45231 .33459 2 .47953 .35050 .44882 .28542 C l CAREERO CAREERl CAREER2 CAREER3 1 .37821 .49886 .40378 .25786 2 .40080 .49990 .41472 .27078 223
Table 103
Summary Data for Discriminant Analysis for Total Case Cn = 12231
Discriminant Function I V a r ia b le s b s Group Centroids
AGE -.1 0 1 5 8 AGE -.3 0 0 7 7 U ndecided .4 4 FRESH .29192 EPLAN1 -.2 8 1 4 0 D ecid ed ■.22 SOPH .35626 SESF4 .25591 JR .28483 EPLAN4 .23383 HUMTY .48690 SCIENG .22851 SOCSCI .20056 EXTYES -.2 1 6 8 0 SC1ENG .52448 GPAC .19239 EXTYES -.2 3 8 8 5 CAREER1 .18679 EXTED2 .06369 SESF2 -.1 8 2 7 6 EPLAN1 -.3 0 0 8 2 HUMTY .17581 EPLAN2 .08677 F2YR -.1 6 9 5 5 EPLAN3 .03501 JPLAN2 .16938 EPLAN4 .19020 GPAA -.1 6 6 3 6 JPLAN1 .13267 MCOLL4 .16200 JPLAN2 .29325 FRESH .15165 JPLAN3 .30881 JPLAN3 .14403 JPLAN4 .16077 EPLAN2 .13420 JPLAN5 -.05811 EXTED2 -.13159 ATTYES -.0 5 7 9 0 FLH .12467 ATTNO -.1 4 8 9 8 HURBAN .11765 SESF1 -.07322 SOPH .11620 SESF2 -.22515 FHS -.1 1 5 9 0 SESF3 -.14331 MHS -.11349 SESF4 .24018 M2YR .11244 SESF5 -.0 7 8 2 2 SOCSCI -.1 1 2 2 9 SESF6 -.09560 EPLAN3 .10867 SESF7 -.0 5 5 3 9 CAREERO -.1 0 8 2 1 FLH .20676 SESF3 -.1 0 3 8 5 FHS .00787 FADVDEG .09100 F2YR -.16884 ATTNO -.08829 FCOLL4 .07374 LURBAN .08673 FADVDEG .16380 SESF6 -.0 8 0 9 2 MLH .14354 JR .07400 MHS .18983 MADVDEG -.0 7 1 6 1 M2YR .15450 JPLAN5 -.0 6 7 7 5 MCOLL4 .32615 CAREER2 -.0 5 8 0 9 MADVDEG .00854 FCOLL4 .05653 LTOWN .23262 JPLAN4 .05565 LURBAN .20688 HTOWN -.05184 HTOWN -.0 6 9 5 3 CAREER3 -.0 4 5 8 6 HURBAN -.0 2 4 6 6 SESF5 -.0 3 3 6 3 GPAA -.3 7 1 3 4 JPLAN1 .02763 GPAB -.3 2 6 5 1 SESF1 -.0 2 4 2 4 GPAC -.0 8 1 7 1 GPAB -.02220 CAREERO .12475 SESF7 -.02111 CAREER1 .40052 LTOWN .01864 CAREER2 .25024 ATTYES -.0 1 3 9 4 CAREER3 .14579 MLH -.0 0 4 6 1
Eigenvalue Rc Lambda E . 0983 .30 .91 <.001 224 (Continued Table 103)
Group Centroids and Discriminating Variables (b>.50 and s >.30) for Total Case (n=12231
-.22 s c o r e .44 Decided Group Undecided Group NOT Humanities College Humanities College NOT Natural Science/Engineering College Natural Science/Eng. O ld e r Age Younger Age Pursue Graduate School Not Pursue Graduate School NOT SESF4 SESF4 (Middle Level of SES)
(* is based on b value >.50) 225 Replication 1. Case A University (n=223)
Hypothesis: The group centroids (means) for the two groups are equal on above selected variables.
The summary Table 104 shows that the value of Wilks’ lambda is .72, which indicates progressively more discrimination compared to the total cases.
However, the statistical hypothesis was failed to be rejected at an alpha .05. This means that the discriminant function for case A university is statistically not significant. Thus, the results indicate that undecided and decided groups for case
A university do not differ significantly on the discriminant function; that is, it is likely that the groups have the same means (groups centroids) on the discriminant function.
Therefore, the conclusion is that the independent variable set does not discriminate between the undecided and decided females for this case. 226
Table 104
Summary Data for Discriminant Analysis for Case A University (n=223)
Discriminant Function I V a r ia b le s b V a ria b le s S Group Centroids
AGE -.0 5 7 6 4 EPLAN1 -.3 3 3 0 4 U ndecided .86668 FRESH .34570 • EPLAN4 .28231 D ecid ed -.4 3 9 2 0 SOPH . .04264 FRESH .25274 JR .26496 AGE -.2 1 1 8 3 HUMTY .49596 HTOWN -.2 0 9 8 0 SOCSCI .00477 MCOLL4 .19837 SCIENG .49107 HUMTY .17549 EXTYES .12719 ATTNO -.1 7 1 9 8 EXTED2 -.1 0 6 8 1 JPLAN1 .16492 EPLAN1 -.2 5 7 0 7 HURBAN .16479 EPLAN2 -.0 0 4 4 7 SOCSCI -.1 5 7 5 8 EPLAN3 .09335 SESF2 -.1 5 5 9 6 EPLAN4 .37608 M2YR .15281 JPLAN1 .39677 ATTYES .14353 JPLAN2 .29478 SESF4 .13219 JPLAN3 .33923 LURBAN .12590 JPLAN4 .25284 SOPH -.1 2 2 7 8 JPLAN5 .02396 SESF6 -.12113 ATTYES .22075 SCIENG .11610 ATTNO -.2 3 6 2 0 FCOLL4 .11610 SESF1 .03391 GPAC .11432 SESF2 -.2 2 5 8 6 F2YR -.1 0 9 5 2 SESF3 -.0 7 3 1 6 MLH -.1 0 8 7 0 SESF4 .22044 FADVDEG -.0 9 2 6 9 SESF5 .00953 EXTYES -.0 8 3 9 2 SESF6 -.17574 GPAA -.0 7 8 5 8 SESF7 .04002 CAREERO .07602 FLH -.24735 SESF1 .07581 FHS -.7 1 5 1 0 MADVDEG -.0 7 0 5 8 F2YR -.2 4 5 7 0 SESF3 -.0 7 0 3 7 FCOLL4 -.7 5 1 7 8 CAREER3 .06373 FADVDEG —. 6 5 4 2 6 LTOWN .05791 MLH .73300 EXTED2 -.0 5 2 8 2 MHS 1.13857 CAREER1 -.0 5 0 1 7 M2YR .28476 FHS -.0 5 0 0 8 MCOLL4 1.15247 SESF7 -.0 4 2 6 3 MADVDEG .49703 EPLAN3 .04081 LTOWN .65245 EPLAN2 -.0 3 7 0 4 LURBAN .54827 JPLAN5 -.0 2 2 6 2 HTOWN -.3 8 7 3 1 MHS -.0 1 9 9 4 HURBAN -.1 2 6 6 6 GPAB .01853 GPAA -.3 3 6 5 4 JR -.0 1 8 1 1 GPAB -.3 8 9 6 6 CAREER2 -.0 1 7 6 9 GPAC -.1 3 7 9 2 JPLAN2 .01220 CAREERO .23718 FLH .00963 CAREER1 .29741 JPLAN3 -.0 0 7 9 4 CAREER2 .31325 SESF5 -.0 0 2 7 4 CAREER3 .16514 JPLAN4 -.0 0 1 7 0
Eigenvalue Rc Wilk's lambda P .3841 .5268 .7225 .0606 227
Table 105
Mean and Standard Deviation for Discriminating Variables for Case A University (n = 2 2 3 )
Number of iCases
Cl Unweighted Weighted Label
U ndecided (1) 75 7 5 .0 Decided (2) 148 1 48.0
T o ta l 223 2 23.0
GrouD means Cl AGE FRESH SOPH JR 1 20.44000 .32000 .21333 .33333 2 20.95270 .18243 .28378 .34459 Cl HUMTY SOCSCI SCIENG EXTYES 1 .54667 .04000 .34667 .68000 2 .43243 .09459 .27703 .72900 Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1 1.54667 .22667 .09333 .52000 2 1.66216 .43243 .10811 .49324 Cl EPLAN4 JPLAN1 JPLAN2 JPLAN3 ' 1 .14667 .54667 .12000 .37333 2 .04730 .43919 .11486 .37838 Cl JPLAN4 JPLAN5 ATTYES ATTNO 1 .02667 .06667 .84000 .02667 2 .02703 .07432 .76351 .08108
Cl SESF1 SESF2 SESF3 SESF4 1 .05333 .00000 .04000 .02667 2 .03378 .02703 .06081 .00676
Cl SESF5 SESF6 SESF7 FLH 1 .06667 .22667 .20000 .16000 2 .06757 .29730 .22297 ' .15541 c l FHS F2YR FCOLL4 FADVDEG 1 .36000 .00000 .34667 .06667 2 .39189 .01351 .27703 .10135 MLH MHS M2YR MCOLL4 1 .24000 .46667 .01333 .24000 2 .30405 .47973 .00000 .14189 Cl MADVDEG LTOWN LURBAN HTOWN 1 .01333 .22667 .70667 .08000 2 .02703 .19595 .62838 .17568 228 (Continued Table 105)
ci HURBAN GPAA GPAB GPAC 1 .65333 .20000 .62667 .13333 2 .54730 .24324 .61486 .08784 Cl CAREERO CAREER1 CAREER2 CAREER3 .20000 .41333 .25333 .12000 .16216 .44595 .26351 .09459 Group Standard Deviations
Cl AGE FRESH SOPH JR 1 1.50889 .46962 .41242 .47458 2 2.00453 .38751 .45236 .47685
Cl HUMTY SOCSCI SCIENG EXTYES 1 .50117 .19728 .47911 .46962 2 .49710 .29365 .44905 .44561 Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1.61323 .42149 .29286 .50296 1.70455 .49710 .31157 .50165
Cl EPLAN4 JPLAN1 JPLAN2 JPLAN3 1 .35616 .50117 .32715 .48695 2 .21299 .49797 .31994 .48663
Cl JPLAN4 JPLAN5 ATTYES ATTNO 1 .16219 .25112 .36907 .16219 2 .16271 .26319 .42637 .27389 Cl SESF1 SESF2 SESF3 SESF4 1 .22621 .00000 .19728 .16219 2 .18129 .16271 .23979 .08220
Cl SESF5 SESF6 SESF7 FLH 1 .25112 .42149 .40269 .36907 2 .25185 .45862 .41765 .36352
Cl FHS F2YR FCOLL4 FADVDEG 1 .48323 .00000 .47911 .25112 2 .48983 .11585 .44905 .30282
Cl MLH MHS M2YR MCOLL4 1 .42996 .50225 .11547 .42996 2 .46157 .50129 .00000 .35012 C l MADVDEG LTOWN LURBAN HTOWN 1 .11547 .42149 .45836 .27312 2 .16271 .39827 .48488 .38184
C l HURBAN GPAA GPAB GPAC 1 .47911 .40269 .48695 .34222 2 .49945 .43050 .48828 .28402
C l CAREERO CAREER1 CAREER2 CAREER3 1 .40269 .49575 .43785 .32715 2 .36985 .49876 .44203 .29365 229
Replication 2. Case B Universitv(n=500)
Hypothesis: The group centroids (means) for the two groups are equal on the selected variables.
