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, substandard margins, 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 comer 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 300 North Zeeb Road. Ann Arbor, Ml 48106-1346 USA 313/761-4700 800/521-0600

Order Number 9427762

Examining the relationship of negative affectivity and subjective well-being to goal-setting processes and task performance

Necowitz, Lawrence B., Ph.D.

The Ohio State University, 1994

UMI 300 N. Zeeb Rd. Ann Arbor, MI 48106

EXAMINING THE RELATIONSHIP OF

NEGATIVE AFFECTIVITY AND SUBJECTIVE WELL-BEING

TO GOAL-SETTING PROCESSES AND TASK PERFORMANCE

Dissertation

Presented in Partial Fulfillment of the Requirements for

the degree Doctor of Philosophy in the

Graduate School of The Ohio State University

by

Lawrence B. Necowitz, B.A., M.A.

*****

The Ohio State University

1994

Dissertation Committee: Approved by

Mary Roznowski

James T. Austin

Robert S. Billings

Howard J. Klein Mary Roznowski Advisor Department of Psychology This project is dedicated in memory of my grandparents

Max and Ida Cohen Israel and Betty Necowitz

whom I miss very much

ii ACKNOWLEDGEMENTS

Although this document represents a single research study, it more accurately reflects the culmination of a greater journey. It is certain that I would not have made it to this point without the guidance of Mary Roznowski.

Mary has truly been a terrific adviser. Her office door was always open and she was willing to assist me whenever her help was requested. Yet, she has also allowed me to branch out on my own as a researcher. I am grateful for the time, assistance, and encouragement she has given me, not only only on this project, but throughout my time at Ohio State.

I would also like to thank the members of my committee:

Jim Austin, Bob Billings and Howard Klein whose comments and suggestions have improved the quality of this project.

This effort would have been difficult to complete without the assistance of Pamela Taylor who not only wrote the computer program to specification, but also had to put up with my "time urgent" personality.

While Pam brought technical knowledge to the project, others provided the financial backing to compensate her efforts. I am grateful to Jim Austin who provided money at the beginning of the project to help defray my own out of

iii pocket expenses. Later, a grant from the Sigma Xi research society and an award from the OSU Department of Psychology helped to fund the project.

I would also like to acknowledge the members of the goal setting seminar at Ohio State who helped me

"brainstorm" an appropriate task for this study.

Finally, I would like recognize my family and friends.

To my parents, Norman and Miriam, who instilled in me the value of education, I thank you for your continual support of my academic endeavors. Without your emotional (not to mention financial) support, I do not think you would be reading this today. To my brother, David, I thank you for making me feel "at home" even when I was away, by putting up with my endless telephone conversations about the Flyers,

Phillies, or Eagles. Your diligence in calling meant more to me than I think you realize. To my sister, Beth, I thank you for brightening my day whenever I heard your voice on the phone or received a picture from you in the mail. Also, thanks for understanding my inability to attend the many events that have been important to you over the last six years.

Although I have made numerous friends during my time at

Ohio State, I would like to acknowledge specifically my two classmates with whom I have shared this journey. Thanks to

Maurya for always being there for me. You blazed the path

for me to follow, and I have missed not having you around

iv here for the last two years. To Kristin, whose own capabilities in this field continually impress me, and who

inspired me to achieve things even I didn't know I could, I

thank you for everything that we have been through together.

Although mere words seem unable to describe my gratitude, always know how special that I think you are.

I have dedicated this project to my grandparents. I thank them for the love they gave me as I was growing up. I only wish that they were here to share this accomplishment with m e .

v VITA

June 12, 1966 ...... Born - Philadelphia, Pennsylvania

1988 ...... B.A., University of Pennsylvania, Philadelphia

1988-1990 ...... Graduate Research Associate, The Ohio State University

1990 ...... M.A. , The Ohio State University

1990-1994 ...... Graduate Teaching Associate, The Ohio State University

PUBLICATIONS

Necowitz, L.B., & Roznowski, M. (in press). Negative affectivity and job satisfaction: Cognitive processes underlying the relationship and effects on employee behaviors. Journal of Vocational Behavior.

MacCallum, R.C., Roznowski, M . , & Necowitz, L.B. (1992). Model modifications in covariance structure analysis: The problem of capitalization on chance. Psychological Bulletin. 111. 490-504.

FIELDS OF STUDY

Major Field: Psychology Studies in Industrial/Organizational Psychology

Minor Field: Quantitative Psychology

vi TABLE OF CONTENTS

DEDICATION ...... ii

ACKNOWLEDGEMENTS ...... iii

VITA ...... vi

LIST OF TABLES ...... x

LIST OF FIGURES ...... xi

ABSTRACT ...... xii

CHAPTER PAGE

I. INTRODUCTION ...... 1

Personality and Goal Setting ...... 2 Purpose of Research ...... 9 Research on Negative Affectivity in I/O Psychology ...... 11 Research on Subjective Well-Being in I/O Psychology ...... 17 On the Distinction Between Negative Affectivity and Subjective Well-Being 19 Examining the Relationship Between Measures of NA and SWB ...... 24 Discussion of Goal-Setting Hypotheses ... 27 Relationship to Self-Set Goals, Goal Commitment and Performance ... 28 Relationship to Satisfaction ...... 29 Relationship to Commitment Following Failure and Self-Set Goals Following Success ...... 38 Relationship to Attributions ...... 40

II. METHOD ...... 46

Subjects ...... 46 Design ...... 46 Procedure ...... 47

vii Manipulations ...... 49 Goal Origin ...... 49 Feedback Type ...... 49 Measures ...... 52

III. RESULTS ...... 58

Part 1 - Examination of Relationship Between NA and SWB ...... 58

Analysis of Convergent Correlations .. 58 Analysis of Confirmatory Factor Analysis Models ...... 60 Comparison of Model 1 and Model 2 61 Comparison of Model 3 and Model 4 64 Construction of an Affect Composite .. 65

Part 2 - Examination of Goal-Setting Hypotheses ...... 68

Manipulation Checks ...... 68 Hypothesis 1 ...... 80 Hypothesis 2 ...... 82 Hypothesis 3 ...... 82 Hypothesis 4 ...... 87 Hypothesis 5 ...... 91 Hypothesis 6 ...... 94 Hypothesis 7a ...... 95 Hypothesis 7b ...... 98

IV. DISCUSSION ...... 102

Discussion of Results for Negative Affectivity and Subjective Well-Being Measures ...... 103 Discussion of Results for Goal-Setting Hypotheses ...... 112 Hypotheses 1 and 6 ...... 112 Hypotheses 2 and 5 ...... 115

viii Hypothesis 3 ...... 121 Hypothesis 4 ...... 125 Hypotheses 7a and 7b ...... 130 Limitations ...... 134 Implications ...... 136

LIST OF REFERENCES ...... 142

APPENDICES ...... 154

A. MEASURES ...... 154 B. INSTRUCTIONS FOR EXPERIMENTERS ...... 164 C. PRELIMINARY INSTRUCTIONS FOR SUBJECTS ...... 168 D. EXPERIMENTAL DEBRIEFING...... 170

ix LIST OF TABLES

TABLE PAGE

1. Correlation Matrix for Personality Scales ...... 59

2. Results of LISREL Analyses for Model 1 and Model 2 63

3. Results of LISREL Analyses for Model 3 and Model 4 66

4. Descriptive Statistics for Variables Assessed at Trial 1, Trial 2, and Trial 3 ...... 71

5. Descriptive Statistics for Personality Scales .... 72

6. Correlation Matrix for All Variables...... 74

7. Correlations Between Affect Scales and Goal Commitment at Trial 1 for the Assigned Condition ...... 83

8. The Mediating Effect of Goal Commitment on the Relationship Between Affect and Task Performance in the Failure Condition ...... 88

9. Hierarchical Regression of Performance Satisfaction on Goal Origin, Feedback Type, and Each Personality Scale ...... 90

10. Hierarchical Regression of Trial 2 Goal Commitment on Goal Origin and Each Personality Scale .... 92

11. Hierarchical Regression of Trial 3 Goal Commitment on Goal Origin and Each Personality Scale ..... 93

12. Correlations Between Composites of Ability, Effort, Difficulty and Luck Attributions With Each Personality Scale in the Success Condition ...... 97

13. The Mediating Effect of Causal Attributions on the Relationship Between Personality and Satisfaction in the Success Condition .... 100

x LIST OF FIGURES

FIGURE PAGE

1. Confirmatory Factor Analysis Models 1 and 2 ...... 25

2. Confirmatory Factor Analysis Models 3 and 4 ...... 26

3. Summary of Hypotheses 1*3 ...... 30

4. Summary of Hypothesis 7b ...... 45

xi ABSTRACT

Negative affectivity (NA; Watson & Clark, 1984; Clark &

Watson, 1991) and subjective well-being (SWB; Diener, 1984)

are two constructs that reflect relatively broad aspects of

an individual's personality. Negative affectivity refers to

the tendency to experience aversive emotional states, as well as to have a negative self-concept. Subjective well­ being refers to an individual's ongoing state of psychological wellness or happiness.

The two purposes for conducting this research were to

examine the empirical commonality between these two constructs, as well as to investigate the influence of NA and SWB on several variables that are central to goal setting theory, including goal level, goal commitment, task performance, satisfaction with performance, and attributions

for performance.

Subjects were 275 introductory psychology students who

completed measures of NA and SWB and then performed several

trials of a computerized perceptual speed task. The manipulated variables were goal origin and feedback type.

Subjects either self-set or were assigned performance goals.

In addition, subjects were either informed that they

surpassed their goals over the course of three trials or

xii were told that they had failed to achieve their goals over the three trials.

Confirmatory and exploratory factor analyses of the personality scales indicated that one general affect factor could adequately describe the data, and that measures of NA could not be subsumed by SWB.

Results for the goal-setting hypotheses demonstrated that those who were lower in NA and higher in SWB were more committed to an initial assigned goal and remained committed to their goal when faced with failure. There was also some evidence that this persistence was related to higher task performance. In addition, those higher in NA and lower in

SWB were more dissatisfied with their performance both under conditions of success and failure. Furthermore, as compared to those with a more negative disposition, those with a more positive disposition tended to take more credit for success by attributing it to ability or effort, rather than luck.

Results are discussed in terms of their implications for the process of goal-setting in organizations as well as the examination of dispositional influences in industrial/ organizational psychology research.

xiii CHAPTER I

INTRODUCTION

Numerous studies have documented the finding that specific, difficult goals lead to higher performance than easy goals or unspecific goals, such as "do your best"

(Locke & Latham, 1990; Locke, Shaw, Saari, & Latham, 1981).

This finding is so robust that goal setting has been described as ... "among the most scientifically valid and useful theories in organizational science" (Locke, Latham, &

Erez, 1988, p. 23). More recently, noting that the strength of goal-setting effects may vary widely across studies, researchers have focused on contextual factors that might influence the goal-setting process. For example, goal commitment has been found to be a necessary condition if individual performance is to be improved through goal- setting (Hollenbeck & Klein, 1987; Hollenbeck, Williams, &

Klein, 1989; Locke et al., 1988). Other factors that have been investigated as influencing the goal-setting process include performance feedback (e.g., Bandura & Cervone, 1983;

Erez, 1977; Kim & Hamner, 1976), incentives (e.g., Campbell,

1984; Erez, Gopher, & Arazi, 1987), origin of the goal

(e.g., assigned vs. participative vs. self-set; Austin,

1989; Dossett, Latham, & Mitchell, 1979; Latham, Erez, &

1 2

Locke, 1988), and task complexity (e.g., Wood, Mento, &

Locke, 1987).

While these factors represent aspects of the

situational context in which goal-setting occurs, one might

also view the process from an individual differences perspective. Such a view would hypothesize that relatively

stable characteristics of the person influence goal-setting processes and outcomes. Note, however, that these two perspectives are neither competing nor incompatible. The

effect of any particular individual difference variable may

depend on the nature of the situational context. For

example, in a study by Hollenbeck and Brief (1987), a positive goal difficulty-performance relationship was found

for all subjects in a self-set goal condition, but only for

those higher in self-esteem when goals were assigned by the

experimenters.

Personality and Goal-Setting

The idea that personality may have an important

influence on goal-setting processes can be traced

historically to level of aspiration research. Frank (1935)

defined level of aspiration as "the level of future

performance in a familiar task which an individual, knowing

his level of past performance in that task, explicitly

undertakes to reach" (p. 119). Thus, the focus of interest

was primarily goal choice. Both Frank (1935) and Gardner

(1940) offered suggestive evidence that level of aspiration

is partially determined by an individual's personality. 3

Frank's conclusions were based on findings that aspiration levels were both constant over time and across type of task.

Thus, he argues that "... certain forms of behavior of the level of aspiration depend upon consistent and general traits of personality" (p. 128). Gardner (1940) examined the relationship between several individual difference variables and level of aspiration. These variables included self-confidence, fear of failure, intellectual achievement need, and ambition. Correlations were low and nonsignificant, but were primarily in the hypothesized direction. The nonsignificant findings could be due to the combination of a small sample size and some low reliabilities of the personality scales.

These researchers introduced questions that are still pertinent today. However, as compared to research examining situational variables, fewer recent studies have examined the role played by individual differences in the goal- setting process. This trend was noted in a review by Kanfer and Kanfer (1991), "Although interest in identifying the role of stable individual differences on goals and self­ regulation has grown over the past decade, this perspective remains relatively unexplored" (p. 316). To date, much of

this research has focused on need for achievement, locus of control, and self-esteem. For example, Latham and Yukl

(1976) examined whether these three variables moderated the relationship between goal origin (assigned vs. participative) and measures of satisfaction and performance 4 in a sample of typists. No evidence was found for the presence of moderator variables. Yukl and Latham (1978) reported that high need achievers and those having an internal locus of control set more difficult goals than low need achievers and those having an external locus of control, although this difference in goal level did not translate into subsequent performance improvement. However, those higher in self-esteem performed at higher levels than those lower in self-esteem. Matsui, Okada, and Kakuyama

(1982) examined the effect of need for achievement on goal level and performance in a sample of undergraduates performing a perceptual speed task. Those higher in achievement need set higher goals and performed better than those lower in achievement need. Finally, Dossett et al.

(1979) examined the effect of need for achievement, need for independence, self-esteem, and locus of control on measures of performance, goal acceptance, and goal attainment in a study of female clerical personnel solving arithmetic problems. Results for the individual differences were primarily negative. The exception was for self-esteem when the dependent variable was goal attainment. Individuals higher in self-esteem who participated in setting their own goals, and who received performance feedback, attained their goals more often than those lower in self-esteem under these same conditions. The lack of significant findings led the authors to conclude that there is ... "limited need for exploring personality variables in goal-setting" (p. 294). 5

These findings indicate that the effects of personality variables in goal-setting research have been quite

inconsistent. Several explanations are offered. Locke et al. (1981) suggested that many of these studies were neither designed to examine specifically the effects of these individual differences nor were the variables included driven by a theoretical rationale. More generally, Weiss and Adler (1984) have noted that the use of personality variables in organizational behavior research is largely atheoretical. They suggest that personality variables can play an important role in organizational behavior when serious thought is given to the theoretical correspondence between the personality variables and the organizational criteria being predicted. Locke et al. (1981) also noted that individual differences are more likely to have an effect when individuals are permitted to set their own goals rather than having the goals assigned to them, because in the latter case, situational demands are strong and would most likely mask individual styles and preferences. Other authors have also argued that when situational demands are

"strong," individual differences will not have large effects

(e.g., Davis-Blake & Pfeffer, 1989; Mischel, 1977; Monson,

Hesley, & Chernick, 1982).

Prior research on personality influences in goal- setting can also be faulted on other design and methodological grounds. Specifically, many studies have used small sample sizes. For example, only 41 typists 6 participated in the Latham and Yukl (1976) and Yukl and

Latham (1978) studies, while Dossett et al. (1979) had a sample size of 60. Compounding this problem are the questionable statistical methods used. For instance, moderators were often tested by dichotomizing the continuous individual difference variable into high and low subgroups and examining means or correlations within each subgroup. A more appropriate technique for investigating the presence of moderators is via moderated regression analysis (Stone &

Hollenbeck, 1984). This method provides higher statistical power as information about individual differences is not

"thrown away" (Cohen & Cohen, 1983).

More theoretically and methodologically sound investigations on individual differences in goal-setting have been conducted recently by Hollenbeck and his colleagues (Hollenbeck & Brief, 1987; Hollenbeck & Klein,

1987; Hollenbeck et al., 1989). Hollenbeck and Brief (1987) argued that the effect of individual differences in the goal setting process is dependent on goal origin. Specifically, they hypothesized that when goals are self-set, individual differences will determine the level of the goal set (cf.

Locke et al., 1981), but not the expectancies or valences for the chosen goal. In contrast, when goals are assigned, individual differences should determine one's reaction to the goal in terms of motivation to pursue the goal, and expectancies and valences for goal attainment. The authors examined their model in a sample of 102 undergraduate 7 business students performing an anagram task. Individual differences included in the study were generalized self­ esteem, self-perceptions of task-specific ability, objective task-specific ability, locus of control, and need for achievement. Results largely supported the hypotheses, although the specific individual difference variables had differing effects. In the self-set condition, the task- specific ability measures (both objective and perceived) were related to goal level, with self-esteem having a weaker relationship. In the assigned condition, self-esteem, need for achievement, and locus of control were related to the valence of attaining the goal, while the task-specific ability measures were related to the expectancy of attaining the goal. In terms of performance, the interactive effect of goal difficulty, goal origin, and self-esteem was mentioned earlier. To reiterate, the goal difficulty- performance relationship was dependent on self-esteem in the assigned condition, but not the self-set condition. In fact, a negative difficulty-performance relationship was found when goals were assigned and self-esteem was low.

Thus, assigning difficult goals to those lower in self­ esteem hampers performance whereas easier goals facilitate performance. In may be that individuals with lower self­ esteem are more susceptible to the negative feedback that follows from assigning difficult goals that are initially not attained than are individuals with higher self-esteem.

In the self-set condition, where individuals are selecting goals that match their expectancies of attainment, a positive difficulty-performance relationship is observed regardless of levels of self-esteem.

The focus of Hollenbeck et al. (1989) was on determining the situational and personal antecedents of commitment to difficult goals. Goal commitment, as previously mentioned, appears to be necessary for difficult goals to lead to higher performance. Thus, determining those factors that lead to commitment is important both practically and theoretically. The Hollenbeck et al. (1989) study was based on a earlier conceptual paper by Hollenbeck and Klein (1987) who argued that goal commitment can be conceptualized in terms of an expectancy theory model.

Specifically, individuals will be committed to difficult goals if they have a high expectancy of attaining the goal and/or if they view the goal as attractive (i.e., highly valent). The authors specified situational and personal factors likely to influence the expectancy and valence of goal attainment. In the empirical study, the personal factors examined were locus of control and need for achievement. Subjects were undergraduate business students who reported their commitment to GPA goals for the academic quarter that were either assigned or self-set at an earlier session. Both an internal locus of control and higher need for achievement were related to increased levels of commitment. Furthermore, the achievement-commitment 9

relation was stronger when goals were self-set rather than

assigned.

In sura, the idea that personality may have an important

influence on goal-setting processes and task performance is not necessarily a recent development, but it is an area that has been largely untapped by organizational researchers.

Most studies that have examined personality in goal-setting can be faulted on either theoretical or methodological

grounds, or both. As Adler (1986) has noted "... the potential usefulness of personality to goal setting research has simply not been adequately examined ... the time is ripe

for a more serious consideration of the role of personality

among goal-setting researchers" (p. 1).

Purpose of Research

To that end, the purpose of this research is to examine

additional personality variables that might play a role in

goal setting processes as well as task performance. I will

focus on several dependent variables that are central to

goal setting theory and practice. These will include: goal

choice/goal level, goal commitment, task performance,

satisfaction with performance, and performance attributions.

With regard to goal commitment, I am particularly interested

in examining commitment to goals following failure feedback.

More specifically, when an individual fails to attain a goal he/she might choose either to lower the goal or maintain the

goal but try harder (cf. Campion & Lord, 1982). Since the

intensification of effort that accompanies persistence will 10

often lead to higher performance levels than if one lowers

the goal, isolating those factors that contribute to goal

persistence is an important area for research.

The two individual difference variables that will be

examined are negative affectivity (NA; Watson & Clark, 1984;

Clark & Watson, 1991), and subjective well-being (SWB;

Diener, 1984). These two constructs were selected for study because they reflect broad aspects of an individual's personality and recent studies have indicated that they have both statistically and practically significant relationships with important organizational criteria, such as job satisfaction and employee withdrawal behaviors (e.g., Judge

& Hulin, 1992; Necowitz & Roznowski, in press).

Additionally, while two separate literatures have developed around the NA and SWB constructs, the conceptual and empirical distinction between them remains a question among organizational researchers (cf. Judge, 1992). Thus, one subgoal of this research is to examine this distinction.

This will be accomplished in a number of ways. First, several confirmatory factor analysis models will be fit to

the data. The specific models are described in a subsequent

section. Next, the extent to which the NA and SWB scales are intercorrelated, and their degree .of relationship with

the dependent variables will be examined. If these measures

are assessing the same construct, one would expect a one-

factor model to fit the data well, and the scales to be highly intercorrelated and show similar relationships to the 11 dependent variables. Conversely, if a two-factor model fits better, and if the NA and SWB scales are not too highly intercorrelated and are differentially related to the dependent variables, one might argue that these are separate constructs.

Research on Negative Affectivity in I/O Psychology

Watson and Clark (1984) reviewed a large body of literature in the personality domain, and came to the conclusion that a number of supposedly distinct personality scales actually measure one common, pervasive, and relatively stable construct. The authors termed this trait negative affectivity (NA), though others have referred to it as neuroticism (Eysenck & Eysenck, 1975; Costa & McCrae,

1987) . NA is defined as the tendency to experience aversive emotional states, as well as to have a negative self- concept. Those higher in NA tend to report more distress, discomfort, and dissatisfaction over time and across situations even in the absence of an objective source of stress. Higher NA individuals also focus more on the negative aspects of other people and the world in general and tend to interpret ambiguous stimuli more negatively

(Goodstein, 1954; Haney, 1973; Phares, 1961). Thus, NA represents a relatively broad personality domain reflecting a syndrome of negative functioning both in terms of affect

(e.g., experiencing unpleasant emotions) and cognitions

(e.g., low self-esteem). NA has also been linked to 12 negative patterns of behavior (e.g., Kemery, 1991; Necowitz

& Roznowski, in press).

In the industrial/organizational psychology literature,

NA has been studied primarily in two domains, its relation

to job attitudes and behaviors, and its influence on

relationships between job stress and job strain. Each will be reviewed briefly in turn.

Recently, there has been a growing interest in examining endogenous sources of variance in job attitudes.

The "dispositional approach," as this movement is labelled, has been strongly influenced by the contributions of Staw and his colleagues (Staw, 1986; Staw, Bell, & Clausen, 1986;

Staw & Ross, 1985). For example, using data from the

National Longitudinal Survey (NLS; Center for Human Resource

Research, 1977), Staw and Ross (1985) found measures of job satisfaction to be significantly correlated over three and

five-year intervals not only when individuals worked for the same employer in the same occupation, but also when

individuals changed both employers and occupations. Gerhart

(1987) replicated these results with an additional sample from the NLS.

Staw et al. (1986) argued that one reason for the attitudinal stability observed by Staw and Ross (1986) and

Gerhart (1987) is that individuals have general positive or negative orientations toward life that influence the way

they perceive and evaluate their job situation. They termed

this orientation the "affective disposition" of the 13 individual and, using archival data, showed that affective disposition assessed during early adolescence was significantly correlated with job attitudes over a 50-year period. The data for this study came from the Berkeley

Intergenerational Studies and trained personality raters who assessed the affective disposition of the subjects.

Following in this line of research were several studies examining the relationship between NA and job satisfaction.

The theoretical rationale for this link follows from the definition of NA. Because higher NA individuals tend to accentuate and focus on the negative in any given situation, and tend to interpret ambiguous stimuli more negatively, they will be more likely to report a greater level of dissatisfaction with their jobs than their lower NA counterparts. Supportive evidence comes from Levin and

Stokes (1989) and Necowitz and Roznowski (in press). In the former study, the inclusion of an NA measure significantly increased the amount of variance accounted for in task/job satisfaction after controlling for the variance explained by task/job characteristics. This was true both in a lab study and in a field sample of professional staff of an international service firm. Necowitz and Roznowski (in press) replicated this finding in a sample of food service managers and employees. In addition, these authors provided evidence that one possible mechanism for NA's influence on satisfaction is via selective perception and evaluation of task relevant information. In an "enriched" task condition in the laboratory, higher NA individuals recalled more negative aspects of a task that they had performed a week earlier than lower NA individuals. In addition, they rated the task lower in identity and autonomy. Finally, the field study indicated NA to be related to negative patterns of behavior. Individuals higher in NA reported that they engaged in a greater frequency of job withdrawal behaviors than those lower in NA. The main effect for NA indicated that this difference was found at all levels of job satisfaction, lending further credence to the view of NA as a general negative syndrome in the affective, cognitive, and behavioral domains.

