CALIFORNIA STATE UNIVERSITY, NORTHRIDGE
THE EFFECTS OF WARNING MESSAGE HIGHLIGHTING
ON NOVEL ASSEMBLY TASK PERFORMANCE
A thesis submitted in partial satisfaction of the requirements for the degree of Master of Arts in
Psychology
Human Factors/Applied Experimental
by
Morris A. Zlotnik
May, 1982 The Thesis of Morris A. Zlotnik is approved:
Tyler Blake
Ma~s. Sanders, Chair
California State University, Northridge
;; ACKNOWLEDGEMENTS
The author would like to acknowledge those persons who have assisted in the conception, development, and implemen tation of this study. I would like to thank Dr. Tyler
Blake and Dr. Bill Wilsoncroft for agreeing to serve on my committiee. I would also like to thank my chairman Dr.
Mark Sanders, for all the assistance he has rendered to me during this task, and especially throughout my (many) years at CSUN. My appreciation is also extended to Xyzyx Infor mation Corporation. Many of the documents reviewed for this study were obtained through my association with Xyzyx.
In addition, the unrestricted use of their technical facil ities has been of invaluable help to me during the prepara tion of this manuscript. Finally, I would like to thank my wife Linda, for all the encouragement, patience, and love she has provided to me throughout this endeavor. This thesis is dedicated to my family and friends.
i i; TABLE OF CONTENTS
Page
List of Figures • • v List of Tables • vi Abstract • vii
INTRODUCTION • • 1 The Warning Process 2 Behavior and warnings • 3 Standardized Warnings 7 Warnings and Complex Written Procedures 10 Statement of the Problem 18
METHOD • • • 22
Independent Variables • 2.2' Dependent Variables • 25) Covariates 28 Subjects 29 Apparatus 30 Procedure 32
RESULTS 38
Multivariate Statistical Assumptions 38 Multivariate Analysis of Covariance 40 Analysis I 41 Analysis II 50
DISCUSSION • • 59 Warning Messages and Task Performance 63 Implications and Limitations • 65
REFERENCES 68
APPENDIX A • • • 72 APPENDIX B 97
APPENDIX C • 122
iv LIST OF FIGURES
Figure
1. Examples of NOTE configurations currently in use. 15
2. Examples of CAUTION configurations currently in 16 use.
3. Examples of WARNING configurations currently in 17 use. 4. Representation of experimental design. 23
5. Mean task completion times for each highlighting 43 condition by cell.
6. Mean total error rate for each highlighting 44 condition by cell.
7 Mean warning message error rate for each 45 highlighting condition by cell.
8. Mean warning message recall rate for experimental 46 highlighting conditions. 9. Mean percentage correct motor operation for each 47 highlighting condition by cell.
10. Mean task completion time for control and 51 experimental groups.
11. Mean total error rate for control and 52 experimental groups.
12. Mean warning message error rate for control and 53 experimental conditions.
13. Mean percentage correct motor operation for 54 control and experimental groups.
v LIST OF TABLES
Table 1. Summary of recommended rules to follow when using precautinary information. 12 2. Overall tests of multivariate analysis of variance for the combined set of dependent 49 variables. 3. Source table of Roy-Bargman Stepdown analysis for experimental versus control conditions. 57 4. Source table of intercorrelations for the dependent variables. 122
vi ABSTRACT
THE EFFECTS OF WARNING MESSAGE HIGHLIGHTING
ON NOVEL ASSEMBLY TASK PERFORMANCE
by
Morris A. Zlotnik
Master of Arts in Psychology
A review of examples and specifications relating to emergency' and non-emergency operating procedures revealed a considerable amount of variation with respect to recom mended standards for highlighting of warning messages. An experiment was therefore conducted to examine the effec tiveness of various methods of highlighting warning mes sages contained in the instructions for a novel assembly task. Seventy-two male subjects screened for manual dex terity were randomly assigned to eight highlighting groups in a 2 X 2 X 2 factorial design. One additional non orthoginal group (no warning) was used as the control. The highlighting factors employed were: dimensionality (2-D vs. 3-D), cross hatching (with diagional lines), and block ing of the warning messages (blocked vs. non-blocked). All conditions reflected highlighting practices currently em- ployed in written procedures. Subjects read a set of as- sembly instructions in which warning message highlighting
vii was varied for each group. Warning message highlighting was omitted from the control group instructions. The sub
jects' task was to assemble a "prototype" device according to the instructions provided. The primary dependent mea
sures were: total task completion time, total task errors,
errors associated with specific warning messages, and re
call of warning messages. In addition, the outcome of the
entire task itself was used as a qualitative indication of
subject performance. Findings indicated that the presence of warning messages shortened task completion time, and re duced total task errors and warning message errors between the experimental and control groups. No significant dif
ferences, however, were obtained between the various high
lighting configurations for any of the dependent measures used. Implications and limitations of this study are discussed, and suggestions for further research are made.
viii INTRODUCTION
Public concern for safety has increased dramatically in recent years. Major "large system" incidents such as the ThreeMile Island {TMI) accident and the DC-10 plane crash, both occuring in 1979, have brought the issue of safety to the public's attention in a dramatic fashion.
Even prior to these events, the public concern for safety had been fostered by 1) the alarming increase in products liability litigation and 2) the sizable awards frequently presented to injured parties {Moll, 1976: Bry, 1978). The advent of the Consumer Products Safety Commission {CPSC) has also increased awareness by publicizing potentially hazardous products and compiling statistical data concern ing product related injuries.
ProdUct manufacturers have both a legal and ethical obligation, given a possible hazard, to provide adequate warnings and instructions with the sale of their products.
This is often accomplished by means of a label or sign at tached to the product and/or warning messages accompanied with the product instructions. It should be noted, howeve~ that the inclusion of these warnings with a product does not automatically relieve the manufacturer from product li ability. Ross {1981) cites several court cases in which the courts have ruled that warnings were inadequate because they " ••• did not adequately communicate the level of
1 2
danger, .. or ..... did not state the severity of the risk and the nature of the hazard 11 (P.35).
Aside from the obvious legal obligation to warn of po tential hazards; warning labels are intended to lessen the risk of personal injury or property damage. Essentially this is accomplished by inducing certain patterns of behav ior and discouraging or eliminating other patterns of be havior (Dorris and Purswell, 1978).
The Warning Process
Previous research on visual warnings has primarily been related to traffic signage. These studies have empha sized such factors as interpretation (Brainard, Campbell, and Elkin, 1961); recognition, comprehension (Walker, Nico lay, and Stearns, 1965); readability, and visibility {King,
1975; Ells and Dewar, 1979).
In many cases it is assumed that once a warning mes sage is received and understood, the user is capable of acting appropriately, in accordance with the message. Un til recently, however, capability on the part of the user was not seen as an intergral part of the warning process.
Dorris and Purswell (1977) suggest that warning messages be designed to meet the following criteria:
1. The warning message must be clearly observable.
2. The warning message must be readily understood by
the user.
3. The user must be willing to act in accordance with 3
the message.
4. The user must be capable of acting appropriately.
The last criteria is an addition to the first three offered by Dorris and Purswell. It would seem logical that any warning could be considered ineffective, if it asked the user to perform some action the user was incapable of doing. Indeed, the courts have recognized this factor as being part of the warning process. For example, Ross
(1981) cites a products liability case dealing with asbes tos fibers. The court ruled that a warning containing the statement to "avoid breathing the dust" was inadequate be cause there was no way for workers to avoid breathing as bestos dust (P.36).
It would seem that a better understanding of the human factors that effect user responses to warnings is required.
Such knowledge might aid in the development of warning mes sages and configurations that are more likely to induce ap propriate responses from users.
Behavior and Warnings
Most of what is known about the design of visual warn ings is concerned with optimizing the perception of infor mation i.e., receiving and understanding the information.
Little research has been conducted regarding the behavioral aspects of warnings. This no doubt stems from the assumed logic that once a warning message is acknowledged by a pro duct user, little can be done about an individual's refusal 4
to act in accordance with the warning. This being the case, the ultimate criterion of a warning message, i.e.,
inducing an appropriate perceptual change in the user's be-
havior, is rarely measured directly. It should be noted
that some research on behavior and warnings is being per
formed by private corporations. Unfortunately, the results
of such studies are often deemed proprietary and thus un
available to researchers outside of the corporation
(Dorris, Note 1). However, there is some evidence to indi
cate few people actually read or even recognize a product
warning.
Dorris and Purswell (1977) conducted an informal study
on warning labels associated with a familiar object, a ham mer. They asked 100 subjects to use the hammer in a simple
assembly task. Virtually no one noticed that warning la
bels were affixed to the hammers. The authors suggest that
these results might be due to the "artificial" nature of
the task. However, a review of the literature indicates
several factors which seem to influence the warning pro
cess. Among these is a clear belief that the behavior of a
product user is modera.ted by the degree of perceived haz
ardness he sees in a product.
Philo and Rine {1977) suggest several key factors in
fluencing the perception of a hazard. Among the most obvi ous are~ familiarity with the product, past experience in terms of injury associated with the product, frequency of exposure to the potential hazard, and accompanied product 5
warnings, if any. There is also some evidence to suggest
that the assessment of potential hazards among certain pop
ulation groups is rather weak.
Martin and Heimstra (1973) administered a perception
of hazard test to elementary school children to determine
what factors affect the degree of hazardousness associated
with several actions. Age, sex, and socioeconomic back
ground were found to relate to children's perception of
hazards. In general, children were found to overestimate
the amount of hazard for a given action more than did a
group of "expert judges."
Dorris and Tabrizi (1978) compared subjects perceived
hazard rating of familiar products with corresponding in
jury data reported by the CPSC. It was found that mean
hazard ratings did not correlate significantly with objec
tive injury frequencies. This suggests that some products
were perceived as being more hazardous and others less haz
ardous than the injury data imply. The authors also note
that most of the products viewed as being more hazardous
had "blades or potential cutting surfaces associated with
them "(P.l61). A plausible explanation for this occurrance
might be that knowledge of the potential hazards of a pro
duct, along with any past experience, moderates a user's
attitude towards the product. Another factor to consider would be the presence of warning messages regarding the hazard. It would seem logical that without prior exper
ience, a user must rely on product warnings to gain 6
knowledge of potential hazards.
A survey conducted by McGuinnes (1977) lends some sup port to these contentions. McGuinnes determined consumer preferences regarding the design of warning signs associ ated with the hazard of spinning blades in power mowers.
He found that perceived effectiveness of the respective warning signs increased as a function of the degree of realism of the hazard. He also found that the vast major ity of individuals (84%) were of the opinion that the addi tion of a warning sign (on power mowers) could prevent in
juries. Although the results of these studies are rather scant, one might conjecture that increased information re garding the potential hazards in a product, in terms of a warning message associated with the hazard, could differen tially effect the user's behavior with the product.