The summary Table 107 shows that the value of Wilks’ lambda is .84, which is 84% of the variance in the discriminate scores not explained by differences between groups. However the statistical hypothesis was rejected at the alpha .05 (p=.0008), which means that the discriminant function for case B university is statistically significant. Thus, the results indicate that undecided and decided differ significantly on the discriminant score; that is, it is unlikely that the groups have the same centroids on the discriminant function.
In this case, the eigenvalue for this function is .20, and the canonical correlation coefficient is .41. Both values are less than these for case A (n=233) university, which means that the discriminant power of this function was not greater than case A university’s, yet it is statistically significant due to the sample size from this university (n=500). In addition the Table of classification results
(Table 106) indicates that 74% of the cases were correctly classified, a result similar to that for case A university.
The values of the standardized discriminant function coefficients show that the following variables contributed most to determine scores on the discriminant function : Standing sophomore (SOPH;b = .51), being the Natural
Science/Engineering College (SCIENG;b=.57), pursuing graduate school 230 (EPLANl;b = -.52) in the educational plan after graduation, having a high-school level for father’s educational background (FHS; b=66), having a 4 year college or university level for father’s education (FCOLL4; b = .860), having an advanced degree for father’s education (FADVDEG; b=.71),and location where most education was completed of females (LTOWN; b=-.67,LURBAN; b=-68).
Table 106
Classification Results for Case B University
No. of Predicted Grouo Actual Group Cases Undecided Decided
Undecided 156 56 100 35.9% 64.1%
Decided 344 32 312 9.3% 90.7%
Percent of •'grouped" cases correctly classified: 73.60%
The positive coefficient direction is the direction more descriptive of the group having the higher mean on the discriminant scores, while negative direction is the direction more descriptive of the group whose mean discriminant score is lower. Therefore, from an examination of the standardized discriminant function coefficients, it can be concluded that the most highlydistinguishing attributes of the undecided group, when compared to the decided group are that females in the undecided proup tend to be sophomores, enrolled in the Natural
Science/Engineering College, have more than a high school level for father’s 231 educational level, do not intend to pursue graduate school, and have completed most of their education in a farm area: females in the decided group tend to display the opposite characteristics.
The value of the structure coefficients indicates that intent to pursue graduate studies (EPLAN1; s = -.39) is the only variable closely related to the function. It can be concluded that undecided females are less likely to pursue graduate studies as their educational plan after graduation, as compared to the decided group. Table 107
Summary Data for Discriminant Analysis for Case B University (n=500)
Discriminant Function I V a r ia b le s b Variables s Group Centroids
AGE .01247 EPLAN1 -.3 8 8 9 4 UNDECIDED .65735 FRESH .28816 EPLAN4 .24147 DECIDED -.2 9 8 1 0 SOPH .50570 GPAC .23728 JR -.0 5 4 0 6 SOPH .22076 HUMTY .40554 SCIENG .21657 SOCSCI .08605 AGE -.1 7 1 9 4 SCIENG .57150 CAREER1 .16356 EXTYES -.3 6 5 9 4 JR -.1 6 1 6 4 EXTED2 .32725 SESF4 .15020 EPLAN1 -.5 1 7 6 7 F2YR -.1 4 7 6 9 EPLAN2 .09574 ATTNO -.14189 EPLAN3 -.0 7 8 3 3 EXTYES -.1 4 1 4 3 EPLAN4 .11202 GPAB -.1 3 3 5 0 JPLAN1 .09728 JPLAN3 .13134 JPLAN2 .06912 PLAN2 .12460 JPLAN3 .44695 CAREERO -.1 2 1 6 6 JPLAN4 .11209 CAREER3 -.11882 JPLAN5 .02069 EPLAN3 .11587 ATTYES -.0 7 0 2 8 FRESH .10350 ATTNO -.2 2 6 9 4 HUMTY .09203 SESF1 .06899 FADVDEG .07876 SESF2 -.1 9 4 0 0 HTOWN .07605 SESF3 -.01258 SESF6 -.07134 SESF4 .18527 SESF2 -.0 6 8 0 0 SESF5 -.1 3 6 0 5 SESF1 .06762 SESF6 .01602 JPLAN4 .06762 SESF7 .04713 M2YR -.0 6 4 6 1 FLH .25457 GPAA -.06359 FHS .65856 SESF7 .06124 F2YR .07513 FCOLL4 .05243 FCOLL4 .86094 SOCSCI -.0 4 4 9 9 FADVDEG .70569 SSSF5 -.0 4 1 0 1 MLH .28779 MLH .03946 MHS .17715 JPLAN2 -.0 2 8 2 0 M2YR -.0 6 2 7 7 MCOLL4 .02714 MCOLL4 .06248 LURBAN -.02714 MADVDEG .03442 JPLAN5 -.0 2 7 1 2 LTOWN -.66692 FLH .02480 LURBAN -.6 7 9 4 6 LTOWN .02461 HTOWN .34003 MHS .02322 HURBAN .28131 JPLAN1 .02058 GPAA -.2 4 8 4 9 SESF3 .02006 GPAB -.27385 CAREER2 .01183 GPAC .08483 MADVDEG .01144 CAREERO .04889 ATTYES .00817 CAREER1 .31553 HURBAN .00620 CAREER2 .19645 EXTED2 -.0 0 4 6 6 CAREER3 .01905 FHS -.0 0 3 4 8
Eigenvalue Rc Wilk' s lam bda p • 1967 .4055 8356 .0008 233 (Continued Table 107)
Group Centroids and Discriminating Variables for Case B University (b>.50 and s>.3CH
- .3 0 D s c o re .66 Decided Group Undecided Group
NOT Sophomore Sophomore NOT Natural Sci/Eng Natural Sci/Eng College Pursue Graduate School NOT Pursue Graduate S ch o o l Father's Ed Level for NOT FHS High-School NOT FCOLL4 4 y e a r C o lle g e /U n v . NOT FADVDEG Advanced Degree Location completed most of Ed. LTOWN NOT LTOWN LURBAN NOT LURBAN
(* is based on s value > .30) 234
Table 108
Mean and Standard Deviation for Discriminating Variables for Case B University (0= 500 ) Number of Cases Cl Unweighted . Weighted Label Undecided (1) 156 156.0 Decided (2) 344 344.0
T o ta l 500 5 00.0
Group Means Cl AGE FRESH SOPH JR 1 20.10897 .53205 .14103 .14103 2 20.39535 .48256 .07849 .20058 C l HUMTY SOCSCI SCIENG EXTYES 1 .21795 .16667 .45513 .75000 2 .18314 .18314 .35465 . .80523 Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1 1.89103 .32051 .10897 .50000 2 1.89826 .50291 .07558 .44477 Cl EPLAN4 JPLAN1 JPLAN2 JPLAN3 1 .08974 .48077 .12179 .44231 2 .03779 .47093 .13081 .38081
Cl JPLAN4 JPLAN5 ATTYES ATTNO 1 .01923 .05769 .81410 .05128 2 .01163 .06395 .81105 .08721 Cl SESF1 SESF2 SESF3 SESF4 1 .01923 .00000 .03846 .00641 2 .01163 .00291 .03488 .00000 Cl SESF5 SESF6 SESF7 FLH 1 .03846 .26282 .27564 .05769 2 .04651 .29360 .25000 .05233 Cl FHS F2YR FCOLL4 FADVDEG 1 .32692 .00641 .45513 .14744 2 .32849 .02616 .43023 .12209 c l MLH MHS M2YR MCOLL4 1 .18590 .52564 .01923 .24359 2 .17151 .51453 .02907 .23256 Cl MADVDEG LTOWN LURBAN HTOWN 1 .01282 .19231 .75641 .21795 2 .01163 .18314 .76744 .18895
Cl HURBAN GPAA GPAB GPAC 1 .68590 .07051 .75000 .15385 2 .68314 .08721 .80233 .08430 235 (Continued Table 108)
Cl CAREERO CAREER1 CAREER2 CAREER3 1 .21154 .52564 .20513 .04487 2 .26163 .44767 .20058 .07267 GrOUTD S ta n d a rd D e v ia tio n s
Cl AGE FRESH SOPH JR 1 1.70227 .50058 .34917 .34917 2 1.76145 .50042 .26933 .40102 Cl HUMTY SOCSCX SCIENG EXTYES 1 .41418 .37388 .49959 .43441 2 .38734 .38734 .47910 .39660
Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1 1.73233 .46818 .31261 .50161 2 1.57369 .50072 .26471 .49766
Cl EPLAN4 JPLAN1 JPLAN2 JPLAN3 1 .28673 .50124 .32810 .49826 2 .19097 .49988 .33769 .48629
Cl JPLAN4 JPLAN5 ATTYES ATTNO 1 .13778 .23391 .39028 .22128 2 .10736 .24503 .39204 .28255 Cl ( SESF1 SESF2 SESF3 SESF4 1 .13778 .00000 .19293 .08006 2 .10736 .05392 .18375 .00000
Cl SESF5 SESF6 SESF7 FLH 1 .19293 .44158 .44828 .23391 2 .21090 .45608 .43364 .22301 Cl FHS F2YR FCOLL4 FADVDEG 1 .47060 .08006 .49959 .35568 2 .47035 .15985 .49583 .32787
Cl MLH MHS M2YR MCOLL4 1 .39028 .50095 .13778 .43063 2 .37750 .50052 .16825 .42308
Cl MADVDEG LTOWN LURBAN HTOWN 1 .11286 .39538 .43063 .41418 2 .10736 .38734 .42308 .39204
Cl HURBAN GPAA GPAB GPAC 1 .46565 .25683 .43441 .36196 2 .46593 .28255 .39883 .27825
Cl CAREERO CAREER1 CAREER2 CAREER3 1 .40971 .50095 .40510 .20769 2 .44016 .49798 .40102 .25998 236 Replication 3. Case C University (n=24ff)
Hypothesis: The group centroids (means) for the two groups are equal on above selected variables.