In the area of job stress, a number of studies have examined the relationship between NA, self-reported job stress (e.g., role conflict, role ambiguity, interpersonal conflict, workload), and self-reported job strains (e.g., job dissatisfaction, somatic complaints at work, perceived frustration). Most of this work has investigated whether NA spuriously inflates the relationship between self-reported stressors and strains. Brief, Burke, George, Robinson, and

Webster (1988) found support for this hypothesis. In their study, NA was significantly correlated with both job stress and job strain measures, and the stress and strain measures were highly intercorrelated. However, these latter correlations were substantially reduced when NA was partialled from these relationships. Thus, the authors argue that the stress-strain relationships may be spuriously 15

inflated when measured by self-report because they are

"contaminated" by NA. They recommended that stress researchers use more objective measures of stresses or

strains or both. Their concern is buttressed by some evidence indicating that while NA is related to self-reports of strain measures, it is unrelated to more objective measures of health such as cancer and mortality (Costa 6c

McCrae, 1987). Others, however, have provided evidence that

NA ,is related to objective physical outcomes (e.g., Anisman

6i LaPierre, 1982; Booth-Kewley & Friedman, 1987).

Subsequent studies have failed to replicate the

findings of Brief et al. (e.g., Chen & Spector, 1991; Jex and Spector, in press). Chen and Spector (1991) argued that

the findings of Brief et al. (1988) could be explained

largely by the content overlap of their NA measure, the

Taylor Manifest Anxiety Scale (TMAS; Taylor, 1953), and the

strain measures. Indeed, several items in the TMAS are

identical to the Somatic Complaints Scale and the Job Affect

Scale used by Brief et al. (1988). Chen and Spector (1991)

found little reduction in stress-strain relationships when

NA was partialled, thus arguing that the relationship

between stressors and strains are only modestly influenced

by NA.

This issue was also studied more recently by

Schaubroeck, Ganster, and Fox (1992) who used structural

equation modeling and confirmatory factor analysis to test

whether NA contaminates the measurement of self-reported 16 stressors and strains, or whether NA causes spurious correlations between stressors and strains due to a negative biasing effect in self reports. Their analysis indicated support for the latter hypothesis. The confirmatory factor analysis supported the notion that NA is a distinct construct from stress and strain; that is, they all measure separate domains. However, they did find a significant attenuation in the strength of relations between stressors and strains when NA was added to the structural model.

Thus, they obtained a similar result to Brief et al. (1988), but were able to rule out the possibility that NA, stressors, and strains are measuring the same construct domain.

One conclusion from this set of studies is that higher

NA individuals have a tendency to report higher levels of both stress and strain perhaps due to differences in their perceptions and interpretations of environmental events.

Indeed, research has shown that higher NA individuals interpret ambiguous stimuli more negatively. Thus, otherwise innocuous work situations may be viewed as potentially threatening and harmful (cf. Brett, Brief,

Burke, George, & Webster, 1990). In addition, because higher NA individuals are more internally focused and ruminative, they may attend to and amplify bodily sensations more, interpreting them as signs of illness (Costa & McCrae,

1987; Watson, 1988; Watson & Pennebaker, 1989). 17

Research on Subjective Well-Being in I/O Psychology

Subjective well-being (SWB), as conceptualized by

Diener (1984), is defined as an ongoing state of psychological wellness or happiness. SWB comprises both affective and cognitive dimensions. Life satisfaction represents the cognitive dimension wherein the individual assesses aspects of his/her life arriving at a single, global satisfaction assessment. The affective dimension of

SWB is referred to as hedonic level, which is defined as the experiencing of positive versus negative emotions. Thus, hedonic level refers to an individual's overall emotional well-being.

While levels of SWB may change over time due to positive and negative life events (Headey & Wearing, 1989), it does appear to be a fairly stable aspect of individuals.

For example, Diener (1984) cites evidence that long-term temporal reliabilities of SWB range between .55 to .70.

Diener and Larsen (1984) found that average levels of affect were also cross-situationally and temporally stable, and that life satisfaction was the most stable variable measured. Costa et al. (1987) have also provided evidence that SWB is fairly stable over a 9-year period.

Nevertheless, other researchers have found the temporal reliability of SWB to be quite a bit lower (Schwarz &

Strack, 1985, cited in Yardley & Rice, 1991), and have shown that SWB is somewhat influenced by momentary mood states 18

(Schwarz & Clone, 1983; Yardley & Rice, 1991). Thus, they

argue that SWB is not a very stable construct.

To address this controversy regarding the stability of

SWB, Headey and Wearing (1989) proposed a dynamic

equilibrium model. They argued that each person has a

"normal" or equilibrium pattern of both life events and SWB which can be predicted from stable person characteristics.

When there are deviations from this normal pattern of

events, either positively or negatively, changes in SWB

occur. However, this is a temporary change because person

characteristics play an equilibrating function, bringing

individuals back to their normal levels. They found preliminary support for this model with a sample of 649

individuals from the Australian Quality of Life Panel Study.

Favorable and adverse life events between 1981-83 predicted measures of SWB in 1983 above the influence of age,

extraversion, neuroticism, and openness to experience.

Nevertheless, the authors could not fully test their model because they did not have enough waves of events data to

construct reliable measures of equilibrium event levels.

Thus, it was impossible to measure whether or not changes in

equilibrium had occurred, the effects of equilibrium change,

and whether or not people revert back to their equilibrium

levels after experiencing "abnormal" positive or negative

events. Nevertheless, the model is provocative in that it helps to explain both the stability and change found in SWB measures. 19

In the I/O literature, SWB has been investigated primarily as a cause of job satisfaction and job withdrawal behaviors. Judge and Hulin (in press) examined the influence of SWB on job satisfaction in a sample of medical office staff (e.g., nurses, lab technicians, etc.). Using structural equation modeling, their results indicated that

SWB was both a significant cause and effect of job satisfaction. In another study, Judge and Hulin (1992) studied the influence of SWB on employee withdrawal behaviors. The sample for this study was the same as the sample that participated in Judge and Hulin (in press). The measure of withdrawal included such diverse behaviors as taking long breaks, chatting with coworkers, missing meetings, as well as more traditional withdrawal behaviors such as turnover and absence. These behaviors are quite similar to those used by Necowitz and Roznowski (in press).

Judge and Hulin (1992) found that SWB significantly predicted the frequency of job withdrawal behaviors above the influence of job satisfaction. Again, the results of the Judge and Hulin studies are quite similar to Necowitz and Roznowski (in press) who found that NA was related to job satisfaction and the frequency of reported withdrawal behaviors.

On the Distinction Between

Negative Affectivity and Subjective Well-Being

The Judge and Hulin results suggest that SWB has similar effects as NA in the area of job attitudes and 20 behaviors, and raises the question as to whether or not there is a distinction between these two constructs. Based on the finding that measures of negative affectivity and positive affectivity (PA) are essentially uncorrelated with each other, Watson and his colleagues (e.g., Watson, Clark,

& Tellegen, 1988; Watson & Tellegen, 1985) have consistently argued that NA is independent of PA. However, other evidence suggests that this finding may be simply an artifact of how affect is measured. For example, Diener,

Larsen, Levine, and Emmons (1985) distinguished between affect frequency and affect intensity. The former refers to how often one feels happy or unhappy, while the latter refers to the degree that emotions are experienced. The frequency of positive and negative affect over time tends to be negatively correlated. That is, individuals who experience frequent pleasant emotions do not also experience frequent unpleasant emotions. Indeed, as Judge (1992) notes, it is rare to be both happy and unhappy at the same time, because one emotion would suppress the other.

However, the intensity of affect is positively correlated over time, such that those who experience pleasant emotions intensely also experience unpleasant emotions intensely.

When measures assess both affect frequency and affect intensity (e.g., Watson et al.'s (1988) Positive and

Negative Affect Schedule), NA and PA will appear unrelated.

It is further argued that affect intensity is really a measure of emotionality or temperament, and not affective or 21 emotional well-being, per se. For example, Diener, Sandvik, and Pavot (1991) found that the frequency of experienced affect was both necessary and sufficient to produce affective well-being. That is, positive affect intensity does not predict well being, and further, intense positive emotions often have negative consequences that can be detrimental to well-being (Diener, Colvin, Pavot, & Allman,

1991).

In addition, Judge (1992) argues that simply because negative items covary more highly with negative items, and positive items covary more highly with positive items does not necessarily mean that NA and PA are separate constructs.

For example, Judge (1992) demonstrated that splitting the

Job Descriptive Index work scale into positive and negative items yielded a significant improvement in fit in the measurement model over a single factor solution, and an odd- even split. Thus, if one argues for separate NA and PA constructs, one must also argue for separate constructs of negative and positive job satisfaction. These results, according to Judge (1992), simply mean that individuals respond differently to positive and negative items. These ideas have also been echoed by Kemery and his colleagues

(Kemery, 1991; Kemery, Mossholder, & McGrath, 1989).

Researchers studying similar personality variables have been faced with comparable questions. For example, Scheier and Carver (1993), in studying dispositional differences in optimism note that factor analyses of their measure of 22 optimism, the Life Orientation Test (LOT), consistently yield two separate factors, one containing the positively worded items (optimistic items), and the other containing the negatively worded items (pessimistic items). As the authors note:

Identification of two factors raises the question of

whether it is better to view optimism and pessimism as

opposite poles of a single dimension, or as

constituting two separate, but correlated dimensions.

We have thus far taken the former view. (p. 27,

underline added)

Furthermore, in the job characteristics literature, the Job

Diagnostic Survey (JDS; Hackman & Oldham, 1975) needed to be revised due to a similar measurement issue (Harvey,

Billings, & Nilan, 1985; Idaszak & Drasgow, 1987). In this case, a six factor model fit the data better than the five factor theoretical model. This sixth factor was an artifact, however, in that it contained all of the negatively worded items, again suggesting that individuals respond differently to positive and negative items. Thus, affectivity may represent a single continuum (cf. Kemery,

1991), and scales assessing NA with both positive and negative items (e.g., Spielberger, 1979; Stokes & Levin,

1990) may better assess this continuum than scales with negative adjectives only (e.g., Watson et al. 1988).

Further, Judge (1992) equates PA and NA to the hedonic level component of SWB, arguing that SWB subsumes NA. 23

However, this may be an overstatement. As has been

discussed, NA encompasses both affective and cognitive

dimensions. It is defined not only as affective well-being, but also in terms of self-concept and self-esteem. Studies

examining NA have shown it to be significantly related to

self-ratings of happiness and life satisfaction (Stokes 6>

Levin, 1990), the cognitive component of SWB. Indeed,

Diener (1984) has noted that high self-esteem is one of the

strongest predictors of SWB, and Campbell, Converse, and

Rodgers (1976) found that self-satisfaction was a strong predictor of life satisfaction. Thus, NA may be as broad a construct as SWB.

Of course, one alternative is that NA and SWB are really measuring the same thing. In fact, Judge and Hulin

(1992) have argued that Staw et al. (1986), Levin and Stokes

(1989), and by extension Necowitz and Roznowski (in press,

since they used Levin and Stokes' measure), assessed

subjective well-being, but called it affective disposition, or negative affectivity.

Thus, an examination of the interrelationship between measures of NA and SWB, as well as their effects on other variables of interest is needed. Though results of studies using different samples indicate that similar effects are found, these relationships have not been examined within the same sample. To that end, this study will measure SWB and

NA using multiple indicators, and examine their

interrelationship as well as whether or not they have 24

differential effects on the dependent variables, in the hope

of clarifying the theoretical and empirical confusion that

presently exists in the literature.

Examining the Relationship Between

Measures of NA and SWB

To examine the interrelationship between NA and SWB,

the magnitude of the intercorrelations between their

respective indicators will be examined. In addition, several confirmatory factor analysis models will be

evaluated. The models that will be tested are presented in

Figure 1 (Model 1 and Model 2) and Figure 2 (Model 3 and

Model 4), and are based on recent theoretical arguments

regarding the distinction, if any, between NA and SWB.

If NA and SWB are measuring the same common construct,

then a single factor model (Model 1) with the seven NA and

SWB indicators loading on that factor should fit the data

better than a two-factor model containing a NA latent factor

and a SWB latent factor (Model 2). In Model 2, only

indicators of NA are permitted to load on the NA latent

factor, while indicators of SWB are permitted to load on the

SWB latent factor only.

Model 3 is based on Judge's (1992) recent argument that

SWB subsumes NA. In this model the two latent factors

represent the two components of SWB: hedonic level and life

satisfaction. The scales loading on the hedonic level

factor are all of the NA scales and the SWB scales assessing

affect. The scale that would load on the life satisfaction 25

NASch

TAS

NAS

NA/SWB UF

FHQ

ASS

Modal 1 SWLS

NASch

NA TAS

NAS

UF

SWB FHQ

ABS

SWLS Modal 2

FIGURE 1 NASch: Nagatlva Affact Schadula TAS: Trait Anxlaty Scala NAS: Nagatlva Affactivity Scala UF: Undarvood & Froalng Frequency of Poaitlva and Nagaelva Affact FHQ: Fordyca Happlnasa Quastlonnalra ABS: Affact Balanca Scala SWLS: Satisfaction With Llfa Scala 26

NASch „ TAS Hedonic Level

UF ABS FHQ

Life Satis­ faction SWLS

Model 3

NASch

UF

Hedonic ABS Level FHQ

TAS NAS

Afe/Self Satis­ faction ^ SWLS

Modal 4 FIGURE 2 NASch: Nagatlva Affact Schadula TAS: Trait Anxlaty Scala NAS: Nagatlva Affectivity Scala UF: Undarwood & Froaing Frequency of Poeltlve and Nagatlva Affect FHQ: Fordyce Happiness Questionnaire ABS: Affect Balance Scala SWLS: Satisfaction With Life Scale 27 factor is the Satisfaction With Life Scale.

Model 4 is derived from two arguments. First, it was stated previously that the NA construct defined by Watson and Clark (1984) is broader in scope than simply affective well-being, but incorporates self-concept and self-esteem.

Second, Judge and Hulin (1992) have argued that Stokes and

Levin's measure of NA, which was developed directly from the conceptual arguments of Watson and Clark (1984) is a measure of subjective well-being. If this is the case, then the

Trait Anxiety Scale (Spielberger, 1979) would appear also to be a measure of SWB, and not NA per se. This would not be the case for the PANAS-NA, as this scale seems to assess affective experience exclusively. Thus, in Model 4, the NAS and TAS load on both the hedonic level and life satisfaction factors. If Model 4 fits the data better than Model 3 this would be evidence that either a) the NAS and TAS assess SWB, and not NA per se, and/or b) the construct that Watson and

Clark (1984) defined conceptually is broader than their operational definition (Watson et al., 1988).

Discussion of Goal-Setting Hypotheses

Though both SWB and NA will be examined, for the purpose of clarity, goal-setting hypotheses will be discussed within the context of NA. Identical hypotheses will be tested for SWB where higher SWB will be considered equivalent to lower NA and vice-versa. 28

Relationship to Self-Set Goals,

Goal Commitment and Performance

Individuals higher in NA tend to rate themselves lower in self-esteem and self-satisfaction than those lower in NA.

For example, higher NA individuals rate themselves lower on a variety of self-descriptive trait scales (e.g., considerate, cooperative, friendly). In addition, other evidence has shown that lower NA individuals are rated by clinicians as more cheerful and self-satisfied, whereas higher NA individuals are rated as having poorer self-images

(Block, 1965). Thus, higher NA individuals seem to be more dissatisfied with themselves and to have greater feelings of inadequacy than those lower in NA.

The positive self-image of those lower in NA is likely to translate into the level of the goals that individuals choose to pursue. Individuals who have higher self-regard should set higher goals than those who do not (Hollenbeck 6

Brief, 1987). Thus, the following two hypotheses are offered:

HI: When individuals have the opportunity to set their own goals, those lower in NA will set higher goals than those higher in NA.

H2: When goals are assigned, individuals lower in NA will be more committed to the assigned goal than those higher in NA.

Along these lines, and given the finding that difficult goals lead to higher performance than easier goals, one 29 would expect lower NA individuals to perform at higher

levels than higher NA individuals:

H3: Goal level and goal commitment mediate the effect of NA on task performance. Individuals lower in NA will have higher levels of task performance than those higher in NA, due to setting higher goals in the self-set condition and greater commitment to goals in the assigned condition.

A summary of hypotheses 1-3 is depicted in Figure 3.

Relationship to Satisfaction

Goal-setting theory, as well as other motivation

theories like control theory (Campion & Lord, 1982; Klein,

1989; Lord & Hanges, 1987) and social-cognitive theory

(Bandura, 1986), argues that experiences of success and failure have implications for affective reactions. All of these theories state that satisfaction does not depend on one's absolute level of performance attainment; rather, satisfaction levels depend on performance relative to some goal or standard. If an individual falls short of his or her goal, then he or she will experience displeasure or dissatisfaction. If an individual meets or exceeds his or her goal, he or she will feel pleasure and satisfaction.

(Note, however, that in some cases performance above goals may be disvalued, for example, if the individual is trying for a specific point goal (Lord & Hanges, 1987)). Thus, the primary determinant of affective reactions are discrepancies between one's goal and one's performance. Further, the greater the positive discrepancy between performance and 30

HI: NA/SWB Goal Level (Self-set)

H2: NA/SWB “ Goal Commitment (Assigned)

H3: NA/SWB ^ Goal Level ► Performance Goal Commitment

Figure 3

Summary of Hypotheses 1-3 31 goals (i.e., the more performance is above goals), the greater the satisfaction, whereas the greater the negative discrepancy between performance and goals (i.e., the more performance is below goals), the lower the satisfaction

(Locke & Latham, 1991).

Numerous studies have found support for the idea that discrepancies between performance and goals determine satisfaction. For example, in a series of five studies by

Locke, Cartledge, and Knerr (1970), subjects were provided with specific point goals in a reaction time task (studies

1,3,4) and an addition task (study 2). Note that in this case deviations above or below the goal should lead to dissatisfaction. The results for these four studies indicated a high negative correlation between goal- performance discrepancies and satisfaction (r's- -.92, -.61,

-.60, -.75, respectively). In study 5, the researchers examined the influence of goal-performance discrepancies for subjects' working toward course grade goals. The difference between subjects minimum grade goal (i.e., the lowest grade a person would be satisfied with on an exam) and performance was positively correlated with satisfaction (r-.84). In this case, high positive discrepancies should be valued (as opposed to studies 1-4 where deviations above or below the goal were negative), and thus the sign of the relationship is positive. 32

Bandura and Cervone (1986) also provided evidence that

discrepancies between goals and performance influence self- evaluative responses. In this study, subjects pursued a goal while riding an ergometer, a type of exercise bicycle.

Subjects attempted to achieve a 50% increase in effort over a practice session. Feedback was manipulated such that subjects were in either a condition that was a large substandard discrepancy (effort 24% above practice session), a moderate substandard discrepancy (effort 36% above practice session), a small substandard discrepancy (effort

46% above practice session), or a small suprastandard discrepancy (effort 54% above practice session). Results

indicated that subjects were most dissatisfied with a large

substandard discrepancy and least dissatisfied with a small

suprastandard discrepancy. Satisfaction with a moderate

substandard and small substandard discrepancy fell in between these extremes.

More recently, Kernan and Lord (1991) found in two

studies that performance goals added significant variance to

the prediction of satisfaction when performance was already

included in the regression model. Thus, they also came to

the conclusion that it is the discrepancy between actual and

desired performance that is the more important determinant of satisfaction, rather than absolute level of performance.

The idea that the discrepancy between goals and performance influences affective reactions is well established. Locke and Latham (1991) report that the 33 average sample-weighted correlation between level of success and satisfaction over all studies examining this relationship is .51. However, before discussing how NA

influences satisfaction, it is important to note that some researchers have argued that another factor instead of, or in addition to, discrepancies influence satisfaction.

Carver and Scheier (1991) refer to this factor as the rate at which discrepancies are reduced. Thus, they contend that

it is the rate of progress with which one is moving toward a standard that determines satisfaction and not goal- performance discrepancies. For example, consider the case where performance is below one's goal. According to Carver and Scheier (1991), this situation will not necessarily lead to negative affect. Negative affect will result if progress toward the goal is below some standard for progress rate.

Conversely, positive affect will result if progress toward the goal is above some standard for progress rate. In short, they posit that it is the rate of discrepancy reduction that determines affect, not the size of the performance-goal discrepancy.

Carver and Scheier (1991) did not provide empirical evidence to support their views. However, two papers have recently examined whether progress rate, in addition to or

instead of discrepancies, influences satisfaction. Hsee and

Abelson (1991) tested this question in two studies, In both cases, subjects were presented with a variety of hypothetical outcomes and asked to rate their satisfaction 34 with each outcome. Results indicated that both performance- goal discrepancy and progress rate predicted satisfaction.

Subjects preferred outcomes that were less negative than more negative relative to a standard, and outcomes that indicated negative discrepancies were being reduced at a faster rate.

Hendrick and Lord (1992) also examined this question in an experimental study where subjects had to rate the effectiveness of actors giving psychology lectures on a set of three videotapes. Subsequently, subjects were given feedback on their rating accuracy relative to "expert" raters where 90% accuracy was assigned as the goal. Goal- performance discrepancy was manipulated by having subjects in either a small negative discrepancy or large negative discrepancy condition. Progress rate was manipulated by having subjects progress over the three trials toward the goal at either a faster or slower rate. Results indicated that both performance-goal discrepancy and progress rate predicted satisfaction significantly. These two papers indicate that both discrepancies and progress rate are important determinants of satisfaction.

In the present study, progress rate is controlled by having subjects in the failure condition experience the same magnitude of failure on each trial. Thus, progress toward the goal is identical for all subjects. In the success condition, progress rate is not applicable because all subjects are experiencing a positive discrepancy. Hence, in 35

this study only discrepancies vary between subjects. That

is, subjects in the success condition will experience a positive discrepancy and subjects in the failure condition will experience a negative discrepancy.

As noted before, there is a high correlation between discrepancies and satisfaction (Locke & Latham, 1991).

Locke and Latham (1991) provide several explanations for why

the relationship is not perfect, including factors such as measurement error, the use of multiple standards, and the significance or importance of success for the individual.

While not stated explicitly, Locke and Latham (1991) suggest

that personality may influence affective reactions when they state, "...different people may evaluate the same actual discrepancy in different ways" (p. 236).

Rohrback and Lord (1991) have recently argued that personality may have an influence on how subjects interpret discrepancies and their subsequent affective, behavioral, and cognitive responses to performance feedback. Thus, different types of individuals may be more or less sensitive

to discrepancies. These researchers assessed several personality variables to determine whether they predicted

the choice of cognitive (i.e., reducing goals) versus behavioral (i.e., increasing effort) strategies to negative discrepancies. They found that individuals with incremental beliefs about intelligence (i.e, that intelligence is malleable and can be developed via effort; Dweck &

Henderson, 1987), and an action orientation (Kuhl, 1985) 36 respond by increasing effort when faced with large negative discrepancies. In contrast, those with entity beliefs about

intelligence (i.e., that intelligence is something that cannot be changed) and a state orientation, respond by decreasing goals when faced with large negative discrepancies. Thus, Rohrback and Lord (1991) argue that personality is a perceptual schema that helps individuals interpret and respond to discrepancies. Their study, however, did not measure satisfaction, so the relationship between personality and satisfaction could not be assessed directly.

It is suggested here that differences in NA are likely to influence the affective reactions that individuals have to their performance. Wright and Mischel (1982) have shown that transient mood states influence individual's evaluations of their own performances. They had subjects perform a mental rotation task in either a positive, neutral, or negative mood state. Feedback was manipulated by the experimenters such that in one condition subjects were informed that they answered 45 problems correct and 12 incorrect (success), while in the other condition subjects were told they had 28 correct and 29 incorrect (failure).

Subsequently, subjects rated their satisfaction with their performance. Positive affect subjects rated their performance more positively than neutral and negative affect subjects and this effect was due to differences in the failure condition. In addition, Watson and Clark (1984) 37 cite evidence showing that individuals higher in NA accept negative information about themselves more easily than do lower NA individuals and tend to ruminate and exaggerate their mistakes and failures (Block, 1965; Zahn, 1960).

Hence, higher NA individuals may be more sensitive to negative performance-goal discrepancies than lower NA individuals. As a result, when provided with information that they have failed to meet their goal (either assigned or self-set), individuals higher in NA will report a lower evaluation of their performance than those lower in NA. The high self-concept of the lower NA individual may buttress him or her from negative self-evaluation perhaps by defensive or ego-enhancing distortion (Wright & Mischel,

1982).

Making an a priori prediction for affective reactions under conditions of success is not as straightforward. One reason is that in the Wright and Mischel (1982) study no significant differences were found for self-evaluations under conditions of success. However, some evidence suggests that even in an otherwise "positive" situation, individuals higher in NA may perceive certain aspects of that situation more negatively than their lower NA counterparts. As stated previously, Necowitz and Roznowski

(in press) reported that higher NA subjects in an "enriched" laboratory condition perceived their task as having less identity and autonomy than lower NA subjects. Furthermore, in a field study, these authors found that when respondents 38 were relatively satisfied with their jobs, higher NA individuals performed significantly more withdrawal behaviors than lower NA individuals. The authors reasoned that the higher NA respondents were perhaps dwelling upon the minor frustrations and irritations that are bound to occur on the job even though they were satisfied on a global level. Lower NA individuals may be more able to shrug off these minor problems. Thus, these results suggest that higher NA individuals may remain focused on the negative aspects of events and situations even under relatively positive conditions. Thus, the following is hypothesized:

H4: Individuals higher in NA will evaluate their performance more negatively than individuals lower in NA under conditions of both success and failure.