Several others have investigated factors that seem to influence safe behavior. Tokuhata, Colflesh, Smith, Ramas wamy, and Digon (1976) conducted a study to determine the incidence of product related injuries and the behavioral factors moderating safe behavior by consumers. The authors found injured consumers were more likely to have purchased reconditioned products, to have assembled the product com ponents, or, repair a damaged product themselves. It is also of interest to note that reading or not reading pro duct instructions, was not related to the frequency of ac cidential injuries. The authors suggest that this finding might be the result of artifactual constraints of the 7
measurements. One might expect that properly designed in
structions could moderate safe behavior by warning of po
tential hazards. This would be especially important during
the assembly or repair of a product. This assumes that if
a hazard warning is perceived, the warning message will be
effective in terms of inducing appropriate behavior.
Optimizing the perception of a hazard warning can
facilitate this process. One method of doing this is by
standardization.
Standardized Warnings
Over the years several professional or industrial as
sociations have generated standardized formats for warning
and accident prevention signs. Among the most widely used with respect to industry are the American National Standard
Institute (ANSI) "Specifications for accident prevention
signs" (Z35.1-1973), and FMC Corporation's "Product Safety
Sign and Label System" ( 1980) •
The primary objective of these publications is to min imize the proliferation of nonstandard signs and to provide criteria for uniform recognition of hazards. The basic
elements of the ANSI accident prevention sign and the FMC warning label system are very similar. However, FMC stres ses the use of symbols or pictograms with warnings, while
ANSI regards symbols as highly desirable, but optional.
Both formats do agree as to the intended purpose of symbols. 8
In general, an optimal warning sign should contain the
following three elements: a signal word, a symbol or picto gram, and a action-emphasis. The signal word (e.g., Warn ing, Caution, or Note) identifies the level of hazard asso ciated with a specified dangerous condition or behavior.
The next portion of a warning sign should contain the sym bol or pictogram. These symbols are designed to communi cate the nature of the hazard and the consequences that can result if the warning is not heeded. The last portion of the warning sign contains the action-emphasis, which speci fies and informs of the hazardous condition or behavior.
It sometimes provides the reason for the action (e.g., men working above-- do not start motor). When symbols are not used in a warning sign, the signal word is intended to call attention to the action-emphasis just below the signal word.
Also included in the standards are recommendations concerning the classification of signs according to use.
In general, WARNINGS refer to hazards or unsafe practices which could result in severe personal injury or death.
CAUTIONS refer to those actions which could result in minor personal injury or product damage.
Within the standards themselves, little mention is made of any empirical testing to validate recommendations.
However, there is some evidence to indicate certain ele ments of the signs are effective with respect to recognition and comprehension.
Bresnahan and Byrk (1975) evaluated elements of ANSI 9
accident prevention signs for effectiveness as visual haz
ard alert cues. The authors view warning signs as a coun
termeasure which can moderate the man/machine interface in
the accident causal sequence. They found that signal words
(e.g., Danger, Caution, Notice, etc.) and color (e.g., Red,
Yellow, Green) both elicited differential hazard associ ation values {HAV's) consistent with their a priori order ing. This suggests that in the case of accident prevention signs color is an effective method of linking, or high lighting, a given hazard with a given warning sign {this highlighting technique would of course be less effective to anyone who is color vision impaired). Also, color may not be an economically desirable highlighting technique for all situations that require attention to differential hazard levels. One such situation would be the use of warnings or precautionary information in product assembly instructions.
If a number of warnings were required within the instruc tions then printing costs might be excessive. Another more serious situation involves the use of precautionary infor mation in complex operating or maintenance procedures.
Such instructions frequently employ non-color highlighting techniques to emphasize critical information. In both of these cases it seems likely that non-color highlighting me thods by which perception of critical information could be optimized would 1) increase the liklihood that appropriate actions are taken at appropriate times and 2) reduce the liklihood of operating errors when actions are performed. 10
Warnings and Complex Written Procedures
Sample specification manuals and/or operating proce dures from a variety of government agencies were obtained.
All of the documents are in current or pending use. Each document can be considered applicable to complex man machine operations (i.e., procedural or maintenance). They include the following:
-NASA Space Shuttle Orbiter Procedures.
- Naval Aviation Training and Operating Procedure
Standardization (NATOPS) flight manuals.
- Department of Defense (DOD) technical manual writing
handbook.
- FAA approved commercial DC-10 flight manual.
- Urban Mass Transit Administration (UMTA) Specifica-
tion for Bus Maintenance Manuals.
NRC Commercial Nuclear Power Plant operating proce
dures.
Typically, the specification manuals cover the prepar ation aspects for all the technical documents reviewed and approved by the agency. They include, at a minimum, the general style, format, and technical content requirements of the documents. Also included in the specifications are recommendations for the preparation of all precautionary information (i.e., Notes, Cautions, Warnings) that may be required.
The amount of detail devoted to precautionary infor mation varies considerably. For example, one specification 11
devotes an entire section (7 pages) to warnings, cautions,
and notes (MIL-HDBK-63038-1, 1977), while another devotes
one page (NATOPS, 1980). At a minimum, specifications will
cover such topics as; definitions, general content and for mat requirements, and some information regarding placement.
More detailed specifications will also include such topics
as; objectives and principles, special hazard conditions
(e.g., Radiation and Laser warnings), and pictorial symbol
requirements. To an extent, there is some general consen
sus among the various specifications as to recommended
rules to follow when precautionary information is used. Table 1 provides a summary of these rules. There are two general rules that virtually all the
specifications examined agree upon. First, cautions and warnings should be placed immediately before the actions to which they apply. Second, notes, cautions, and warnings
should not contain procedural (i.e., command) steps. With
regard to the first rule, it seems clear that any other lo
cation increases the probability of the warning or caution being read too early or too late for their most effective
utilization. With regard to the second rule, any intended
action or decision should have its own sequence number in order to attract the users attention and allow easier tracking of actions taken. One situation in which these rules would seem most applicable is with emergency opera ting procedures for nuclear power plants. Unfortunately the available evidence suggests that these rules have in 12
Table 1
Summary of recommended rules to follow when using pre- cautionary information (Adapted from MIL-HDBK-
63038-1, 1977).
• \vARNINGS and CAUTIONS should always precede the text or procedural steps to which they apply.
• NOTES may precede or follow applicable text depending on the material to be highlighted.
• A NOTE should always precede a procedural step to which it applies.
• WARNINGS, CAUTIONS, and NOTES should not be numbered or procedural steps. e Wnen a WARNING, CAUTION, or NOTE consists of two or more paragraphs, the heading WARNING, CAUTION, NOTE should not be repeated above each paragraph.
• If it is ever necessary to precede a paragraph by both a WARNING and a NOTE, or a CAUTION and a NOTE, etc. they should appear in the sequence noted, namely, WARNINGS, CAUTIONS, NOTES. e When several WARNINGS, CAUTIONS, or NOTES appear together, the heading should appear only once. e All the lines of text for WARNING and CAUTION statements shall appear on one page.
• NOTE statements may be continued from one page to the next. 13
the past been virtually ignored by the nuclear industry.
Fuchs, Engelschall, and Imlay (1981) conducted a detailed
evaluation of emergency procedures from nine NRC commercial
nuclear power plants. All procedures were evaluated in
terms of~ presentation style, level of detail, and admin
istrative control. According to the authors, administra
tive control referred to the usability and availability of critical procedures by the control room operators. The
study found examples of misplaced cautions and notes and inclusion of procedural data (with the cautions and notes) in all the emergency procedures that were evaluated.*
One area in which these specifications do not agree concerns the recommended highlighting for precautionary in formation. Highlighting is frequently employed to empha size (i.e., focus attention or differentiate) the signal
* The NRC is currently developing a specification for emer gency operating procedures (NRC,July 1981). The NRC has basically adopted the recommended rules covered in the
Technical Manual Writing Handbook, which are detailed in Table 1. 14
words and action-emphasis contained in Notes, Cautions, and
Warnings. The principle means of presenting the recommend ed highlighting formats is by depicting examples of them.
Visual inspection of these examples revealed a considerable amount of variation with respect to recommended warning message highlighting. Examples of the configurations cur- rently in use are depicted in Figures 1, 2, and 3. As can be seen the greatest amount of variation exists among the
Cautions and Warnings than among the Notes.
Most of the source documents reviewed are consistent with respect to one key factor. In general, as the hazard level of a particular task increases and the signal words change (e.g., Caution to Warning), the amount of highlight ing present with the signal word also increases. Unfortu nately, the specifications do not always recommend the same amount or type of highlighting for a given signal word.
For example, NATOPS manuals utilize a Caution signal word which is depicted inside a wavy rectangular box. NASA em ploys a 3-dimensional higlighting technique for the Cau tions used in the Space Shuttle System Malfunction Proce dures Requirements. In contrast, NATOPS uses dimensional highlighting for Warnings, while NASA depicts the Warning signal word using cross hatching inside a rectangular box.
With respect to the action-emphasis, the only type of highlighting used involves isolating the information inside a message block. When used as a form of highlighting it is 15
SOURCE NOTE CONFIGURATION -ffi Se ect/dese- t U4J'S jS necess4ry • ItEM ll:XEC NASA: System Malfunction a ITEM X EXEC Procedures Requirements. NOTE During reconftgu. ration •lways •aintatn at least ~~~ ~~ ~~ =~ !:d
Note DoD: Manuals, NATOPS Circuit breakers can be reset alter touch- Flight Requirements For down on field landings to enable ground roll braking and antiskid,
a. IIOTI
DoD: Technical Manual b. Writing Handbook. NOTE
c. NOTE
FAA Approved Commercial !!£!.L. IF AUTOPILOT IS NOT NEEDED. DC-10 Flight Manual. CONSIDER TURNING RUDDf~ S'BY ?'."~~ SW OFF UNTIL APPROACH
~ UMTA: Specification For Bus Maintenance Manuals. Injector is closed when rocker arm is up and injector spring is not compressed.
NOTE: The conditions given above for stopping reactor coolant pumps NRC: Commercial Nuclear ·Power Plant Operating should be continuously monitored Procedures. throughout this instruction.
NOTE NASA: Space Shuttle Flight When each of follow1ng CRT displays Data File Preparation is first called, verify all LRUs Standards. are des e 1ected
Figure 1. Examples of NOTE configurations currently in use. 16
SOURCE CAUTION CONFIGURATION
jCAUTIONl Pro I onqed ops NASA: System Malfunction > 243 degf •• 11 Procedures Requirements. will ~sult tn eventua 1 FC damage
DoD: Manuals, NATOPS [~~ur!~~ l Flight Requirements For With AIR SOl'RC£ OFF se-lected, rem.un below 300 KIAS 0.8 Thl:-:.
a. ( CAUTION DoD: Technical Manual I Writing Handbook. ... ("AUTION
c. f~lfTlqN:
FAA Approved Commercial ' ICAU!1CN~I f1..APS MAY NOT EXTEND TO JS•UNTil SPEED IS REDUCED. OC-10 Flight Manual.
CADTION UMT A: Specification For Bus Maintenance Manuals. Make sure to turn camshaft clock- wise. Turning counterclockwise will loosen camshaft bolt.
NRC: Commercial Nuclear CAUTiON: Do not perform the following steps Power Plant Operating Procedures. until the above verification is made.
I r:n:uTroN NASA: Space Shuttle Flight Stop at vent pas it ion to avoid Data file Preparation damage to hatch/hatch actuator. Standards. I Continue rotation after vent
Figure 2. Examples of CAUTION configurations currently in use. 17
SOURCE WARNING CONFIGURATION
~W,ARN):f.l Los• of ESSlBC NASA: System Malfunction (OAt) causes loss Procedures Requirements. of E.SS lBC power". 1 ACl, and eventual J loss of all Ma1n A po~er.