The summary Table 109 shows that the value of Wilks’ lambda is .77, which means that 77 % of variance in the discriminate scores is not explained by differences between the groups. Testing the significance of Wilks’ lambda failed to be rejected at the alpha .05 (p=.18). This meas that the discriminant function of case C university (n=246) is not statistically significant. Thus, the results indicate that undecided and decided do not differ significantly on the discriminant score; that is, it is likely that the groups have the same centroids on the discriminant function.
Therefore, the conclusion is that the independent variable set does not discriminate between undecided and decided students for this case. Table 109
Summary Data for Discriminant Analysis for Case C University (n=246)
Discriminant Function I Variables b Variables s Group Centroids
AGE -.1 8 3 0 6 JPLAN2 .32509 UNDECIDED .80218 FRESH -.0 4 3 5 3 MHS -.26352 DECIDED -.36549 SOPH .20461 FLH .25990 JR .24509 GPAA -.2 3 1 6 9 HUMTY .25444 SESF4 .22449 SOCSCI .35750 EXTYES -.20736 SCIENG .27881 F2YR -.19743 EXTYES -.4 0 5 2 2 JPLAN1 -.1 9 2 5 9 EXTED2 .25045 SOCSCI .16813 EPLAN1 -.09255 SESF3 -.16647 EPLAN2 -.1 1 9 4 5 FHS -.1 6 4 6 5 EPLAN3 -.2 6 0 7 8 M2YR .16462 EPLAN4 -.0 5 0 1 5 LURBAN .16022 JPLAN1 -.12488 MCOLL4 .15744 JPLAN2 .42500 EPLAN3 -.12945 JPLAN3 .13964 GPAB .12904 JPLAN4 -.0 7 3 5 5 AGE -.1 2 2 8 6 JPLAN5 -.0 1 7 0 1 SESF1 -.1 1 5 7 3 ATTYES .10521 HURBAN .11379 ATTNO .06449 SESF2 -.1 1 2 6 1 SESF1 -.07345 SCIENG .10805 SESF2 .00029 FADVDEG -.1 0 6 7 5 SESF3 .09578 SOPH .10201 SESF4 .25489 JR .09576 SESF5 .11466 JPLAN3 .09348 SESF6 .27070 MLH .08852 SESF7 -.03508 SESF6 .08852 FLH -.3 4 5 2 0 ATTNO .08803 FHS -1.02932 GPAC .07996 F2YR -.49636 FCOLL4 .07596 FCOLL4 -.84542 CAREER1 .07259 FADVDEG -.6 4 5 8 0 JPLAN5 -.0 4 4 0 5 MLH .03142 HTOWN -.0 4 0 8 4 MHS -.0 5 4 5 8 SESF7 -.0 4 0 8 4 M2YR .31983 HUMTY -.0 3 8 3 1 MCOLL4 .23321 EXTED2 -.0 3 3 0 7 MADVDEG .12461 MADVDEG -.0 3 2 1 4 LTOWN .74876 EPLAN2 -.0 2 7 8 4 LURBAN .83083 CAREERO -.02622 HTOWN -.2 3 1 5 3 CAREER2 -.0 2 6 0 6 HURBAN -.16018 ATTYES .02274 GPAA -.2 9 3 2 3 SESF5 .02054 GPAB -.05917 EPLAN4 -.02042 GPAC -.0 8 8 2 6 JPLAN4 -.0 1 8 5 3 CAREERO .04654 EPLAN1 .00875 CAREER1 .41353 LTOWN -.0 0 8 3 4 CAREER2 .27133 CAREER3 -.0 0 7 2 5 CAREER3 .19627 FRESH -.00099
E ig e n v a lu e Rc Wilk’s Lambda p .2956 .4777 .7718 .1758 238
Table 110
Mean and Standard Deviation for Discriminating Variables for Case C University rn=246)
Number o f C ases
C l U nw eighted Weighted Label Undecided (1) 77 77.0 Decided (2) 169 169.0
T o ta l 246 2 4 6.0
Group means Cl AGE FRESH SOPH JR 1 20.68831 .19481 .36364 .27273 2 20.94083 .19527 .30769 .22485
C l HUMTY SOCSCI SCIENG EXTYES 1 .29870 .20779 .20779 .64935 2 .31953 .13609 .15976 .75740
Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1 1.48052 .38961 .09091 .41558 2 1.53846 .38462 .10059 .49112 Cl EPLAN4 JPLAN1 JPLAN2 JPLAN3 1 .06494 .40260 .27273 .38961 2 .07101 .51479 .13018 .33728 Cl JPLAN4 JPLAN5 ATTYES ATTNO 1 .02597 .09091 .84416 .07792 2 .02959 .10651 .83432 .05325 C l SESF1 SESF2 SESF3 SESF4 1 .01299 .00000 .02597 .03896 2 .03550 .01183 .07101 .00592
Cl SESF5 SESF6 SESF7 FLH 1 .06494 .33766 .23377 .19481 2 .05917 .28994 .25444 .09467 Cl FHS F2YR FCOLL4 FADVDEG 1 .33766 .00000 .32468 .06494 2 .43195 .03550 .28402 .10059
Cl MLH MHS M2YR MCOLL4 1 .33766 .35065 .03896 .18182 2 .28994 .50296 .01183 .11834
Cl MADVDEG LTOWN LURBAN HTOWN 1 .01299 .22078 .72727 .23377 2 .01775 .22485 .63905 .25444 239 (Continued Table 110)
Cl HURBAN GPAA GPAB GPAC 1 .61039 .09091 .72727 .12987 2 .54438 .18935 .65680 .10059 Cl CAREERO CAREER1 CAREER2 CAREER3 1 .14286 .54545 .19481 .10390 2 .15385 .50296 .20710 .10651
Grouo Standard Deviations
Cl AGE FRESH SOPH JR 1 1.58330 .39865 .48420 .44828 2 1.83454 .39758 .46291 .41873
Cl HUMTY SOCSCI SCIENG EXTYES 1 .46069 .40839 .40839 .48030 2 .46768 .34391 .36748 .42993 Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1 1.66710 .49086 .28936 .49605 2 1.41842 .48795 .30168 .50141
Cl EPLAN4 JPLAN1 JPLAN2 JPLAN3 1 .24803 . .49364 .44828 .49086 2 .25760 .50127 .33750 .47419
Cl JPLAN4 JPLAN5 ATTYES ATTNO 1 .16010 .28936 .36509 .26981 2 .16995 .30940 .37290 .22521 C l SESF1 SESF2 SESF3 SESF4 1 .11396 .00000 .16010 .19477 2 .18560 .10846 .25760 .07692 C l SESF5 SESF6 SESF7 FLH 1 .24803 .47601 .42600 .39865 2 .23665 .45508 .43684 .29363 Cl FHS F2YR FCOLL4 FADVDEG 1 .47601 .00000 .47132 .24803 2 .49682 .18560 .45229 .30168
Cl MLH MHS M2YR MCOLL4 1 .47601 .48030 .19477 .38822 2 .45508 .50148 .1 0846 . .32397 Cl MADVDEG LTOWN LURBAN HTOWN 1 .11396 .41749 .44828 .42600 2 .13244 .41873 .48170 .43684 Cl HURBAN GPAA GPAB GPAC 1 .49086 .28936 .44828 .33836 2 .49951 .39295 .47619 .30168 Cl CAREERO CAREER1 CAREER2 CAREER3 1 .35222 .50119 .39865 .30713 2 .36187 .50148 .40643 .30940 240 Replication 4. Case D University (n=254)
Hypothesis: The group centroids (means) for the two groups are equal on above selected variables.
As can be seen in the summary Table 111, the value of Wilks’ lambda is
.78, which means 78% of the variance in the discriminate scores is not explained by differences between groups. Testing the significance of wilks’ lambda failed to reject at the alpha .05(p=.20). This means that the discriminant function of case
D university (n=254) is not statistically significant. Thus, the results indicate that undecided and decided do not differ significantly on the discriminant score; that is, it is likely that the groups have the same means (groups centroids) on the discriminant function.