Relationship to Commitment Following Failure

and Self-Set Goals Following Success

The next dependent variable of interest will be goal commitment, with particular interest in whether NA influences goal commitment following failure. Viewed in terms of control theory (Campion & Lord, 1982; Klein, 1989;

Lord & Ranges, 1987), when an individual receives feedback that he or she is not meeting his or her goal, a decision is made to either change one’s behavior, that is, to increase one's effort, or to lower one's goal. An important research question is the degree to which the propensity to increase effort or to lower goals is a function of individual 39 differences, as it is the intensification of effort that will lead to higher performance levels.

To date, little research has examined whether personality influences individual's choices to reduce discrepancies by cognitive or behavioral means. As mentioned previously, Rohrback and Lord (1991) studied this issue and found that differences in state versus action orientation, and entity versus incremental beliefs about intelligence predicted cognitive or behavioral responses when faced with negative discrepancies. It is argued here that differences in NA will also predict responses to negative discrepancies. Because individuals higher in NA are more likely to ruminate over and exaggerate their failure experiences than those lower in NA, they may be more likely to lower their goal rather than increase their effort. The higher self-confidence of the lower NA individual should translate into increased effort to achieve the goal. Operationally, lower NA individuals should report higher levels of commitment to goals that are not initially obtained than higher NA individuals. Indeed, Campion and

Lord (1982) conceptualize goal commitment as ... "an unwillingness to subsequently reduce goals to a lower level when confronted with error signals" (p. 268).

H5: Following feedback indicating that they have failed to reach their goal, individuals lower in NA will be more committed to the unmet goal on subsequent trials than those higher in NA. 40

As stated above, evidence suggests that individuals higher in NA focus selectively on the negative aspects of

situations even under relatively pleasant conditions. Thus,

it was argued that they may evaluate their performance more negatively than lower NA individuals under conditions of

success. This discounting of their own successes may have

consequences for the goals they subsequently attempt to pursue. In other words, because higher NA individuals

evaluate the same "success" level more negatively they may choose to pursue lower goals on subsequent trials than lower

NA individuals.

H6: When individuals succeed in attaining their goals, higher NA individuals will set lower subsequent goals than

lower NA individuals.

Relationship to Attributions

The final dependent variable is the types of attributions individuals make for their performance. In general, people tend to have a bias toward perceiving

themselves favorably. Social psychologists refer to this as the self-serving bias (Brown & Rogers, 1991; Miller & Ross,

1975; Zuckerman, 1979). One manifestation of this bias is that individuals tend to take credit for their success, attributing it to internal and/or stable causes, like ability or effort, whereas they tend to deny blame for failure, by attributing it to external and/or unstable causes like the difficulty of the task or bad luck. For example, Chacko and McElroy (1983) had subjects perform a 41 proofreading task and presented them with bogus feedback that they had either surpassed or failed to meet their goal.

They found that subjects who had succeeded tended to make more internal and stable attributions for their success, while those that failed attributed their performance to more external, unstable causes. Similarly, Mone and Baker (1992) found that positive goal-performance discrepancies among students pursuing course grade goals were related positively to internal, stable attributions, whereas negative goal- performance discrepancies were related to external, unstable attributions.

Although self-serving appears to be a general bias, it is not universal. For example, research in the depression literature has shown that depressed individuals tend to make different attributions for personal events than non­ depressed individuals. Specifically, depressed individuals tend to habitually attribute bad events (e.g., failures) to causes that are internal, stable, and global, and good events (e.g., successes) to causes that are external, unstable, and specific (Abramson, Seligman, &Teasdale,

1978; Peterson & Seligman, 1984). Although this type of

"explanatory style" is proposed to be specific to depressives, other evidence suggests that its pervasiveness is more general. For example, Heimberg, Klosko, Dodge,

Becker, & Barlow (1989) found that non-depressed, but anxious patients, exhibited this same type of explanatory style as depressed patients. In addition, other researchers (e.g., Nezu, Nezu, & Nezu, 1986) have found high

correlations between explanatory style, scales measuring NA,

and depression, suggesting that this habitual style may be more related to NA, than specific to depression only.

Perhaps to reflect this greater generality, what Seligman and his colleagues once termed the "depressive explanatory style" (e.g., Peterson & Seligman, 1984), is now being referred to as the "pessimistic explanatory style" (e.g.,

Burns & Seligman, 1991; Peterson, Seligman, & Vaillant,

1988). Clark and Watson (1991) suggest that this style represents one important cognitive component of NA.

Hence, when individuals higher in NA succeed on a task

they may be more likely to explain their performance as being due to the ease of the task, or luck, and their failure as due to their lack of ability or effort.

Conversely, those lower in NA may take more credit for their successes and externalize blame for their failures.

H7a: Individuals higher in NA will be more likely to attribute their successes to the ease of the task or luck rather than their ability or effort, and will be more likely to attribute their failures to ability or effort, rather than to the difficulty of the task or luck. Lower NA

individuals will show the opposite pattern.

Furthermore, while goal-setting and control theory researchers have argued that affective reactions are determined by the direction and magnitude of the goal- performance discrepancy, Mone and Baker (1992) have recently 43 posited that individual's causal attributions moderate the relationship between discrepancies and satisfaction. They suggest that satisfaction with success will be higher if an individual attributes success to internal (e.g., ability, effort) rather than external causes (e.g., task difficulty, luck). Conversely, individuals will be more dissatisfied with failure to the extent failures are attributed to internal, rather than external causes. Thus, there should be a stronger positive relationship between goal-performance discrepancies and satisfaction for those who attribute performance to internal as opposed to external causes. The authors found support for their hypothesis. More positive goal-performance discrepancies led to higher satisfaction for those who attributed success to internal causes, and more negative goal-performance discrepancies led to lower satisfaction for those who attributed failure to internal causes.

These results imply that the influence of NA on satisfaction might be partially mediated by attributions.

For example, in the success condition, NA might have a direct influence on satisfaction, as well as being partially mediated by attributions if higher NA individuals attribute their success to external causes, and lower NA individuals attribute success to internal causes. Similarly, in the failure condition, partial mediation might be implicated if higher NA individuals attribute their performance to 44 internal causes and lower NA individuals attribute their performance to external causes.

H7b; The influence of NA on satisfaction is partially mediated by causal attributions. Higher NA individuals who succeed will be less satisfied than lower NA individuals who succeed partially because the former will make external attributions (i.e., task difficulty, luck), while the latter will make internal attributions (i.e., ability, effort).

Higher NA individuals who fail will be less satisfied than lower NA individuals who fail partially because the former will make internal attributions, and the latter will make external attributions. This hypothesis is depicted in

Figure 4. 45

Performance Satisfaction t

Attributions NA/SWB (Internal/External)

Figure 4

Summary of Hypothesis 7b CHAPTER II

METHOD

Subjects

Subjects were 275 undergraduate students enrolled in an

introductory psychology course at the Ohio State University.

Participation in the experiment fulfilled part of a course requirement. Subjects were run in groups of 4 individuals, and groups were randomly assigned to experimental conditions.

Design

This study examined the relationship of negative affectivity and subjective well-being to goal-setting processes and task performance. There were two manipulated variables, goal origin and feedback type. Goal origin had two levels. Half of the subjects set their own goals. The other half of the subjects were assigned goals by the experimenter. Feedback also had two levels. Half of the subjects were provided with feedback informing them that they had surpassed their goal on each trial. The other half were Informed that they had failed to meet their goal on each trial. Negative affectivity and subjective well-being were measured with several self-report scales described below.

46 47

Procedure

Subjects signed up for the experiment at a session time posted by the experimenter. The maximum number of subjects run at one time was four. Subjects were told that the study involved examining the influence of goals on performance.

Each subject was seated at a computer terminal. Before beginning the task, subjects were asked to complete several self-report questionnaires. They were told that this would help the experimenters determine whether certain individuals approach tasks of this type differently. Subjects were asked two demographic questions (sex, year in school) and then completed scales assessing negative affectivity and subjective well-being. The NA and SWB measures appeared in alternating order.

Next, subjects were introduced to the task. They were told that they would be completing several performance trials of a perceptual speed task. This task has been used in a number of other goal-setting studies (e.g., Matsui,

Kakuyama, & Onglatco, 1987; Matsui, Okada, & Kakuyama,

1982). In this task, subjects are presented with a string of 20 single-digit, randomly generated numbers, and asked to count how many of a certain number is in each string (e.g.,

"How many 3's are there?"). Subjects enter their answer and the computer presents another string for them to solve. At the end of each trial subjects are told the number of strings that they solved correctly. 48

To familiarize the subjects with the task, they were asked to complete two practice trials. Subjects were told to try to solve correctly as many strings as they could in each trial. Trials were 2 minutes in length. After subjects completed the two practice trials they were provided with feedback on their practice performance in the form of the number of strings solved correctly on each trial. To control the baseline for self-set subjects setting their goals, all subjects were told that they solved

14 strings correctly on the first practice trial, and 17 strings correctly on the second practice trial. This feedback represented the average number solved correctly by a sample of pilot subjects.

Subjects then completed a series of three performance trials. Each trial proceeded in an identical manner.

Subjects set a goal for themselves or were assigned a goal for the number of strings to solve correctly on each trial.

Then, they completed a goal commitment questionnaire and performed the task for two minutes. After performing, feedback (success/failure) was provided and subjects completed a questionnaire assessing their satisfaction with

their performance, and several questions about their attributions for their performance. After the third performance trial, subjects completed several manipulation check items, and questions assessing their comfort with using computers. Finally, subjects were debriefed and dismissed. 49

Manipulations

Goal Origin. The two levels of goal origin were goals either set by the subjects or assigned by the experimenter.

After the practice trials, subjects in the self-set condition were asked to set a challenging, but attainable goal for themselves in terms of the exact number of strings that they would attempt to solve correctly on the next trial. Subjects in the assigned condition were assigned goals by the experimenter. So as not to confound goal origin with goal difficulty, and to match subjects on ability, each subject in the assigned condition was "yoked" with a subject in the self-set condition. This was done by dividing the self-set subjects into ability subgroups based on their performance on practice trial 2 (high, medium, low). The computer then determined for each assigned condition subject which of these ability groups he/she was in (by using his/her score on practice trial 2). Then, the computer randomly assigned to the subject the goal sequence set by a self-set subject in that subgroup. Thus, subjects in the assigned condition had a similar goal difficulty distribution as subjects in the self-set condition, and were matched on ability based on their practice performance.

Feedback Type. The two levels of feedback were either success or failure. Subjects in the failure condition were

told that they fell short of their goal by a certain number of strings on each trial, and subjects in the success condition were told that they surpassed their goal by a 50 certain number of strings on each trial. The magnitudes of success and failure were 3, 2, and 3 strings for trials 1,

2, and 3 respectively. The magnitudes were the same regardless of the actual number of strings solved correctly.

For example, a subject in the failure condition with a goal of 20 would have been told, "You answered 17 strings correctly on this trial. This is 3 below your goal." A subject with a goal of 25 would have been told the same thing except the number 17 would have been replaced by the number 22.

This particular sequence was selected because pilot studies indicated that the discrepancy between goals and performance was large enough so that subjects felt a sense of success or failure, but was not too large as to be not credible.

The only variability between the success and failure conditions was that in the success condition, subjects were asked to set, or were assigned, another goal on each subsequent trial, while in the failure condition subjects continued to try for the goal that they failed to meet.

Because success subjects thought that they had surpassed their goal, it did not make sense to ask their commitment to

that goal on the upcoming trial. It was more interesting to see how personality influenced goal level on the next trial.

In the failure condition, it was valid to assess goal commitment as this provided a measure of goal persistence

that has been advocated as an alternative to a subsequent 51 goal change discrepancy score (e.g., self-set-goal-2 minus self-set-goal-1; Hollenbeck et al., 1989). Logistically, the former measure was simpler to assess with the current design.

To summarize, subjects in this experiment performed a series of 5 trials of a perceptual speed task. Subjects first completed several measures assessing negative affectivity and subjective well-being. Then, they performed two practice trials to familiarize themselves with the task and to provide the self-set condition subjects with a reference point for setting their goals. Subjects completed three performance trials and answered questionnaires pertaining to goal commitment, satisfaction with performance, and attributions for performance. Subjects in the self-set condition set their own performance goals, while subjects in the assigned condition were provided with goals to pursue (subjects in the self-set condition set the goals which were "yoked" to subjects in the assigned condition based on ability). Subjects were told either that they failed to achieve their goal over the three trials, or were told that they surpassed their goal over the trials.

Subjects in the failure condition tried for the same goal over the three trials, whereas subjects in the success condition had goals that varied over the three trials. 52

Measures

Those measures not shown in their entirety in the text are presented in Appendix A.

Demographics. Subjects were asked to provide their sex and their year in school.

Subjective well-being. Subjective well-being was assessed with four measures. These included:

The Affects Balance Scale-Bradburn (1969). This scale contains a list of 22 adjectives describing affective states (e.g., nervous, bitter, joyous, contented). Subjects respond to the question "In general, how often do you have each of these feelings?" Scores can be derived for both positive and negative affect, as well as a balance score reflecting the difference between positive and negative affect. This latter measure was used in keeping with the definition of SWB. The balance score was calculated by averaging responses to the positive and negative items and subtracting the latter from the former. This resulted in a scale ranging from -4 to +4. A constant of 5 was added to each subject's score yielding a scale ranging from 1 to 9, with lower scores reflecting more negative affect and higher score reflecting more positive affect. A recent study estimated coefficient alpha of this scale to be .92 (Judge &

Hulin, 1992).

Fordvce's (1977) Happiness Scale. This scale contains two items. The first item asks subjects how happy or unhappy they usually feel from O-extremely unhappy 53

(utterly depressed, completely down) to 10-extremely happy

(feeling ecstatic, joyous, fantastic!). The second item asks respondents to estimate the percent of time out of 100%

that they feel happy. The percent item was divided by 10, and added to the first item, yielding a two-item scale ranging from 0-20, with higher scores representing more happiness.

Frequency of Positive and Negative Affect Scale-

Underwood and Fromine (1980). This is a six-item scale assessing the extent to which subjects generally experience positive and negative affective states. Subjects indicate the degree to which they agree or disagree with each item

(1-strongly disagree, 7-strongly agree). Two sample items are: "I am cheerful less often than other people." and "I often feel down in the dumps." Higher scores indicate more positive affective states. Coefficient alpha has been estimated at .91 (Judge & Hulin, 1992).

The Satisfaction With Life Scale (SWLS)-Diener.

Emmons. Larsen, and Griffin (1985). The SWLS is a five-item scale with subjects indicating the extent to which they agree or disagree with each statement (1-strongly disagree,

5-strongly agree). Two sample items are: "I am satisfied with my life." and "The conditions of my life are excellent.” Higher scores reflect greater satisfaction.

Coefficient alpha for this scale has been estimated at .87

(Diener et al,, 1985) and .83 (Pavot, Diener, Colvin, &

Sandvik, 1991). 54

Negative affectivity. Negative affectivity was operationalized with three self-report scales.

The Positive and Negative Affect Schedule (PANAS)-

Watson. Clark, and Tellegen (1988). The PANAS is a twenty-

item measure that asks respondents the extent to which they have felt various emotional states (1-very slightly or not at all, 5-extremely). Ten items are used to assess negative affectivity (e.g., distressed, nervous), and 10 items assess positive affectivity (e.g., interested, active).

Coefficient alpha has been estimated at .88 for PA and .87 for NA. Scores on this scale indicate adequate stability over time (Watson et al., 1988).

The Negative Affectivity Scale (NAS)-Stokes and

Levin (1990^. This is a 21-item measure where subjects

indicate the extent to which they agree or disagree with each item on a seven-point scale (1-disagree strongly,

7-agree strongly). Two sample items are "I am hopeful and optimistic about the future" and "I always expect the worst to happen." Higher scores indicate higher levels of NA.

Stokes and Levin (1990) report alpha estimates of .86, .84, and .85 in three validation samples. A six-week test-retest reliability was reported at .88.

The Trait Anxiety Scale (TAS)-Spielberger 0,979).

This scale is from the larger State-Trait Personality

Inventory. The scale contains 10 descriptive statements and subjects respond with the frequency that they feel as

indicated in the statement (1-almost never, 4-almost 55 always). Two sample items are, "I feel nervous and restless" and "I lack self-confidence." Higher scores indicate higher levels of NA. Internal consistency for this scale ranges from .80 to .85 (Spielberger, 1979).

Self-set goals. In the self-set condition, subjects' goals were determined by asking the following, "On the next trial, we would like you to set a goal for yourself. In other words, please set a challenging goal for the number of strings that you will be trying to solve correctly on the upcoming trial. The goal that you set should be a challenging one, yet one you think you can achieve."

Goal Commitment. Goal commitment was assessed with

Hollenbeck, Klein, O'Leary, & Wright's (1989) self-report measure. This scale contains seven items and subjects report the extent to which they agree or disagree with each statement (1-strongly disagree, 5-strongly agree). Higher scores indicate higher levels of goal commitment.

Coefficient alpha is estimated at .80 (Hollenbeck et al.,

1989). The measure correlates significantly with other operationalizations of goal commitment, including a discrepancy measure between self-set goals and assigned goals, as well as measures of subsequent goal change.

Performance. Performance was operationalized in terms of the number of strings the subject solved correctly during each performance trial.

Performance satisfaction. Performance satisfaction was assessed with a 5-item scale taken from Klein (1987). 56

Subjects indicate the extent to which they agree or disagree with each statement on a 5-point scale (1-strongly disagree,

5-strongly agree). Higher scores indicate higher levels of satisfaction. Coefficient alpha of the scale is estimated at .94 (Klein, 1987).

Performance attributions. Four questions were used to determine subjects' attributions for their performance on each trial. These attributions are based on Weiner and his colleagues' research (Weiner, 1974; Weiner, Frieze, Kukla,

Reed, Rest, & Rosenbaum, 1971) indicating that in achievement situations individuals attribute their performance to four causal sources that differ on the dimensions of internality-externality and stability-

instability. These four causes are ability (internal and

stable), effort (internal and unstable), the difficulty of

the task (external and stable), and luck (external and unstable). Responses were on a 5-point scale (1-not at all

likely, 5-extremely likely) to the following question:

To what extent was your performance on this trial due to: a) your ability. b) your effort. c) the nature of the task. d) luck or chance.

Manipulation checks. Several questions were asked to determine the success of the task manipulations. These

questions were on a 5-point scale (1-strongly disagree,

5-strongly agree). The items were: 57

1) I was able to personally decide how many strings I would try for on AT LEAST ONE trial in this experiment.

2) I solved more strings than my goal on this task.

3) The feedback I received was an accurate evaluation of my performance.

4) I am good at this task.

5) The computer gave me a goal of how many strings to solve.

6) The feedback I was given did NOT reflect my true performance.

7) Relative to my goals, I performed poorly on this task.

8) I fell short of my goal on each trial.

Computer anxiety. Subjects were asked to respond to three items that assessed their comfort with using computers. Items were on a five-point scale (1-strongly disagree, 5-strongly agree), and included:

1) I am confident in my computer skills.

2) I have computer "anxiety."

3) Working with computers does NOT concern me.

Higher scores indicate greater comfort with using computers. CHAPTER III

RESULTS

Part 1-Examination of the Relationship Between NA and SWB

Analysis of convergent correlations.

Table 1 presents a correlation matrix for the seven NA and SWB indicators. Cronbach alpha scale reliability estimates are presented on the diagonal. Alphas ranged from

.73 to .90, indicating adequate scale reliabilities.

All correlations are statistically significant, and in many cases are rather high. To examine these correlations further, averages were calculated. The average correlation between the NA scales (3 correlations) was .63 (SD-.14).

Notable here was a particularly high correlation between the

Negative Affectivity Scale and the Trait Anxiety Scale of

.83. The average correlation between the Subjective Well-

Being scales (6 correlations) was .56 (SD-.09). These averages indicate a substantial amount of commonality between scales assessing these constructs.

Next, the correlations between the NA and SWB scales were averaged. If NA and SWB are distinct constructs, then this correlation should be low when compared to the within construct correlations. The correlation between each NA scale and each SWB scale (12 correlations) averaged -.53

(SD-.14). Thus, this average is only reduced slightly from

58 59

Table 1

Correlation Matrix for Personality Scales

Variable 1 2 3 4 5 6 7

1. PANAS-NA .86

2. NAS .50 .87

3. TAS .56 .83 .86

4. UF -.38 -.68 -.58 .79

5. FHQ -.39 -.54 -.51 .58 .73

6. AB - .48 -.71 -.71 .71 .59 .90

7. SWLS -.24 -.59 -.52 .48 .42 .58 .83

Note. N-275. All correlations significant at pc.01. PANAS-NA-Negative Affect Scale (Watson et al., 1988); NAS-Negative Affectivity Scale (Stokes & Levin, 1990); TAS-Trait Anxiety Scale (Spielberger, 1979); AB-Affect Balance Scale (Bradburn, 1969); UF-Underwood & Froming (1980) Frequency of Positive and Negative Affect Scale; FHQ-Fordyce (1977) Happiness Scale; SWLS-Satisfaction with Life Scale (Diener et al., 1985). 60 the within construct averages of .63 and .56. This high average correlation provides fairly strong evidence that these seven scales may be assessing one common construct.

Analysis of confirmatory factor analysis models.

The four confirmatory factor analysis models discussed previously were fit to the observed data using the maximum likelihood method provided in the LISREL Version 7.16 computer program (Joreskog & Sorbom, 1989). The matrix of correlations between scales served as input to the program.

Each model's goodness of fit to the data was evaluated, and specific model comparisons were made.

Several goodness of fit indices are available. The most common is the chi-square test. The chi-square test evaluates the null hypothesis that the model holds exactly in the population. If the chi-square is significant, the null hypothesis of exact fit is rejected. If the chi-square is not significant, then it is plausible that the model holds exactly in the population.

The chi-square test has been criticized on several grounds. First, the null hypothesis is not tenable. No model holds exactly in the population. Rather, models fit more or less well. Thus, the null hypothesis cannot be considered valid (Browne & Cudeck, 1993). In addition, the value of the chi-square statistic is dependent on sample size. The larger the sample size, the higher the chi-square value, and the more likely one will reject the null hypothesis. As a result of these problems, the chi-square 61 should not be the sole criterion by which the fit of models is evaluated.

A number of other fit statistics have been proposed.

The three measures that will be used here are the goodness of fit index (GFI; Tanaka & Huba, 1985), the non-normed fit index (NNFI; Bentler & Bonnett, 1980), and the root mean squared residual (RMSR; Steiger, 1989). The GFI represents the ratio of the sum of squares accounted for by the model to the sum of squares obtained from the data matrix. It ranges from 0 to 1 with values above .90 indicating that the model fits well. The NNFI compares the improvement in fit from the specified model compared to the null model. That is, how much better is the substantive model compared to a model specifying that the measured variables are uncorrelated in the population. The NNFI ranges from 0 to 1 with models having values of .90 or above considered to fit the data well. The RMSR measures the degree to which covariances generated by the hypothesized model fit the observed covariances. The smaller the RMSR, the better the fit of the model. This statistic can be interpreted on the scale of the correlation coefficient when using a correlation matrix as input. RMSR values less than .05 indicate that the model fits well.

Comparison of Model 1 and Model 2 .

A comparison was made between the fit of Model 1 and

Model 2. Recall that Model 1 hypothesized that the seven scales could be represented adequately by a single factor, 62 while Model 2 specified two latent factors, one for the NA

scales and one for the SWB scales. Results for these

analyses are presented in Table 2.

The table indicates that both models fit the data well.

For the simpler one-factor model, all of the fit measures

fall within the guidelines for good model fit. The GFI is above .90, and the NNFI is .899. In addition, the RMSR is below .05. Furthermore, all factor loadings were high,

ranging from a low of .569 for the PANAS-NA scale to a high

of .901 for the NAS scale. Model 2 fits the data slightly better than Model 1, as evidenced by higher GFI and NNFI

statistics, and a slightly lower RMSR.

There are several reasons, however, to prefer Model 1.

First, since Model 1 fits the data adequately according to

conventional requirements, it is preferred due to its parsimony. Second, Model 2 fits better than Model 1 due primarily to permitting the two latent factors to be correlated. The correlation between the factors is -.884, which is rather large. To examine how much influence this path had on the fit statistics, a model was examined where

the correlation between the two factors was constrained to 2 zero. This model fit the data quite poorly (x -301.11,

df-14, p-.OOO, GFI-.827, NNFI-.623, RMSR-.356). Thus, a

two-factor independent model is not tenable. The poor fit

of this model, coupled with the extremely large correlation

between the factors, points to Model 1 as the best

representation of the data. 63

Table 2

Results of LISREL Analyses for Model 1 and Model 2

2 Model X df P GFINNFIRMSR

1 91.14 14 .000 .909 .899 .046

2 42.00 13 .000 .958 .959 .031

Hull 1162.29 21 .000 .343 — .490

Note. N-275. GFI-Goodness of Fit Index (Tanaka & Huba, 1985); NNFI-Non-Normed Fit Index (Bentler & Bonett, 1980); RMSR-Root Mean Squared Residual (Steiger, 1989). 64

To support this view further, an exploratory factor

analysis was conducted. Principal axis factoring with

squared multiple correlations as communality estimates was

used to fit the common factor model to the observed data.