DoD: Manuals, NATOPS WARNING Flight Requirements For I I lli not use APC if high idle rpm still exists.
a. I WAIHIHG I DoD: Technical Manual Writing Handbook. 0. WAIINING
c. LWARNING?
FAA Approved Commercial DC·lO Flight Manual. NOT USED
UMT A : Specification For Bus Maintenance Manuals. NOT USED NRC: Commercial Nuclear Power Plan:t Operating OPTIONAL Procedures .
NASA: Space Shuttle Flight ~ARNING Data File Preparation Side hatch shou1d not be opened until Standards. Ground confirms external atmosphere
is clear of contamination I
Figure 3. Examples of WARNING configurations currently in use. 18
consistently employed with all three signal words. No
doubt this form of highlighting is intended to separate
precautionary information from procedural information. In
some instances where a variety of configuration alterna
tives are presented (e.g., MIL-HDBK-63038-1, 1977), the fi
nal selection of a configuration set is often based on the
typesetting equipment available when the document is final
ly printed.
It would seem that the critical factor for any set of
signal words is to present a consistent highlighting pat tern to the user. Only then would a user be able to asso
ciate a given signal word with a specific highlighting for mat. Such a method would seem to be consistent with Bres nahan and Byrk's Hazard Association Valves for Accident
Prevention Signs {1975). This of course assumes the user
is always exposed to a consistent highlighting pattern
(e.g., all NATOPS) throughout all the procedures he uses.
This would seem unlikely given the many different configur ations currently in use. Another factor to consider is the level of salience associated with the various highlighting techniques. With respect to warning messages, salience re
fers to the degree to which a particular highlighting for mat is noticeable or conspicuous to the user.
Statement of the Problem
It is interesting to note that little empirical evi dence is presented as support for any of the recommended 19
highlighting configurations. As a result, there is no
available information from research to indicate which con
figuration is in fact optimal for a particular signal word.
Without such data, decision makers would likely be reticent
to make any changes in recommended highlighting configura
tions. Even with hard data to support specific recommenda
tions, adoption of new design standards is never guaran
teed. Nevertheless, an evaluation of this type could pro
vide consumer product manufacturers with guidelines by which an optimal (i.e., most salient) highlighting format
could be selected for specific product instructions. This
type of evaluation would have an additional benefit to man
ufacturers since it provides them with a means of defense
against poorly designed warnings in the event of a product
liability suit (Peters, 1978).
This study proposes to examine the effects of warning message (i.e. signal word and action-emphasis) highlighting techniques on behaviors associated with a novel assembly task. The highlighting techniques to be evaluated are dimensionality and cross hatching for the signal words, and blocking of the action-emphasis. A detailed description of each highlighting technique and the apparatus is contained in the Method section. Two sets of hypotheses have been
formulated regarding the effects of warning messages and warning message highlighting on a novel assembly task. The first set of hypotheses concern the effects of the differ ent warning message highlighting techniques among the 20
experimental groups themselves. The second set concern the overall effects of warning messages between the no warning control group and the combined experimental groups. The following hypotheses were advanced regarding the effects of warning message highlighting.
Perception of critical information should increase wi·th
greater highlighting emphasis in assembly task
warnings. Consequently:
As the amount of highlighting increases the speed of assembly operations should decrease.
In general, increased highlighting should reduce the
number of errors subjects make while performing the as
sembly task.
Recall of warning messages should be facilitated by in
creasing amounts of highlighting.
The percentage of correct task outcomes (i.e. the de
vice operates as intended) should not vary signifi
cantly among these groups.
With respect to the three highlighting techniques, the
most salient cues are expected to be dimensionality
(3-D), cross hatching (crossed), and blocking
(blocked), respectively. Moreover, all highlighting
configurations are expected to have an additive effect
on the dependent measures. The following highlighting
configurations are listed in there anticipated order of
effectiveness from most salient to least salient.
They are as follows: 21
3-Dimensional, Blocked, Crossed
3-Dimensional, Blocked, Non-Crossed
3-Dimensional, Non-Blocked, Crossed
3-Dimensional, Non-Blocked, Non-Crossed
2-Dimensional, Blocked, Crossed
2-Dimensional, Blocked, Non-Crossed
2-Dimensional, Non-Blocked, Crossed
2-Dimensional, Non-Blocked, Non-Crossed
The following hypotheses have been formulated with re
gard to the expected differences between the control and
the experimental groups. Due to the presence of warning messages:
Task completion time should be greater in the experi
mental group when compared to the control group.
Fewer errors are expected to occur in the experimental
groups, for both the overall task and specifically with
regard to warning message errors when compared to the
control group.
The percentage of correct task outcomes is expected to
be higher in the experimental groups compared to the
control group. METHOD
Independent Variables
Three independent variables were utilized in the exper iment. They were dimensionality, cross hatching, and blocking. The first two were used to highlight the signal word in the warning message while the latter was used for the action-emphasis. All three reflected standard high lighting practices used in written instructions.
The study employed a 2 X 2 X 2 factorial design with all three variables serving as between subject factors. In addition, the study used a non-orthoginal control group in which subjects were presented task instructions containing no warning signal words. A representation of the design together with examples of the highlighting configurations employed is provided in Figure 4.
Dimensionality refers to the presence or absence of thick shading around two adjacent sides of the signal word.
The presence of this shading alters the perspective of the signal word and makes it appear to be 3-dimensional, on pa per. Without the shading there is no perspective and the signal word appears to be 2-dimensional. Current users of dimensional highlighting are the NASA System Malfunction
Procedures Requirements (1979) and NATOPS (1980) depicted in Figures 2 and 3, respectively.
In the context of the present experiment, cross hatch ing refered to the presence or absence of diagional lines
22 23
ACTION EMPHASIS DIMENSIONALITY
BLOCKED NON-BLOCKED
~ffEJa CROSSED I I ~~~ 2-D
I WARNING I NON-CROSSED I WARNING I I I
CROSS HATCHING
BLOCKED NON-BLOCKED Wc~fd?%1 CROSSED wtzzf%1 I I 3-D
I WARNING ' NON-CROSSED I I l WARNING '
NON ORTHOGINAL CONTROL
Figure 4. Representation of experimental design.
Note: Highlighting configurations reduced from actual size. 24
around the signal word. Such distinctive markings are
typically placed around critical labels to make them more
conspicious than other non-critical labels (Woodson and
Conover, 1973). Another more frequent (and possibly more
appropriate) use of cross hatching is to place it around
the emergency procedure sections of operating manuals
(NATOPS, P. 98). Both military and commercial aviation op
erating manuals frequently contain such distinctive borders to facilitate quick referencing of critical information.
The NASA warning configuration depicted in Figure 3 shows an example of the cross hatching investigated.
Blocking relates to a separate distinctive border en closing the action-emphasis portion of the warning message.
It is used primarily to separate the action-emphasis from the rest of the text. Since the action-emphasis is usually
in sentence format blocking provides a visual break from any text below the warning message. Figures 2 and 3 show an example of the blocking format used by NASA and employed in the study.
Originally, color had been considered as a means of highlighting (e.g., red, yellow, etc.). Although color is frequently utilized in accident prevention signs (Z35.1-
1973; FMC, 1980), additional printing costs typically pro hibit its widespread use in consumer product instructions.
Also, color coding has been found to be highly effective as a means of differentiating the degree of perceived hazard ness (Bresnahan and Byrk, 1975). The established 25
population stereotypes associated with color (Woodson and Conover, 1973) might tend to confound the results, since the objective of the present study was to differentiate warnings from the rest of the text only, and not from other warnings.
Dependent Variables
A total of four dependent measures were used to assess subjects performances: specific task completion time, to- tal task assembly errors, assembly errors associated with warning messages, and recall of warning messages.
Specific completion times refer to the number of sec onds a subject spent reading the instructions and perform ing the steps on each page. These individual times were summed in order to get a single measure for the entire task
(TASKTIME). It was expected that subjects would spend more time refering to task instructions containing warnings.
Also, those actions refered to in the warning message were expected to be performed slower (and with fewer errors) than similar instructions containing no warnings. Specific task assembly times, by page, were measured by the experi menter with a digital stop clock (Lafayette model number
54030). This was done by having the subjects operate are mote start/stop switch that was connected to the stop clock. The clock itself was concealed from the subject be hind a table divider. As the subjects turned to the first page of instructions with one hand, they simultaneously 26
activated the remote switch with the other hand. The sub
jects then read the instructions and completed all the
steps on the page. After completing the last step on the page, subjects activated the remote switch again and waited
for instructions from the experimenter. Page completion time was recorded from the stop clock by the experimenter and the stop clock was then reset to 0. After a specified rest interval of 30 seconds, subjects were told to turn to the next page of instructions and operate the switch in the same manner. This was done until the entire instruction booklet was completed.
No attempt was made to obtain a separate measure of reading time and task assembly time. Due to the gramma tical style (imperative form) in which the assembly in structions were written it would be very difficult to sep arate reading time from assembly time. This was confirmed in a pilot study when it was found that subjects typically read and performed an instructional step simultaneously.
Two measures of assembly errors were obtained. They were the total number of errors made by the subject for the entire task (TOTALERS) and errors associated with the warn ing messages only (WARNERS). Thus 'vARNERS was a subset of
TOTALERS. TOTALERS was suggested as a means of assessing the impact of warnings on other steps both before and after the placement of a warning. It should be noted that in the context of the present study assembly errors refered to both precision errors and identification errors. Precision 27
errors were defined as those instances where a particular
component was correctly selected but inappropriately used
(e.g., installing the right size screw, but not turning it
the required number of times). Identification errors were defined as those instances where an inappropriate component was selected (commissive) or where a component was not sel
ected when required (omissive). Fewer identification er rors were expected to occur since all components were lo
cated in a named parts bin. Experimenter correction of
subjects• identification errors was dependent on the impact these errors made in subsequent steps. Only errors that interfered with subsequent steps were corrected. Subjects that made frequent errors of this type (more than two) were not included in the analysis. Data from two subjects had to be discarded because of this.
Subject recall of warning messages was assessed using the instruction booklet from the control group. These in structions contained no warnings or isolated actionemphasis
(the action-emphasis refered to one of the sentences in the steps presented). Subjects were asked to locate the warn ing messages by first finding the appropriate step, and then, stating the particular sentence relating to the action-emphasis. Since all the signal words used in the instructions were the same (i.e., warning) the subject was actually being asked to locate the action-emphasis only.
However, scoring was based on both location of the warning and identification of the action-emphasis in the step. 28
Covariates Three covariates were used in the study. The first was
obtained from an initial screening test for finger dexter
ity using the Purdue Pegboard. The critical score was the mean of two trials for Both-hand Assembly Dexterity. The
second covariate was obtained from a query of the subjects
experience with the apparatus or similar apparatus. This was obtained a£ter the administration of the Purdue Peg board test and prior to the subjects introduction to the apparatus. Based on the subjects response to the query they were ranked by the experimenter into 3 ordinal levels of experience (i.e., experienced, somewhat experienced, and no experience).