Therefore, the conclusion is that the independent variable set does not discriminate between undecided and decided females for this case. 241
Table 111
Summary Data for Discriminant Analysis for Case D University(n=254)
Discriminant Function I V a r ia b le s b Variables s Group Centroids AGE -.0 1 1 4 7 EPLAN2 .31311 UNDECIDED .65340 FRESH .34578 EXTED2 -.31109 DECIDED -.41047 SOPH .36975 JR .30753 JR .48174 SOCSCI -.2 7 5 0 0 HUMTY .07826 M2YR .26785 SOCSCI -.1 7 4 5 0 ATTYES -.2 5 9 1 7 SCIENG .11107 AGE -.2 4 8 1 3 EXTYES -.1 3 8 3 3 HUMTY .21969 EXTED2 -.0 9 0 4 2 CAREER1 .20275 EPLAN1 .14830 EXTYES -.1 9 9 9 7 EPLAN2 .34709 FADVDEG .19741 EPLAN3 .30810 EPLAN3 .19257 EPLAN4 .02301 SESFl -.16729 JPLAN1 .14230 JPLAN2 .16483 JPLAN2 .21675 FCOLL4 -.1 5 3 9 6 JPLAN3 .13163 CAREER2 -.1 4 6 2 1 JPLAN4 -.0 2 6 7 7 EPLAN1 .13813 JPLAN5 -.0 4 0 0 3 ATTNO .13452 ATTYES -.2 6 4 1 5 JPLAN4 .12155 ATTNO -.10838 F2YR .12155 SESFl -.19201 SESF4 .12155 SESF2 -.12342 FRESH .11944 SESF3 -.15789 CAREERO -.1 1 3 2 0 SESF4 .06838 SCIENG .11168 SESF5 -.05186 JPLAN3 .10898 SESF6 -.3 1 2 0 9 MADVDEG -.1 0 4 1 9 SESF7 -.1 4 9 8 2 SESF2 -.0 9 5 9 6 FLH .40429 JPLAN5 -.0 7 8 7 5 FHS .80589 FLH .07021 F2YR .39640 LURBAN -.0 6 6 7 7 FCOLL4 .83756 SOPH .05941 FADVDEG 1.02425 GPAA -.05818 MLH -.3 9 5 7 6 SESF6 -.0 5 5 9 4 MHS -.4 1 0 6 7 SESF7 -.0 4 5 4 3 M2YR .08479 MLH .03958 MCOLL4 -.4 7 2 1 4 FHS -.0 3 5 7 0 MADVDEG -.3 8 8 6 9 MHS -.0 3 3 3 5 LTOWN -.3 8 4 3 7 GPAC -.02700 LURBAN -.52520 SESF3 -.02268 HTOWN .19507 SESF5 -.0 2 2 6 8 HURBAN .19271 JPLAN1 .02106 GPAA -.0 8 4 1 1 MCOLL4 -.0 1 4 2 2 GPAB .00836 HURBAN -.0 1 1 0 0 GPAC -.1 4 2 8 7 HTOWN -.0 0 8 5 7 CAREERO -.1 8 7 9 5 GPAB .00568 CAREER1 -.00575 LTOWN -.00540 CAREER2 -.1 6 7 5 5 CAREER3 -.0 0 1 1 1 E ig e n v a lu e Rc W ilk 's Lambda p .27 .4613 .7872 .2061 242 Table 112
Mean and Standard Deviations for Discriminating Variables for Case D University (n=254)
Number of Cases
Cl Unweighted Weighted Label Undecided (1) 98 98.0 Decided (2) 156 156.0 Total 254 254.0
G roup means C l AGE FRESH SOPH JR 1 20.14286 .26531 .22449 .37755 2 20.51282 .21154 .19872 .23077
C l HUMTY SOCSCI SCIENG EXTYES 1 .27551 .12245 .27551 .86735 2 .17949 .23718 .22436 .92949
Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1 1.85714 .48980 .11224 .55102 2 2.45513 .41667 .03205 .44872
Cl EPLAN4 JPLAN1 JPLAN2 JPLAN3 1 .05102 .58163 .18367 .42857 2 .05128 .57051 .12179 .37179 JPLAN4 JPLAN5 ATTYES ATTNO 1 .02041 .06122 .78571 .06122 2 .00641 .08333 .88462 .03205
C l SESFl SESF2 SESF3 SESF4 1 .00000 .00000 .01020 .02041 2 .01923 .00641 .01282 .00641 Cl SESF5 SESF6 SESF7 FLH 1 .01020 .22449 .17347 .03061 2 .01282 .25000 .19231 .01923 C l FHS F2YR FCOLL4 FADVDEG 1 .18367 .02041 .46939 .26531 2 .19872 .00641 .55128 .17949
C l MLH MHS M2YR MCOLL4 1 .08163 .36735 .03061 .47959 2 .07051 .38462 .00000 .48718 C l MADVDEG LTOWN LURBAN HTOWN 1 .01020 .13265 .80612 .16327 2 .02564 .13462 .83333 .16667 243 (Continued Table 112)
Cl HURBAN GPAA GPAB GPAC 1 .74490 .10204 .71429 .08163 2 .75000 .12179 .71154 .08974 Cl CAREERO . CAREER1 CAREER2 CAREER3 1 .11224 .66327 .17347 .05102 2 .15385 .55769 .23718 .05128 G roup S ta n d a rd D e v ia tio n s
Cl AGE FRESH SOPH JR 1 1.44308 .44377 .41939 .48727 2 1.37483 .40971 .40032 .42268 Cl HUMTY SOCSCI SCIENG EXTYES 1 .44907 .32949 .44907 .34094 2 .38500 .42672 .41850 .25683
Cl EXTED2 EPLAN1 EPLAN2 EPLAN3 1 1.47837 .50247 .31729 .49995 2 1.98492 .49459 .17670 .49897 C l EPLAN4 JPLAN1 JPLAN2 JPLAN3 1 .22117 .49583 .38921 .49742 2 .22128 .49660 .32810 .48484 Cl JPLAN4 JPLAN5 ATTYES ATTNO 1 .14212 .24097 .41244 .24097 2 .08006 .27728 .32051 .17670 Cl SESFl SESF2 SESF3 SESF4 1 .00000 .00000 .10102 .14212 2 .13778 .08006 .11286 .08006 Cl SESF5 SESF6 SESF7 FLH 1 .10102 .41939 .38060 .17315 2 .11286 .43441 .39538 .13778 Cl FHS F2YR FCOLL4 FADVDEG 1 .38921 .14212 .50163 .44377 2 .40032 .08006 .49897 .38500 C l MLH MHS M2YR MCOLL4 1 .27521 .48456 .17315 .50215 2 .25683 .48807 .00000 .50145 Cl MADVDEG LTOWN LURBAN HTOWN 1 .10102 .34094 .39737 .37151 2 .15857 .34241 .37388 .37388 Cl HURBAN GPAA GPAB GPAC 1 .43816 .30426 .45408 .27521 2 .43441 .32810 .45451 .28673 CAREERO CAREER1 CAREER2 CAREER3 Cl 1 .31729 .47502 .38060 .22117 2 .36196 .49826 .42672 .22128 CHAPTER V
SUMMARY, DISCUSSION AND RECOMMENDATIONS
This study was designed to investigate the utility of Bandura’s self-efficacy theory in relationship to the understanding and treatment of the career indecision of female students in Korea. Focuses were upon investigating relationships between career indecision and self-efficacy expectations as measured at the task-specific level.
The major purposes of the study were to 1) investigate the relationship between women’s career indecision and self-efficacy expectations, and other selected background characteristics such as age, grade level, major field, social economic status (SES), parent’s educational background, additional enrichment educational activities, previous work experience, location where most education was completed, hometown, educational and career plan after completion of college and attitude regarding planning family and its influence on their career plans, and 2) to describe the differentiation between decided and undecided students on the suggested variables.
244 More specifically, this study was conducted to answer the following research
1. Are the four sub-scales of CDS related to the four sub-scales of
TSOSS ?
2. Are the four sub-scales of CDS related to age, grade level, college
area, educational and job plan after completion of college, attitude
regarding family and career, socio economic status(SES), parent’s
educational background, additional enrichment educational activity,
location where most education was completed, home town, previous
work experience, and grade point average(GPA) ?
3. Are the four sub-scales of TSOSS related to age, grade level, college
area, educational and job plan after completion of college, attitude
regarding family and career, socio economic status(SES), parent’s
educational background, additional enrichment educational activity,
location where most education was completed, home town, previous
work experience, and grade point average(GPA) ?
4. Are the four sub-scales of CDS and the four sub-scales of TSOSS
related to the students’ decided or undecided career decision status ?
If the groups differ, what variables are important in discriminating the
two groups ?
5. Are there any discriminant variables to distinguish both decided and
undecided groups on the suggested variables such as age, grade level, 246 college area, educational and job plan after completion of college,
attitude regarding family and career, socio economic status(SES),
parent’s educational background, additional enrichment educational
activity, location where most education was completed, home town,
previous work experience, and grade point average(GPA) ? If the
groups differ, what variables are important in discriminating the two
groups ?
Limitations of the Study
This study was relational in nature. A relational study cannot establish cause and effect relationships between variables (Miller, 1986). Therefore, the investigator sought only to explain and predict relationships among variables.
This study was subject to the limitations associated with self-reported data, such as question asking grade point average, since obtaining the grade point average of each student from the academic records of the four universities was too cumbersome.
In addition, the data utilized in this study were obtained from different university environments such as co-ed universities and single sex universities, and different level of universities such as competive and non-competitive universities in terms of admission. These dissimilarities of school environment may result in some bias in the findings. Therefore, further study is needed to investigate the differentiation of those four universities. Sample.
This study was repeated four times with female students from two women’s
universities and two co-ed universities in Seoul. The total subjects were 1,275 female
undergraduate students enrolled in 1993 Autumn semester at four universities in
Seoul, the capital of Korea.
Instrument
1. The translated Korean version of the Career Decision Scale (CDS) - 3rd
Rev. (Osipow, Carney, Winer, Yanico, & Koschier, 1976) was used to measure dependent variables.
2. The translated Korean version of the Task Specific Occupational Self-
Efficacy Scales developed by Osipow and Rooney (1989) based on Bandura’s theoretical framework and the work characteristics of the occupations in the
Dictionary of Occupational Titles (DOT. 1977) was used to measure independent variables.
3. A short form of background characteristics questionnaire was used to measure extraneous variables.
Variables
A. Independent Variables (4 sub-scales of TSOSS)
1. Verbal/interpersonal related skills 2. Quantitative, scientific and business related skills 3. Physical strength and agility related skills 4. Aesthetic skills
B. Dependent Variables (4 sub-scales of CDS, and Career Status)
1. Confusion 248
2. Uncertainty 3. Classical approach-approach conflict 4. External barriers 5. Career status : undecided or decided
C. Extraneous Variables were age, grade level, college area, educational and job plan after completion of college, attitude regarding family and career, socio economic status(SES), parent’s educational background, additional enrichment educational activity, location where most education was completed, home town, previous work experience and grade point average(GPA).
Data Analysis
The Statistical Package for the Social Sciences was used to analyze the data.
(SPSSx-Users Guide, 1985). Frequencies, mean and standard deviation were used to describe the background characteristics of respondents. , In order to meet research questions 1, 2 and 3, canonical correlation analysis was performed.
Discriminant analysis was also used for questions 4 and 5.
In addition, due to the many dummy variables, one technical strategy for deciding what background variables to include in the independent variable set for this canonical correlation was recommended by Dr. Warmbrod. The strategy was to run the multiple regression on each of the dependent variables in order to decide which of the independent variables are important (significant). These were 8 regressions of 4 sub-scales of CDS and 4 sub-scales of TSOSS with selected background characteristics. Only the sets of dummy variables that were demonstrated to be important by the regression analysis were included in the canonical correlation.
When using multiple regression, regression assumptions regarding multicollinearity and residuals were checked. There was not evidence that the assumptions were violated. 249 Summary of Findings
A. Relationship between the four sub-scales of CDS and the four sub-scales of TSOSS.
Statistical Hypothesis : There is no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients(Rc 2) are equal to zero.
Results of five analyses indicate that there is a statistically significant relationship between the four sub-scales of CDS and the four sub-scales of TSOSS.