The number of factors to retain was determined by the

discontinuity technique. This method is recommended over

"rule of thumb" approaches, such as eigenvalues greater than

one, because it is less arbitrary and more accurate (Ford,

MacCallum, & Tait, 1986). A scree plot of factors by

eigenvalues clearly indicated that a strong first factor

could adequately describe the data. This first factor had

an eigenvalue of 3.96, with the second factor having an

eigenvalue of .22. The first factor also explained nearly

100% of the variance. All factor loadings were high,

ranging from a low of -.56 for the PANAS-NA scale to a high of -.88 for the NAS.

Comparison of Model 3 and Model 4 .

Model 3 and Model 4 were compared to examine whether or not SWB is a broader construct than NA. That is, if NA can be subsumed by SWB, Model 4 should not fit significantly better than Model 3. In addition to examining descriptive

goodness of fit indices, one can use a chi-square difference

test to examine whether or not Model 4 fits better. This is because Model 3 is nested in Model 4, meaning that the free parameters in Model 3 are a subset of the free parameters in

Model 4. The null hypothesis is that the two models are equivalent. The test statistic is the difference between 65 2 2 the chi-squares in each model (x^-X^). with degrees of freedom, df^-df^. If this statistic is significant, the null hypothesis of model equivalence can be rejected and one can conclude that Model 4 fits significantly better than

Model 3. If the chi-square difference statistic is not significant, then Model 3 is preferred as it is more parsimonious.

Results for the analysis are presented in Table 3.

Both models fit the data well. The measures of goodness of fit are above the acceptable ranges (.90 or higher for the

GFI and the NNFI, .05 or lower for the RMSR). Model 4's fit

Indices are slightly better than Model 3's fit statistics.

The difference in chi-squares between the models is 42.92

(91.14-48.22), and the difference in degrees of freedom is 3

(15-12). A chi-square of 42.92 with 3 degrees of freedom is 2 significant at a .01 level of significance (x q ^-5.841).

Thus, Model 4 fits significantly better than Model 3. This means that the hypothesis that NA is subsumed by SWB cannot be supported. Further implications of this finding will be discussed in Chapter IV.

Construction of an Affect Composite.

Because the findings indicated that a one-factor model could describe the data rather well and was parsimonious, a composite measure was created to use in testing the goal- setting hypotheses, in addition to examining each scale individually. This composite included the seven scales used in the confirmatory factor analyses, as well as the PANAS-PA 66

Table 3

Results of LISREL Analyses for Model 3 and Model 4

2 Model X df P GFI NNFI RMSR

3 91.14 15 .000 .909 .907 .046

4 48.22 12 .000 .953 .944 .037

Null 1162.29 21 .000 .343 -- .490

Note. N-275. GFI-Goodness of Fit Index (Tanaka & Huba, 1985); NNFI-Non-Normed Fit Index (Bentler & Bonett, 1980); RMSR-Root Mean Squared Residual (Steiger, 1989). 67

scale. The PANAS-PA scale was not used in the confirmatory

analyses because it is not a measure of NA according to

Uatson and Clark (1984, 1991). Nevertheless, an examination

of the correlations between the PANAS-PA scale and the other

seven scales (see Table 6) indicates high correlations between the PANAS-PA and all other personality scales,

except the PANAS-NA scale. Thus, it appears that this scale

is also measuring the common affect construct. A unit-

weighted composite, hereafter called the affect composite,

was constructed by reverse scoring the NA scales,

standardizing all eight scales, and then summing the

standardized scores. Thus, higher scores reflect more

positive affect and lower scores reflect more negative

affect. This composite will assess the underlying affect

construct better than any individual scale alone. In

addition, the reliability of the composite should be higher

than for the individual scales because there are a greater

number of items in the composite covarying highly with each

other than there are in each individual scale (Allen & Yen,

1979). Indeed, coefficient alpha for the composite scale

was .96 whereas alpha ranged from .73 to .90 for the

individual scales. 68

Part 2-Examination of Goal-Setting Hypotheses

Subjects' responses to the manipulation check items indicated that the experimental manipulation was successful.

Goal origin. Two items were used to assess the goal origin manipulation. Subjects in the self-set goal condition had a higher mean response to the item asking whether they were able to decide personally how many strings they would try for on at least one trial (M - 3.90, M - S S cLS

2.67, t(272)- 8.75, e <.0001). Subjects in the assigned condition had a higher average response than subjects in the self-set condition to the item asking whether the computer gave them a goal of how many strings to solve (M - 2.28, s s M -4.24, t(272)- -14.32, £<.0001). cLS Feedback type. Four items assessed whether subjects perceived accurately the success/failure manipulation. Two of these items were descriptive in nature and two were evaluative. Subjects in the success condition agreed to a greater degree than those in the failure condition that they solved more strings than their goal on the task (M - 4.30, SU M^a» 1.66, t(272)— 26.14, £<.0001), while subjects in the

failure condition more often indicated than subjects in the

success condition that they fell short of their goal on each

trial (Msu- 1.68, Mffl- 3.99, t(272)- -20.08, £<.0001). For

the evaluative items, success subjects had a higher mean

response than those in the failure condition to the item

asking whether they were good at the task (Mgu- 3.70,

3.12, t(272>- 5.66, £<.0001), while failure condition 69

subjects more often agreed than success condition subjects

that they had performed poorly on the task relative to their

goals (M - 1.87, ~ 2.73, t(272)- -7.35, £<.0001). SU la Feedback accuracy. Two items assessed whether subjects

thought the feedback they received was an accurate

evaluation of their performance. After reverse scoring the

second item, these questions were combined to yield a 2 -item

feedback accuracy scale ranging from 2 to 10. Higher scores

indicated greater perceptions of accuracy. Coefficient

alpha for the scale was .77. A 2x2 ANOVA was performed on

the scale scores to examine whether perceptions of accuracy

differed by condition. The ANOVA indicated that there was no difference between the self-set/assigned conditions (Mss“

6.50, Mfls- 6.27, F(l,270) - .82, ns), and no difference between the success/failure conditions (M - 6.49,

6.36, F(l,270) - .48, ns). The interaction was also not

significant (Mss s u - 6.52, Mssfa- 6.49, Mfls su- 6.43,

M f - 6.11, F(1,270) - .34, ns). Thus, subjects in each

condition perceived the feedback to be equally accurate. In

addition, all means are above the midpoint of the scale.

Thus, subjects generally thought the feedback was accurate.

Computer anxiety. To examine whether computer anxiety

differed by experimental conditions, subjects were also

asked to respond to three questions that assessed how

comfortable they were with using computers. A unit-weighted

composite was created from the three items. Coefficient

alpha for the scale was .40. This low reliability was due 70 to low correlations between item #3, the negatively worded item, and the other two items. Item #3 was subsequently dropped from the composite, yielding a two-item scale ranging from 2 to 10 with higher scores indicating less computer anxiety. The revised scale had an alpha of .64. A

2x2 ANOVA was performed on the scale to determine if computer anxiety differed by condition. The ANOVA indicated no difference in anxiety between the self-set/assigned conditions (M - 7.08, M - 7.42, F ( 1 ,270) - 1.98, ns), and S S 3S no difference between the success/failure conditions (Msu~

7.23, 7.16, F(l,270) - .09, ns). The interaction was also not significant (M - 7.11, M c - 7.04, M - sSySU s s ,13 a s ,su 7.46, M f - 7,38, F(1,270) - .00, ns). Thus, subjects did as j zb not differ in their computer anxiety by experimental conditions. The means indicated that subjects generally felt comfortable working with computers. (A 2x2 ANOVA was performed on the 3-item scale. Similar results were found.)

Table 4 contains descriptive statistics for the full sample at each trial, with the exception of goal level where data is based on self-set goal condition subjects only.

Table 5 presents descriptive data for the personality scales and is based on the full sample. Both Tables 4 and 5 indicate little range restriction for the variables. The observed scale ranges are close, or equivalent, to the possible scale ranges. Additionally, all variables have moderate to large standard deviations. 71

Table 4

Descriptive Statistics for Variables Assessed at Trial 1.

Trial 2. and Trial 3

Variable Possible Observed Mean SD Range Range

Trial 1

Goal Level N/A 0-40 17.98 4.47 Performance N/A 1-29 17.77 3.60 Number Attempted N/A 8-39 19.59 3.73 Goal Commitment 7-35 14-35 25.79 4.15 Satisfaction 5-25 6-25 17.87 3.99 Ability Attribution 1-5 1-5 3.42 1.26 Effort Attribution 1-5 1-5 3.72 1.29 Difficulty Attribution 1-5 1-5 3.63 .94 Luck Attribution 1-5 1-5 2.61 1.27

Trial 2

Goal Level N/A 2-45 20.16 5.05 Performance N/A 1-29 19.03 3.60 Number Attempted N/A 9-41 21.14 3.78 Goal Commitment 7-35 13-35 25.37 4.25 Satisfaction 5-25 7-25 18.62 3.68 Ability Attribution 1-5 1-5 3.34 1.26 Effort Attribution 1-5 1-5 3.55 1.35 Difficulty Attribution 1-5 1-5 3.53 .94 Luck Attribution 1-5 1-5 2.62 1.29

Trial 3

Goal Level N/A 2-45 21.68 6.62 Performance N/A 6-30 18.80 3.78 Number Attempted N/A 11-42 21.62 3.59 Goal Commitment 7-35 11-35 25.20 4.41 Satisfaction 5-25 5-25 17.99 3.93 Ability Attribution 1-5 1-5 3.31 1.36 Effort Attribution 1-5 1-5 3.47 1.40 Difficulty Attribution 1-5 1-5 3.56 1.03 Luck Attribution 1-5 1-5 2.66 1.33

Note. JJ-274, except for goal level where n-181. 72

Table 5

Descriptive Statistics for Personality Scales

Scale Possible Observed Mean SD Range Range

PANAS-PA 10-50 13-50 33.98 6.59

PANAS-NA 10-50 10-45 20.03 6.68

NAS 21-147 31-126 68.98 16.99

TAS 10-40 10-38 19.24 5.36

AB 1-9 2.58-8.24 6.19 .83

UF 5-30 9-30 21.77 3.85

FHQ 0-20 0-18.6 14.24 3.27

SWLS 5-35 5-35 23.22 5.61

Note. N-274. PANAS-PA-Positive Affect Scale (Watson et a l., 1988); PANAS- NA-Negative Affect Scale (Watson et al., 1988); NAS-Negative Affectivity Scale (Stokes & Levin, 1990); TAS-Trait Anxiety Scale (Spielberger, 1979); AB-Affect Balance Scale (Bradburn, 1969); UF-Underwood & Froming (1980) Frequency of Positive and Negative Affect Scale; FHQ-Fordyce (1977) Happiness Scale; SWLS-Satlsfaction with Life Scale (Diener et a l., 1985). 73

Mean responses for goal commitment and satisfaction

with performance were moderately high, as indicated by

averages that are above the scale midpoint. The same is

true for the ability, effort, and difficulty attributions, while the luck attribution was slightly below the midpoint

of 3. Goal level increased significantly over the three

trials (tr12(179)- 9.75, £<.0001; tr23(179)- 7.02, p<.0001),

as did the number of strings attempted (tr^2(272)- 13.95,

£<.0001; tr23(272)- 4.90, £<.0001). Task performance,

defined as the number of strings solved correctly, increased

from trial 1 to trial 2 (t(272)- 9.52, £<.0001), but

performance on trial 2 was not significantly different from

performance on trial 3 (t(272)- -1.37, ns).

Overall, the sample can be described as moderately high

in subjective well-being and moderately low in negative

affectivity. Mean responses were above the midpoint on the

SWB scales and below the midpoint on the NA scales.

Table 6 contains variable intercorrelations and

Cronbach's alpha scale reliabilities for multi-item scales.

Correlations are based on the full sample, with the

exception of correlations between the goal level variables

and the personality scales which are provided for self-set

subjects only. Reliabilities for all scales were high.

Alpha ranged from .73 to .90 for the personality scales.

For the attitudinal scales, alpha ranged from .75 for

commitment at trial 1 to .83 for each of the satisfaction

scales. 74

Table 6

Correlation Matrix for All Variables

Variable 1 2 3 4 5 6 7 8 9

1. PANAS-PA .86

2. PANAS-NA - .01 .86

3. NAS -.53 .50 .87

4. TAS -.36 .56 .83 .86

5. UF .46 - .38 -.68 -.58 .79

6. FHQ .37 - .39 -.54 -.51 .58 .73

7. AB .44 - .48 -.71 -.71 .71 .59 .90

8. SWLS .38 - .24 -.59 -.52 .48 .42 .58 .83

9. GL-TRl .13 .04 -.15 -.14 .04 .03 .01 .01 —

10. GL-TR2 .01 .10 -.09 -.06 - .02 -.04 - .05 -.01 .81

11. GL-TR3 .02 .18 -.02 .04 -.08 -.04 - .09 - .01 .64

12. Perf-TRl .15 - .04 -.09 -.05 .04 .06 .09 .03 .12

13. Perf-TR2 .09 - .07 -.04 -.01 .03 .08 .03 - .03 .11

14. Perf-TR3 .12 .04 -.02 .02 .00 .10 - .01 -.03 .02

15. Num-TRl .15 .02 -.07 -.01 .08 .02 .08 .04 .11

16. Num-TR2 .15 - .03 -.11 -.06 .09 .03 .10 .04 .14

17. Num-TR3 .12 - .03 -.08 -.05 .05 .04 .06 .05 .07

18. Com-TRl .27 - .03 -.25 -.15 .22 .15 .20 .22 .14 75

Table 6 (continued)

Variable 1 2 3 4 5 6 7 8 9

19. Cora-TR2 .30 -.10 -.29 -.21 .28 .20 .25 .21 .11

20. Cora-TR3 .28 -.01 -.26 -.16 .23 .13 .23 .19 .10

21. Sat-TRl .19 .06 -.14 -.11 .16 .10 .10 .18 .22

22. Sat-TR2 .22 .00 -.19 -.15 .19 .20 .16 .24 .12

23. Sat-TR3 .18 .02 -.14 -.11 .22 .15 .20 .17 .09

24. Abil-TRl .05 .00 -.06 -.03 -.02 .06 .00 .03 .11

25. Abil-TR2 .03 -.10 -.12 -.09 .02 .09 .04 .09 .08

26. Abil-TR3 .05 -.10 - .14 -.13 -.02 .10 .10 .07 .01

27. Efrt-TRl .08 - .06 -.11 - .13 .01 .16 - .02 .08 .04

28. Efrt-TR2 .08 -.01 -.07 -.06 .00 .13 - .04 .10 .05

29. Efrt-TR3 .08 -.01 -.09 -.08 .00 .17 .02 .05 -.01

30. Diff-TRl -.03 -.07 .00 .02 .03 .07 - .02 .00 .12 O H 31. Diff-TR2 -.09 -.04 -.01 .03 -.04 i l .07 - .01 .05

32. Diff-TR3 .08 -.01 -.06 -.01 .05 .07 .10 .04 .05

33. Luck-TRl .00 .06 .08 .08 .09 -.01 .03 - .01 - .17

34. Luck-TR2 .01 .07 .09 .12 .04 -.03 - .02 - .01 - .09 1 1 35. Luck-TR2 .00 .06 .04 .10 .08 0 .03 - .03 - .13 76

Table 6 (continued)

Variable 10 11 12 13 14 15 16 17 18

10. GL-TR2 —

11. GL-TR3 .91 —

12. Perf-TRl .11 .08 —

13. Perf-TR2 .10 .08 .81 —

14. Perf*TR3 .07 .07 .70 .72 —

15. Num-TRl .13 .09 .76 .62 .49 —

16. Num-TR2 .15 .12 .70 .73 .52 .88 —

17. Num-TR3 .11 .09 .62 .62 .50 .81 .90 —

18. Com-TRl .15 .13 .22 .18 .18 .21 .25 .24 .75

19. Com-TR2 .15 .14 .18 .15 .16 .17 .21 .21 .78 CM o Com-TR3 .15 .13 .16 .13 .18 .20 .22 .24 .73 CM Sat-TRl .38 .42 .13 .12 .13 .16 .16 .13 .15

22. Sat-TR2 .34 .40 .09 .11 .11 .16 .16 .15 .19

23. Sat-TR3 .28 .33 .09 .10 .13 .15 .16 .12 .23

24. Abil-TRl .26 .32 .01 .08 .05 .00 .04 .02 .19 O 00 25. Abil-TR2 .23 .26 .06 .19 .14 .17 .14 .19

26. Abil-TR3 .14 .15 .07 .15 .11 .11 .13 .09 .14 77

Table 6 (continued)

Variable 10 11 12 13 14 15 16 17 18

27. Efrt-TRl .26 .33 -.03 .00 -.02 -.07 -.03 -.05 .04

28. Efrt-TR2 .25 .33 .01 .06 .07 -.01 .03 .05 .07

29. Efrt-TR3 .20 .22 .06 .08 .13 .05 .06 .03 .12

30. Diff-TRl .11 .12 -.02 .02 -.01 .00 .02 .04 .02

31. Diff-TR2 .11 .13 -.04 .01 .03 -.01 .05 .06 .04

32. Diff-TR3 .11 .13 .06 .07 .10 .07 .11 .08 .08

33. Luck-TRl -.24 -.23 -.08 -.10 .00 -.05 -.07 -.10 -.16

34. Luck-TR2 -.18 -.17 -.07 -.13 -.05 -.06 -.08 -.09 -.12

35. Luck-TR3 -.20 -.22 -.06 -.11 -.06 -.04 -.07 -.09 -.11 78

Table 6 (continued)

Variable 19 20 21 22 23 24 25 26 27

19. Com-TR2 .81

20. Com-TR3 .83 .80

21. Sat-TRl .26 .28 .83

22. Sat-TR2 .32 .34 .79 .83

23. Sat-TR3 .36 .41 .69 .80 .83

24. Abil-TRl .24 .22 .31 .32 .28 —

25. Abil-TR2 .30 .27 .30 .34 .33 .68 —

26. Abil-TR3 .25 .24 .26 .31 .48 .55 .69 —

27. Efrt-TRl .11 .04 .30 .31 .27 .47 .40 .40 —

28. Efrt-TR2 .10 .10 .34 .36 .33 .41 .54 .42 .72

29. Efrt-TR3 .18 .18 .35 .38 .49 .36 .45 .64 .60

30. Diff-TRl .12 .11 .23 .22 .26 .27 .19 .19 .06

31. Diff-TR2 .13 .10 .19 .24 .21 .28 .31 .22 .09

32. Diff-TR3 .18 .16 .25 .29 .32 .21 .22 .34 .01

33. Luck-TRl -.18 -.21 .18 -.22 -.20 -.25 -.29 - .23 -.22

34. Luck-TR2 -.15 -.26 - .18 -.26 -.29 -.25 -.30 - .25 -.16

35. Luck-TR3 -.16 -.20 .12 -.21 -.25 -.21 - .31 -.26 -.17 79

Table 6 (continued)

Variable 28 29 30 31 32 33 34 35

28. Efrt-TR2 —

29. Efrt-TR3 .72 —

30. Diff-TRl .04 .03 —

31. Diff-TR2 .16 .09 .61 —

32. Diff-TR3 .07 .23 .44 .64 —

33. Luck-TRl -.26 ■-.19 -.04 ■-.06 .09 —

34. Luck-TR2 -.25 ■-.25 -.04 ■-.05 - .05 .76 —

35. Luck-TR3 -.21 ■-.20 .00 •-.04 .03 .67 .77 ---

Note. N-274, except for correlations between goal level and personality scales where n-181. Diagonals contain scale reliabilities (alphas). For r> |.12|, £<.05; for r>[.15|, £<.01. For goal level/personality correlations: r>|.14|, p<.05; r> |.17|, £<•0 1 . PANAS-PA-Positive Affect Scale (Watson et al., 1988); PANAS- NA-Negative Affect Scale (Watson et al., 1988); NAS-Negative Affectivity Scale (Stokes & Levin, 1990); TAS-Trait Anxiety Scale (Spielberger, 1979); AB-Affect Balance Scale (Bradburn, 1969); UF-Underwood & Froming (1980) Frequency of Positive and Negative Affect Scale; FHQ-Fordyce (1977) Happiness Scale; SWLS-Satisfaction with Life Scale (Diener et al., 1985); GL-Goal Level; Perf-Task Performance; Num-Number Attempted; Com-Goal Commitment (Hollenbeck et al., 1989); Sat-Performance Satisfaction (Klein, 1987); Abil-Ability Attribution; Efrt-Effort Attribution; Diff-Difficulty Attribution; Luck-Luck Attribution; TRl-Trial 1; TR2-Trial 2; TR3-Trial 3. 80

Tests of Hypotheses

Separate analyses by gender and year in school yielded no differences in results. Thus, results presented are based on the entire sample.

Hypothesis 1.

Hypothesis 1 predicted that individuals lower in negative affectivity and higher in subjective well-being would set higher goals than those higher in NA and lower in

SWB. This hypothesis was tested by correlating goal level

at Trial 1 for those in the self-set goal condition with

each personality scale and the affect composite. Results

for each scale are shown in Table 6. All of the SWB scales were not related significantly to goal level. The Negative

Affectivity Scale (NAS) and the Trait Anxiety Scale (TAS) were both correlated significantly with goal level. The

correlation between the NAS and goal level was -.15 (e<05),

and the correlation between the TAS and goal level was -.14

(£<.05). The PANAS-NA scale was not related to goal level.

The composite affect scale was also not related to goal

level (r-.08, e>.10). This latter result is not surprising

given the nonsignificance of all of the SWB measures.

Though it is tempting to conclude from this analysis

that lower NA is related to setting higher goals, this may

be an overstatement. When the data was examined more

closely, it was discovered that three individuals had set a

goal of 0 on the first trial, one had set a goal of 1, and

one had set a goal of 3. These data points presented a 81 concern as they were clear outliers. They were also influential data (Sackett & Larson, 1990; Stevens, 1984) in the sense that that they affected the significance of the correlation coefficients. Two options for handling these data points were considered. Orr (1988) noted that 67% of the researchers he surveyed suggest removing outliers of this nature only if there is a clear reason for the researcher to think they are invalid. Kruskal (1960), however, has suggested that investigators report findings including and excluding influential data.

These suggestions were combined in the following manner. First, data presented above include the influential data. Next, a subsequent analysis was conducted where those who set a goal level of 0 were dropped, while those who set a goal level of 1 or 3 were included. It was thought that a goal of 0 may have been more likely the result of misunderstanding the instructions. Without those who set a goal of 0, the significant correlations dropped to nonsignificance. The correlation between the NAS and goal level was -.10 (j>>.10), and the correlation between the TAS and goal level was -.07 (e>.10). Thus, when this secondary analysis in considered, hypothesis 1 is not supported.

Of the three individuals who set a goal of 0, one was in the failure condition, while the other two were in the success condition. Because the failure condition subject would have received feedback below 0 on the first trial 82

(0-3- -3), this subject was dropped from all subsequent

analyses. Because the two subjects in the success condition

would have received feedback above 0 on the first trial

(0+3- 3), they were included in subsequent analyses. The

subjects who had set goals of 1 and 3 were also in the

success condition.

Hypothesis 2 .

Hypothesis 2 predicted that when goals are assigned,

lower NA and higher SWB individuals would be more committed

to their goals than higher NA and lower SWB individuals.

This hypothesis was tested by correlating commitment at

Trial 1 for subjects in the assigned condition with each

personality scale. Results are summarized in Table 7. The hypothesis was supported. All correlations between

personality scales and commitment are significant. The

exception is the PANAS-NA scale, where the correlation is

not significant, but is in the hypothesized direction. The

largest correlation is for the affect composite. Thus,

these findings indicate that lower NA individuals and higher

SWB individuals are more committed to their initial assigned

goals than are higher NA and lower SWB individuals.

Hypothesis 3.

Hypothesis 3 predicted that goal level and goal

commitment mediate the influence of NA and SWB on task

performance. That is, it was predicted that individuals

higher in SWB and lower in NA would have higher levels of

task performance, due to their setting higher goals in the 83

Table 7

Correlations Between Affect Scales and Goal Commitment at

Trial 1 for the Assigned Condition

Scale Goal Commitment

PANAS- PA .36

PANAS-NA - .14

NAS -.35**

TAS -.24* ** AB .31

UF .33

FHQ .29

SWLS .39 ** Affect Composite .41

Note. n-93; £<.05; £<.01. PANAS-PA-Positive Affect Scale (Watson et al., 1988); PANAS- NA-Negative Affect Scale (Watson et al., 1988); NAS-Negative Affectivity Scale (Stokes & Levin, 1990); TAS-Trait Anxiety Scale (Spielberger, 1979); AB-Affect Balance Scale (Bradburn, 1969); UF-Underwood & Froming (1980) Frequency of Positive and Negative Affect Scale; FHQ-Fordyce (1977) Happiness Scale; SWLS-Satisfaction with Life Scale (Diener et al., 1985). 84 self-set goal condition and higher goal commitment in the assigned goal condition. In order to test this mediation hypothesis, NA/SWB must first be related significantly to task performance. Thus, the NA and SWB measures were correlated with task performance within the self-set and assigned conditions. Task performance was defined as the number of strings solved correctly. Performance was examined within each trial separately as well as averaged over the three trials. None of these correlations was significant. As a result, hypothesis 3 could not be tested in its original form.