It can be argued that in addition to the type o£ high
lighting presented, (i.e., dimensionality, cross hatching, and blocking) recall of warning messages might also be moderated by subject exposure to the specific task instruc tions. Since subjects may view a page of instructions for the duration of the task {plus a constant rest interval be tween pages), exposure is a function of specific page com pletion time. Thus the effect of highlighting formats on subject recall might be masked by the varying exposure times to the task instructions. This being the case, spe cific task completion time could be used as a covariate for recall. 29
Subjects
To insure that the subjects exposure to the task appa ratus would be completely novel, only subjects reporting no past experience (i.e., assembly and/or operation) with the apparatus were used. Results from a pilot study with the task instructions found that some female subjects took longer than 45 minutes to complete the task and some failed to assemble the apparatus correctly. The instructions uti lized contained no warning messages. Since it is hypothe sized that instructions with warnings would take longer to complete, the decision was made to use male subjects only.
Male subjects were drawn from undergraduate psychology classes at California State University, Northridge. Sub jects were first screened for finger dexterity using the
Purdue Pegboard Test. Subjects with scores in excess of +
1.5 standard deviations of the mean for male hourly produc tion workers (Purdue Research Foundation, 1968) were ex cused. As a result, two subjects were excused from the study. Prior to the actual experiment, prospective sub jects were informally interviewed to determine there exper ience level with similar kinds of apparatus. Any subjects reporting direct previous experience with the actual appa ratus were excused. One subject was excused because of their direct prior experience with the apparatus. Eight subjects were randomly assigned to each of the experimental conditions and the control condition. Thus, a total of 72 subjects were required. 30
Apparatus
The apparatus used in the novel assembly task was based on the following selection criteria.
Assembly - only common hand tools such as screwdrivers, wrenches, pliers, etc. should be necessary for assembly.
In addition, parts should be reusable and the apparatus must be easily disassembled for reuse.
Tasks - tasks should be readily discrete and moderately difficult to perform. Errors should be somewhat notice able. Tasks involving both tolerances and measurements should be included.
Instructions - instructions should be precise and dif ferentiate clearly between tasks. Whenever practical, in structions will be presented in accordance with the guide lines established by project PIMO (Grieme, Cleveland,
Chubb, et.al., 1969). Insertion of potential hazard warnings must be readily possible for numerous subtasks.
In addition, each action-emphasis utilized must be directly related to a specific task in order to be measurable (e.g.,
Warning - Do Not Tighten Contact Brush Nut More Than 1/4
Turn).
Size - all parts to be used for assembly must be of sufficient size to be accessable and visible to both the subject and experimenter.
Assembly Time - for those conditions in which warning messages with the greatest amount of highlighting are in corporated, the device should take no more than 45 minutes 31
to assemble. Function the apparatus should have a demonstratable operation when assembled, if its function is to be used as a means of assessing assembly errors.
Hazards - Since no real hazard can be ethically or le gally used, the experimental tasks must be designed so as to make subjects perceive that a potential hazard exists.
Hazards were presented to subjects in terms of verbal in structions prior to the task. The instructions warned sub jects about the consequences of assembling the apparatus incorrectly.
Based on these selection criteria, an apparatus was de veloped for the experiment using several hobby kits pur chased from a retail electronics store. None of the kits purchased had any warnings or cautions associated with the assembly or use of them. Concurrently, assembly instruc tions were prepared in accordance with the preestablished criteria using the instruction manuals supplied with the hobby kits as a basis. The written instructions used for the control group is presented in appendix A. Written in structions for one of the experimental groups (Crossed,
Blocked, 3-Dimensional) is presented in Appendix B.
The prototype apparatus was pilot tested using the con trol set of written instructions in order to evaluate its utility for the experiment. Determination and placement of warning messages was also accomplished by the pilot study utilizing a control set of task instructions. The 32
frequency of errors and their potential severity (as deter mined by the impact of the error on successful completion of the task) were used as selection criteria for the pro posed warnings. This was done to take advantage of any
systematic errors that might occur when subjects perform a novel task for the first time. The pilot test served a secondary purpose in that it allowed for the assessment of potential warning messages (i.e., action-emphasis) in the instructions. Further studies indicated the need for some basic tools and parts orientation. To facilitate assembly, a tool familiarization page was included in the instruc tions and a nomenclatured parts-bin was used to separate and display the various equipment pieces.
Procedure
Individual subjects were seated in an experimental room at a table containing the experimental apparatus and assem bly instructions. A table divider was used to prevent sub jects from seeing the apparatus before the experiment be gan. Prior to the experiment, subjects were tested for manual dexterity abilities and screened for previous exper ience with technical hobby kits.
Instructions from the test manual for the "Both-hands"
Purdue Pegboard finger dexterity test were then read to subjects. Two one-minute trials were given to each sub ject. Scoring was done by the experimenter at the comple tion of each trial. The mean of the two trials was 33
calculated and compared with appropriate standards for sub jects tested. Those subjects exceeding predetermined mini mum and maximum scores for the selected test (i.e., + 1.5 standard deviations from a standardized mean) were excused.
These subjects were replaced with others that did not ex ceed the critical score. In addition, subjects were ques tioned about their past experience with technical hobby kits (i.e., Radio Shack hobby kits and Heathkits) and their knowledge of basic electronics. Subjects responses to the queries were used by the experimenter to rank them on the experience covariate. Successfully screened subjects were then randomly assigned to either the non-orthoginal control group, or one of the eight warning message groups. Each warning message group represented a different combination of the independent variables investigated; cross hatching, blocking, and dimensionality.
Six different warning messages were incorporated into a set of written instructions for the experimental groups.
All instructions were presented to subjects in notebook form with text on the left-hand page and supporting illus trations on the right-hand page. All warning messages were placed directly before the procedural step to which it ap plied. The messages themselves referred to a specific ac tion in a step (e.g., Make sure that brush does not move or bend). In the control condition, the action was included as part of the step (see Appendix A). In the experimental conditions, the action was extracted from the same step and 34
used as the warning message (see Appendix B). Thus, all the instruction booklets contained the same actions. In additions, phrasing for a particular warning message was identical throughout the eight experimental conditions. No page in the experimental group's assembly instructions con tained more than one warning message. However, some pages contained no warnings.
Prior to the actual assembly task all subjects were told the following:
"I want you to imagine that you work for a company
that manufactures electric motors for industrial
use. Your task is to assemble and operate a mini
ature "prototype" version of an electric motor ac
cording to a set of written instructions. The in
structions in this notebook will give you precise
step-by-step directions on how to assemble and
operate the motor. Before we begin, I want you to
become familiar with these tools and supplies you
will use to assemble the motor. The tools and
supplies you will use are pictured on the opposite
page. I want you to take a few moments to become
familiar with these items and their names."
The experimenter then pointed to each item on the page and read their names aloud for the subjects.
"Most of the parts you will need are contained in
this parts bin. Their names are indicated inside
each parts bin. In some cases, there may be more 35
than one type or size of a motor part in a bin.
Each step in the instructions is supplemented with
pictures. The text is on the left side and the
related pictures are on the right. An item in the
step is usually followed by a number in parenthe
sis. Search the picture for the part that is
pointed at by an arrow with that same number. You
should then find that part in the parts bin or on
the table and perform the action that you are in
structed to do. Follow the entire instructions
carefully, step-by-step. Remember that you are
working for a company that makes these motors for
industrial use. If the motor is put together in
correctly, it might not work. Also remember that
this is a miniature version of the actual motor.
A real motor, incorrectly assembled, might cause
injury to the operator. Try to work as quickly
and accurately as you can. There is more than
enough time for you to assemble and operate the
motor. However, once you turn a page there is no
going back to the previous page."
Subjects were also given instructions on how to operate the toggle switch that was on the table. Each subject was told the following:
"Every time I say 'go to the next page' I would
like you to turn to the next page of instructions.
As you do so, I would also like you to press this 36
switch to the ON position. After you have fin
ished all the steps you were instructed to do on
that page, I want you to then press the switch to
the OFF position. Do not turn to the next page until you are told to do so. There will be a 30
second rest interval between pages. Please do not
read ahead or do anything else with the apparatus
until I t.ell you to go to the nex·t page of in
structions~ When I do, press the switch again to
the ON position. We will do this for every page.
Are there any questions?"
Questions pertaining to the instructions and the use of the toggle switch were answered. As each subject performed the assembly task the experimenter noted assembly times for each single page of instructions, along with any obvious assembly errors. A table divider was used to prevent subjects from observing the experimenter's actions during the assembly task. Subject errors that prevented comple tion of subsequent actions were corrected by the experimen ter during the rest interval. Subjects that made more than two such errors were omitted from the analysis. Once the entire assembly task was completed by the subjects the ap- paratus was tested to see if it would operate. Warning message recall was then determined for those subjects in the experimental conditions. Subjects were first shown ex amples of the various highlighting formats ( without the action-emphasis messages) and asked to identify the exact 37
format utilized in their instructions. Then, using the control group instructions they were asked to recall the location of the warnings (i.e., specific steps in which warnings were used) and the messages they contained. Once completed, subjects were fully debriefed as to the purpose of the experiment and the various hypotheses. They were then issued credit for the experiment, thanked and excused. RESULTS
Multivariate Statistical Assumptions
Data obtained from the study was analyzed using a Mul tivariate Analysis of Covariance (MANCOVA) procedure. The dependent variables employed in the analysis were TASKTIME,
TOTALERS, WARNERS, RECALL, and MOTOR. Prior to the multi variate procedure, several general tests of assumptions were conducted to insure the appropriateness of the statis tical analysis selected. In the paragraphs that follow a summary of this analysis is presented.
Multivariate Normality. The experimental design in volved a sample size of 72 subjects. Eight subjects were employed in each cell of a 2 X 2 X 2 factorial design, plus a non-orthoginal control group. Thus, the sample size pro duced far more than the 10 degrees of freedom for error needed to assure multivariate normality of the sampling distribution. In addition, the equal sample sizes among the experimental conditions and the provision of two-tailed tests enhance the robustness of the MANCOVA procedure.
Therefore, the assumption of normality was accepted.
Homogeneity of Dispersion Matricies. Sample variances from the dependent measures were compared in each of the nine groups. A 20:1 critical cutoff ratio was used as the upper limit of the ratio of largest to smallest variance
(Tabachnick and Fidell, in press) • The largest ratio found was 7:1 when comparing the variances for TOTALERS between
38 39
the Crossed, Blocked, 2-Dimensional group and the control group. Tests for homogeneity of covariance matricies were performed using SPSS (Statistical Package for the Social
Sciences) MANOVA. This was done for the measures associa ted with the 3 independent variables between themselves and for the non-orthoginal experimental versus control condi tions. Results for the 3 IVs alone show that F (196,3707) = 1.09, P > .05 and for the non-orthoginal comparison F (28,517) = 1.40, P > .05 for Boxes M. These statistics indicate a non-significant deviation from homogeneity of variance-covariance matricies for the measures employed.
Outliers. A test for multivariate outliers was run us ing the Mahalanobis o2 test for outliers (BMDP7M). Using a critical cutoff level of .001, no outliers were identified in the data.