The nature of the relationship is summarized in the Table 113. 250
Table 113
Summarized Relationship between 4 sub-scales of CDS and TSOSS
Relationship between 4 sub-scales of CDS and TSOSS CDS TSOSS Total Case Higher Lower (n = 1135) CONFUS All 4 sub-scales of TSOSS Case A Univ Higher Lower (n=204) CONFUS, EXTRNL All 4 sub-scales of TSOSS Case B Univ Higher Lower (n=476) * CONFUS, UNCERT All 4 sub-scales of TSOSS Case C Univ Higher Lower (n=233) CONFUS, EXTRNL 1,2, and 3 sub-scales
Case D Univ Higher Lower (n=204) CONFUS 1,2, and 3 sub-scales
1. For the combined cases (n=1135), females who had higher confusion (first sub-scale of CDS) in making a career decision typically had lower scores on the four sub-scales of TSOSS, which are verbal/interpersonal, quantitative scientific/business, physical strength and agility, and aesthetics related skills; females who had lower confusion in making a career decision typically had higher scores on the four sub scales of the TSOSS.
2. For case A university (n=204), females who had higer confusion and greater external barriers (first and fourth sub-scales of the CDS) in making a career 251 decision typically had lower scores on the 4 sub-scales of the TSOSS; females who had less confusion and external barriers in making their career decision typically had higher scores on overall four sub-scales of the TSOSS.
3. For case B university (n=476), females who had higher confusion and uncertainty (first and second sub-scales of CDS) in making a career decision typically had the lower on the four sub-scales of the TSOSS; females who had lower confusion and uncertainty (first and second sub-scales of the CDS) in making a decision typically had the higher scores on four sub-scales of the TSOSS.
4. For case C university (n=233), females who had higher confusion and greater external barriers (first and fourth sub-scales of CDS) in making a career decision ranked lowest scores on verbal/interpersonal, quantitative scientific/business, and physical strength related skills (first, second, and third sub scales on TSOSS); females who had lower confusion and smaller external barriers in making a career decision ranked higher scores on first, second and third sub-scales on the TSOSS.
5. For case D university (n=204), females who had higher confusion ( first sub-scale of CDS) in making a career decision ranked lowest scores on verbal/interpersonal, quantitative scientific/business, and physical strength related skills (first, second, and third sub-scales of TSOSS); females who had lower confusion in making a career decision typically ranked higher in scores on verbal/interpersonal, quantitative scientific/business, and physical strength related skills on the TSOSS. 252
B. Relationship between the foursub-scales ofCDS and the selected variables
Statistical Hypothesis : There is no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients(Rc 2) are equal to zero.
The decision for statistical hypothesis (Table 114) and the nature of relationship between (Table 115) the four sub-scales of CDS and selected background characteristics are as follows.
Table 114
Decision Regarding Ho
Decision Regarding Ho Total Case Reject Ho, But not (n=1232) meaningful due to Rc2 <.10 Case A Univ Reject Ho (n=229) Case B Univ Reject Ho (n=498) Case C Univ Reject Ho (n=251) Case D Univ Fail to Reject Ho (n=251) 253 Table 115
Summarized Relationship between four sub-scales of CDS and Selected Variables
Relationship between 4 sub-scales of CDS and Selected Variables CDS Selected Variables Total Case No meaningful Relationship (n=1232) Case A Univ High CONFUS, CLASIC, Freshman, No plan to (n=229) EXTRNL go to graduate school, Full/part time employment plan Case B Univ High UNCERT,but Lower Freshman, Plan to go to (n=498) CONFUS and EXTRNL graduate school and to get a job after graduation, Not the middle level on SES Case C Univ High CONFUS CLASIC Soc/Sci College, No (n=251) EXTRNL, but Lower plan for further ed, but UNCERT Plan for work
Not Humanities College High CONFUS UNCERT and Juniors, No plan to CLASIC EXTRNL work, Marriage Plan Case D Univ No Relationship (n=251)
1. For the combined cases (n=1232), even though canonical roots 1 and 2 were found to be statistically significant, no practical explanation of the variance was gained. Possibly the results were significant only because of the large sample size for the combined cases (n = 12321.
2. For case A university (n=229), females who were freshman, who had no 254
plan to go to graduate school, who wished to have full time/part time employment
after graduation typically had higher confusion, conflict, and felt greater external
barriers in making a career decision; females who were not freshman, who wished
to pursue graduate study, and who did not have a job plan after graduation had
lower confusion, conflict and felt less external barriers in making career decision.
3. For case B university (n=498), females who were freshman, had plans to
go to graduate school, had plans to get a job after graduation, and did not place in
the middle level on SES typically had lower confusion, felt lower external barriers in
making a career decision, but they had higher scores on uncertainty : females who
were not freshman, wished to pursue graduate study, who did not want to further
their education, and who fell in the middle level on SES typically had higher
confusion and external barriers, but they less uncertainty in making a career decision.
4. For case C university (n=251), females who were pursuing graduate school
after completion of college, and had no job plan, and were not enrolled in the Social
Science College typically had lower scores on confusion, conflict and external
barriers, but higher scores on uncertainty; females who were enrolled in the Social
science college, had no plan for further education, but had a plan for work after
graduation, typically had higher scores on confusion, conflict and external barriers,
but lower scores on uncertainty.
In addition, females who were enrolled in Humanities College and were juniors, who had full-time employment or any job plan after graduation rather than
marriage typically had lower scores on overall four sub-scales on CDS; females who 255 were not enrolled in the Humanities College and were not juniors, who had no plan to work and had marriage plans typically had higher scores on confusion, uncertainty, conflict, and external barriers.
5. For case D University (n=251), no statistical hypothesis was rejected among the canonical roots. Thus the conclusion is that the 4 sub-scales of CDS and selected variables have no relationship for this case.
C. Relationship between the four sub-scales of TSOSS and the selected variables.
Statistical Hypothesis : There is no relationships between the dependent (criterion) and the independent (predictor) variable sets.
Ho : All squared canonical correlation coefficients(Rc 2) are equal to zero.
The nature of relationship between (Table 116) the four sub-scales of TSOSS and selected background characteristics are as follows. 256
Table 116
Summarized Relationship between the 4 sub-scales of TSOSS and selected variables
Relationship between the 4 sub-scales of TSOSS and selected variables
TSOSS Selected variables
Total Case High SCIENCE Science/Engineering College (n=1168) Low AESTHE Science/Engineering but Not High SCIENCE, STRENGTH, Humanities College, High- AESTHE school education for mother
Case A Univ High SCIENCE Science/Engineering (n=208) Low AESTHE Less than a high school ed. for mother
Humanities, but not Sci/Eng High VERBAL College, Less than a high school Low SCIENCE, AESTHE ed. for mother
Case B Univ High SCIENCE Science/Engineering (n=476) Low AESTHE
High VERBAL Humanities College Low AESTHE
Case C Univ High STRENTH, AESTHE Not Humanities, Not enriching (n=233) ed. experiences, Negative attitude regarding family and career, More than high school ed. for mother
Science/EngineeringLower no. of enriching ed. experience, High VERBAL, SCIENCE, Positive attitude regarding AESTHE family and career, 4 year of college ed. for mother, GPA of "A"
Case D Univ High SCIENCE Science/Engineering (n=251) Social/Science, Not Humanities
Humanities, Social/Science, Not High VERBAL Sci/Eng. Not 2 yr vocational ed. Low AESTHE for mother 257
1. For the combined cases (n = 1168), females who were enrolled in the
Natural Science and Engineering College typically had higher scores on science related skills, but lower scores on aesthetic skills on the TSOSS; females who were not enrolled in the Natural Science and Engineering College typically had lower scores on science related skills, but higher scores on aesthetic skills on the TSOSS.
In addition, females who were enrolled in the Natural Science and
Engineering College, not enrolled in the Humanities College, and females whose mothers had a high school education typically had higher scores on science, physical strength, and aesthetic related skills on the TSOSS; females who were enrolled in the
Humanities College and not enrolled in the Natural Science and Engineering
College, and females whose mothers had less than a high school education typically had lower scores on science, physical strength, and aesthetic related skills on the
TSOSS.
2. For case A university (n=208), females who were enrolled in the Natural
Science and Engineering College, and females whose mothers had less than a high school education had higher scores on science, but lower on Aesthetic related skills on the TSOSS;females who were not enrolled in the Natural Science and Engineering
College, and females whose mothers had more than a high school education had lower scores on science, but higher on aesthetic related skills on the TSOSS.
In addition females who were enrolled in the Natural Science and
Engineering College and not enrolled in the Humanities College, and females whose mothers had more than a high school education, and females were urban area home 258 town typically had higher scores on science and aesthetic, but lower scores on verbal related skills on TSOSS; females who were enrolled in the Humanities College and not enrolled in the Natural Science and Engineering College, and females whose mothers had less than high school education typically had lower scores on science and aesthetic, but higher scores on verbal related skills on the TSOSS.
3. For case B university (n= 476), females who were enrolled in the Natural
Science and Engineering College typically had higher scores on science but lower scores on aesthetic related skills on the TSOSS; females who were not enrolled in
Natural Science and Engineering College typically had lower scores on science but higher aesthetic related skills on the TSOSS. In addition, females who were enrolled in the Humanities College typically had higher scores on verbal but lower scores on aesthetic related skills on the TSOSS; females who were not enrolled in the
Humanities College typically had lower scores on verbal but higher scores on aesthetic related skills on the TSOSS.
4. For case C university (n=233), females who were enrolled in the
Humanities College, and did not have additional enriching educational experiences, and who had a positive attitude regarding family and career, and females whose mothers had less than high-school education typically had lower scores on physical strength and aesthetic related skills on the TSOSS; females who were not enrolled in the Humanities College, did not have additional enriching educational experiences, who had a negative attitude regarding family and career, and females whose mothers had more than high school education typically had higher scores on physical strength 259 and aesthetic related skills on the TSOSS. In addition, females who were enrolled in the Natural Science and Engineering College, had a lower number of additional enriching educational experience, had a positive attitude regarding career and family, whose mothers had 4 year of college education, and females who reported a GPA of "A" typically had higher scores on verbal, science, and aesthetic related skills on the TSOSS; females who were not enrolled in the Natural Science and Engineering
College, had large number of additional enriching educational experience, had a negative attitude regarding career and family, whose mothers did not have 4 year of college education, and females who reported a GPA lower than "A" typically had lower scores on verbal, science and aesthetic related skills on the TSOSS.