A secondary analysis was conducted where correlations between NA/SWB and task performance were examined within each feedback condition. This was done because results, to be presented below, suggested that certain effects for NA and SWB were operating differently within each feedback condition. Specifically, under conditions of failure those

lower in NA and higher in SWB were more committed to their goals on successive trials than those higher in NA and lower

in SWB. However, under conditions of success differences in

NA and SWB did not predict differences in goal level on

subsequent trials. Commitment toward goals following

failure should translate into greater effort and higher performance. However, because NA and SWB did not influence

goal level in the success condition one would not anticipate

subsequent differences in performance. 85

As expected, none of the correlations between NA/SWB and task performance were significant in the success condition, while several correlations were significant in the failure condition. The significant correlations were with average task performance (i.e., the mean number of strings solved correctly over the three trials). The correlation between the PANAS-PA scale and average task performance was .26 (j><.01), and the correlation between the affect composite and average task performance was .16

(j>-.05). Marginally significant results were obtained for the NAS and SWLS. For the NAS, the correlation was -.15

(E<.07), and for the SWLS the correlation was .15 (e<.07).

One reason that individuals higher in SWB and lower in

NA may have greater task performance in the failure condition is that they might work harder following failure than their higher NA/lower SWB counterparts. That is, they may be more committed to their goals, and this commitment may translate into greater effort and higher performance.

Evidence presented below does suggest that those higher in

SWB and lower in NA are more committed to their goals following failure.

One strategy that lower NA/higher SWB individuals may have used was trying to work more quickly on each trial.

Thus, subjects may have attempted more strings (i.e., quantity), at the expense of obtaining more correct (i.e., quality). Perhaps then, the marginally significant findings above would be stronger if the number of strings attempted 86

was used as the measure of task performance. Indeed, the

correlation between the NAS and the average number of

strings attempted over the three trials was significant (r-

-.17, £<.05), but the correlation between the SWLS and the

average number of strings attempted was .15 (e<-07). The

correlation between the PANAS-PA scale and the average

number of strings attempted was .24 (e<.01), while for the

affect composite it was .16 (e~*05).

Analyses were conducted to examine whether any of the

significant effects described above were mediated by goal

commitment. The three-step procedure described by James and

Brett (1984) was used. In each case, the dependent variable was either the mean number of strings solved correctly over

the three trials or the mean number of strings attempted

over the three trials. The independent variable was the

personality scale, and the mediating variable was the

subject's mean goal commitment over the three trials. To

test for mediation, the mediator is regressed first on the

independent variable. If this result is significant, the

dependent variable is regressed on the mediator. Next, the

independent variable is added to the equation and the change 2 2 in R is examined. If this change in R is not significant,

the mediator completely accounts for the relationship

between the independent and dependent variables. If the 2 change in R is significant then the independent variable

has a direct effect on the dependent variable. A change in 2 2 R less than the original R between the independent and 87 dependent variables provides evidence for partial mediation. 2 2 A change in R equal to the original R between the

independent and dependent variables, indicates that the proposed mediator does not mediate the effect.

Results for the five analyses performed are presented in Table 8. For each analysis, the hypothesis of some type of mediation of goal commitment is supported. The influence of the PANAS-PA on the number solved correctly and the number attempted is partially mediated by goal commitment.

Thus, the PANAS-PA has some direct influence on the dependent variable. The influence of the affect composite on the number solved correctly and the number attempted is completely mediated by goal commitment. The influence of the NAS on number attempted is also completely mediated by goal commitment. Thus, there is some evidence that affect is related to performance following failure. Individuals higher in SWB or lower in NA may perform better than those lower in SWB or higher in NA because they are more committed to their goals. This argument, however, must be considered tentative as the majority of the scales were not significantly related to performance.

Hypothesis 4 .

Hypothesis 4 predicted that individuals higher in NA and lower in SWB would evaluate their performance more negatively than individuals lower in NA and higher in SWB under conditions of both success and failure. To test this hypothesis, a composite satisfaction measure was constructed 88

Table 8

ine neaiacing tnecc or uoai uommiumenr on tne Keiacionsmp

Between Affect and Task Performance in the Failure Condition

Original Relationship: PANAS-PA --- > Number solved correctly r2-.067 (e < .01)

Step 1: PANAS-PA --- > Commitment r2-.118 (e < .01) Step 2: Commitment --- > Number solved r -.084 (e < .01)

Step 3: Add PANAS-PA to Step 2 equation Ar2-.029 (£<.05)*

Original Relationship: Affect composite --- > Number solved r2-.026 (£-.05) Correctly

Step 1: Affect Composite — > Commitment r2-.069 (e < .01) Step 2: Commitment --- > Number Solved r -.084 (£<.01) Step 3: Add Affect Composite to 2 'k'k Step 2 equation Ar -.008 (ns)

Original Relationship: 2 Affect composite ----> Number attempted r -.027 (e <.05) 2 Step 1: Affect Composite — > Commitment ^-.069 (£<.01) Step 2: Commitment — > Number attempted r -.070 (£<.01) Step 3: Add Affect Composite to « ^ Step 2 equation Ar -.010 (ns)

Original Relationship: „ NAS --- > Number attempted r -.028 (e <.05) 2 Step 1: NAS > Commitment ^-.072 (£<.01) Step 2: Commitment — > Number attempted r -.070 (£<.01) 2 Step 3: Add NAS to Step 2 equation Ar -.010 (ns)

Original Relationship: 2 PANAS-PA ----> Number attempted r -.058 (£<.01) 2 Step 1: PANAS-PA ----> Commitment ^ - . 1 1 8 (£<.01) Step 2: Commitment — > Number attempted r -.070 (£<.01)

Step 3: Add PANAS-PA to Step 2 equation Ar2-.025 (£-.05)

' k ' k Note. partial mediation; complete mediation. 89

by averaging each subject's reported performance

satisfaction at trials 1, 2, and 3. This composite was

created for two reasons. First, satisfaction measures for

each trial were highly intercorrelated. The average

correlation was .76. Second, when hypothesis 4 was analyzed

by trial, no differences were found across trials. Thus,

the dependent variable is the average satisfaction reported

over the three trials. A separate hierarchical regression

equation was run for each personality scale. The

independent variables to enter each equation were, in order:

goal origin (self-set or assigned), feedback type (success

or failure), and then the personality scale. Results are

summarized in Table 9. Goal origin made no contribution to

satisfaction. Feedback type explained 32.5% of the variance. All of the personality scales, with the exception

of the PANAS-NA, added unique variance above the influence

of feedback type. Percentage contribution for the scales

that were significant ranged from a low of 1.5% for the

Fordyce Happiness Questionnaire to 3.4% for the Affect

Composite. No higher order interactions were significant.

Thus, hypothesis 4 was supported. All of the affect scales,

except one, made statistically significant unique

contributions to predicting performance satisfaction above

the contributions of goal origin, and the type of feedback

(success or failure). 90

Table 9

Hierarchical Regression of Performance Satisfaction on Goal

Origin. Feedback Type, and Each Personality Scale

Variable 0 t AR2 o CM Step 1: Goal Origin .36 .000

Step 2: Feedback Type -.57 -11.41 .325*

Step 3a : PANAS-PA .17 3.43 .028*

PANAS-NA -.04 -.57 .000 1

NAS -.16 00 .027*

TAS -.14 -2.64 .017* 00 UF *-* 3.60 .031*

FHQ .12 2.45 .015*

AB .16 3.24 .025*

SWLS .17 3.46 .029*

Affect Composite .19 3.80 .034*

Note. N-274; *£<.01.

For each personality scale, the variance explained represents the contribution when the scale is entered third in separate regression equations. PANAS-PA-Positive Affect Scale (Watson et al., 1988); PANAS- NA-Negative Affect Scale (Watson et al., 1988); NAS-Negative Affectivity Scale (Stokes & Levin, 1990); TAS-Trait Anxiety Scale (Spielberger, 1979); AB-Affect Balance Scale (Bradburn, 1969); UF-Underwood & Froming (1980) Frequency of Positive and Negative Affect Scale; FHQ-Fordyce (1977) Happiness Scale; SWLS-Satisfaction with Life Scale (Diener et al., 1985). 91

Hypothesis 5.

Hypothesis 5 predicted that individuals lower in NA and higher in SWB would be more committed to a goal following failure than those higher in NA and lower in SWB. This hypothesis was tested at trial 2 and trial 3 separately, in order to examine whether commitment persisted across trials. A separate hierarchical regression equation was run for each personality scale. The independent variables to enter each equation were, in order: goal origin (self-set or assigned), followed by each personality scale. Results for trial 2 are summarized in Table 10. Goal origin made a statistically significant contribution in explaining variance in trial 2 commitment. Subjects in the self-set goal condition were more committed to their goal following failure than subjects in the assigned goal condition. In addition, 5 of the 8 personality scales, and the affect composite, made statistically significant unique contributions. For the scales that were significant, percentage contribution ranged from a low of 3.6% for the

SWLS to a high of 9.0% for the PANAS-PA scale. No higher order interactions were significant.

Results for trial 3 are summarized in Table 11. Again, goal origin made a statistically significant unique contribution to goal commitment. Self-set goal condition subjects remained more committed to their goal than assigned condition subjects. The percentage contribution for the personality scales were reduced slightly at trial 3. Four 92

Table 10

Hierarchical Regression of Trial 2 Goal Commitment on Goal

Origin and Each Personalitv Scale

Variable fi t AR2

Step 1: Goal Origin -.17 -2.05 .030*

Step 2a : PANAS-PA .31 3.72 .090**

PANAS-NA -.13 -1.58 .017 ** NAS -.27 -3.23 .069

TAS -.11 -1.31 .012 * UF .20 2.42 .040

FHQ .08 .90 .006 * AB .22 2.68 .048

SWLS .19 2.31 .036*

Affect Composite .25 3.05 .062**

Note. n-139; j><.05; £<.01.

For each personality scale, the variance explained represents the contribution when the scale is entered second in separate regression equations. PANAS-PA-Positive Affect Scale (Watson et al., 1988); PANAS- NA-Negative Affect Scale (Watson et al., 1988); NAS-Negative Affectivity Scale (Stokes & Levin, 1990); TAS-Trait Anxiety Scale (Spielberger, 1979); AB-Affect Balance Scale (Bradburn, 1969); UF-Underwood & Fronting (1980) Frequency of Positive and Negative Affect Scale; FHQ-Fordyce (1977) Happiness Scale; SWLS-Satisfaction with Life Scale (Diener et al., 1985). 93

Table 11

Hierarchical Regression of Trial 3 Goal Commitment on Goal

Origin and Each Personalitv Scale

2 Variable fi \1 AR

Goal Origin -.17 -2.08 .031*

PANAS- PA .28 3.39 .075**

PANAS-NA -.01 - .09 .000

NAS - .20 -2.40 .039*

TAS -.02 -.32 .000

UF .13 1.47 .015

FHQ .02 .26 .000 * AB .21 2.52 .043

SWLS .19 2.30 .036*

Affect Composite .18 2.11 .036*

St ‘ic'Jc Note. n-139; £<.05; £<.01.

For each personality scale, the variance explained represents the contribution when the scale is entered second in separate regression equations. PANAS-PA-Positive Affect Scale (Watson et al., 1988); PANAS- NA-Negative Affect Scale (Watson et al., 1988); NAS-Negative Affectivity Scale (Stokes & Levin, 1990); TAS-Trait Anxiety Scale (Spielberger, 1979); AB-Affect Balance Scale (Bradburn, 1969); UF-Underwood & Froraing (1980) Frequency of Positive and Negative Affect Scale; FHQ-Fordyce (1977) Happiness Scale; SWLS-Satisfaction with Life Scale (Oiener et al., 1985). 94

of the 8 measures, and the affect composite, made

statistically significant unique contributions, ranging from

a low of 3.6% for the SWLS and the affect composite, to a

high of 7.5% for the PANAS-PA scale. The Underwood and

Froming (1980) scale, which made a significant contribution

at trial 2, was not significant at trial 3.

These results indicate a moderate amount of support for hypothesis 5. In both trial 2 and trial 3, over half of the

affect measures made significant unique contributions to

goal commitment. The results of trial 3 indicate that this

persistence effect continues over the short term, but that

it may dissipate after repeated failure experiences. Trial

3 results mirrored those of trial 2, except that the variance contributions were reduced.

Hypothesis 6.

Hypothesis 6 predicted that when individuals succeed in

attaining their goals, those higher in SWB and lower in NA would be more likely to set higher goals on subsequent

trials than those lower in SWB and higher in NA. This hypothesis was tested by correlating each personality measure, and the affect composite, with goal level at trial

2 and trial 3 for subjects in the self-set, success

condition. Results indicated little support for the hypothesis. For trial 2, both the Negative Affectivity

Scale and the Trait Anxiety Scale were correlated with goal

level at a marginal level of significance (r^ g “ -.19,

I><.08; rT£

from hypothesis 1, the magnitude of these correlations seem

to be due to two outlier points. One of these is from a

subject low on the NAS and TAS who set a goal of 45. The

other is from a subject high on the NAS and TAS who set a

goal of 2. When these data points were removed, both

correlations became non-significant (rj^g“ •■03, ns;

rTAS“ '*05, ns).

No other scales had a significant correlation with goal

level at trial 2. In addition, none of the personality

scales was correlated significantly with goal level at trial

3.

Hypothesis 7a.

Hypothesis 7a predicted that individuals higher in NA

and lower in SWB would be more likely to attribute their

successes to the ease of the task or luck rather than

ability or effort, and their failure to ability or effort

rather than the difficulty of the task or luck. Those lower

in NA and higher in SWB were predicted to show the opposite

pattern.

To test this hypothesis, attributions for performance were averaged across trials to create a composite

attribution for ability, effort, task difficulty, and luck.

Composites were created because each type of attribution was highly correlated across trials. The average correlations were .64, .68, .56, and .73 for ability, effort, difficulty,

and luck, respectively. In addition, analyses by trial

indicated no differences from the composite results. Each 96

type of attribution was correlated with each personality

scale and the affect composite within the success and

failure conditions separately.

The hypothesis was generally supported in the success

condition but not the failure condition. Results for the

success condition are presented in Table 12. For the majority of these scales, individuals lower in NA and higher

in SWB were more likely to attribute their success to either

ability or effort. Two of the scales (the NAS and TAS)

indicated that they were also less likely to attribute their

success to luck (the affect composite was marginally

significant). No significant results were found for

difficulty attributions. Thus, these results indicate that higher NA/lower SWB individuals may take less credit for

their success than their lower NA/higher SWB counterparts;

rather they may attribute some portion of their success to

luck.

Results for the failure condition did not support the hypothesis. The only significant correlation was for the

Underwood and Froming (1980) scale. Those scoring higher on

this scale were more likely to attribute their failure to

luck than those lower on this scale (r-.21, £><-05). All

other correlations in the failure condition between

attributions and personality scales were low and

nonsignificant. 97

Table 12

Correlations Between Composites of Ability. Effort.

Difficulty, and Luck Attributions with Each Personality

Scale in the Success Condition

Ability Effort Difficulty Luck • o PANAS-PA .18** .08 .13 00

** PANAS-NA -.17** -.21 -.02 .07

NAS -.21** -.22*** -.13 .23

TAS -.22*** -.27*** - .06 .26*** * i—i AB .11 .10 -.11

UF .16** .17** .09 -.06

FHQ .17** .27*** .10 - .04 o 00 SWLS .06 .05 - .04

Affect ** Composite .21 .24*** .11 -.15*

‘jc'fc'Jc Note. £-135; £<.10; £<.05; £<.01. PANAS-PA-Positive Affect Scale (Watson et al., 1988); PANAS- NA-Negative Affect Scale (Watson et al., 1988); NAS-Negative Affectivity Scale (Stokes & Levin, 1990); TAS-Trait Anxiety Scale (Spielberger, 1979); AB-Affect Balance Scale (Bradburn, 1969); UF-Underwood & Froming (1980) Frequency of Positive and Negative Affect Scale; FHQ-Fordyce (1977) Happiness Scale; SWLS-Satisfaction with Life Scale (Diener et al., 1985). 98

Hypothesis 7b.

Hypothesis 7b predicted that the influence of NA/SWB on satisfaction would be partially mediated by causal attributions. Specifically, individuals higher in NA and lower in SWB were predicted to be less satisfied than those lower in NA and higher in SWB when they succeeded, partially because they would attribute their success to external, rather than intenal causes. When failing, individuals higher in NA and lower in SWB were predicted to be less satisfied than those lower in NA and higher in SWB, partially because they would attribute failure to internal, rather than external causes.

To test this hypothesis, the three-step mediation procedure described earlier was used (James and Brett,

1984). The dependent variable was the measure of average satisfaction used to analyze hypothesis 4, the independent variable was each personality scale, and the mediating variable was an internal-external attribution measure similar to one used by Chacko (1982). Specifically, an external attribution score was calculated for each individual by averaging his/her task difficulty and luck attribution score, and then subtracting this from an internal attribution score that was calculated by averaging each subject's ability and effort attribution scores. The resulting scale ranged from -4 to +4 and a constant of 5 was then added yielding a scale ranging from 1 to 9. Higher scores indicate a greater tendency to attribute performance 99 to internal causes like ability or effort. Lower scores indicate a greater tendency to attribute performance to external causes like the nature of the task or luck.

Since the results for hypothesis 7a indicated that

NA/SWB influenced attributions in the success condition only, hypothesis 7b was analyzed for success condition subjects only. Results for those scales that supported the hypothesis are summarized in Table 13.

Three of the seven scales, and the affect composite indicated that the effect of personality on satisfaction was mediated by the tendency to make internal versus external explanations for performance. In the case of the Fordyce

Happiness Questionnaire, the Negative Affectivity Scale, and the Affect Composite the results indicated that the effect was one of partial mediation; that is, each scale had a direct effect on satisfaction independent of the indirect effect on satisfaction via attributions. For the Trait

Anxiety Scale, the effect of NA on satisfaction was completely mediated by attributions. In each of these cases, those higher in NA or lower in SWB tended to make external attributions for their success whereas those lower in NA and higher in SWB tended to make internal attributions for their success. Results for the scales that did not support the hypothesis can be explained by one of three reasons. The PANAS-PA scale and the Underwood and Froming scale were significantly related to satisfaction but not to the internal-external attribution scale. The PANAS-NA was 100

Table 13

ine raeaiacing tiiecc or Lausai AtcriDUtions on tne Keiationsnio

Between Personality and Satisfaction in the Success Condition

Original Relationship: NAS --- > Satisfaction r2-.066 (e <.01)

Step 1: NAS --- > Internal-External r2-.058 (£<.01) Attributions

Step 2: Attributions --- > Satisfaction r2-.106 (£<.01)

Step 3: Add NAS to Step 2 equation Ar2-.034 (E<.05)*

Original Relationship: TAS --- > Satisfaction r2-.047 (e <-05) Step 1: TAS — > Internal-External r -.089 (e <-01) Attributions Step 2: Attributions --- > Satisfaction r2-.106 (e < -01)

Step 3: Add TAS to Step 2 equation Ar -.015 (ns)

Original Relationship: FHQ --- > Satisfaction r2-.063 (e < 0 1 )

Step 1: FHQ — > Internal-External r2-.023 (e <-09) Attributions Step 2: Attributions — > Satisfaction r2*.106 (e<•01) * Step 3: Add FHQ to Step 2 equation Ar -.042 (e < - 05)

Original Relationship: Affect Composite --- > Satisfaction r2-.073 (e <-01)

Step 1: Affect --- > Internal-External r2-.042 (e <.05) Attributions

Step 2: Attributions — > Satisfaction r2-.106 (e <-01)

Step 3: Add Affect Composite to Step 2 equation Ar2-.043 (e <-01)*

'ic Note. partial mediation; complete mediation. 101 significantly correlated with the attribution measure, but was not related to satisfaction. Finally, the SWLS was neither related to satisfaction nor the attribution scale. CHAPTER IV

DISCUSSION

There were two purposes for conducting this research.

The first objective was to examine the empirical commonality between negative affectivity and subjective well-being.

Both constructs have gained considerable attention recently among researchers interested in examining dispositional influences on organizational attitudes and behaviors.

Individual researchers have preferred to investigate one of these constructs primarily at the expense of the other, with the consequence being a lack of clarity in the literature over the distinction between them. In an attempt at clarification, confirmatory factor analysis models were developed from conceptual arguments in the literature and applied to data from 275 introductory psychology students who completed several measures of negative affectivity and subjective well-being. In addition, an analysis of the magnitude of the correlations among these measures was conducted, as well as determining whether the measures differentially predicted the dependent variables of interest.

The second objective was to investigate the influence of negative affectivity and subjective well-being on several variables that are central to goal-setting theory. The 102 103 examination of personality variables in goal-setting research has received scant attention in comparison to situational variables, yet investigations of personality influences have been recognized as being able to provide additional knowledge of those factors that influence goal- setting processes and task performance (e.g., Adler, 1986).

Negative affectivity and subjective well-being were selected because an analysis of the nature of the constructs suggested that they might influence motivated behavior. In addition, these constructs reflect relatively broad aspects of personality and have been shown to be related to other organizational criteria.

Subjects in the experiment performed a perceptual speed task after completing the negative affectivity and subjective well-being measures. Subjects either self-set or were assigned performance goals. Over three trials, subjects were either informed that they had surpassed their goals, or had failed to achieve their goals. Primary interest was in how the personality traits influenced goal level, goal commitment, satisfaction with performance, attributions, and task performance.

Discussion of Results for Negative Affectivity

and Subjective Well-Being Measures

An analysis of the correlations between the NA and SWB measures provided initial evidence that the measures assess a common construct. The between-construct average correlation of -.53 was only slightly smaller than the 104

within-construct NA average correlation of ,63 and within-

construct SWB average correlation of .56. A second piece of

evidence that these scales measure a common construct comes

from the analysis of confirmatory models 1 and 2. Model 1

argued that the scales could be defined by a single factor, whereas Model 2 suggested that two factors would best

represent the data. Although Model 2 fit the data better

than Model 1, the fit statistics for Model 1 did reach an

acceptable level according to conventional standards. Thus,

it was preferred over Model 2 due to its parsimony.

Other evidence also suggested that Model 1 was a better

representation of the data. First, in the two-factor model

(Model 2) the correlation between the factors was -.884.

When a subsequent model was analyzed constraining the correlation between the factors to zero, the fit statistics were quite poor. Next, an exploratory factor analysis was

conducted with the scree plot clearly indicating that one

factor could adequately describe the data.

Models 3 and 4 examined whether SWB was a broader construct than NA. That is, Judge (1992) has suggested that

SWB subsumes NA, because SWB assesses affective well-being and life satisfaction, whereas NA assesses affective well­ being exclusively (according to Judge). If this were true,

Model 4 should not have fit the data better than Model 3 because measures of NA should have loaded on the hedonic

level factor only. However, a comparison of these models 2 using the x difference test for nested models indicated 105

that Model 4 did fit significantly better than Model 3.

Thus, Judge's argument that NA is subsumed by SWB cannot be supported by these data.

The empirical evidence from these analyses offers several conclusions. First, measures of NA and SWB have been shown to have a high degree of overlap, both in terms of item wording and how the measures relate to one another.

The analysis of confirmatory factor models indicates that one general affect factor can adequately describe the data, and that measures of NA cannot be subsumed by SWB. Thus, the question can be raised: What is the difference between these constructs?

If this question is to be answered, one must move away from the empirical level and examine the conceptual definitions of the constructs. Two criticisms can be made, both directed at those researchers who study negative affectivity. The first is the general lack of acknowledgement by NA researchers of the subjective well­ being construct. That is, while subjective well-being researchers have discussed how negative affectivity and positive affectivity fit into the definition of subjective well-being, NA researchers, most notably Watson and his colleagues, do not acknowledge the SWB construct in their writings. This is a critical flaw because, as has been shown, measures of the two constructs are highly related.

If one is going to use NA as the dispositional construct of interest, it should be made clear how NA is different from 106 other related constructs (cf. Campbell & Fiske, 1959). To date only Judge (Judge, 1992; Judge & Hulin, in press; Judge

& Hulin, 1992, Judge & Locke, 1993) has offered an explanation that distinguishes subjective well-being from negative affectivity in suggesting that NA is subsumed by

SWB, or in other words, that SWB is the broader construct.