Linearity. Utilzing SPSS SCATTERGRAM, plots of the in tercorrelations among the DVs and correlations between the
DVs and the covariates were obtained. There were no indi cations of non-linear relationships among the data. In addition, inspection of the DV intercorrelations revealed significant relationships for the task outcome and error measures. (Table 4, in Appendix C, presents a matrix of the intercorrelations obtained for the dependent measures.)
Thus, the data was considered to be sufficiently linear to preclude the need for any data transformations.
Homogeneity of Regression. The covariates used in this study were selected as a means of controlling for subject 40
relevant variables (i.e., manual dexterity, and prior ex perience) that could influence the dependent measures. A failure~ to adequately compensate for these factors in the analysis could lead to false conclusions regarding the ef fectiveness of the IVs. Tests for homogeneity of regres sion evaluate the degree of interaction between the IVs and the covariates. The presence of such an interaction may indicate that the covariate adjustments would be different for each of the IV conditions. Results from an overall test of homogeneity of regression were found to be non significant. Further analysis using SPSS ANOVA failed to reveal any significant interactions between the IVs and the covariates. Thus the decision was made to analyze the data using a MANCOVA procedure.
Multivariate Analysis of Covariance
A preliminary MANCOVA procedure failed to attribute any statistical significance to the manual dexterity covariate,
PURDUE. In addition, no evidence was found to support a correlation between task completion times and warning mes sage recall. However, the experience level covariate was found to be statistically significant among several of the dependent measures. Thus, the decision was made to elimi nate the two ineffective covariates to prevent a corres ponding loss in degrees of freedom. A MANCOVA procedure was therefore conducted using experience level as the only covariate. It should be noted that the statistical 41
limitations regarding a MANOVA procedure are applicable to those of MANCOVA. Consequently, the MANCOVA procedure was conducted without further consideration of the statistical assumptions.
Due to the inclusion of a non-orthoginal control in the experimental design, two separate analyses were performed on the data. In the first analysis, a 2 X 2 X 2 MANCOVA was performed to test the effects of different warning mes sage highlighting techniques between the experimental groups themselves. The dependent measures for this anal ysis were task completion time, WARNERS, TOTALERS, MOTOR, and RECALL. The MOTOR dependent measure was an objective determination of the outcome of the entire task i.e., whe ther or not the assembled device operated. In the second analysis, a 1 X 8 MANCOVA was performed to test the overall effects of warning messages between the control group and the various experimental groups. The dependent measures for this analysis were task completion time, WARNERS,
TOTALERS, and MOTOR. RECALL was not measured in the con trol group.
Analysis I
A 2 X 2 X 2 MANCOVA was performed to determine the dif ferential effectiveness of warning message highlighting.
The independent variables were dimensionality and cross hatching of the signal words, and blocking of the action emphasis. For reasons of brevity, treatment groups are 42
sometimes referred to in their acronymic form. For example
Crossed, Blocked, 3-Dimensional (CB3) and Non-Crossed, Non
Blocked, 2-Dimensional (NCNB2). Treatment means for the various dependent measures investigated are graphically presented in Figures 5 thru 9 for each DV. Since the con trol group was not included in the present analysis their means are presented for reference purposes only. Control group differences are discussed in the second analysis.
With the exception of the task completion time measure, all the DVs seem to indicate a rather narrow range of values.
This would suggest that the various factorial combinations of crossing, blocking, and dimensionality did not produce the differential effectiveness that was hypothesized (the lack of variability with the task outcome measure was ex pected),. The task completion time measure does however show some evidence of these anticipated differences. Fig ure 5 clearly shows the widest degree of variation among the DVs measured. In fact there is some evidence to sug gest the presence of a moderate 3-way interaction. As can be seen, task completion times are dependent on the level of the third factor, in this case dimensionality. Subject scores seem to be inverting as a function of this factor.
For example, subjects in the noncrossed conditions tended to have lower scores when warning messages were non-blocked and 2-dimensional. However, when warnings were 3-dimen sional scores were much higher. This same sort of effect is present for subjects in the crossed conditions, except 43
fr---A Crossed ~ Non-Crossed ~ Control
1800
1700
1600
1500
1400 -Ul '0 s:: 0 0 1300 (!) Ul 2-Dimensional 'r I I Blocked Non-Blocked ~ Highlighting Conditions H E-1 ::.::: Ul < *· Control E-1 I 1800
1700
1600
1500
1400
1300 )-Dimensional 't I I -Blocked Non-Blocked Highlighting Conditions Figure 5. Mean task completion times for each highlighting condition by cell. 44
~Crossed o--o Non-Crossed 16 it Control 15 14 13 12 11 10 9 s. 8 : 7 6 5 2- Dimensional ~ [/) Blocked Non-Blocked 0::: ~ Highlighting Conditions ~ E-f 0 E-f 'it Control 16 15 14 13 12 11 10 9 8 7 6 5 I( )-Dimensional 1 Blocked Non-Blocked Highlighting Conditions
Figure 6. Mean total error rate for each highlighting condition by cell. 45
8. 4.0 ~ Non-Crossed
J.O ~ Control
...... \.0 II 2.0 Q) $..! 0 0 tiJ s 1.0 : e ~ ·r-1 >< cU .._.s 2-Dimensional
t/) Blocked Non-Blocked o:::; Highlighting Conditions o:::;~ ex: ::s: 4.0
J.O *Control
2.0
1.0
3-Dimensional Blocked Non-Blocked Highlighting Conditions Figure 7. Mean warning message error rate for each highlighting condition by cell. 46
l::r---l:l. Crossed o--o Non-Crossed n 6 ts.
...... 0 s:: 0 0 5 ·.-I +' ctl C) 4 0 ...-i "d s:: 3 ctl
Q) QD ctl 2 [/) [/) sQ) 1 QD s:: ·.-Is:: 2-Dimensional ~ ctl Blocked Non-Blocked a= Highlighting Conditions -Q) +' ctl ~ 0 0 H 6 H <:t:: u ~ r.r.l ts. 0::: 5 ...-i ctl +' 0 4 +' s:: ctl 3 ~
2
1
3-Dimensional Blocked Non-Blocked Highlighting Conditions Figure 8 . Mean warning message recall rate for experimental highlighting conditions. 47 I '
~Crossed
100 ~ Non-Crossed 88 75 tsr------~ 63 50
38
25 * Control 13 2-Dimensional 0 Blocked Non-Blocked Highlighting Conditions
100 88
75 ls=------~ 63 o---.______50 --o 38
25 1' Control 13 3-Dimensional 0 Blocked Non-Blocked Highlighting Conditions Figure 9 . Mean percentage correct motor operation for each highlighting condition by cell. 48
that it is in the opposite direction.
An overall multivariate analysis for these experimental conditions is summarized in table 2. Using Wilk's criter ion, none of the main effects or higher order interactions proved to be significant. The largest F' score yielded was
1.82 for the C X B X D interaction. Due to the absence of any overall multivariate significance it was de~med inap propriate to assess the effects of the IVs on the indivi dual DVs. Therefore, a Roy-Bargman Stepdown Analysis was not performed.
It should be noted that if the present analysis were to be conducted in a univariate fashion (i.e. ANOVA) results would have been somewhat different. The univariate analy sis of variance for task completion time indicated a signi ficant C X B X D interaction, F(l, 55) = 9.28, P < .004.
Given that five dependent measures were employed in the present analysis, one could expect overall multivariate significance if fewer DVs were used because more degrees of freedom would be available for the analysis.
The present results also suggest that no essential dif ferences in effects existed among the IVs. This would seem to imply that all the highlighting conditions moderated the
DVs about equally. Thus, degree of visual highlighting for warning messages seemed to be of little consequence when compared to one another. 49
Table 2 Overall tests of multivariate analysis of variance for the combined set of.dependent variables.
SOURCE DF F SIGNIFICANCE
CROSSING( C) 5/51 .841 .526
BLOCKING(B) 5/51 .630 .677 DMENSION(D) 5/51 .792 .559 c X B 5/51 .JJ5 .890 c X D 5/51 1.68 .157 B X D 5/51 .JJJ .890 c X B X D 5/51 1.82 .127 50
Analysis II
The second MANCOVA analysis was an overall test of the
effects of warning messages between the control and experi mental groups. In this case, the independent variable was
the presence of warning messages in the instructions, re gardless of the highlighting technique employed. Treatment, means for the DVs used in the present analysis are illus
trated in Figures 10 thru 13.
With the exception of the task completion time measure, all of the treatment means were found to be in accordance with the hypothesized differences between the control and experimental groups. Figure 10 shows that task completion
time was actually greater in the control condition (~ =
1871 sec.) than the experimental (~ = 1630 sec.). This is opposite to what was expected since it was reasoned that the inclusion of warning messages would slow task assembly times. Figure 11 depicts the mean total error rate for the two groups. As expected, subjects made more total errors
in the control group (~ = 16.37) than the experimental (~ =
9.23). Figure 12 depicts the mean error rate associated with warning messages. Again as with TOTALERS, subjects made more warning errors in the control (~ = 3.00) than the experimental (~ = 1.25). Figure 13 shows the effects of motor outcome between the two groups. As predicted, the percentage of correct motor operation was greater in the experimental group(~= 75%) than the control (~ = 25%).
Using Wilk's criterion, the overall multivariate 51
1900
1800
,...... 1700 [I) 't:l ~ 0 1600 (.) Q) [I) - 1500 ~ H E-1 ::.::: 1400 {J) c::x: E-1 1300
1200
Control Experimental * Highlighting Condition Figure 10. Mean task completion time for control and experimental groups.