5. For case D university (n=251), females who were enrolled in the
Humanities College and not enrolled in the Social Science and the Natural Science
Engineering College typically had lower scores on science related skills on the
TSOSS; females who were enrolled in the Social Science and the Natural Science and Engineering Colleges, but not enrolled in the Humanities College typically had higher scores on science related skills on the TSOSS. In addition, females who were enrolled in the Natural Science and Engineering College, not enrolled in the
Humanities and the Social Science Colleges, and females whose mothers had 2 year vocational education typically had lower scores on verbal, but higher scores on aesthetic related skills on the TSOSS ; females who were enrolled in the Humanities and the Social Science Colleges, and not enrolled in the Natural Science and
Engineering, and females whose mothers had NOT 2 year vocational/technical 260 education typically had higher scores on verbal, but lower scores on aesthetic related skills on the TSOSS.
D. Differences between groups of undecided and decided with four sub-scales of CDS and TSOSS.
Results of the five analyses support the conclusion that the four sub-scales of
CDS and TSOSS discriminate the career status (undecided or decided) of Korean women enrolled in the universities. The nature of the relationship (Table 117) is summarized as follows :
Table 117
Summarized Discriminating Variables on the Career Status
Relationship between Career Status and 4 sub-scales of CDS and TSOSS Decided Undecided Total Case L:CONFUS, UNCERT, H:CONFUS, UNCERT, (n=1135) EXTRNL EXTRNL H:CLASSIC, VERBAL L:CLASSIC, VERBAL Case A Univ L:CONFUS, EXTRNL H:CONFUS, EXTRNL (n=204) H;UNCERT,VERBAL, L:UNCERT, VERBAL, STRENGTH, AESTHE STRENGTH, AESTHE Case B Univ L:CONFUS, CLASSIC, H:CONFUS, CLASSIC, (n=469) EXTRNL EXTRNL H:UNCERT L;UNCERT Case C Univ L: CONFUS, H;CONFUS, CLASSIC (n=222) CLASSIC, EXTRNL EXTRNL H:UNCERT, VERBAL L:UNCERT, VERBAL
Case D Univ L: CONFUS, H:CONFUS, CLASSIC, (n=240) CLASSIC, EXTRNL EXTRNL H:UNCERT, VERBAL L:UNCERT, VERBAL 261
1. For the combined cases (n=1135), undecided females were more likely to
have confusion, uncertainty and external barriers, but less likely to have classic factor,
which represents a classical approach conflict compared to decided females. In
addition undecided females tended to have lower scores on verbal/interpersonal
related skills on the TSOSS compared to decided females.
2. For case A university (n=204), undecided females were more likely to have
confusion and external barriers, as compared to decided females, who were more
likely to have uncertainty. In addition undecided females were less likely to
demonstrate verbal/interpersonal, physical strength, and aesthetic related skills on
the TSOSS as compared to decided females.
3. For case B university (n=469), undecided females were more likely to be
confused, in conflict and feeling external barriers to career choice as compared to the
decided group. In addition decided females were more likely to be uncertain and
need more support in making decisions compared to undecided females.
Interestingly, for case B university overall self-efficacy expectations failed to
discriminate both groups.
4. For case C university (n=222), undecided females were more likely to have confusion, conflicts and external barriers compared to decided females. However,
the decided group females were more likely to need support which represent
uncertainty about how to proceed in making decisions, and more likely to have higher scores on verbal/interpersonal related skills on the TSOSS compared to undecided females. 262 5. For case D university (n=240), undecided females were more likely to
have confusion, external barriers, and conflicts, but they were less likely to have
verbal/interpersonal related skills on the TSOSS. However, decided group females
were more likely to need additional support and show uncertainty about their choice.
E. Differences between groups of undecided and decided and with selective variables.
Results of the two analyses support the conclusion that such selective variables
as college area, age, grade level, educational plan, location where most of the
education was completed and father’s educational background discriminate the career
status (undecided or decided) of Korean women enrolled in the universities.
1. For the combined case (n=1223), undecided females were more likely to
be of younger in age, to not pursue graduate school as a further educational plan, and to be enrolled in the Humanities, and the Natural Science/Engineering colleges; decided females were more likely to be of older in age and pursue graduate school as a further educational plan. However, females in the undecided group were more likely to have a middle level of social economic status compared to the decided group.
2. For case A university (n=223), the statistical hypothesis failed to be rejected at the alpha .05. This means that the discriminant function for case A university was statistically not significant. The conclusion is that the independent variable set does not discriminates between the undecided and decided females for this case. 263
3. For case B university (n=500), females in the undecided group tended to be sophomores, enrolled in the Natural Science/Engineering College, have fathers who had more than a college level education, did not intend to pursue graduate school, and had completed most of their education in a farm area as compared to the decided group; females in the decided group tended to score in the opposite range on the above mentioned variables.
4. For case C university (n=246), the statistical hypothesis failed to be rejected at the alpha .05. This means that the discriminant function for case C university was statistically not significant. The conclusion is that the independent variable set does not discriminates between the undecided and decided females for this case.
5. For case D university (n=254), the statistical hypothesis failed to be rejected at the alpha .05. It means that the discriminant function for Case D university was statistically not significant. The conclusion is that the independent variable set does not discriminates between the undecided and decided females for this case.
In summary, from three replication studies, the statistical hypothesis failed to be rejected at the alpha .05 level. The results indicate that the groups, undecided and decided on case A, C, and D universities, do not differ significantly on the discriminant scores; that is, it is likely that the groups have the same centroids
(groups means) on the discriminant function. Therefore, the conclusion is that the independent variable set does not discriminate between undecided and decided females for these three cases.
However, from two studies, some variables were found to discriminate between the two groups as shown in the Table 118. 265
Table 118
Summarized Discriminating Variables on the Career Status
Variables in Discriminating Groups of undecided and decided Decided Undecided Total case NOT Humanities College Humanities Colie (n = 1223) NOT Natur Sci/Eng Coll Natur Sci/Eng Older age Younger age EPLAN1* NOT EPLAN1 NOT SESF4 SESF4 Case A Univ No Discriminating Variables (n=223) Case B Univ NOT Sophomore Sophomore (n=500) NOT Natur Sci/Eng Natur Sci/Eng EPLAN1* NOT EPLAN1 NOT FLH FLH* NOT FCOLL4 FCOLL4* NOT FADVDEG FADVDEG* LTOWN * NOT LTOWN LURBAN* NOT LURBAN Case C Univ No Discriminating Variables (n=246) Case D Univ No Discriminating Variables (n=254)
* EPLAN1 : Pursue graduate study after completion of college. * FLH : Less than high-school level for father’s ed. * FCOLL4 : 4 year College or University level for father’s ed. * FADVDEG: Advanced degree for father’s ed. * LTOWN : Where most of the education was completed in a town. * LURBAN : Where most of the education was completed in an urban setting. 266 Discussion of findings
Originally this research was inspired by the recommendations of vocational
psychologists. They pointed out that research is needed for different cultural
backgrounds to provide empirical data to test the validity of westem-society oriented
theories of career development. According to vocational psychologists, current
theories of career development cannot be adequately applied to women (Holland,
1966; Osipow, 1983; Brown and Brooks, 1990) because most studies are based on
empirical data generated from studies examining men’s careers. They also suggest
that separate theories may be needed for different culture because the results of the
unique socialization due to race and sex have never been adequately considered in
the present theories, which were developed to explain the career development and
occupational choice making of western society.
From the above perspectives, self-efficacy expectation theory measured by the
TSOSS and career indecision measured by the CDS are discussed based on the
findings as follows of a study including women in four college in Korea.
First, in addressing the relationships between the four sub-scales of CDS and
the four sub-scales of TSOSS, the statistically significant results of four replication
studies offer strong support for the research hypothesis that there is a significant relationship between self-efficacy and career indecision for females in Korea.
However, the magnitude of this relationships is limited due to the small amount of variance that is explained. It might be speculated that according to theoretical formulations of Fassinger(1985, 1990), and Farmer (1985), females, 267 specifically about young Korean women enrolled in universities, are influenced by more factors in making career decisions than males so that any single factor might not contribute enough to the decision-making process to yield a relationship of great magnitude.
In addition, the results of cases A and C studies suggest that females enrolled in co-ed universities were more likely to have greater external barriers compared to females on single-sex universities. It can be speculated that females enrolled in co ed universities are more likely to perceive either internal or external barriers to decision making due to experience of competition with males in their school life.
Second. in addressing the relationships between the four sub-scales of CDS and the selective background characteristics, for case A and C studies, females who have more definite employment plans had higher levels of confusion and conflict and felt greater external barriers in making a career decision. On the other hand, for case B university, females who plan to pursue graduate school after completion of college had higher scores on confusion and external barriers. When considering the higher level of sex discrimination in work place in Korea, it can be speculated that pursuing graduate school might be the secure and temporary choice of females when asked about future educational plans after completion of graduation. In addition, for case C university, females who plan to pursue graduate school had a higher score on uncertainty which represents the need for additional support for their initial decisions.
Third, by examing the above relationships between the four sub-scales of CDS, 268 TSOSS, and selected background characteristics, the findings suggest that self-
efficacy expectation measured by the TSOSS can be explained more precisely on
career indecision for females compared to selected variables. In addition, this
finding partly supports Bandura’s Self-Efficacy theory, which explains that career
choice and aspiration are natural extensions of the concept of self-efficacy, and the
self-efficacy is an important variable among the many that contribute to how career
decisions and aspirations unfold in the process of career development.
This finding also partly supports Hackett and Betz (1981) study, which pointed
out that self-efficacy theory could be useful in helping one to understand the career
development of women.
Fourth, in addressing relationships between the four sub-scales of TSOSS and
the selected variables, findings indicate that self-efficacy varies depending upon the college field of enrollment of the respondents. Females who were enrolled in the
Natural Science and Engineering College had higher scores on quantitative, scientific and business related skills on the TSOSS, while females who were enrolled in the
Humanities College had higher scores on verbal and interpersonal related skills on the TSOSS. Therefore these findings partly support Hackett and Betz (1981) theory that self-efficacy expectations determine the range of perceived career and academic options and people’s perseverance and ultimate success in their careers.
Fifth. in addressing differences between the groups of undecided and decided and the four sub-scales of CDS and TSOSS, findings indicate that undecided females were more likely to have higher scores on confusion, uncertain, and external barriers 269 on the CDS, while decided females were more likely to have higher classical approach-approach conflicts when several possible careers were attractive. When considering the unique characteristics of females, it can be speculated that decided females decisions tend to be only temporary;they change over time or based on situations such as marriage or parenthood plans. It was also concluded that the four sub-scales of CDS (diffusion, support, approach, and external barriers) possess relatively good discriminate validity.