A second issue concerns the translation from the conceptual definition to the operational definition. When one first examines Watson and Clark's (1984) conceptual definition of NA and then their measure of NA, there seems to be some slippage, as the definition is rather broad, while their measure is extremely narrow. That-is, NA is defined not simply in terms of negative mood. Watson and colleagues have argued that those higher in NA have negative attitudes about themselves and low self-esteem, as well as negative attitudes about the world in general. Thus, although some writers might equate NA with trait anxiety,

Watson and Clark (1984) themselves argue that trait anxiety

"is too narrow a labelling of the construct" (p. 465). What is puzzling however, is that in terms of measuring NA they choose ten negatively worded adjectives. While negative adjectives may adequately assess aspects of nervousness and worry, this is only one component of the construct.

Measures that include negative adjectives only do not incorporate the cognitive orientation that distinguishes those higher in NA from those lower in NA. This shortcoming has been identified by other researchers, most notably 107

Stokes and Levin (1990). They suggest that although aspects

of NA can be assessed with existing measures, no measure

includes all of the dimensions described by Watson and Clark

(1984). Stokes and Levin's (1990) measure includes three

dimensions--nervous/calm, satisfied/dissatisfied with

oneself, and pessimistic/optimistic about the future (a

fourth aspect, cynical/trusting of others, loaded on a

separate component in a principal components analysis, and

items representing this component were eliminated).

Empirically, it is clear from this study that not all measures of NA lead to identical relationships with

dependent variables. Typically, a study will use a single measure of NA only, and base conclusions on results from

that measure. This research was unique in that it

incorporated multiple measures of NA and SWB. An

examination of the results indicates that conclusions as to

the relationship between NA and the goal-setting variables would have been rather different if one were using Watson's measure only versus Levin and Stokes' measure only. The NAS was related to goal commitment, performance satisfaction,

and commitment following failure while the PANAS-NA scale was not related to these measures. The results for

Spielberger's TAS were m o r e similar to the NAS. One possibility is that effects on commitment are due to the

cognitive aspect of NA; that is, how self-confident and

optimistic people are rather than differences on the

affective aspect of NA. This argument could explain the 108 results found here, because the PANAS-NA scale simply does not capture this cognitive aspect.

Thus, it is recommended that researchers assess NA with a broader measure than simply negative adjectives.

Conceptually, NA is a broad construct and measures that capture this breadth should be preferred over more narrow measures. The choice of a particular measure of NA is important because different measures can show varying relationships to the dependent variables of interest.

Another issue concerns the NA-PA distinction. As mentioned in the introduction, Watson and Clark (1984, 1991) argue that NA and PA are separate, orthogonal dimensions.

Thus, they measure PA with a second set of positive adjectives. Indeed, in this study, the PANAS-PA scale was uncorrelated with the PANAS-NA scale (r--.01). However, other researchers, most notably Diener (1990), Judge (1992), and Kernery (1991), have argued that NA and PA are not orthogonal. They suggest that the distinction is artifactual, due primarily to differences in item-wording direction. It also seems that the independence might be due to the specific items on the Watson et al. (1988) scale.

For instance, the Affects Balance Scale also contains negative items to measure NA and positive items to measure

PA. An examination of the correlation between an NA scale and a PA scale developed from these items yielded a significant relationship (r--.50, jK.OOl), In addition, the

NA scale from the Affects Balance Scale was correlated with 109 the PANAS-PA scale (r--.34, jK-OOl)! an<* the PA scale from the Affects Balance Scale was correlated with the PANAS-NA scale (r--.30, jK-001). Thus, the independence between NA and PA found by Watson and colleagues may also be due to the specific adjectives they use, in addition to the wording direction issue.

It should be noted also that information about positive affect is useful. Indeed, in this study some of the largest effect sizes were found for the PANAS-PA scale. Due to these considerations, Kemery (1991) has suggested that NA and PA should be measured on a single, continuous dimension, which he refers to as affective disposition, a term also used by Staw et al. (1986). This seems to be the position of those researchers who have measured the hedonic level component of subjective well-being with a balance score between positive affect and negative affect, and those who use measures like Froming's Frequency of Positive and

Negative Affect Scale.

Recently, it has been recognized that scales like the

NAS and TAS are perhaps best considered measures of affective disposition, rather than negative affectivity per se. Items on these scales are both positively and negatively worded, and indeed the NAS does not show discriminant validity with experienced positive affect

(Stokes & Levin, 1990). However, Stokes and Levin (1990) argue that the construct defined by Watson and Clark (1984) probably taps a combination of positive and negative affect. 110

For example, they note that someone who disagrees with an item like "I am hopeful and optimistic about the future" most likely experiences lower levels of positive emotions than someone who agrees with the item. Again, the implication is that the NAS captures more of the conceptual breadth of the construct, as well as suggesting that Watson and Clark may have misnamed the construct.

Thus, these results suggest that negative affectivity

(or affective disposition) cannot be subsumed by subjective well-being. In addition, I have tried to argue that Watson and Clark's operational definition of negative affectivity is more narrow than their conceptual definition. Watson and

Clark (1984) measure state negative affect and trait negative affectivity with the identical scale, except that they change the wording of the question (e.g., How do you feel right now? vs. In general, how do you feel?). It may be that it is inadequate to simply change the wording of the question and still capture the breadth of the trait. If one accepts these arguments one is left with the conclusion that there is not much difference between subjective well-being and trait negative affectivity (or trait affective disposition). Both measure emotional well-being, while SWB measures life satisfaction and NA assesses self- satisfaction. Certainly, one would expect life and self- satisfaction to be highly related and indeed Campbell et al.

(1976) found self-satisfaction to be the best predictor of life satisfaction, while Stokes and Levin (1990) found Ill happiness and life satisfaction ratings to have some of the highest convergent validity correlations with their NA scale

(r--.509, and r--.455, respectively).

Judge and Hulin (1992) have argued that Stokes and

Levin's (1990) measure is actually an alternative operationalization of subjective well-being. This is an

interesting claim because it would mean that, based on how

the NAS was constructed, the trait defined by Watson and

Clark (1984) is subjective well-being. The present study suggests that Judge and Hulin (1992) may be correct.

Measures of negative affectivity, such as the NAS and TAS, are highly related to, and behave similarly with respect to dependent variables as measures of subjective well-being.

It is also difficult to differentiate negative affectivity and subjective well-being from a conceptual standpoint.

Although dispositional research has gained much momentum in the last several years, results from the present study suggest that researchers need to put more care into deciding which particular dispositional measures to study, and distinguishing them from other constructs. Negative affectivity and subjective well-being are not the only dispositional constructs available for examination. Another

related construct is dispositional optimism, which Scheier and Carver (1993) define as holding positive expectancies

for one's future and the belief that good, rather than bad

things will occur in one's life. This conceptualization bears resemblance to the optimism/pessimism about the future 112 component of NA. Indeed, items on the Life Orientation

Test, used to measure dispositional optimism, such as "I hardly ever expect things to go my way" and "In uncertain times, I usually expect the best" bear striking similarity to items on the NAS like "Things rarely work out the way I want them to" and "I always expect the worst to happen."

Also, the NAS captures optimism with items such as "The future seems rather bleak and unpromising" and "I am hopeful and optimistic about the future." Thus, perhaps one could argue that optimism is subsumed by affective disposition.

In short, there needs to be more dialogue between researchers about the similarities and differences between the dispositional constructs being studied (cf. Scheier &

Carver, 1993). I am optimistic that the ultimate results of such a dialogue will clarify the problems currently present in the literature and will show that dispositional research can make an important contribution to the field of industrial and organizational psychology.

Discussion of Results for Goal-Setting Hypotheses

Hypotheses 1 and 6 .

Hypotheses 1 and 6 were concerned with whether or not differences in negative affectivity and subjective well­ being predicted the level of the goals individuals set.

Hypothesis 1 was analyzed using all individuals in the self­ set goal condition, while hypothesis 6 was tested using individuals in the self-set, success condition. Little support was found for either hypothesis. Initially, results 113 indicated that the NAS and TAS were related significantly to goal level at Trial 1, and related to goal level at Trial 2 at a marginal level of significance. However, these correlations became nonsignificant when several outliers were removed from the data. In addition, none of the subjective well-being scales was related to goal level at either Trial 1, 2, or 3, and no negative affectivity scale was related significantly to goal level at Trial 3.

These two hypotheses were based on the thinking that individuals who had more positive self-images would set higher goals than those with lower self-images. However, it may be that the design of the study vitiated the chance of finding support for these hypotheses. Specifically, in order to provide individuals with some knowledge of the task, they performed two practice trials prior to completing the experimental trials and were given feedback on their performance. Perhaps this preliminary information resulted in subjects using a more specific judgment of their capability in setting their goals, like self-efficacy.

Indeed, self-efficacy, an individual's perception of his or her capability to perform a certain action in a particular context, has been found to be a determinant of goal difficulty levels (Locke & Latham, 1990). Self-efficacy is a more specific construct than self-esteem, for example, in that it may differ from one domain to the next and is more changeable through experience (cf. Hollenbeck & Brief,

1987). Locke and Latham (1990) report a mean correlation 114

between self-efficacy and goal difficulty of .38.

Hollenbeck and Brief (1987) also reported that goal

difficulty was more strongly predicted by measures of task-

specific ability (i.e., objective ability and perceptions of

task-specific ability) than by more general personali ty

measures like generalized self-esteem, need for achievement,

and locus of control.

One influence on an individual's level of self-efficacy will be his/her prior performance in that domain (Bandura,

1986). Thus, providing feedback on practice trial

performance will likely influence self-efficacy for the

experimental trials, and therefore self-efficacy may have

more of an effect on goal level variables than more general

trait measures. Unfortunately, this study did not include a

measure of self-efficacy, and comparisons between self-

efficacy and the NA/SWB measures could not be made.

Does this necessarily mean that general trait measures

will never be important in predicting the performance levels

that individuals attempt to pursue? The answer is most

likely no. It may be the case that when individuals have no

experience with a particular task then measures that focus

on general expectations about one's capabilities will

explain more variance in goal levels than a task-specific

measure like self-efficacy. Indeed, in their discussion of

dispositional optimism Scheier and Carver (1987, 1993) note

that individuals hold expectancies at varying levels of

generality, and that general outcome expectancies (which 115 they call dispositional optimism, but which is also one component of negative affectivity) may play a greater role than more specific outcome expectancies in novel situations.

One implication of this argument is that the measures used in the current study may have predicted goal level to a greater degree at Trial 1 had subjects not been given the two practice trials. Instead, the task could have been merely described to the subjects and then the subjects could have been asked to set goals for themselves. This would also imply that the influence of NA/SWB on succeeding trials would have dissipated and that more specific measures like self-efficacy would play a more important role. In sum, when tasks are novel, it may be that more general beliefs about one's competence and expectancies for success are important in determining attainment levels that individuals pursue, but as individuals gain experience with the task, more specific measures play the primary role.

Hypotheses 2 and 5 .

Hypothesis 2 and hypothesis 5 were concerned with how differences in negative affectivity and subjective well­ being influenced goal commitment. Hypothesis 2 tested initial commitment to an assigned goal while hypothesis 5 focused on reactions to goal commitment following failure experiences. Strong support was found for hypothesis 2.

Except for the PANAS-NA scale, all of the personality scales were significantly related to goal commitment, and indicated that those lower in NA/higher in SWB were more committed to 116

the Initial assigned goal than those higher in NA/lower in

SWB.

These results extend the line of inquiry of Hollenbeck

and Klein (1987) and Hollenbeck et al. (1989), Neither

negative affectivity nor subjective well-being were

considered specifically by Hollenbeck and Klein (1987) nor

tested in Hollenbeck et al. (1989), yet the results here

suggest that they are important antecedents of goal

commitment. Indeed, the effect sizes found in this study

are at least as high as those found by Hollenbeck et al.

(1989) for need for achievement and locus of control. The

effect sizes in this study in terms of variance explained in

goal commitment ranged from 6% to 17%, while Hollenbeck et

al. (1989) found a total effect size of 9% due to the two

personality variables (6% due to need for achievement, 3%

due to locus of control).

Future research might examine whether the effects found

here are because those lower in NA/higher in SWB have

greater expectancies for reaching the assigned goal, higher

valence for the assigned goal, or both. It would appear

that results might be due to expectancies primarily.

Because those lower in NA tend to have higher self-esteem

and greater optimism in terms of future events than those higher in NA, they would probably hold more positive

expectancies for goal attainment. Similarly, those higher

in SWB who have a greater level of life satisfaction than

those lower in SWB should hold more optimistic expectancies 117 about: the future in general, and goal attainment in particular. Furthermore, Hollenbeck and Klein (1987) argue that self-esteem should affect goal commitment by increasing expectancies for goal attainment. Nevertheless, these ideas are counter to the findings of Hollenbeck and Brief (1987), who found that a generalized self-esteem factor consisting of a measure of self-esteem, locus of control, and need for achievement was related more to the valence of goal attainment than the expectancy of goal attainment. Clearly, more research is needed on how personality variables, both individually and in combination, influence goal commitment as well as expectancies and valences for goal attainment.

Support was also found for hypothesis 5 for several of the personality scales. This hypothesis was tested using only those who received failure feedback. On Trial 2, 5 of the personality scales, as well as the affect composite, added unique variance to the prediction of goal commitment after controlling for the influence of goal origin.

Percentage contribution for the significant personality scales ranged from 3.6% to 9.0%. On Trial 3, 4 of the 5 personality scales that made significant contributions at

Trial 2, and the affect composite, remained significant in terms of adding unique variance to the prediction of goal commitment. Percentage contribution ranged from 3.6% to

7.5%.

The first comment about these results concerns findings for goal origin. In these analyses, goal origin made a 118 significant contribution to predicting goal commitment both at Trial 2 and Trial 3. In each case, the percentage of variance explained was 3%, with self*set condition subjects more committed to their goals than assigned condition subjects. These results are in opposition to those found by

Hollenbeck et al. (1989) where goal origin did not account for significant variance in students' commitment to course grade goals. Though the reason for this discrepancy is not entirely clear, conceptually it might be expected that goal commitment would be higher for those in self-set goal conditions than those in assigned goal conditions. For example, Hollenbeck and Brief (1987) argue that in self-set goal conditions an individual will choose a goal that he or she values and has a high expectancy of achieving. Thus, goal commitment will be uniformly high. In assigned goal conditions, Hollenbeck and Brief (1987) argue that goal commitment will be lower and more variable than in self-set conditions because low ability persons may be assigned difficult goals, and high ability persons may be assigned easier goals than they otherwise would have chosen. In the present study, this effect in the assigned condition should have been mitigated somewhat because subjects were yoked to those in the self-set condition based on ability.

Nevertheless, commitment at Trial 2 and Trial 3 was higher in the self-set than in the assigned condition (Trial 2:

Mss-25.29, Mas-23.72, t(137)-2.05, £<.05; Trial 3:

M -25.14, M -23.55, t(137)-2.08, £<.05), but did not SS ES 119

differ at Trial 1. Furthermore, commitment was also more variable in the assigned condition than the self-set

condition at Trial 1 (sdss-3.84, sdas~4.62, F-1.47, pc.05),

and Trial 2 (sd -3.89, sd -4.93, F-1.61, j><.05), but not SS AS at Trial 3.

Another explanation for the effect of goal origin on

goal commitment may be that allowing individuals the

autonomy to select their own goals influences commitment.

For example, Hollenbeck and Klein (1987) argue that volition, or the extent to which an individual is free to

engage in a behavior (Salancik, 1977), is likely to

influence commitment. When individuals freely choose a

goal, they cannot later say that the goal was unrealistic, which might happen in an assigned goal condition.

Finally, the fact that individuals in the self*set

condition had to type their own goals into the computer while the goals for those in the assigned condition were

merely presented on the screen may have made the self-set

goals psychologically more "public" than the assigned goals

(Salancik, 1977). That is, while publicness refers to whether others are aware of one's goals (Salancik, 1977),

and although this did not differ objectively between goal

origin conditions, the mere process of typing in the goals versus reading one's goals on the computer screen could have made one's goal subjectively more public. Publicness is

also hypothesized to influence goal commitment (Hollenbeck &

Klein, 1987) and Hollenbeck et al. (1989) found that public 120 goals led to more commitment than goals that were not made public. One method that could test this publicness explanation would be for assigned condition subjects to retype their goals after the goals appear on the computer screen.

Regarding the results for the personality scales, individuals lower in NA/higher in SWB tended to remain more committed to their goals following failure than those higher in NA/lower in SWB. These findings complement those of

Rohrback and Lord (1991) who also found that personality variables affect responses to performance feedback.

Investigations like these are important because they suggest that different types of individuals may respond differently when faced with the same objective goal-performance discrepancy. To date, few researchers have examined whether personality predicts cognitive or behavioral changes in response to discrepancies. Rather, the focus has been on situational variables like the magnitude and frequency of failure (e.g., Campion & Lord, 1982).

These results suggest that those lower in NA and higher in SWB remain confident in their ability to achieve their performance goals even in the face of successive failures.

Although individuals may tend to doubt themselves when they fail to achieve a goal, Bandura and Cervone (1986) argue that what is more important than this immediate reaction is how people bounce back from failures, or their resiliency.

It is interesting and not surprising that the effects for 121 personality diminish over time. In the present study, the percentage of variance in goal commitment accounted for by the personality scales decreased over the three trials.

Even the most optimistic individual will likely set his or her sights lower when faced with several repeated failures.

In this study, both the magnitude and frequency of failure were a controlled aspect of the research design. However, in the "real world", subsequent failures may be avoided by this short-term persistence, and the present results suggest that short-term persistence may be dependent on levels of negative affectivity and subjective well-being.

Hypothesis 3.

Hypothesis 3 was concerned with whether differences in

NA and SWB predicted task performance. Specifically, because those lower in NA and higher in SWB were predicted to set higher goals in the self-set condition, and be more committed to goals when they were assigned, it was thought that they would have higher levels of task performance.

Task performance was defined as the number of strings solved correctly during each two-minute trial. The first step in a test for mediation is to demonstrate that the independent variable and dependent variable are related. Thus, NA and

SWB were correlated with task performance within the self­ set and assigned conditions separately. Hypothesis 3 was not supported as none of these correlations was significant.

A subsequent analysis was conducted within each feedback condition separately because the influence of NA 122 and SWB appeared to be operating differently within the feedback conditions. That is, differences in NA/SWB predicted goal commitment following failure, but not subsequent goal level following success. When individuals remain committed to their goals in the face of failure, they should work harder toward them, resulting in higher performance. Analyses for the failure condition indicated that two personality scales (the PANAS-PA scale and the affect composite) were correlated significantly with task performance, and two other personality scales (the NAS and the SWLS) were related to task performance at a marginal level of significance. In addition, the relationship between the PANAS-PA scale and task performance was partially mediated by goal commitment, and the relationship between the affect composite and task performance was completely mediated by goal commitment.

In considering the nature of the task used in this study, it seemed as if individuals could take one of two strategies in an attempt to increase their performance. One of these strategies might be to be particularly meticulous and make sure that each response is correct. However, subjects were prompted away from this strategy by directions asking that they work as fast as they were able. Thus, a second strategy would be to work more quickly on subsequent trials. One implication of this strategy is that quality

(i.e., number correct) might be sacrificed somewhat for quantity (i.e., attempting more strings). Hence, the 123 analysis was performed again using the number attempted on each trial as a measure of task performance, rather than the number correct. In this analysis, the relationship between the affect composite and the average number attempted over the three trials remained significant, as did the relationship between the PANAS-PA scale and the number of strings attempted. In addition, the result for the NAS, which had been marginally significant using number correct as a measure of task performance, was now significant

(£<.05) using number attempted as the measure of task performance. Again, the result for the PANAS-PA scale was partially mediated by goal commitment, while the result for the affect composite was completely mediated by goal commitment. Goal commitment also completely mediated the relationship between the NAS and the average number of strings attempted.

These results indicate that when individuals fail to reach their goals those higher in SUB and lower in NA tend to perform at higher levels than those lower in SWB and higher in NA. This is due either partially or completely to the fact that the lower NA/higher SWB individuals are more committed to their goals when faced with failure than higher

NA/lower SWB individuals. As mentioned in the results section, these findings for performance must be considered tentative at the present time, as the majority of the personality scales were not related significantly to performance. It should be noted however, that the composite 124 measure derived from the personality scales was related to

task performance both when it was defined as the average number of strings solved correctly and the average number of strings attempted.

Nevertheless, future research should examine the

influence of NA/SWB on performance using different tasks or different variations of the present task. Toward the latter end, it may be that the time limit per trial used in the present design was not long enough for motivational differences to be translated into performance differences.

However, lengthening the time per trial must be balanced against the prospect of subjects simply becoming bored with, and withdrawing from, the task. That is, this task is not one that all subjects find particularly exciting. It is, though, one that lends itself well to the manipulation of success and failure in the laboratory, as subjects are generally unaware of how well they are doing. Thus, one cannot simply double the time per trial, for example, without thinking about these other factors.

The influence of NA and SWB on task performance should also be examined in field settings. Indeed, some critics of dispositional research argue that the practical implications of recent studies examining personality effects are minimal because the personality measures affect attitudinal variables primarily and not behavioral indicators (e.g.,

Gerhart, 1991). However, recent research has found that NA and SWB are related to the frequency with which individuals 125 engage in withdrawal behaviors (e.g., Judge and Hulin, 1992;

Necowitz & Roznowski, in press). In addition, George (1989) found that mood states, which were partially determined by dispositional affectivity, predicted the frequency of absences among department store salespeople. Nevertheless, the influence of dispositional variables on both objective and subjective job performance indicators is an interesting area for future research.

Hypothesis 4 .

Hypothesis 4 examined whether differences in NA and SWB predicted satisfaction with task performance. This hypothesis received strong support. Seven of the 8 personality scales, and the affect composite, made statistically significant unique contributions to the prediction of performance satisfaction.

Previous research in goal-setting has focused primarily on the size of the discrepancy in explaining performance satisfaction, finding that the larger the negative discrepancy the lower is one's satisfaction with performance, and the larger the positive discrepancy the higher is one's satisfaction with performance (Locke &

Latham, 1991). Results for this hypothesis further support the idea that discrepancies are important in predicting satisfaction as feedback condition explained 32.5% of the variance in performance satisfaction. As expected, those who succeeded were more satisfied with their performance than those who failed. 126

Recently, both Locke and Latham (1991) and Rohrback and

Lord (1991) have argued that personality variables may influence one's satisfaction with his or her performance because the same objective discrepancy may not be perceived in an identical manner by all individuals. Support was found for this idea. Individuals higher in NA and lower in

SVB were less satisfied with their performance both under conditions of success and failure.

Results for the failure condition support previous research indicating that those with a more negative disposition tend to exaggerate and ruminate about their mistakes and failures (Block, 1965; Zahn, 1960). In addition, those with pleasant dispositions may distort the negative feedback or may be more easily able to put negative experiences out of their minds (cf. Watson & Clark, 1984).

Results for the success condition complement more recent research showing that even in an otherwise pleasant situation, those with a more negative disposition tend to perceive that situation more negatively than those with a more positive disposition. As mentioned previously,

Necowitz and Roznowski (in press) found that higher NA individuals performed more withdrawal behaviors than lower

NA individuals when they were satisfied with their jobs, and that higher NA individuals were less satisfied with an

"enriched" laboratory task than lower NA individuals.

The present results suggest that even when they succeed, those higher in NA and lower in SWB are less 127 satisfied with their performance than those lower in NA and higher in SWB. One explanation for this finding points to the types of attributions these individuals make for their performance (to be discussed in more detail below). When individuals succeeded, those higher in NA/lower in SWB tended to attribute their success to external causes whereas those lower in NA/higher in SWB tended to make internal attributions for their success.

Results for this hypothesis extend the research literature in two important ways. For goal-setting research, these results suggest that personality variables can have an influence not only on variables such as goal level and goal commitment, which have been the primary focus of personality research in goal-setting, but also on affective reactions to one's performance. Although previous research has focused on situational variables as explanations of performance satisfaction (e.g., discrepancies or progress rate), results here argue that affective disposition is an important construct to include when explaining an individual's satisfaction with his or her performance.

Although these results suggest that perceptions of discrepancies may differ depending on NA and SWB, future research might examine the influence of these constructs on variables related to progress rate, the second factor hypothesized to influence performance satisfaction. For example, Carver and Scheier (1990) argue that individuals 128

have a certain standard for their rate of progress toward a

goal when faced with negative goal-performance

discrepancies. It may be that those with more negative

dispositions hold more unrealistic standards for what this

progress rate should be than those with more positive

dispositions. Alternatively, one might manipulate progress

rate in a laboratory and test whether those with more negative dispositions perceive the same objective rate of progress more negatively than those with more positive dispositions.

Researchers studying the influence of dispositional variables on organizational attitudes have limited

themselves to job or task satisfaction. Results here

suggest that affective disposition influences not only one's satisfaction with the job or task, but also satisfaction with one's performance in that job or task. Indeed, because affective disposition is a general trait construct, it should influence satisfaction in a variety of situations.

Importantly, these research findings have resulted when the nature of the task or the level of performance is a controlled aspect of the research design. Previous studies that have reported significant correlations between dispositions and job satisfaction (e.g., Brief et al., 1988;

Judge & Hulin, 1992) often contain a sample of individuals

that perform a variety of jobs. Thus, one possible explanation for the relationship between dispositions and satisfaction is that those with a more negative disposition 129 are in lower status or lower paying jobs, and hence are less satisfied. Similarly, those with a more negative disposition may be less satisfied with their performance because they perform at lower levels.