* Represents the mean of all experimental highlighting conditions combined. 52
16 15 14 13 12 11 10 9 rn eg0::: 8 Control Experimental ~ Highlighting Condition Figure 11. Mean total error rate for control and experimental groups. * Represents the mean of all experimental highlighting conditions combined. 53 ...... ~ II ().) ~ 0 (.) rJl E ;:s E •r-i >< ttl J.O ...... E w 0:: ~ 0:: <:r: ~ 2.0 1.0 Control Experimental * Highlighting Condition Figure 12. Mean warning message error rate for control and experimental groups. * Represents the mean of all experimental highlighting conditions combined. 54 ~ r-l +' 0 . (]) 100 $-! ~ 0 0 '1j (]) +'m ~ 75 (]) Pi 0 0::: 0 E-i 0 :E 50 . (]) mtiD +' ,:::: (]) 0 ~ (]) p.. 25 ,:::: m ~ - Control Experimental* Highlighting Condition Figure 13· Mean percentage correct motor operation for contr~l and experimental groups. * Represents the mean of all experimental highlighting conditions combined. 55 analysis revealed a highly significant main effect, !(4,66) = 6.55, P < .0002. This indicates that a substantial rela tionship existed between the presence of warning messages and the combined DVs. Consequently, a step-down analysis was performed to determine the significance of each depen dent variable separately. In a step-down analysis, the highest priority DV is tested with a univariate analysis of variance. Each subse quent dependent measure is evaluated in turn with the high er priority measures acting as covariates. This is in ad dition to any covariates previously used in the overall analysis. The MOTOR dependent measure was entered into the equation first, followed by TOTALERS, WARNERS, and task completion time (RECALL was of course not employed in this analysis since data on this measure was not collected from subjects in the control condition). MOTOR was given the highest priority because of the expected impact of subject errors on the outcome of the entire task. That is, since errors were allowed to be cumulative (except for those that prevented required subsequent actions) their effects would be reflected on the operability of the device. TOTALERS was entered into the equation second because it was thought that warning messages would moderate subject error rate over the entire task. Since WARNERS is considered to be a subset of TOTALERS, it was entered into the equation after TOTALERS. Task completion time was given the lowest prior ity since subject variations in overall performance could 56 easily confound this measure. An experimentwise signifi cance level of .05 was maintained by evaluating each DV at the .01 level. Table 3 shows the step-down analysis con ducted to test the effects of warning messages between the experimental and control conditions. MOTOR. Correct motor operation was found to be signi ficantly related to the presence of warning messages (step down F (1,69) = 9.24, p < .004; "J?. = .11). As was indica ted in Figure 13, three times the number of subjects in the experimental conditions had the assembled motor operate, compared to the control group. It should be noted that this ratio did not change, even after the treatment means were adjusted for the experience level covariate. TOTALERS. The relationship between total error rate and warnings was found to have the strongest IV-DV associ ation (step-down F (1,68) = 12.13, p < .001, ~1. = .22). Again, as with the MOTOR DV, covariate adjustments produced only minor differences in the treatment means for the two groups. WARNERS. The third highest priority dependent measure, warning errors, was found to be non-significant in the step-down analysis. It should be noted that this result must be viewed within a multivariate context in which total error was partialled out of this measure. Thus it would be inappropriate to draw any conclusions regarding warning errors, per se. TASKTIME. Total task completion time, the lowest 57 Table J Source table of Roy-Bargman Stepdown analysis for experimental versus control conditions. SOURCE DF UNIVARIATE STEPDOWN ETA2 F F MOTOR( 1) 1/69 9.24*· 9.24* .11 TOTALERS(2) 1/68 19.99*** 12 .1J** .22 WARNERS(J) 1/67 14.85*** .?5 .18 TASKTIME(4) 1/66 J.5? 2.25 .05 NOTE: Numbers in parentheses indicate the order in which variables were entered into the analysis. * .J: <. 004 ** P<.001 *** P <.OOOJ --- 58 priority dependent measure, was not significantly related to the presence of warning messages. This was confirmed by examining both the step-down and univariate F. DISCUSSION The primary objective of this study was to determine which highlighting configuration, if any, was optimal for displaying precautionary information in text. A secondary objective was to determine the effects of warning messages, per se, on a novel assembly task. As a set, the various hypotheses regarding differential warning message high lighting could not be supported from the data. As a conse quence, task completion time, total error rate, warning message error rate, and recall were not shown to be signi ficantly different between the various warning message con ditions. However, a majority of the hypotheses concerning the overall effects of warning messages were supported. The statistical analysis yielded significant effects for motor operability and total error. rate. The assembly time measure was not significant. Within the context of the present study, it would seem that the highlighting techniques and the concept of sali ence, did not produce the desired differential effects among the experimental conditions. Results do not confirm the a priori ordering of their relative effectiveness. More importantly, no conclusive evidence could be found to suggest that any of the highlighting methods employed was superior to any other of the methods. It was hoped that results from this study could be offered as support for one highlighting method over another (e.g., dimensionality over .-..· 59 ~ ---- 60 cross hatching), however, this was not the case. No clear trends were established with any of the DVs employed for the experimental conditions. There are several possible explanations for this occur rence. The most reasonable explanation is that all the highlighting conditions do moderate subject performance about equally. Thus, the degree of visual highlighting present was not critical. The presence of the warning mes sages themselves could have been the determining factor. A considerable amount of evidence from the second analysis supports this contention. With the exception of the task completion time measure, all of the treatment means for the DVs support this conclusion. It should be noted that the time measure suggests that subjects took longer to perform the assembly task in the control condition than the experi~ mental conditions. However, this difference was not signi ficant. This is totally opposite to what was expected. Instead of slowing performance times, warning messages ten ded to quicken them. A possible reason for this could be due to the method by which warnings were placed into the instructions. As was stated earlier, the frequency and severity of subject errors (from a pilot study) were used as selection criteria for the warnings. This was done to take advantage of any systematic errors associated with the assembly task. The warning messages were, therefore, loca ted at critical points throughout the instructions. Thus it is possible that the warnings allowed subjects to .,_....------ 61 process critical information faster and perform the re quired actions more efficiently. This could account for control subjects taking longer to perform the task than the experimental subjects. A potential explanation for the task completion time results and the overall homogeneity of the experimental conditions could be due to the lack of a "stressing" factor associated with the task. In the context of the present study, stress could take the form of a limited amount of time (speed stress) to perform the task. Typically, single trial assembly tasks like those of the current study do not have time limits associated with them. In fact, subjects in this study were specifically told they would have "••• more than enough time to assemble and operate the motor." Assembly time was therefore not perceived by subjects as being a critical factor in the task. However, procedures used in emergency situations will frequently have critical time limits associated with key steps and, in some cases, warning messages (NASA, 1978). Such warning messages are intended to help the user make accurate and timely deci sions regarding a critical step or action. Thus, results might have been different if subjects were given specific time limits to perform certain tasks. One might then expect that certain highlighting conditions, e.g., Dimen sionality to be superior over others. With regard to task completion time, it would be expected that overall perfor mance times between the experimental and control groups 62 would be opposite from what they are presently. It is interesting to note some of the unexpected find ings regarding warning message saliency and recall. No evidence was found to indicate that recall was differen tially enhanced by any of the highlighting formats. Sub jects were given ample time to recall warnings and were allowed to guess. However, few subjects could recall more than half the warnings and none could recall the entire set of warnings correctly. During the debriefing period, sub jects were shown representative examples of each highlight ing configuration used in the experiment. Each variable was fully explained along with the specific differences be tween each format, e.g., CB2 versus CB3. Subjects were then asked to identify the specific highlighting format they were exposed to during the assembly procedure. There was a tendency for subjects to add a degree (or more) of saliency to the highlighting configuration to which they were exposed. In other words, many subjects overestimated the amount of highlighting they actually received. For ex ample, subjects might say that their warnings appeared to be 3-dimensional or blocked, when in fact they were 2-dimensional or non-blocked. Forty-six percent of the subjects who were asked to recall their own highlighting configuration, incorrectly added a degree of saliency. Only 30% of those surveyed could accurately remember the warning message format they were exposed to. In total, 70% of the subjects that were ___..--~- 63 queried could not accurately recall their own highlighting format. This finding might also explain why the various warning message formats did not produce the expected dif ferences in subject performance. The tendency to attribute increasing amounts of sali ency to warnings is interesting. This finding also tends to agree with McGuinnes' (1977) survey of warning labels for power lawnmowers. In that study he found the perceived effectiveness of different warning labels increased as the degree of realism of the hazard increased. Of course these findings are in no way indicative of any particular warning message format superiority. However, it does seem to sug gest that there was a tendency among subjects to attribute greater saliency to the warning than actually existed. Warning Messages and Task Performance Few studies can be cited which relate directly to be havior and performance aspects of warning messages asso ciated with assembly tasks. Preliminary results indicated that perception of critical information was increased by the use of warning messages. This runs counter to an in formal study conducted by Dorris and Purswell (1977) in which warning labels affixed to hammers were virtually unnoticed by subjects. Part of this is no doubt due to the high familiarity these subjects had with the apparatus and the task. The assembly apparatus and tasks utilized in the present study were intended to be completely novel to ,.._...- --- - 64 subjects. This novelty factor may have accounted for the differences. Results might have been different if a famil iar assembly task was utilized. However, generalizability of results would have been affected. 