In addition undecided females tended to have lower only verbal/interpersonal related skills on the TSOSS compared to other sub-scales of the TSOSS from the replication studies; decided females tended to have higher verbal/interpersonal related skills on the TSOSS. Only the study of case A university found that lower verbal/interpersonal, physical strength and agility, and aesthetic skills were related to undecided status.
These findings can be speculated according to Hackett and Betz (1981) study.
Differential sex role socialization prevents women from gaining equal access to information from which self-efficacy expectations are acquired. This socialization results in women having lowered overall self-efficacy and lower career self-efficacy for traditionally male occupations. Thus they postulate that women, compared to men, possess lower and weaker career-related efficacy expectations, and that these differences help explain women’s vocational behavior.
Sixth, in addressing differences between the groups of undecided and decided and the selected background variables, findings fail to support that females’ career 270 decision status was related to many selected variables. However, two of the studies
indicated that females who were younger in age and enrolled in the Natural
Science/Engineering college were an undecided group.
Findings of above relationships also partly support western oriented research
findings in career indecision. These findings were conflicting and confusing and
provided no empirical basis to permit the reliable differentiation of decided and
undecided students on the basis of personal, social, or academic characteristics
(Harman, 1973; Gordon, 1984; Lunneborg, 1975; Holland & Holland, 1977; Slaney,
1988).
Lastly, since this study used translated instruments, several questions were raised based upon Osipow’s observations (Osipow, 1994). First is the question: was the translation accurate and appropriate ? Second: can all the concepts be translated linguistically from culture to culture ? Third: will several translators agree that the translation is similar and adequate ? In order to satisfy all above questions, the researcher tried to take certain steps in the translation process as has been explained in an earlier chapter. However, it is possible that some bias has resulted from the translation of the instruments.
In summary, the study is a significant attempt toward integration of a large body of research literature and provides a comprehensive exploratory study of the relationships among the many variables predictive of females’ career indecision from
Korean women enrolled in university. Generally these findings agree with the theory and research conducted in western societies. 271 from Korean females enrolled in universities. It contributes to advancement of western oriented theoretical understanding of the process of career development in women.
Recommendations
A. Recommendations for Counselors at Higher Education in Korea
Originally, the reason for using four sub-scales of CDS as well as TSOSS in this study was to look at differentiating between different types of undecided females as well as between undecided and decided females in order to provide certain counseling information regarding the different reasons for indecision.
According to the findings, there are different sub-types of undecided students.
Therefore, a counselor is recommended to proceed differently with students who are undecided because they are confused and lack information about occupations (high score on first sub-scale of CDS) than with those who are undecided because several occupations have great appeal for them (high score on third sub-scale of CDS).
Similarly, females who need support and reassurance for a tentative decision (higher score on the second sub-scale of the CDS) will be counseled differently than those who cannot reach a decision because they perceive either internal or external barriers to decision making (higher score on external barriers). In addition, from the theoretical perspectives, low expectations of self-efficacy would lead to avoidance of those tasks and behaviors and consequently, continued indecision. The implications of these findings for the counseling of career undecided individuals might be extremely important by suggesting increasing self-efficacy expectations related to 272 those tasks and behaviors.
B. Recommendations for Education and Business/Industry Leaders in Korea
Especially, an educational policy maker should pay attention to the fact that
females placed on decided group tended to show higher needs for support and
external barriers in making career decisions. Therefore, educational policy makers
are recommended to provide adequate policy for additional consistent support to
implement their career decisions in the college school life. The policy should show
a broadening experience that leads to expanded options. For example, the work
experiences or internship program in the college school year should encourage the
females to practice their intended career direction by providing a broadening
opportunity. In addition business policy makers are recommended to provide equal
opportunity to females in hiring for the world of work after completion of college.
Therefore, the three major parties — government, business/industry and higher
education should work together and jointly participate in the development of higher
educated female manpower.
C. Recommendations for further study
The present relational study between career indecision, self-efficacy expectations, and selected variables is the first attempt to test western oriented self- efficacy theory in Korea. It also represents a significant attempt toward integration of a large body of western oriented research literature and provides a comprehensive exploratory study of the relationships among the many variables predictive of Korean females’ career indecision. Therefore findings should be interpreted cautiously. 273 Further study is recommended with integration to refinement and replication by testing different variables such as marital or parental status. At the same time sex differences research on the relationship between self-efficacy expectations and career indecision is needed in order to explain how male and female approach the career decision process. In addition, the current study used self-reported perceptions of career status by using one item of the CDS. Further study is recommended to use career indecision scale scores to identify the career status according to the CDS manual. This will represent a more accurate measure of career status.
Lastly, comparative research is recommended to determine the extent to which these findings are applicable in other education settings and different cultures. APPENDIX A
Career Decision Scale (CDS)
274 PLEASE NOTE
Copyrighted materials in this document have not been filmed at the request of the author They are available for consultation, however in the author’s university library.
2 7 5 -2 7 6 2 7 8 -2 8 1 2 8 3 -2 8 6
University Microfilms International APPENDIX B
Task-Specific Occupational Self-Efficacy (TSOSS)
277 APPENDIX C
Selected Background Characteristics Questionanire
282 REFERENCES
Appel, V.,Haak, R.A., & Witzke, D.B. (1970). Factors associated with indecision about collegiate major and career choice, 78th Annual Convention. APA. pp. 667 - 668.
Ashby, J.D.,Wall, H.W.,&Osipow, S.H. (1966). Vocational certainty and indecision in college freshmen, 1966. Personnel and Guidance Journal. 44.1037 - 1041.
Bandura, A. (1977). Self-efficacy:Toward a unifying theory of behavioral change. Psychological Review. 84. 191 - 215.
Bandura, A. (1978a). Reflections on self-efficacy. Advances in Behavioral Research and Therapy. 1. 237 - 269.
Bandura, A. (1978b) Self-efficacy : Toward a unifying theory of behavioral change. Advances in Behavioral Research and Therapy. 1. 139 - 161.
Bandura, A. (1986) The explanatory and predictive scope of self-efficacy theory. Journal of Social and Clinical Psychology. 4. 359 - 375.
Betz, N.E. & Hackett, G, (1981). The relationship of career-related self-efficacy expectations to perceived career options in college women and men. Journal of Counseling Psychology. 28. 399 - 410.
Betz, N.E., & Hackett, G. (1983). The relationship of mathematics self-efficacy expectations to the selection of science-based college majors. Journal of Vocational Behavior. 23. 329 - 345.
Betz, N.E., & Hackett, G. (1986). Applications of self-efficacy theory to understanding career choice behavior. Journal of Social and Clinical Psychology. 4. 279 - 289.
Betz, N.E.,& Hackett, G. (1987). Concept of agency in educational and career development. Journal of Counseling Psychology. 34.299 - 308.
287 288 Bores-Rangel, E., Church, A.T.,Szendre, D.,and Reeves, C. (1990). Self-efficacy in relation to occupational consideration and academic performance in high school equivalence students. Journal of Counseling Psychology. 31.407 - 418.
Borgen, F.H. (1986). New approaches to the assessment of interests. In W.B. Walsh & S.H. Osipow (eds.), Advances in Vocational Psychology - Volume 1: The assessment of interests (pp. 83 - 125). Hillsdale, NJ:Lawrence Erlbaum.
Brown, D. & Brooks, L. (1990). Career Choice and Development:Applying Contemporary Theories to Practice. Jossev-Bass Publishers:San Francisco. Second edition.
Branch, L.E. & Lichtenberg, J.W. (1987, August). Self-efficacv and career choice. Paper presented at the annual meeting of the American Psychological Association, New York, NY.
Brachter, W.T. (1982). The Influence of The Family On Career Selection: A family Systems Perspective. Personnel and Guidance Journal. 6 1 .87—91.
Crites, J.O. (1969). Vocational psychology. New York:NcGraw-Hill.
Crites, J.O. (1981). Career counseling : Models. Methods and Materials. New Y ork: McGraw-Hill, 1981.
Dawis, R.V.,& Lofquist, L.H. (1984). A Psychological Theory of Work Adjustment. Minneapolis:University of Minnesota Press.
Frederick G. Lopez & Scott Andrews (1987). Career Indecision: A Family Systems Perspective. Journal of Counseling and Development . Feb, 1987, V. 65.
Fitzgerald, L.F. and Betz, N.E. (1983) "Issues in the Vocational Psychology of Women", in Handbook of Vocational Psychology. Volume 1: Foundations. Walsh. W.B. and Osipow. S.H..eds.
Fuqua, D.R., Seaworth, T.B.,& Newman, J.L. (1988). The relation of state and trait anxiety to different components of career indecision. Journal of Counseling Psychology. 35. 154 - 158.
Goodman, J.,& Waters, E. (1986). Adult career development : Concepts, issues, and practices. Alexandria, Virginia:National Career Development Association. 289 Goodson, D. (1981). Do career development needs exist for all students entering college or just the undecided major students? Journal of College Student Personnel. 22. 413-417.
Goodstein, L.D. (1965). Behavior theoretical views of counseling. In B. Stefflre & W.H. Grant (eds.), Theories of counseling (1st ed.,pp. 156 - 157, 2nd ed., pp. 243-303). New York:McGraw Hill.
Gordon, V.N. (1984). The undecided college student. Springfield, IL:Charles C. Thomas.
Gordon, V.N. (1982). Are undecided students changing ? Vocational Guidance Quarterly. 30.265-271.
Gordon, V.N. & Poison, C. (1985). Students needing academic alternative advising: A national survey. NACADA Journal. 5. 77-84.
Gordon, V.N.,& Kline, D.I. (1989). Ego-identity statuses of undecided and decided students and their perceived advising needs. NACADA journal.5-15.
Hackett, G. & Betz, N.B. (1981). A self-efficacy approach to the career development of women. Journal of Vocational Behavior. 18. 326 - 339.
Hawkins, J.G., Bradley, R.W., & White, G.W. (1977). Anxiety and the process of deciding about a major and vocation. Journal of Counseling Psychology. 24. 398 -403.
Hartman, B.W., & Hartman. P.T. (1982). The Concurrent and Predictive Validity of The Career Decision Scale Administered to High School Students. Psychological Reports. 52. 95 - 100.
Hartman, B.W.,Fuqua, D.R.,&Jenkins, S.J.(1986). The reliability/Generalizability of the Construct of Career Indecision. Journal of Vocational Behavior. 28*142 - 148.