However, when the task is identical for all subjects

(e.g., Levin & Stokes, 1989; Necowitz and Roznowski, in press), and the feedback is controlled, as in the present study, these alternative explanations cannot explain the data. That is, while these explanations are plausible (see hypothesis 3), they cannot explain totally why those higher in NA and lower in SWB are less satisfied with their jobs or their performance. Nevertheless, these alternatives do provide interesting questions for future research. As mentioned previously, researchers need to investigate the influence of affective disposition on both objective and subjective performance indicators. In addition, whether those higher in NA/lower in SWB self-select, or are selected into, lower paying or lower status jobs is an interesting question. Furthermore, these results may have implications for self-appraisals. That is, because those higher in NA and lower in SWB tend to be dissatisfied with their performance under conditions of success and failure one might hypothesize that they will provide more negative self­ appraisals than those lower in NA and higher in SWB. 130

Hypotheses 7 a and 7b.

Hypothesis 7a predicted that those higher in NA and lower in SWB would be more likely than those lower in NA and higher in SWB to attribute their success to the ease of the task or luck instead of their ability or effort, and their failure to ability or effort rather than the difficulty of the task or luck. Support for this hypothesis was obtained in the success condition only. Individuals with a more positive disposition tended to attribute their performance more to internal factors (i.e., ability and effort) than those with a more negative disposition. In addition, two of the negative affectivity scales (the NAS and the TAS) indicated that those with a more negative disposition were more likely to attribute their success to luck. None of the scales were correlated significantly with task difficulty.

One reason for this might be that subjects may have misinterpreted the item that assessed task difficulty. This item asked the extent to which subjects' performance was due to "the nature of the task." An examination of the descriptive statistics leads to the conclusion that subjects interpreted this to mean "the fact that the task was easy."

The mean for this item was 3,71 on the 5-point scale indicating that success condition subjects generally felt that their positive performance might be because the task was simple.

Results for the success condition support the idea that a pessimistic explanatory style may be symptomatic of those 131 higher in NA/lower in SWB rather than specific to

depressives. Those higher in NA/lower in SWB tend to take

less credit for their success than those lower in NA/higher

in SWB. Instead, there was some indication that they externalize the reasons for their success, perhaps by

attributing it to luck. This type of explanatory style is

likely to have negative ramifications for these individuals.

For example, it would seem that success experiences might change one's level of NA/SWB. However, results here suggest

that this may not be sufficient. If those with more negative dispositions do experience success, yet attribute

that success to external causes, the negative disposition is

likely to continue. It may be necessary for these

individuals to go through some type of cognitive therapy,

like that proposed by Beck and his colleagues (e.g., Beck,

Rush, Shaw, & Emery, 1979). As part of this therapy for

depression, individuals are given reattribution training in which they are taught not to blame themselves for negative events, but to look for other factors that may have

contributed to the bad event. Results here suggest that it may be necessary to also provide reattribution training for positive events; that is, teach these individuals to take

credit when good things happen to them. Indeed, making

internal attributions for success increases one's self­

esteem (Abramson et al., 1978).

Hypothesis 7a was not supported in the failure

condition. In this condition, subjects generally tended to 132 externalize the causes for their failure. Mean scores were highest for attributions to task difficulty and lowest for attributions to ability. It is not entirely clear why support was not found for the hypothesis in this condition.

One might speculate that the failure condition did not have a great deal of impact on the subjects, as they were introductory psychology students who were required to perform the experiment for course credit. Thus, they may not have been very ego-involved in the task. Such an argument must be tempered by several considerations. First, there was an influence of personality on attributions in the success condition where subjects performed the same task.

Second, given the results for hypotheses related to goal commitment and performance satisfaction, it appears that reactions to the failure condition did differ depending on levels of NA and SWB. Nevertheless, future research should examine whether differences in NA and SWB predict attributions for failure using different tasks and different subject samples. Support might be found when the task is more ego-involving or otherwise engages the subject.

Since hypothesis 7a was supported in the success condition but not the failure condition, hypothesis 7b was examined for success condition subjects only. Again, this hypothesis suggested that differences in satisfaction with performance due to NA and SWB might be due to attributional differences. Results for half of the personality scales (4 of 8) supported the hypothesis that the tendency for those 133 higher in NA and lower in SWB to make more external rather than internal attributions for success mediated the relationship between performance feedback and satisfaction with performance. That is, individuals with more negative dispositions are less satisfied than those with more positive dispositions when they succeed because the former attribute their success to external causes while the latter attribute success to internal causes. For three of the scales (the FHQ, NAS, and affect composite), the effect was one of partial mediation while the influence of the TAS on satisfaction was completely mediated by the attribution measure. These findings support the view of Mone and Baker

(1992) who suggested that causal attributions are important to consider in examining the relationship between discrepancies and satisfaction. They also suggest that one way for lower NA/higher SWB individuals to become more satisfied with their accomplishments is for them to alter their attributions for their performances. As mentioned previously, reattribution training as provided by cognitive therapy might be helpful in this regard. Those who take credit for their successes are more likely to feel competent about their abilities to perform the particular task than those who attribute their success to luck. In addition, internal attributions for success can lead to more general positive outcome expectancies and perhaps changes in levels of NA and SWB. 134

Limitations

The major limitation with this study concerns

generalizability issues that need to be addressed in any

laboratory investigation. It is simply a fact that in organizations, individuals do not perform perceptual speed tasks trying to attain a certain goal for the number of strings they can solve correctly in a designated time period. In addition, the extent to which introductory psychology students are representative of typical organizational members continues to be an open question

(Gordon, Slade, & Schmitt, 1986, 1987; Greenberg, 1987).

It must be remembered however, that the primary consideration in selecting a task for this study was not

that it possess what Aronson, Brewer, and Carlsmith (1985) refer to as mundane realism, or that it be superficially

similar to everyday situations. Rather, the situation needed to have experimental realism. That is, it needed to absorb and engage the participants. This presented a dilemma primarily due to the manipulation of success and

failure. Simply put, subjects needed to believe the feedback manipulation for the results to be considered valid. This was especially tricky given that failure and

success was manipulated over three trials.

A further complication was that the feedback provided was relative to the individual's own performance. Other studies that have incorporated a success/failure feedback manipulation often compare the subject's performance to a 135

reference group or group norm (e.g., Johnson, Turban, & Ng,

1992). In the latter case, feedback Is also a complex

issue. Consider the Johnson et al. (1992) study. In this

investigation, subjects' goals for a word search task were assessed by asking them how many more/less words they would attempt to find than that found by the reference group

(i.e., subjects from an earlier pilot study). Success and

failure were manipulated by telling subjects that they found approximately 5 more/less (strong positive/negative

feedback) or 2 more/less (mild positive/negative feedback) than the average. This would imply that someone who had set a goal of 4 more than the average in the mild positive feedback condition might be considered a success subject relative to the group, but in reality he/she failed to meet his/her personal goal. Thus, considering the subject a

"success" subject is incorrect.

Although it is true that individuals often compare

their performances and accomplishments to others, it seems that if one is interested in the influence of personal goals on task performance, feedback should be in terms of those personal goals. Thus, I chose to provide feedback not with reference to a group norm, but relative to the actual goal

level set by, or assigned to, the subject.

All of these factors--feedback over successive trials, believability of the feedback, and feedback with reference to one's personal goal level--led to a consideration of a number of tasks. The particular task chosen was judged to 136 be the best at balancing these various criteria.

Specifically, the perceptual speed task used is one where subjects, because they are instructed to work quickly, generally are unable to count the number of strings they are completing as the trial progresses. As a result, the feedback seems credible. Also, the number of trials selected needed to balance the desire to test for persistence over trials with the inevitable loss of believability as the number of trials increases. Pilot studies indicated that more than three experimental trials decreased credibility to an unacceptable level.

Thus, though the particular task does not necessarily mimic that of an actual organizational situation, it was one that was carefully selected to meet certain experimental criteria. As such, it seems likely that the psychological processes and effects identified may generalize to a field setting. Indeed, goal-setting effects have been found to generalize across laboratory and field settings (Latham &

Lee, 1986; Locke & Latham, 1991). Furthermore, given that subjects in this study were required to perform the task for course credit and participated for only a short duration, findings here may underestimate those found in an organizational setting.

Implications

Results from this study offer suggestions for the process of goal-setting in organizations. For example, complementing the findings of Hollenbeck and Brief (1987), 137 it was found that merely allowing individuals the autonomy to set their own goals led to greater commitment than having goals assigned to them. Thus, one way for supervisors or managers to increase goal commitment would be to allow employees some involvement in the process of setting performance goals. Nevertheless, these findings for goal origin are counter to the results of Hollenbeck et al.

(1989). It may be that in the latter case, the outcomes of performance were more salient in terms of their importance to the subjects (i.e., course grade goals) than in the former cases (i.e., a perceptual speed task and solving anagrams), and that this situational factor overwhelmed the possible influence of commitment differences due to goal origin. Hence, if managers can link goal accomplishment to salient external rewards, assigned goals may have similar influences on commitment as self-set goals.

In addition, results suggest that the process of assigning goals may need to differ depending on an individual's affective disposition. Those with a more negative disposition were found to have lower initial commitment to an assigned goal, and lower subsequent commitment to goals following failure than those with a more positive disposition. Thus, for those with a more positive disposition, managers may be able to assign goals that slightly or moderately exceed their initial capabilities without detrimental effects on performance. These individuals will be committed to the goal even if they 138 should fail initially. Hence, they may work harder toward the goals on subsequent attempts, with eventual goal accomplishment as the result.

On the other hand, for individuals with a more negative disposition, goals should be set at, or slightly below, what a manager believes to be that person's capability. This is because any initial failure may lead to a lowering of one's commitment and a decrease in performance. Indeed,

Hollenbeck and Brief (1987) have shown that assigning difficult goals to those lower in self-esteem results in a negative effect on performance. For these individuals, easier initial goals that are surpassed may lead to an enhanced belief in their ability to perform the task and reach higher subsequent performance goals.

However, even if those with a more negative disposition are afforded the opportunity to succeed by providing them with easier goals, their perceptions of their ability to perform the task may not increase if they attribute success to external rather than internal causes. Thus, results here have implications for how performance feedback is delivered.

That is, managers should provide feedback not simply in a manner that indicates that the individual succeeded but also information that will allow him or her to make internal attributions for that performance. For example, a manager might present information that the behavior or performance is high in consistency (i.e., the person usually behaves or performs that way in that situation), low in distinctiveness 139

(i.e., the person behaves or performs similarly in other

situations), and low in consensus (i.e., other people behave

or perform differently in that situation) (cf. Chacko &

McElroy, 1983). This particular combination of consistency,

distinctiveness, and consensus information leads people to make internal attributions for behavior (Kelley, 1973).

Feedback provided in this manner may have positive effects

on subsequent performance as well as performance

satisfaction. In addition, positive effects on goal

commitment and subsequent performance are likely if managers

can aid subordinates in making external attributions for

failure.

It was also argued that individuals with a more negative disposition, because they tend to evaluate the same objective level of performance more unfavorably, may provide more negative self-appraisals than those with a more positive disposition. As such, they may not fall prey to

traditional self-serving biases. Although attributing

success and failure in a self-serving manner may have positive effects on goal commitment and performance, it may

also be the case that the self-appraisals of those higher in

NA and lower in SWB are more accurate than those lower in NA and higher in SWB. Indeed, research in depression shows

that those who are more depressed may have more accurate perceptions of reality than those who are less depressed

(e.g., Alloy & Abramson, 1979). If those with more negative dispositions are more accurate in appraising their 140

performance than those with a more positive disposition,

perhaps situational forces that induce accurate self­

appraisals, like accountability (Boyle, 1991), may not be

necessary for these individuals.

In sum, this research has shown that negative

affectivity and subjective well-being can have important

influences on variables that are central to goal setting

theory. As such, not only does it extend our knowledge of

those factors that influence goal-setting processes, but it

also provides further evidence that affective disposition

has effects on organizational criteria other than job

satisfaction. Future research should examine whether the

findings here generalize across samples and settings.

Although there was some evidence that affective

disposition influenced performance following failure,

further studies should investigate whether NA and SWB

influence both subjective and objective performance measures. Other research suggests that personality variables can influence performance. For example, meta-

analytic results (e.g., Barrick & Mount, 1991) indicate that

the personality trait conscientiousness is related

significantly to performance in a variety of occupational

groups using a number of performance criteria. It has also been found that those high in conscientiousness, as compared

to those lower on this trait, have a greater propensity to

set goals and are more committed to them (Barrick, Mount, &

Strauss, 1993). Thus, these tendencies partially explain 141 differences in performance based on this trait. Studies such as the present one and Barrick et al. (1993) are important because not only do they show that personality can influence important organizational criteria, but they also provide evidence about the processes by which personality effects occur. As such, they give further credence to the importance of personality constructs in industrial/ organizational psychology research and practice. LIST OF REFERENCES

Abramson, L.Y., Seligman, M.E.P., & Teasdale, J. (1978). Learned helplessness in humans: Critique and reformulation. Journal of Abnormal Psychology. 87. 32-48.

Adler, S. (1986, July). A role for personality in goal-setting research. Paper presented at the 21st International Congress of Applied Psychology. Jerusalem, Israel.

Allen, M.J., & Yen, W.M. (1979). Introduction to measurement theory. Monterey, CA: Brooks/Cole.

Alloy, L.B., & Abramson, L.Y. (1979). Judgment of contingency in depressed and nondepressed subjects: Sadder but wiser? Journal of Experimental Psychology: General. 108. 441-485.

Anisman, H., & LaPierre, Y. (1982). Neurochemical aspects of stress and depression: Formulations and caveats. In R.W. Neufeld (Ed.), Psychological Stress and Psychopathology (pp. 179-217). New York: MacGraw-Hill.

Aronson, E., Brewer, M., & Carlsmith, J.H. (1985), Experimentation in social psychology. In G. Lindzey & E. Aronson (Eds.), Handbook of Social Psychology. Vol. 1. Hillsdale, N.J.: Erlbaum.

Austin, J.T. (1989). Effects of shift in goal origin on goal acceptance and attainment. Organizational Behavior and Human Decision Processes. 44. 415-435.

Bandura. A. (1986). Social foundations of thought and action: A social-cognitive theory. Englewood Cliffs, N.J.: Prentice-Hall.

Bandura, A., & Cervone, D. (1983). Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems. Journal of Personality and Social Psychology. 45, 1017-1028.

Bandura, A., & Cervone, D. (1986). Differential engagement of self­ reactive influences in cognitive motivation. Organizational Behavior and Human Decision Processes. 38, 92-113.

Barrick, M.R., & Mount, M.K. (1991). The Big Five personality dimensions and job performance: A meta-analysis. Personnel Psychology. 44, 1*26.

142 143

Barrick, M.R., Mount, M.K. , & Strauss, J.P. (1993). Conscientiousness and performance of sales representatives: Tests of the mediating effects of goal setting. Journal of Applied Psychology. 78. 715- 722.

Beck, A.T., Rush, A.J., Shaw, B.F., & Emery. G. (1979). Cognitive therapy of depression. New York: Guilford Press.

Bentler, P.M., & Bonett, D.G. (1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin. 88, 588-606.

Block, J. (1965). The challenge of response sets: Unconfounding meaning, acquiescence, and social desirability in the MMPI. New York: Appleton-Century-Crofts.

Booth-Kewley, S., & Friedman, H.S. (1987). Psychological predictors of heart disease: A quantitative review. Psychological Bulletin. 101. 343-362.

Boyle, K.A. (1991). Improving the quality of self-assessments: The role of accountability. Unpublished master's thesis, Ohio State University, Columbus.

Bradburn, N.M. (1969). The structure of psychological well-being. Chicago: Aldine.

Brett, J.F., Brief, A.P., Burke, M.J., George, J.M., & Webster, J. (1990). Negative affectivity and the reporting of stressful life events. Health Psychology. 9, 57-68.

Brief, A.P., Burke, M.J., George, J.M., Robinson, J.S., & Webster, J. (1988). Should negative affectivity remain an unmeasured variable in the study of job stress? Journal of Applied Psychology. 73. 193-198.

Brown, J.D., & Rogers, R.J. (1991). Self-serving attributions: The role of physiological arousal. Personality and Social Psychology Bulletin. 12, 501-506.

Browne, M.W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K.A. Bollen & J.S, Long (Eds.), Testing Structural Equation Models (pp. 136-162). Newbury Park, CA: Sage.

Burns, M.O., & Seligman, M.E.P. (1991). Explanatory style, helplessness, and depression. In C.R. Snyder & D.R. Forsyth (Eds.), Handbook of Social and Clinical Psychology (Ch. 14, pp. 267-284). New York: Pergamon Press.

Campbell, A., Converse, P.E., & Rodgers, W.L. (1976). The quality of american life. New York: Sage. 144

Campbell, D.J. (1984). The effects of goal-contingent payment on the performance of a complex task. Personnel Psychology. 37. 23-40.

Campbell, D.T., & Fiske, D.W. (1959). Convergent and discriminant validation by the multimethod-multitrait matrix. Psychological Bulletin. 56, 81-105.

Campion, M.A., & Lord, R.G. (1982). A control systems conceptualization of the goal-setting and changing process. Organizational Behavior and Human Performance. IQ, 265-287.

Carver, C.S., & Scheier. M.F. (1990). Origins and functions of positive and negative affect: A control-process view. Psychological Review. 97. 19-35.

Chacko, T.I. (1982). An examination of the affective consequences of assigned and self-set goals. Human Relations. 35. 717-726.

Chacko, T.I., & McElroy, J.C. (1983). The cognitive component in Locke's theory of goal-setting: Suggestive evidence for a causal attribution interpretation. Academy of Management Journal. 26. 104-118.

Chen, P.Y., & Spector, P.E. (1991). Negative affectivity as the underlying cause of correlations between stressors and strains. Journal of Applied Psychology. 76. 398-407.

Clark, L.A., & Watson, D. (1991). General affective dispositions in physical and psychological health. In C.R. Snyder & D.R. Forsyth (Eds.), Handbook of Social and Clinical Psychology. (Ch. 12, pp. 221-245). New York: Pergamon Press.

Cohen, J., & Cohen, P. (1983). Applied multiple regression/ correlation analysis for the behavioral sciences. 2nd edition. Hillsdale, NJ: Erlbaum.

Costa, P.T., & McCrae, R.R. (1987). Neuroticism, somatic complaints, and disease: Is the bark worse than the bite? Journal of Personality. 55. 301-316.

Costa, P.T., Zonderman, A.B., McCrae, R.R., Cornoni-Huntley, J., Locke, B.Z., & Barbano, H.E. (1987). Longitudinal analyses of psychological well-being in a national sample: Stability of mean levels. Journal of Gerontology. 42, 50-55.

Davis-Blake, A., & Pfeffer, J. (1989). Just a mirage: The search for dispositional effects in organizational research. Academy of Management Review. 14. 385-400.

Diener, E. (1984). Subjective well-being. Psychological Bulletin. 95, 542-575. 145

Dlener, E., Colvin, C.R., Pavot, W.G., & Allman, A. (1991). The psychic costs of intense positive affect. Journal of Personality and Social Psychology. 61. 492-503.

Diener, E., Emmons, R.A., Larsen, R., & Griffin, S. (1985). The satisfaction with life scale. Journal of Personality Assessment. 49, 71-75.

Diener, E., & Larsen, R.J. (1984). Temporal stability and cross- situational consistency of affective, behavioral, and cognitive responses. Journal of Personality and Social Psychology. 47, 871-883.

Diener, E., Larsen, R.J., Levine, S., & Emmons, R.A. (1985). Frequency and intensity: The underlying dimensions of positive and negative affect. Journal of Personality and Social Psychology. 48, 1253-1265.

Diener, E., Sandvik, E., & Pavot, W. (1991). Happiness is the frequency, not the intensity, of positive versus negative affect. In F. Strack, M. Argyle, & N. Schwarz (Eds.), Subjective well­ being: An interdisciplinary perspective. (pp. 119-139). New York: Pergamon Press.

Dossett, D.L., & Latham, G.P., & Mitchell, T.R. (1979). Effects of assigned vs. participatively set goals, knowledge of results, and individual differences on employee behaviors when goal difficulty is held constant. Journal of Applied Psychology. 64. 291-298.

Dweck, L.S., & Henderson, J.L. (1987). Theories of intelligence: Background and measures. Unpublished manuscript.

Erez, M. (1977). Feedback: A necessary condition for the goal-setting performance relationship. Journal of Applied Psychology. 62, 624-627.

Erez, M . , Gopher, D., & Arazi, N. (1987). Effects of self-set goals and monetary rewards on the performance of complex tasks under time-sharing conditions. Unpublished manuscript.

Eysenck, H.J., & Eysenck, S.B.G. (1975). Eysenck personality questionnaire manual. San Diego: Educational and Industrial Testing Service.

Ford, J.K., MacCallum, R.C., & Tait, M. (1986). The application of exploratory factor analysis in applied psychology: A critical review and analysis. Personnel Psychology. 39, 291-314.

Fordyce, M.W. (1977). Development of a program to increase personal happiness. Journal of Counseling Psychology. 24, 511-521.

Frank, J.D. (1935). Individual differences in certain aspects of the level of aspiration. American Journal of Psychology. 42, 119- 128. 146

Gardner, J.W. (1940). The relation of certain personality variables to level of aspiration. The Journal of Psychology. 9, 191-206.

Gerhart, B. (1991). The doubtful practical relevance of dispositional effects on lob satisfaction. Working Paper #90-06, Center for Advanced Human Resource Studies, Cornell University.

Gerhart, B. (1987). How important are dispositional factors as determinants of job satisfaction? Implications for job design and other personnel programs. Journal of Applied Psychology. 72, 366-373.

Goodstein, L.D. (1954). Interrelationships among several measures of anxiety and hostility. Journal of Consulting Psychology. 18, 35-39.

Gordon, M.E., Slade, L.A., & Schmitt, N. (1987). Student guinea pigs: Porcine predictors and particularistic phenomena. Academy of Management Review. 12, 160-163.

Gordon, M.E., Slade, L.A., & Schmitt, N. (1986). The "science of the sophomore" revisited: From conjecture to empiricism. Academy of Management Review. 11. 191-207.

Greenberg, J. (1987). The college student as guinea pig: Setting the record straight. Academy of Management Review. 12,, 160-163.

Hackman, J.R., & Oldham, G.R. (1975). Development of the job diagnostic survey. Journal of Applied Psychology. 60, 159-170.

Haney, J.N. (1973). Approach-avoidance reactions by repressors and sensitizers to ambiguity in a structured free-association task. Psychological Reports. 33. 97-98.

Harvey, R., Billings, R., & Nilan, K. (1985). Confirmatory factor analysis of the job diagnostic survey: Good news and bad news. Journal of Applied Psychology. 70. 461-468.

Headey, B., & Wearing, A. (1989). Personality, life events, and subjective well-being: Toward a dynamic equilibrium model. Journal of Personality and Social Psychology. 57, 731-739.

Heimberg, R.G., Klosko, J.S., Dodge, C.S., Becker, R.E., & Barlow, D.H. (1989). Anxiety disorders, depression, and attributional style: A further test of the specificity of depressive attributions. Cognitive Therapy and Research, 13, 21-Hendrick,

Hendrick, H.L., & Lord, R.G. (1992). The effects of progress rate and performance-%oal discrepancy on satisfaction. Paper presented at the Seventh Annual Conference of the Society for Industrial and Organizational Psychology, Montreal. 147

Hollenbeck, J.R., & Brief, A.P. (1987). The effects of individual differences and goal origin on goal setting and performance. Organizational Behavior and Human Decision Processes. 40, 392- 414.

Hollenbeck, J.R., & Klein, H.J. (1987). Goal commitment and the goal- setting process: Problems, prospects, and proposals for future research. Journal of Applied Psychology. 72. 212-220.

Hollenbeck, J.R., Klein, H.J., O'Leary, A.M., & Wright, P.M. (1989) Investigation of the construct validity of a self-report measure of goal commitment. Journal of Applied Psychology. 74. 951-956.

Hollenbeck, J.R., Williams, C.R., & Klein, H.J. (1989). An empirical examination of the antecedents of commitment to difficult goals. Journal of Applied Psychology. 74. 18-23.

Hsee, C.K., & AbeIson, R.P. (1991). Velocity relation: Satisfaction as a function of the first derivative of outcome over time. Journal of Personality and Social Psychology. 60, 341-347.

Idaszak, J.R., & Drasgow, F. (1987). A revision of the job diagnostic survey: Elimination of a measurement artifact. Journal of Applied Psychology. 22, 69-74.

Isen, A.M., & Gorglione, J.M. (1983). Some specific effects of four affect-induction procedures. Personality and Social Psychology Bulletin. 9, 136-143.

James, L.R., & Brett, J.M. (1984). Mediators, moderators and tests for mediation. Journal of Applied Psychology. 69. 307-321.

Jex, S.M., & Spector, P.E. (in press). The impact of negative affectivity on stressor-strain relations: A replication and extension. Work and Stress.