'I'he present findings indicated that both subject error rate (TOTALERS} and task outcome {motor operability) were significantly moderated by incorporating warning messages into the assembly procedures. Clearly, the location of warning messages in assembly instructions is critical to proper task performance. But warnings, in and of them selves, are not sufficient to promote this. Although warnings may be reasonably expected to reduce subject error rate, there mere inclusion in assembly proce dures may not promote successful task outcome. Desired task outcome may be promoted only when warnings are system atically worded and placed within the procedures. In the present case, correct motor operation was enhanced because the warnings were intentionally located at critical points throughout the assembly task. In other words, the warning messages were linked to specific actions that could effect the operability of the device. Although the interpretation of these results is admit tedly speculative, it does suggest that warnings properly designed and located may 1} significantly reduce subject error rate when performing a novel task, and 2) promote desirable task outcomes. Unfortunately, present experimen tal design considerations prevented the author from 65 quantifying the effects of warning message location on per formance outcome. Implications and Limitations The case can be made that for any given warning message format used in this experiment, there were at least two different highlighting techniques present. 'I'his is in ad dition to the actual highlighting methods evaluated in the present study. 'l'he first one relates to the act ion emphasis, while the second relates to the signal word. Specifically, differences between the control and experi mental groups could be attributed to the action-emphasis (without the signal word), or, the signal word (without the action-emphasis). Although the combination of these two factors, i.e., NCNB2 proved to be effective, they were not evaluated as individual factors. Additional research could be conducted to resolve this discrepancy. However, this would have to be weighed against certain practical consid erations. For instance, a signal word, e.g., "Warning", without an action-emphasis provides no direction to the user. In a similar sense, an actionemphasis without a sig nal word provides no level of criticality to the user re garding his actions. Such configurations would probably not conform to the standardized formats for warning mes sages currently being promulgated. There are several limitations to the present study that should be mentioned. The experiment relied heavily on the 66 role playing ability of subjects with regard to the hazard ous nature of the assembly task. This was required since no real assembly hazards, i.e., bodily injury are ethically possible. Thus it would not be entirely valid to general ize these results to actual assembly situations. The fact that the study was done using a student subject population in a laboratory environment also limits generalizability. These results might not be applicable if subjects had been familar with the apparatus, as with most assembly line op erations. However, the results do seem to be applicable to one time assembly situations such as those associated with many consumer products on the market today. Alternative methods should be investigated for further research in the area of warning messages. If possible, methods should be developed for direct observations in com mercial/industrial situations. The effectiveness of warn ing messages placed on actual equipment should also be in vestigated. Manufacturers will frequently place warning messages directly onto the equipment. Typically this is done by manufacturers to assure that the legal obligations to warn of potential hazards are met, even if assembly or operating instructions are not available. As a preliminary effort, warning messages could be placed on selected compo nents used in the present study. Information on this matter could then be easily collected and reported. One situation in which a change in warning message highlighting would be appropriate is with the commercial 67 nuclear industry. This is suggested for two reasons. First, the NRC is in the process of drafting a specifica tion on emergency operating procedures (NUREG-0799, 1980). A portion of this new specification will be concerned with highlighting precautionary information. Second, visual in spection of available Nuclear Power Plant operating proce dures indicates very little in the way of elaborate warning message highlighting (see Figures 1 and 2 for examples). Thus, any highlighting changes that might be effected would not be expected to interfere with established user expec tations. 68 References American National Standards Institute. ANSI Z35.1 Standard Specification for Accident Prevention Signs. New York, N.Y: American National Standards Institute, 1973. Brainard, R.W., Campbell, R.J. and Elkin, E.H •• Design and Interpretability of Road Signs. Journal of Applied Psychology, 1961, 45, 130-136. Bresnahan, T.F., and Bryk, J •• The hazard association values of accident prevention signs •. Professional Safety, 1975, January, 17-25. Bry,B •• Product Liability: Firms Face Rising Costs as Injury Awards Swell. Los Angeles Times, October 22, 1978, Business Section, pp 1~2~9. Dorris, A.L., and Purswell, J.L •• Warnings and Human Behavior: Implications for the Design of Product Warnings. Journal of Products Liability, 1977, 1, 207-220. Dorris, A.L., and Tabrizi, M.F •. An Empirical Invesiti gation of Consumer Perception of Product Safety. Journal of Products Liability, 1978, ~, 155-163. Ells, J.G., and Dewar, R.E •• Rapid comprehension of verbal and symbolic traffic sign messages. Human Factors, 1979, 21, 161-168. FMC Corporation. Product Safety and Label System - 3rd Edition. Santa Clara, CA.: FMC Corporation, 1980. 69 Fuchs, F., Engelschall, J., Imlay, G •• Evaluation of Emergency Operating Procedures for Nuclear Power Plants (NUREG/CR-1875) U.S. Nuclear Regulatory Commission, Washington, D.C.: u.s. Government Printing Office, April 1981. Inaba, K., Engelschall, J., Barr, w., and Gold, D •• A Specification For Bus Maintenance Manuals (UMTA-IT-06- 0235-81-l) Urban Mass Transit Administration, Washington, D.C.: November 1980 (available through NTIS). King, E.L •• Recognition of Symbol and Word Traffic Signs. Journal of Safety Research, 1975, 2, 80-84. Martin, G.L., and Heimstra, N.W •• The Perception of Hazard by children. Journal of Safety Research, 1973, 5, 238-246. McGuinnes, J •• Human Factors in Consumer Product Safety. Proceedings of the Human Factors Society - 21st Annual Meeting, 1977, 292-294. MIL-HDBK-63038-l (TM) Technical Manual Writing Handbook. Washington D.C.: u.s. Department of Defense, l May 1977. MIL-M-85025A (AS) Manuals, NATOPS Flight: Requirements For the Prepartion of. Washington, D.C.: U.S. Department of Defense, 8 December 1980. Moll, R.A •• Product Liability: A Look at the Law. Engineering Education, 1976, 66, 326-331. 70 Nie, N.H., Hull, C.H., Jenkins, J.G., Steinbrenner, K. and Bent, D.H •• Statistical Package for the Social Sciences. 2nd Edition, New York: McGraw Hill, 1975. Peters, G.A. New Product Safety Legal Requirements. Jour nal of the Systems Safety Society, 1978, ~(1), 21-23. Philo, H.M., and Rine N.J •• The Danger was never obvious. Journal of Products Liability, 1977, !' 12-19. Ross, K •• Legal and Practical Considerations For The Creation of Warning Labels and Instruction Books. Journal of Products Liability, 1981, 4, 29-45. Tabachnick, B., and Fidell, L •. Use and Interpretation of Multivariate Statistics. In Press. Tokuhata, G.K., Colflesh, v., Smith M., Ramaswamy, K., and Digon, E •• consumer behavior and product injuries. Journal of Safety Research, 1976, 8, 116-125. u.s. National Aeronautics and Space Administration, Space Shuttle Flight Data File Preparation Standards (JSC- 09958). Appendix H Crew Procedures Management Plan. December 14, 1979. u.s. National Aeronautics and Space Administration, System Malfunction Procedures Requirements (JSC-10528). Appendix B Crew Procedures Management Plan. April 15, 1978. u.s. Nuclear Regulatory Commission, Draft Criteria for Preparation of Emergency Operating Procedures (NUREG- 0799 For Comment). Washington, D.C.: U.S. Government Printing Office, July 1981. 71 Walker, R.E., Nicolay, R.C., and Stearns, C.R •• Comparative accuracy of recognizing American and international road signs. Journal of Applied Psychology, 1965, 49, 322-325. Woodson, W.E., and Conover, P.W •• Human Engineering Guide for Equipment Designers. Berkeley: University of California Press, 1973. Reference Notes 1. Dorris, A.L. Product Safety Manager, J.I. Case Co., Personal Communication, September 4, 1981. 72 APPENDIX A CONTROL GROUP INSTRUCTIONS NOTE: Written instructions used in the experiment were printed on both sides of the page. Bureaucratic idiosyncrasies forced the author to present them on single pages. -- --- 73 INSTRUCTIONS FOR ASSEMBLY OF A PROTOTYPE ELECTRIC MO'l'OR I want you to imagine that you work for a co.npany that manufactures electric motors for indusLri.'\ l atse. Your task is to assemble and operate a miniat_._n·e "prototype" version of an elect.ric motor in this not.ebook according to a set of written instructions. The instr·r~tions will give you precise step-by-step directions on how to assemble and operate the motor. Before we beg in, I w.:mt you to become fanli 1 iar with these t~ools and supplies you 111ill use t.o assemble t_he motor. The tools and supplies you will use are pictured on the opposite page. I want you to take a few moments to become famil in.r with these it.e;ns ·'tnd t~hei r names. Most of tl1e parts you will need are contained in tJ1 is parts bin. Their names are indicated inside each part~ bi~. In 5ome cases, there may be more than one typ·~ or s i.. ~~~ •) E :noLor part in a bin. Each step in the instructions is S!l!:_)i_Jle·nented with pict11res. The text is on the left side Clnd the related pictures are on the right. An ite;n in t_he ste;? is usually f:Jllowed by a number in bracket.s. Search t:L:l ,,?ict_lr:-e for t~&e part that is pointed at by an ar!"ow wi tn that same nu:nber. You should then find that [>art in the part bin or on U1e table :'lnd perform the st.ep that you are inst-ructed to ·1·::>. Follow the entire instructions carefully, ste~-by-stap. Remember that you are working for a company that m.:i"'-<·:!-3 t::1ese :notors for industrial use. If the motor is ptlL L·:l·Jel':ler incorrectly, it might not work. Also remember that this is a miniature version of the actual motor. A real motor, incorrectly assembled, might cause injury to the operator. Try to work as quickly and accurately as you can. Thet:'e is ·nore than enough time for you to assemble and operate f·."'ne •notor. However, once you turn a page, there is no going back to a previous page. If there are no questions, you may begin ~1en you turn the page. .,.., ...... ------~ 74 POINTED PLIERS NUTDRIVER WRENCH 8~~ WRENCH SCREWDRIVER OIL --- '------~ ------ 75 ASSEMBLY INSTRUC'riONS FOR PROTOTYPE MOTOR St-ap l: Position motor console (5) so that arr0~ (l) is on top side and points toward you. St~p 2: Push VII·:> 1-3/4" screws (3) through holes #25 (2) and #28 (4) fran botlo~ of motor console (5). 76 • • 1 5 MOTOR CONSOLE ---- 77 ASSEMBLY INSTUCTIONS FOR PROTOTYPE MOTOR Step 3: Lower curved pole piece (2) onto screws so that curved side of pole faces down and extra screw hole (1) points left. Step 4: Install nuts (3) on 1-3/4" screws finger tight. Using box wrench (4), tighten each nut 1/4 of a turn. 78 ---- ~------..,.- ' 2 4 ------ ------~------79 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 5: Push 7/8" screw (4) through single hole on angled bracket (3). Install nut (2) and tighten securely. Set bracket assembly (1) to the side. Step 6: Repeat Step 5 on second angled bracket. Then, continue. Step 7: Place screw (6) of first bracket assembly (1) through hole #47 on motor console. Make sure that bracket is flush against left side (5) of console slot. Install washer (7) and nut (8) from below console and tighten securely with nutdriver wrench. 80 2 3 4 1 -- ANGLED BRACKETS 5 6 BRACKET INSTALLATION 81 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 8: Push 7/8" screw (1) through hole #36 from bottom of motor console. Install three nuts (2) on screw finger-tight. Using nutdriver wrench, tighten nuts 1/4 of a turn. Step 9: Use the following chart to identify upper and lower contact brushes (see chart). MOTOR CONTACT BRUSHES POSITION SHAPE UPPER ~ LOWER .. .-·._v . c:J Step 10: Select the lower contact brush (4 ) . Do not attempt to bend the brush until after it is installed on the console. Place brush on top of screw (1) at hole #36. Make sure that brush tip (5) points up and screw passes through center hole (3). 82 ' . 