Hartman, B.W.,Fuqua, D.R., Blum, C.R.,& Hartman, P.T. (1985). A Study of The Predictive Validity of The Career Decision Scale in Identifying Longitudinal Patterns of Career Indecision. Journal of Vocational Behavior. 2 8 .142 - 148.
Hartman, B.W.,& Fuqua, D.R., & Blum, C.R. (1985). A Path-analytic Model of Career Indecision. Vocational Guidance Quarterly. 33. 231 -240. 290 Hartman, B.W., Fuqua, D.R., & Jenkins, S.J. (1986). The reliability and generalizability of the construct of career indecision. Journal of Vocational Behavior. 28. 142 - 146.
Hershenson, D., & Roth, R. (1966). A decisional process model of vocational development. Journal of Counseling Psychology. 13. 368 - 370.
Holland, J.L. (1973a). Making Vocational Choices: A Theory of Careers. Englewood Cliffs, N.J.:Prentice-Hall.
Holland, J.L. (1985a). Making Vocational Choices :A Theory of Vocational Personalities and Work Environments. (2nd ed.) Englewood Cliffs, N.J.:Prentice-hall.
Holland, J.L.,& Holland, J.E. (1977). Vocational indecision:More evidence and speculation. Journal of Counseling Psychology. 24. 404 - 414.
Holland, J.L.,Magoon, T.M.,&Spokane, A.R.(1981). Counseling psychology:Career interventions, research and theory. Annual Review of Psychology. 32. 279 - 395.
Hollender, J. (1972). differential parental influences on vocational interest development in adolescent males. Journal of Vocational Behavior. pp. 67 - 76.
Jyung, C.Y. (1989). Predictors of students’ career maturity in central Ohio high schools. Unpublished doctoral dissertation. The Ohio State University.
Korean Educational Development Institute (KDEI), 1992. Educational Indicators in Korea.
Krumboltz, J.D. (1979). A social learning theory of career decision making. In A.M. Mitchell, G.B. Jones and J.D. Krumboltz (eds.) Social Learning and Career Decision Making Cranston. R.I.:Carroll Press, 1979.
Larson,L.M.,Heppner, P.P.,Ham, T.,& Dugan, K. (1988). Investigating Multiple subtypes of Career Indecision Through Cluster Analysis. Journal of Counseling Psychology. 35. 439 - 446.
Layton, P.L. (1984). Self-efficacy, locus of control, career salience and women’s career choice. Unpublished thesis, University of Minnesota, Minneapolis, MN. 291 Lent, R.W., Brown, S.D.,& Larkin, K.C. (1986). Self-efficacy in the predition of academic performance and perceived career options. Journal of Counseling Psvcholog. 33. 265 - 269.
Lent, R.W.,& Hackett, G. (1987). Career Self-efficacy:Empirical Status and Future Directions. Journal of Vocational Behavior. 30. 347 - 382.
Martin, F., Sabourin, S., Laplante, B., & Jean-Claude Coallier(1991). Diffusion, Support, Approach, and External Barriers as Distinct Theoretical Dimensions of The Career Decision Scale : Discontinuing Evidence ? Journal of Vocational Behavior 38. 187 - 197.
Matsui, T. and Tsukamato, S. I. (1991). Relations between career self-efficacy measures based on occupational titles and Holland codes and model environments: A methodological contribution. Journal of Vocational Behavior. 38. 78 - 92.
Mitchell, L.K., & Krumboltz, J.D. (1990). Social Learning Approach to Career Decision Making:Krumboltz’s Theory. In Brown & Brooks (eds), Career Choice and Development: Applying Contemporary Theories to Practice, Jossey- Bass Publisher. Sanfrancisco.
Osipow, S.H. (1968). Theories of career development . New York: Appleton-Century- Crofts.
Osipow, S.H. (1973). Fundamental Ideas of Career Education. In Senech. L. New paths in social science curriculum design. chicago:SRA.
Osipow, S.H., Carney, C.G., Winer, J., Yanico, B., & Koschier, M. (1976). The Career Decision Scale (3rd rev.). Psychological Assessment Resources, Inc., Odessa, Florida.
Osipow, S.H.,Carney, C.G.,& Azy Barak (1976). A scale of Educational-Vocational Undecideness: A Typological Approach. Journal of Vocational Behavior 9. 233 - 243.
Osipow, S.H. 09831. Theories of career development (3rd ed). Englewood Cliffs, NJ:Prentice-Hall.
Osipow, S.H., & Reed, R. (1985). Decision Making Style and Career Indecision in college students, Journal of Vocational Behavior 27. 368 - 373. 292 Osipow, S.H. (1986). Manual for the Career Decision Scale Odessa, FL:Psychological Assessment Resources.
Osipow, S.H .,& Rooney, R.A. (1989). Task Specific Occupational Self-Efficacv Scale. Columbus. OH.
Osipow, S.H., & Rooney, R.A. (1990). Task-specific Occupational Self-efficacy Scales rrSOSl. Columbus. OH: Authors.
Osipow, S.H. (1990). Convergence in Theories of Career Choice and Development: Review and Prospect. Journal of Counseling Psychology. 1990. 122 - 131.
Osipow, S.H.,& Rooney, R.A. (1990). Factor analysis and reliability assessment of the Task Specific Occupational Self-Efficacy Scale. Unpublished raw data.
Osipow, S.H. (1990). Careers:Research and Personal or How I think an Individual’s Personal and Career Life Intertwine:A Personal Example. The Counseling Psychologist. Vol 18. No.2. April. 1990.
Osipow, S.H. (1991). Developing Instruments for Use In Counseling. Journal of Counseling & Development. Nov./Dec. 1991, Vo. 70.
Osipow, S.H., & Rooney, R.A. (1992). Task-Specific Occupational Self-efficacy Scale: The development and validation of a prototype. Journal of Vocational Behavior. 40. pp. 14 -32.
Osipow, S.H.,Temple, R.D., & Rooney, R.A. (1993). The short form of the Task- Specific Occupational Self-Efficacy Scale. Journal of Career Assessment. 1. 13 - 20.
Osipow, S.H. (1994). The Career Decision Scale:How Good Does it Have to be ? Journal of Career Assessment. 2. winter, pp. 15 - 18.
Parsons, F. (1909). Choosing a Vocation. Boston:Houghton Mifflin.
Prestholdt, P.H.,Lane, I.M.,&Matthews, R.C. (1987). Nurse turnover as reasoned action development of a process model. Journal of Applied Psychology. 72. 221 - 227.
Temple, R.D. (1991). The relationship between task-specific occupational self- efficacv and career indecision. Unpublished master’s thesis, The Ohio State University, Columbus, OH. 293 Temple, R.D., & Osipow, S.H. (1994) The relationship between task-specific self- efficacy egalitarianism and career indecision for females. Journal of Career Assessment. 2. winter. 82 - 90.
Rounds, J.B.,& Tracy, T.J. (1990). From trait-and factor to person-environment fit counseling : Theory and process. In W.B. Walsh, & S.H. Osipow (eds.) Career counseling :Contemporary topics in vocational psychology. Hillsdale, NJ:Lawence Erlbaum.
Rooney, R.A. (1990). Task-specific occupational self-efficacy, general occupational self-efficacy and career undecidedness in college males and females. Unpublished master’s thesis, Department of Psychology, The Ohio State University, Columbus, OH.
Rooney, R.A. and Osipow, S.H. (1992). Task-specific occupational self-efficacy : The development and validation of a prototype scale. Journal of Vocational Behavior. 40. 14 - 32.
Savickas, L.,& Jaijoura, D. (1991). The Career Decision Scale as A Type Indicator. Journal of Counseling Psychology. 38 (11.85 -90.
Slaney, R.B. (1988). The assessment of career decision making. In W.B. Walsh & S.H. Osipow (EDs)., Career decision making (pp.33- 76) Hillsddale, NJ: Erlbaum.
Slaney, R.B., Palko-Nonemaker, D., & Alexander, R. (1981). An Investigation of Two Measures of Career Indecision. Journal of Vocational Behavior. 18. 92-103.
Schulenberg, J.E., Vondracek, F.W.,& Hostetler. M. (1989). Short-term Changes in Adolescents’ Work Values : Effects of Career Indecision, Grade Level, and Gender. Paper Presented at The Annual Meeting Of The American Educational Research Association, San Francisco.
Schulenberg, J.E., Vondracek, F.W.,& Crouter, A.C. (1984). The influence of the family on vocational development. Journal of Marriage and the Family. 10. 129 - 143.
Spokane, A.R. (1987). Conceptual and methodological issues in person-environment fit research, a special issue of the Journal of Vocational Behavior. 1987.31. 217 - 361.
Spokane, A.R. (1991). Career intervention. Englewood Cliffs, NJ:Prentice Hall. 294 Super, D.E. (1955). The dimensions and Measurement of Vocational Maturity. Teachers College Record. 57. 151 - 163.
Super, D.E. (1957). The Psychology of Careers : An introduction to vocational development. New York:Harper & Row.
Super, D.E.,& Overstreet, P.L. (1960). The vocational maturity of ninth grade bo vs. NY:Teachers College Press.
Shimizu, K.,Vondracek, F.W.,Schulenberg, J.E.,& Hostetler, M. (1988). The Factor Structure of The Career Decision Scale -.Similarities across Selected Studies. Journal of Vocational Behavior, 32, 213 - 225.
Vondracek, F.W., Hostetler, M., Schulenberg, J.E., & Kazuaki Shimizu (1990). Dimensions of Career Indecision, Journal of Counseling Psychology. 1990. Vol.37. No.l. 98 - 106.
Vondracek, F.W.,Lemer, R.M .,& Schulenberg, J.E. (1986). Career development : A Life-span developmental approach. Hillsdale. NJ:Lawrence Erlbaum.
Titley, R., & Titley, B. (1980). Initial choice of college major: Are only the "undecided" undecided ? Journal of College Student Personnel. 2 1 .293-298.
United States Department of Labor. (1981). Selected Characteristics of Occupations Defined in the Dictionary of Occupational Titles. U.S. Government Printing Office, Washington, D.C.
Vondreacek, F.W.,Hostetler, M., & Schulenberg, J.E. (1989, March). A typology of vocational indecision and implications for counseling. Paper presented at the annual meeting of the American Educational Research Association, San Francisco.
Vondracek, F.W.,Lemer, R.M., & Schulenberg, J.E. (1986). Career development: A life-span developmental approach. Hillsdale, NJ:Erlbaum.
Walsh, W.B. & Osipow, S.H. (eds.) (1983). Handbook of vocational psychology. Volume 1 Foundations. Hillsdale, New Jersey:Lawrence Erlbaum.