Johnson, D.S., Turban, D.B., & Ng, Y.N. (1992, May). Effects of feedback sign, strenzth. and controllability on intrinsic motivation, self-efficacv. self-set goals, and performance. Paper presented at the 7th Annual Conference of the Society for Industrial and Organizational Psychology, Montreal.

Joreskog, K.G., & Sorbom, D.G. (1989). LISREL 7: A guide to the program and applications (2nd ed.). Chicago: SPSS

Judge, T.A. (1992). The dispositional perspective in human resources research. In G.R. Ferris & K.M. Rowland (Eds.), Research in Personnel and Human Resources Management (Vol. 10, pp. 31-72). Greenwich, CT: JAI Press.

Judge, T.A., & Hulin, C.L. (in press). Job satisfaction as a reflection of disposition: A multiple-source causal analysis. Organizational Behavior and Human Decision Processes. 148

Judge, T.A., & Hulin, C.L. (1992). Job satisfaction and subjective well-being as determinants of lob adaptation. Manuscript submitted for publication.

Judge, T.A., & Locke, E.A. (1993). Effect of dysfunctional thought processes on subjective well-being and job satisfaction. Journal of Applied Psychology. 78. 475-490.

Kanfer, R . , & Kanfer, F. (1991). Goals and self-regulation: Applications of theory to work settings. In Advances in Motivation and Achievement. T_* 287-326.

Kelley, H.H. (1973). The process of causal attribution. American Psychologist. 28. 107-128.

Kemery, E.R. (1991). Affective disposition, role stress, and job withdrawal. Journal of Social Behavior and Personality. 6, 331- 347.

Kemery, E.R., Mossholder, K.W., & McGrath, R. (1989). Negative and positive affectivity: Reality or method artifact? Paper presented at the 1989 Academy of Management meeting. Washington, DC.

Kernan, M.C., & Lord, R.G. (1991). An application of control theory to understanding the relationship between performance and satisfaction. Human Performance. 4, 173-185.

Kim, J.S., & Hamner, W.C. (1976). Effect of performance feedback and goal setting on productivity and satisfaction in an organizational setting. Journal of Applied Psychology. 61, 48- 57.

Klein, H.J. (1987). Reactions to goal-setting and feedback: A test of a control thoerv model of work motivation. Unpublished manuscript.

Klein, H.J. (1989). An integrated control theory model of work motivation. Academy of Management Review. 14. 150-172.

Kraiger, K . , Billings, R.S., & Isen, A.M. (1989). The influence of positive affective states on task perceptions and satisfaction. Organizational Behavior and Human Decision Processes. 44, 12-25.

Kuhl, J. (1985). Volitional mediators of cognitive-behavior consistency: Self-regulatory processes and action control versus state orientation. In Kuhl, J. & Beckmann, J. (Eds.), Action control: From cognition to behavior, (pp. 101-128). New York: Springer-Verlag. 149

Latham, G.P., Erez, M., & Locke, E.A. (1988). Resolving scientific disputes by the joint design of crucial experiments by the antagonists: Application to the Erez-Latham dispute regarding participation in goal setting. Journal of Applied Psychology (Monograph), 22. 753-772.

Latham, G.P., & Lee, T.W. (1986). Goal-setting. In E.A. Locke (Ed.), Generalizing from laboratory to field setting. Lexington, MA: Lexington Books.

Latham, G.P., & Yukl, G.A. (1976). Effects of assigned and participative goal setting on performance and job satisfaction. Journal of Applied Psychology. 61, 166-171.

Levin, I., & Stokes, J.P. (1989). Dispositional approach to job satisfaction; Role of negative affectivity. Journal of Applied Psychology. 24, 752-758.

Locke, E.A., Cartledge, N., & Knerr, C.S. (1970). Studies of the relationship between satisfaction, goal-setting, and performance. Organizational Behavior and Human Performance. 2. 135-158.

Locke, E.A., & Latham, G.P. (1990). A theory of goal-setting and task performance. Englewood Cliffs, N.J.: Prentice-Hall.

Locke, E.A., Latham, G.P., & Erez, M. (1988). The determinants of goal commitment. Academy of Management Review. 12, 23-39.

Locke, E.A,, Shaw, K.H., Saari, L.M., & Latham. G.P. (1981). Goal setting & task performance: 1969-1980. Psychological Bulletin. 90, 125-152.

Lord, R . G . , & Hanges, P.J. (1987). A control system model of organizational motivation: Theoretical development and applied implications. Behavioral Science. 32., 161-178.

Matsui, T., Kakuyama, T., & Onglatco, M.L. (1987). Effects of goals and feedback on performance in groups. Journal of Applied Psychology. 22, 407-415.

Matsui, T., Okada, A., & Kakuyama, T. (1982). Influence of achievement need on goal setting, performance and feedback effectiveness. Journal of Applied Psychology. 62, 645-648.

Miller, D.T., &Ross, M. (1975). Self-serving biases in attribution of causality: Fact or fiction? Psychological Bulletin. 82. 313- 325.

Mischel, W. (1977). The interaction of person and situation. In D. Magnusson and N.S. Endler (Eds.), Personality at the crossroads: Current issues in interactional psychology, (pp. 333-352). Hillsdale, N.J.: Erlbaum. 150

Mone, M.A., & Baker, D.D. (1992). A social-cognitive, attributional model of personal goals: An empirical evaluation. Motivation and Emotion. 16, 297-321.

Monson, T.C., Hesley, J.W., & Chernick, L. (1982), Specifying when personality traits can and cannot predict behavior: An alternative to abandoning the attempt to predict single-act criteria. Journal of Personality and Social Psychology. 43. 385-399.

Necowitz, L.B., & Roznowski, M. (in press). Negative affectivity and job satisfaction: Cognitive processes underlying the relationship and effects on employee behaviors. Journal of Vocational Behavior.

Nezu, A.M., Nezu, C.M., & Nezu, V.A. (1986). Depression, general distress, and causal attributions among university students. Journal of Abnormal Psychology. 95. 184-186.

Pavot, W . , Diener, E. , Colvin, C.R., & Sandvik, E. (1991). Further validation of the Satisfaction With Life Scale: Evidence for cross-method convergence of well-being measures. Journal of Personality Assessment. 57. 149-161.

Peterson, C., & Seligman, M.E.P. (1984). Causal explanations as a risk factor for depression: Theory and evidence. Psychological Review. 91. 347-374.

Peterson, C . , Seligman, M.E.P., & Vaillant. G. (1988). Pessimistic explanatory style is a risk factor for physical illness: A thirty-five year longitudinal study. Journal of Personality and Social Psychology. 55. 23-27.

Phares, E.J. (1961). TAT performance as a function of anxiety and coping-avoiding behavior. Journal of Consulting Psychology. 25, 257-259.

Rohrback, M.R., & Lord, R.G. (1991). Control theory, discrepancies, and individual differences. Paper presented at the Sixth Annual Conference of the Society for Industrial and Organizational Psychology, St. Louis.

Sackett, P.R., & Larson, J.R. (1990). Research strategies and tactics in industrial and organizational psychology. In M.D. Dunnette & L.M. Hough (Eds.), Handbook of Industrial and Organizational Psychology (2nd ed., Vol 1) (pp. 419-490). Palo Alto, CA: Consulting Psychologists Press

Salancik, G. (1977). Commitment and the control of behavior and belief. In B.M. Staw & G.R. Salancik (Eds.), New directions in organizational behavior (pp. 1-54). Chicago: St. Claire Press. 151

Schaubroeck, J., Ganster, D.C., & Fox, M.L. (1992). Dispositional affect and work-related stress. Journal of Applied Psychology. 12, 322-335.

Scheier, M.F., & Carver, C.S. (1993). On the power of positive thinking: The benefits of being optimistic. Current Directions in Psychological Science. 2, 26-30.

Scheier, M.F., & Carver, C.S. (1987). Dispositional optimism and physical well-being: The influence of generalized outcome expectancies on health. Journal of Personality. 55, 169-210.

Schwarz, N., & Clone, G. (1983). Mood, misattribution, and judgments of well-being: Informative and directive functions of affective states. Journal of Personality and Social Psychology. 4£, 513- 523.

Schwarz, N., & Strack, F. (1985). Cognitive and affective processes in judgements of subjective well-being: A preliminary model. In H. Brandstaetter 6c E. Kirchler (Eds.), Economic psychology. Proceedings of the 10th IAREP Colloquium (pp. 439-447). Trauner, Linz, Austria.

Spielberger, C.D. (1979). Preliminary manual for the State-Trait Personality Inventory (STPI). Unpublished manuscript, Department of Psychology, University of South Florida.

Staw, B.M. (1986). Organizational psychology and the pursuit of the happy/productive worker. California Management Review. 28(4), 40-53.

Staw, B.M., Bell, N.E., 6i Clausen, J.A. (1986). The dispositional approach to job attitudes: A lifetime longitudinal test. Administrative Science Quarterly. 31, 56-77.

Staw, B.M., & Ross, J. (1985). Stability in the midst of change: A dispositional approach to job attitudes. Journal of Applied Psychology. 70. 469-480.

Steiger, J.H. (1989). EzPATH: A supplementary module for SYSTAT and SYSGRAPH. Evanston, IL: SYSTAT.

Stevens, J.P. (1984). Outliers and influential data points in regression analysis. Psychological Bulletin. 95, 334-344.

Stokes, J.P., 6c Levin, I.M. (1990). The development and validation of a measure of negative affectivity. Journal of Social Behavior and Personality. I, 173-186.

Stone, E.F., & Hollenbeck, J.R. (1984). Some issues associated with the use of moderated regression. Organizational Behavior and Human Performance. 34, 195-213. 152

Tanaka, J.S., & Huba, G.J. (1985). A fit index for covariance structure models under arbitrary GLS estimation. British Journal of Mathematical and Statistical Psychology. 38. 197-201.

Taylor, J.A. (1953). A personality scale of manifest anxiety. Journal of Abnormal and Social Psychology. 48, 285-290.

Underwood, B . , & Fronting, W.J. (1980). The mood survey: A personality measure of happy and sad moods. Journal of Personality Assessment. 44. 404-414.

Watson, D. (1988). Intraindividual and interindividual analyses of positive and negative affect: Their relation to health complaints, perceived stress, and daily activities. Journal of Personality and Social Psychology. 54, 1020-1030.

Watson, D . , & Clark, L.A. (1984). Negative affectivity: The disposition to experience aversive emotional states. Psychological Bulletin. 96, 465-490.

Watson, D., Clark, L.A,, 6c Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology. 54, 1063-1070.

Watson, D ., 6c Pennebaker, J.W. (1989). Health complaints, stress, and distress: Exploring the central role of negative affectivity. Psychological Review. 96. 234-254.

Watson, D . , & Tellegen, A. (1985). Toward a consensual structure of mood. Psychological Bulletin. 98. 219-235.

Weiner, B. (1974). Achievement motivation and attribution theory. Morristown, N.J.: General Learning Press.

Weiner, B., Frieze, I., Kukla, A., Reed, L., Rest, S., 6c Rosenbaum, R.M. (1971). Perceiving the causes of success and failure. In E.E. Jones, D. Kanouse, H.H. Kelley, R.E. Nisbett, S. Valins, and B. Weiner (Eds.), Attribution: Perceiving the causes of success and failure. Morristown, N.J.: General Learning Press.

Weiss, H.M., 6c Adler, S. (1984). Personality and organizational organizational behavior. In Research in Organizational Behavior. vol. 6 . (pp. 1-50). JAI Press.

Wood, R.E., Mento, A.J., 6c Locke, E.A. (1987). Task complexity as a moderator of goal effects: A meta-analysis. Journal of Applied Psychology. 72. 416-425.

Wright, J., 6c Mischel, W. (1982). The influence of affect on cognitive social learning person variables. Journal of Personality and Social Psychology. 42, 901-914. 153

Yardley, J.K., & Rice, R.W. (1991). The relationship between mood and subjective well-being. Social Indicators Research. 24, 101-111.

Yukl, G.A., & Latham, G.P. (1978). Interrelationships among employee participation, individual differences, goal difficulty, goal acceptance, goal instrumentality, and performance. Personnel Psychology. 31. 305-323.

Zahn, T.P. (1960). Size estimation of pictures associated with success and failure as a function of manifest anxiety. Journal of Abnormal and Social Psychology. £1, 457-462.

Zuckerman, H. (1979). Attributions for success and failure revisited, or: The motivational bias is alive and well in attribution theory. Journal of Personality. 42, 245-287. APPENDIX A

MEASURES

154 155

Affects Balance Scale (Bradburn, 1969)

Directions: Below is a list of words that describe the way people feel. We would like you to tell us IN GENERAL how often you have each of these feelings. Please indicate the degree to which to feel each emotion using the following scale:

1-Never 2-Rarely 3-Sometimes 4-Frequently 5-Always

1. Nervous 12. Friendly

2. Sad 13. Anxious

3. Irritable 14. Miserable

4. Happy 15. Enraged

5. Pleased 16. Delighted

6. Hopeless 17. Afraid

7. Resentful 18. Unhappy

8. Glad 19. Bitter

9. Tense 20. Joyous

10. Angry 21. Contented

11. Cheerful 22. Warm 156

Fordyce Happiness Scale (Fordyce, 1977)

Directions: Use the list below to answer the following question: IN GENERAL, HOW HAPPY OR UNHAPPY DO YOU USUALLY FEEL? Please check the one statement below that best describes your AVERAGE happiness, that is how you feel most of the time.

______10. Extremely happy (feeling ecstatic, joyous, fantastic!).

______9. Very happy (feeling really good, elated!).

______8. Pretty happy (spirits high, feeling good).

______7. Mildly happy (feeling fairly good and somewhat cheerful).

______6. Slightly happy (just a bit above neutral).

______5. Neutral (not particularly happy nor unhappy).

______4. Slightly unhappy (just a bit below neutral).

______3. Mildly unhappy (just a little low).

______2. Pretty unhappy (somewhat "blue", spirits down).

______1. Very unhappy (depressed, spirits very low).

______0. Extremely unhappy (utterly depressed, completely down).

Directions: Consider your emotions a moment further. On the average, what percent of the time out of 100% do you feel happy? On the average:

The percent of time I feel happy % 157

Frequency of Positive and Negative Affect. (Underwood & Froming, 1980)

Directions: The following questions ask about your happiness. Indicate your degree of agreement with each statement by placing a number from the following scale in the blank preceding each item. Please consider how you feel USUALLY, OR IN GENERAL, In arriving at a response.

The 5-point scale is:

1-strongly disagree 2-disagree 3-neither agree nor disagree 4-agree 5-strongly agree

1. ______I am usually quite cheerful.

2. ______I generally look at the sunny side of life.

3. ______I usually feel as though I'm bubbling over with energy.

4. ______Compared with my friends, I think less positively about life in general.

5. ______I am cheerful less often than most people.

6. ______I often feel down in the dumps. 158

The Satisfaction With Life Scale (Diener, Emmons, Larsen, & Griffin, 1985)

Directions: Below are five statements with which you may agree or disagree. Using the 1-7 scale below, indicate your agreement with each item by placing the appropriate number on the line preceding that item. Please be open and honest in your responding.

The 7-point scale is:

1-strongly disagree 2-disagree 3-slightly disagree 4-neither agree nor disagree 5-slightly agree 6-agree 7-strongly agree

1. ______In most ways my life is close to ideal.

2. ______The conditions of my life are excellent.

3. ______I am satisfied with my life.

4. ______So far I have gotten the important things in life.

5. If I could live my life over, I would change almost nothing. 159

Positive and Negative Affect Schedule (Watson, Clark, & Tellegen, 1988)

This scale consists of a number of words that describe different feelings and emotions. Read each item and then mark the appropriate answer in the space next to that word. Indicate to what extent YOU GENERALLY FEEL THIS WAY, THAT IS, HOW YOU FEEL ON THE AVERAGE. Use the following scale to record your answers.

1 - very slightly, or not at all

2 - a little

3 - moderately

4 - quite a bit

5 - extremely

1. _____ interested 11. ______irritable

2. _____ distressed 12. ______alert

3. _____ excited 13. ______ashamed

4. _____ upset 14. ______inspired

5. _____ strong 15. ______nervous

6. _____ guilty 16. ______determined

7. _____ scared 17. ______attentive

8. _____ hostile 18. ______jittery

9. _____ enthusiastic 19. ______active

10. ____ proud 20. afraid 160

Negative Affectivity Scale (Stokes & Levin, 1990)

Directions: Please indicate the extent to which you agree or disagree with the following statements. Please use the following response format for these items and place the appropriate number in the blank.

1 - disagree strongly 2 - disagree 3 - disagree slightly 4 - neutral 5 - agree slightly 6 - agree 7 - agree strongly

1. After an embarassing experience, I worry about it for days. 2. I know that things will continually improve in my life. 3. I feel that I have a great deal to be proud of. 4. I often feel restless and jittery for no apparent reason. 5. Things rarely work out the way I want them to. 6. 1 am not as well liked as most people. 7. Every day seems exciting, new, and different. 8. My feelings are more easily hurt than most other people. 9. I can easily concentrate on things for as long as I like. 10. Whenever someone criticizes me, I think about it for days. 11. I am hopeful and optimistic about the future. 12. When things go wrong, I blame myself. 13. I rarely lose sleep over worrying about something. 14. I am a person of worth, at least as good as other people. 15. I always expect the worst to happen. 16. I am more content and happy than most other people. 17. Happy endings only occur in the movies and in fairy tales. 18. I am not as self-confident as most other people. 19. When I meet people for the first time I am tense and uptight. 20. If I could live my life over, I would do many things differently. 21. The future seems rather bleak and unpromising. 161

Trait Anxiety Scale (Spielberger, 1979)

On the following scale, please indicate the extent to which you generally feel the way the item states. That is ON AVERAGE, how often do you feel that way. Use the following scale to record your answers.

1- Almost never

2- Sometimes

3- Often

4- Almost always

IN GENERAL.....

1. I am a steady person

2. _____ I feel satisfied with myself.

3. _____ I feel nervous and restless

4. _____ I wish I could be as happy as others seem to be.

5. _____ I feel like a failure.

6. _____ I get in a state of tension or turmoil as I think over my recent concerns and interests.

7. _____ I feel secure.

8. _____ I lack self-confidence.

9. _____ I feel inadequate.

10. ____ I worry too much over something that really does not matter. 162

Goal Commitment (Hollenbeck, Klein, O'Leary, & Wright, 1989)

1. It's hard to take this goal seriously. 2. It's unrealistic for me to expect to reach this goal. 3. It is likely that this goal may need to be revised. 4. Quite frankly, I don't care if I achieve this goal or not. 3. I am strongly committed to pursuing this goal. 6. It wouldn't take much for me to abandon this goal. 7. I think this goal is a good goal to shoot for.

(1-strongly agree; 5-strongly disagree) 163

Performance Satisfaction (Klein, 1987)

1. Considering my skills and the effort I put forth, I am very satisfied with my performance on this trial. 2. I am very unhappy with my performance on this trial. 3. My performance level on this trial troubles me. 4. My performance on this trial gives me a great deal of personal satisfaction. 5. I am proud of the way I performed on this trial.

(1-strongly disagree, 5-strongly agree) APPENDIX B

INSTRUCTIONS FOR EXPERIMENTERS

164 165

INSTRUCTIONS FOR EXPERIMENTERS

(The summary sheet for subjects should be placed face up ON the desk not lying in front of the monitor.)

When the subjects have been seated at the terminals, introduce yourself and ask them to read the summary of instructions sheet.

"My name is ______and I will be the experimenter for today's session. The first thing that I would like you to do is to read the summary sheet that is lying face-up on your desks. After you have read the sheet, please step over to the terminal here and I will lead you through an example of the task that we will ask you to do for the next 45 minutes or so.”

When all subjects have come over to the terminal, and before you start the example say:

"As stated in the summary sheet, this experiment involves investigating the effects of goal-setting on task performance. The first thing that we will ask you to do is complete eigjht short questionnaires that will help us assess how people view this task. I would like to show you some of the questionnaires you will be asked to complete, as well as an example of the number recognition task that you will be performing. In order to get the program started all you need to do is press the ENTER key."

When the program starts stress the fact that:

a) they will have a longer amount of time to read the instructions than the computer is giving you in the example. b) the instructions screens are timed so they will scroll by themselves. They don't need to press any keys.

c) However, after they enter any data, they must hit the ENTER key.

When the first questionnaire appears, tell them that they should enter an answer that it appropriate and say that there are no right or wrong answers only their opinions, and that all of their repsonses are both anonymous and confidential. 166

"This is one example of a questionnaire that you will see. It has a seven-point scale anchored with two adjectives. Simply enter the number of the answer that is appropriate for you and PRESS the enter key. Keep in mind that their are no right or wrong answers to these questionnaires, only your opinions. Also, all of your responses are both anonymous and confidential, so don't be concerned about being open and honest with us."

Show them that they need to enter a number within the scale range and that if they don’t the computer will beep. Remember to say that they need to respond with something within the scale. DO NOT say that their response is "incorrect" and they need to enter a "correct" one!!

When the second questionnaire appears show them that it is different from the first. You might say something like the followuing:

"This is another type of questionnaire where the scale now ranges from 1-5 and appears toward the top of the screen, the items that you respond to are toward the middle of the screen, and the flashing cursor at the bottom of the screen is where you input your response."

Use your finger or a pen to point to the screen as you illustrate this, if needed. Show them that once they input for #1 and press enter, the computer puts up the 2. and then they respond to item 2. Show them again that they need to enter something within the response scale.

Tell them that on subsequent questionnaires the response scale will change and they should read each questionnaire and response scale carefully before answering.

After the 2nd questionnaire is done, show them the example of the task. Tell them that the task is a number recognition task (DO HOT CALL IT A TEST!!!), and that they will be asked to do 2 practice trials to get them familiar with the task, and 3-5 performance trials, which the computer will determine at random (This is so they don’t ask in the middle how many are left and so they don’t know which is the last one). Tell them each trial will last TWO minutes and that before and after each trial they will be asked several questionnaires to assess their reactions to the task. 167

Make sure they realize that the strings and numbers they will be looking for will change after they enter their answers (Show them that they need to press ENTER after each response).

Tell them that when they are done they should sit and wait until all students have completed the experiment and then you will sign their cards and will ask them some other questions.

When they have finished, give out the debriefing sheet and sign their cards, giving them one credit. The experiment number is QRO-1. Ask them if they have any other questions, and have them leave all sheets on the desk. APPENDIX C

PRELIMINARY INSTRUCTIONS FOR SUBJECTS

168 169

SUMMARY OF INSTRUCTIONS FOR EXPERIMENT QRO

In this experiment you will be performing a computer simulation of a perceptual speed task. You will be shown a string of single-digit numbers. The computer will ask you to examine each string and find how many of a certain number there are in each string (e.g., how many 3's are there?). You will enter your answer and then press the ENTER key. Then a new string will appear. You will be performing a series of "trials" of this task. Each trial will last two minutes. Before and after each trial you will be asked to respond to several questionnaires assessing your reactions to the task.

Before beginning the task, we would like you to complete several questionnaires, which will also help us to determine how individuals perceive this task. The questionnaires will be presented on the computer screen. When you are responding to the questionnaires, please be as open and honest as possible. Your responses are both anonymous and confidential. There is no way for us to match your answers with your name. In addition, please be careful when you respond as the response scale may change on different questionnaires, and once you have entered a response, you will not be able to go back and change that response.

The experiment will take one hour. If you have questions while the experiment is being conducted, do not hesitate to ask the experimenter. S/he will be happy to answer any questions that you may have.

Again, please try to involve yourself seriously in the experiment so that the data collected can provide us with meaningful interpretation.

Thank you for your cooperation. APPENDIX D

EXPERIMENTAL DEBRIEFING

170 171

Experimental Debriefing

In this experiment you filled out several questionnaires dealing with your emotional experiences and completed several trials of a perceptual speed task. Our purpose in running this experiment was to examine how the setting of goals influences individual's performance on work tasks.

It has been suggested that the effect of goal-setting may be dependent on characteristics of individuals. In this experiment we wanted to determine if people's habitual emotional experiences influenced not only task performance, but also how goals were set, satisfaction with performance, and persistence in reaching a goal.

In order to accomplish these objectives, it was necessary for us to manipulate the feedback that you saw on the computer screen. That is, after you performed each trial, the computer told you how many strings you solved correctly. The feedback information you received was similar to that of other subjects in this experiment. That is, your feedback was not necessarily an accurate indicator of your actual performance. In research experiments, what we did is known as experimental control. In order to make comparisons between experimental groups, it is necessary that one group experience the experiment one way, and the second in another way. In this manner, firmer scientific conclusions can be drawn. Please be assured that the method we used was needed to examine the research questions that we have. In fact, we know that college students, like yourself, perform rather well on tasks of this nature.

We would be happy to discuss this experiment further with you and/or send you a report of our findings when the data has been analyzed. As this experiment is ongoing, we ask that you not discuss it with other OSU students.

Again, we would like to thank you for participating in this experiment.

Larry Necowitz Ph.D. Student Industrial/Organizational Psychology 292-8175

Dr. Mary Roznowski Assistant Professor 292-3746