1 2 .. ~<,. LOWER CONTACT BRUSH INSTALLATION ------83 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 11: Install nut (3) on top of lower contact brush (4) finger-tight. Make sure that brush is parallel with installed pole piece (1) and extra hole (2) is on left side of screw. Step 12: While holding lower contact brush (4) in place, tighten nut (3) exactly 1/8 turn. Make sure that brush does not move while tightening nut. 84 • • .! ./< LOWER CONTACT BRUSH INSTALLATION 85 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR St ep 13: While holding motor console at eye level, see if edge (1} of contact brush can be seen through upper hole (2} of angled bracket. If edge of brush can be seen through upper hole, go to Step 15. If not, continue. Step 14: Carefully bend contact brush up or down, as needed, so that edge (1} of brush is seen through upper hole (2) of angled bracket. Step 15: Place screw (3) of second angled bracket through hole #7 on motor console. Loosely install washer (4) and nut (5) from below console so that bracket can turn around freely. Turn bracket so that long side (7) is even with outer edge (6) of console. --- - 86 1 2 MOTOR CONSOLE- SIDE VIEW ,' - -~> ~t "" -----·• ; 7 -----·- ~ -- 87 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 16: While holding armature by iron plates (3) insert spacer (2) onto spindle (1) as far as possible. Do not touch copper windings or commutator. Step 17: While slightly bending down tip of contact brush (5) with wrench, insert spindle (1) into upper hole (4) of bracket. Remove wrench. Step 18: Make sure that tip of contact brush (5) touches commutator (6). -- ...... ------88 ARMATURE INSTALLATION ------.... 89 • • ASSEMBLY INS'l'RUCTIONS FOR PROTOTYPE MO'rOR Step 19: Turn second angled brackeL (2} around and insert opposite end of spindle {3) into upper hole of bracket. Step 20: While holding second angled bracket (2} in place, rotate longer pact of s2indle (1} to see if armature (4} turns freely. Do not tigbten the second angled bracket securely until you are suee d.r~ature t11rns freely. If armature turns freely, go to Step 22. If not, continue. Step 21: Make sure t .hal: SL) i11d le ( 3} moves freely ins ide upper hole of aDg 1 e<1 br~c:ke t ( 2} and tip of contact brush (5) is not stu~k ins i_ , l~ co•nrnutator. Step 22: vfuile holding angled brrtc%1-:!L (8} wiU1 finger, raise motor console. Tighten nut n!1der console wi t~1 nutdri ver wrench. Make sure that bracket i3 3traight, inside console slot (6}. Apply one drop of oil to upper hole (7} of bracket. - 90 1 2 3 MOTOR CONSOLE AND ARMATURE 7 ..-- ... - -· --- ..... 91 ASSEMBLY INSTRUCTIONS FOR PROTOTYPl'~ rw1V'.L'OR Step 23: Select the upper contn.ct bC•l.,l 1. ( 1). Make sure that brush is securely mounled lo br~cket (2) so that brush does not move sideways. Loosely install nut (3) onto contact screw. Step 24: Insert l/2" screw (6) into hole #38 from bottom of motor console. Place bracket (5) over screw so that upper contact bc•Jsh ( 1) touches top of commutator (4). Install nut (7) onto screw finger-tight. Step 25: Make sure lhat upper contact brush touches middle of commutator (4) surface. Then, tighten nut (7) securely with acrew~rivec and box wrench. .... ---- 92 - f '· . - .. .,.._ 1 2 3 7 UPPER CONTACT BRUSH INSTALLATION 93 ASSEMBLY INSTRUCTIOI~~ FOR PROTOTYPE MOTOR St_~p 26: Select two round bar magnets (l) and two flat magnets (2). Connect magnets so that they form two back to back "L's". Make sure that similar ends of the magnet 1v1irs repell each other. Step 27: Attach magnets Lo both sides of curved pole piece (3) so that fl~t magnets (2) touch pole piece. ~ - - -- 94 1 3 t - MAGNET ASSEMBLY AND INSTALLATION 95 ASSEMBLY INST~JCTIONS FOR PROTOTY?E MOTU~ Step 28: Select second curved pole piec8 ( '2). T-'ower pole piece onto bar magnets (4) and long screws. Make sure that top of bar magnets are directly unaer pole 2iece. Install nuts (1,3} onto long screws and tightan securely. Step 29: Loosely install nut (6) on lowec contact brush screw. Select a 10" BLACK wire and a 10" BLdE wire. Attach BLACK wire to •.lpper brush screw (7) and BLUE wire to lower brush screw (5). Tighten nuts. E~· m 0? l?ROCEDURE 96 .' 4 , I, ...... _.,.;...... ;. _...___.___ ·~ ,...... 6 1 MOTOR CONSOLE WIRE ASSEMBLY 97 , . APPENDIX B EXPERIMENTAL GROUP (CROSSED, BLOCKED, 3- DIMENSIONAL) INSTRUCTIONS 98 INSTRUCTIONS FOR ASSEMBLY OF A PROTOTYPE ELECTRIC 1110·fOR I want you to imagine that y.'Ju work for a co.npany t"hat manufactures electric motors for industci-"1.1 11se. Your task is to assemble and operate a miniaL.lt:'e "prototype" version of an electric motor in this notebook according to a set of written instructions. T'ne instr:.1::tions will give you precise step-by-step directions on how to assemble and operate the motor. Before we beg in, I want you to become fa111 il iar with these tools and supplies you wi 11 use to as semble t_he motor. The tools and supplies you will use are pictured on the op~osite page. I want you to take a few moments to become familiar with tnese items -:lnd t·.heir names. Most of the parts you will need are contained in 81is parts bin. Their names are indicated inside e~ch part8 bi~. In some cases, t~ere may be more than one type or si~8 of :notor- part in a bin. Each step in the instructions is Sl\<1 :>lenented with pictures. The text is on the left side "ir'ld t~1e celated .rictures are on the right. An it-em in t-_he st.~? is t.lsually f'Jllowed by a number in brackets. Search t'1 ~ ~)Let_ 1 ce for L'1e part that is pointed at by an arrow with tnat same nunber. You should then find that part in the part bin or on Ll1e ta.~::>l~ :=md perform the stap that you are instr11cte..-l Lo ·1·:.>. Follow the entire instructions carefully, step-by-st8p. Remember that you are working for a company that ma.~ . ~s L~1ese notors for industria 1 use. If the motor is p•.ll tJ.Jet'1er incorrectly, it might not work. Also remember that this is a miniature version of the actual motor. A real rnot:.or, incorrectly assembled, might cause injury to the operator. Try to work as quickly and accurately as you can. T'n.ere is more than enough time for you to assemble and operate L'r1e 1notor. However, once you turn a page, tnere is no going back to a previous page. If there are no questions, you may begin when you turn the page. 99 POINTED PLIERS = NUTDRIVER WRENCH ------·- · -----::::-,. '-·------0 8t~ WRENCH SCREWDRIVER OIL 100 ASSEMBLY INSTRUC'riONS FOR PROTOTYPE MOTOR St:p 1: Position motor console (5) so that arrow (1) is on top side and points toward you. Steo 2: Push t'N'v l-3 /4" screws ( 3) through holes #25 (2) and #28 (4) frryn botlom of motor console (5). l 01 1 5 3 MOTOR CONSOLE 102 ASSEMBLY INSTUCTIONS FOR PROTOTYPE MOTOR Step 3: Lower curved pole piece (2) onto screws so that curved side of pole faces down and extra screw hole (1) points left. In the following step, make sure that you do not tighten nuts more than 1/4 of a turn Step 4: Install nuts (3) on 1-3/4" screws finger tight. Using box wrench (4), tighten each nut. 103 2 ' 3 104 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 5: Push 7/8" screw (4) through single hole on angled bracket (3). Install nut (2) and tighten securely. Set bracket assembly (1) to the side. Step 6: Repeat Step 5 on second angled bracket. Then, continue. In the following step, be sure to use nutdriver wrench to tighten nut Step 7: Place screw (6) of first bracket assembly (1) through hole #47 on motor console. Make sure that bracket is flush against left side (5) of console slot. Install washer (7) and nut (8) from below console. Tighten nut securely. 105 2 3 4 1 \ I ci.=O '· . - f ~-·,· .. -····_: _· . .. - "' · -,.. . r ' ': '' ANGLED BRACKETS 5 6 ,... ·. .. 7 •I :-8 @ BRACKET INSTALLATION 106 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 8: Push 7/8" screw (1) through hole #36 from bottom of motor console. Install three nuts (2) on screw finger-tight. Using nutdriver wrench, tighten nuts 1/4 of a turn. Step 9: Use the following chart to identify upper and lower contact brushes (see chart). MOTOR CONTACT BRUSHES POSITION SHAPE UPPER ~ LOWER t::J ~ Step 10: Select the lower contact brush (4). Do not attempt to bend the brush until after it is installed on the console. Place brush on top of screw (1) at hole #36. Make sure that brush tip (5) points up and screw passes through center hole (3). 107 1 2 4 5 LOWER CONTACT BRUSH INSTALLATION 108 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 11: Install nut (3) on top of lower contact brush (4) finger-tight. Make sure that brush is parallel with installed pole piece (1) and extra hole (2) is on left side of screw. In the following step, make sure that brush does not move or bend Step 12: While holding lower contact brush (4) in place, tighten nut (3) exactly 1/8 turn. 109 • • - ' ;': .d'-l 'f 2 3 4 LOWER CONTACT BRUSH INSTALLATION • • ···•·-·- · .. - -. .•. ·- · • • .-·- ··"'··-·-•.... -- 110 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 13: While holding motor console at eye level, see if edge (1) of contact brush can be seen through upper hole (2) of angled bracket. If edge of brush can be seen through upper hole, go to Step 15. If not, continue. Step 14: Carefully bend contact brush up or down, as needed, so that edge (1) of brush is seen through upper hole (2) of angled bracket. In the following step, make sure that bracket can still turn around freely Step 15: Place screw (3) of second angled bracket through hole #7 on motor console. Loosely install washer (4) and nut (5) from below console. Turn bracket so that long side (7) is even with outer edge (6) of console. 111 1 2 MOTOR CONSOLE- SIDE VIEW ~- ' • : 112 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 16: While holding armature by iron plates (3) insert spacer (2) onto spindle (1) as far as possible. Do not touch copper windings or commutator. Step 17: While slightly bending down tip of contact brush (5) with wrench, insert spindle (1) into upper hole (4) of bracket. Remove wrench. Step 18: Make sure that tip of contact brush (5) touches commutator (6). 113 k· r ! - -·,. - ~ ·'\ 5 ARMATURE INSTALLATION 114 ASSEMBLY INS'l'RUCTIONS FOR PROTOTYPF.: M<.)'rOR Step 19: Turn second angled brack8L (2) around and insert opposite end of spindle (3) into upper hole of bracket. Step 20: While holding second r1ngled bracket (2) in place, rotate longer p a rt of 52indle (l) to see if armature (4) turns freely. Do not tighten the second angled bracket securely until you are Stlt"~ dr:nature turns freely. If armature turns freely, go lo St~~ 22. If not, continue. Step 21: Make sure that spi~ile (3) moves freely inside upper hole of angl~<1 'rJ:r .-1:-::~·~t ( 2) and tip of contact brush (5) is not st~~k ir1s i. :lP- ::::o•n•nutator. Step 22: \fuile holdi~g angled hrnck~L (8) wit~ finger, raise motor console. Ti~ht~n nuL n~der console wi L1 nutdri ver wrench. 1'1-:tke sure U1dt bracket i3 ~tr~ight, inside console slot (6). Apply one drop of oil t.J upper hole (7) of bracket. 115 1 2 3 MOTOR CONSOLE AND ARMATURE 7 -,..... - - ,/ ,.___ . _ 116 .. ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR Step 23: Select the upper contact brush (1). Make sure that brush is securely mounted to bracket (2) so that brush does not move sideways. Loosely install nut (3) onto contact screw. Step 24: Insert 1/2" screw (6) into hole #38 from bottom of motor console. Place bracket (5) over screw so that upper contact brush (1) touches top of commutator (4). Install nut (7) onto screw finger-tight. In the following step, be sure to use screwdriver and box wrench together Step 25: Make sure that upper contact brush touches middle of commutator (4) surface. Then, tighten nut (7) securely. 117 .. ~--- ~: ~ ~ . 1 2 3 UPPER CONTACT BRUSH INSTALLATION 118 ASSEMBLY INSTRUCTIONS FOR PROTOTYPE MOTOR In the following step, make sure that similar ends of magnet pairs repell each other Step 26: Select two round bar magnets (1) and two flat magnets (2). Connect magnets so that they form two back to back "L's". Step 27: Attach magnets to both sides of curved pole piece (3) so that flat magnets (2) touch pole piece. 119 1 3 MAGNET ASSEMBLY AND INSTALLATION 120 ASSEMBLY INSTRtJCTIO~S FOR PROTOTYPE lvlO'l'OK Step 28: Select second curved pole piece (2). Lower pole piece onto bar magnets (4) and long screws. Make sure that top of bar magnets are directly under pole ~ieee. Install nuts ( 1, 3) onto long screws and light.~n securely. Step 29: Loosely install nut (6) on lower contact brush screw. Select a 10" BLACK wire and a 10" BLUE: wire. Attach BL..:z\.CK wire to upper brush screw (7) and BLUE wire to lower brush screw (5). Tighten nuts. ENO OF PROCEDURE 121 1 4 ' ...... 6 7 - "< .... MOTOR CONSOLE WIRE ASSEMBLY 122 APPENDIX C Table 4 Source table of intercorrelations for the dependent variables. TASKTIME MOTOR TOTALERS WARNERS RECALL TASKTIME X MOTOR -.03 X TOTALERS .09 -.41 X WARNERS .14 -.so .54 X RECALL .15 .20 -.26 -.36 X