You are on page 1of 15

Perceptual and Motor Skills, 1996,82,1187-1201.

O Perceptual and Motor SkiUs 1996

HIERARCHICAL MENU DESIGN: BREADTH,


DEPTH, AND TASK COMPLEXITY '
JULIE A. JACK0 AND GAVRIEL SALVENDY
Depar~mentof Engineering School of Itzdrrstrial Engineering
T r e n t o ~Stale College Purdue University

Szrmmary.--In this research a relationship berween an hierarchical menu's depth


and the perceived complexity of a task involving menu retrieval was proposed and
validated. 12 subjects were asked to use six ddferent hierarchical menus of varying
breadth and depth. The dependent variables were response time and accuracy. The
independent variables were depth and breadth of the hierarchy. Subsequent to exped-
mentation, the subjects were asked to complete a questionnaire on users' perceptions
of the complexity of the ddferent menu structures. As depth increased, perceived com-
plexity of the menus increased significantly. These phenomena are h k e d to an exist-
ing theory of task complexity. We suggest that the cognitive component influencing
users' perceptions of task complexity was short-term memory load.

A task has been d e f i e d as a set of conditions that elicit specific activi-


ties or processes (Fleishman & Quaintance, 1984). A comprehensive charac-
terization of any task requires a statement about its complexity, which is
thought to affect both mental workload-and performance in man-machine
systems (Zhao, 1992). Thus, it is imperative that the complexity of a task be
assessed when designing tasks that are part of a man-machine system.
It is important to make a distinction between the difficulty and com-
plexity of a task. The two terms are often used interchangealily when they ac-
tually refer to two different things. The difficulty of a task has been defined
as the amount of effort required to complete the task (Shaw, 1976). A task
can be difficult because it requires a large amount of effort without neces-
sarily being complex; other tasks are ddficult because they are complex, that
is, they require high skill or knowledge (Campbell, 1988; Locke, Shaw, Saari,
& Latham, 1981). The Subjective Workload Assessment Technique, develop-
ed for s c h g air transport pilots' mental workload, indicated that tasks of
low complexity are typically performed automatically. Moderate complexity
is determined by uncertainty, unpredictabihty, or unfamiliarity. Very corn-
plex tasks require extensive mental effort, skilled planning, and total atten-
tion to task (Eggemeier, McGhee, & Reid, 1983; Reid, Shingledecker, & Eg-
gemeier, 1981).
Traditionally, the complexity of a task has been expressed on a contin-

'This study was made possible throu h the support of the NEC Fellowshi to J A Jacko and
the NEC Proiessorship to G. ~ a l v e n 8Address enquiries to J. A Jacko, &.D., b$anrnent of
Engineering, Trenton State College, W w o o d Lakes, CN 4700, Pennington Road, renton, NJ
08650-4700.
1188 J. A. JACK0 & G . SALVENDY

uum ranging from simple to complex, and researchers have assumed that the
underlying attributes of tasks that determine complexity are understood. An
examination of the relevant literature suggests that in most cases these attri-
butes have not been identified and are not understood. The first objective of
this research was to address this lack of understanding by examining com-
plexity in a domain in which it has previously not been examined. Once the
attributes of the task that drive the users' perceptions of complexity have
been identified, the second objective of this research was to provide a theo-
retically based justification of the contribution of these attributes to
complexity.
Task Complexity
It is imperative that the complexity of a task be assessed when design-
ing hurnan-machine systems because complexity plays a significant role in
the assessment of human workload (Lysaught, Hd, Dick, Plamondon, Lin-
ton, Wierwdle, Zaklad, Bittner, & Wherry, 1989; Sheridan & Simpson,
1979). The challenge for researchers is to identify a suitable operational defi-
nition for the complexity of a task so that a comprehensive characterization
of tasks may be developed. A review of literature (Table 1) indicates many
operational definitions for complexity exist (Jacko, Salvendy, & Koubek,
1995).
From this review of literature it became apparent that Campbell's
(1988) study is the most comprehensive approach to characterizing the com-
plexity of a task. The study's objective was to generate an integrative frame-
work for researchers' varied conceptualizations of complexity. Campbell or-
ganized the various approaches to complexity by indicating that complexity
has been treated in the literature primarily as a psychological experience, an
interaction between task and personal characteristics, and a function of the
objective characteristics of a task. However, there are deficiencies inherent in
psychologically based measures of complexity (Schwab & Cummings, 1976).
Therefore, Campbell's (1988) research focused upon objective characteristics
of tasks and the interaction of those characteristics with attributes of the hu-
man.
The underlying foundation of Campbell's (1988) framework of com-
plexity was the work of Schroder, Driver, and Streufert (1967) who identi-
fied three aspects of a complex task as the number of dimensions of infor-
mation requiring attention, the number of alternatives associated with each
dimension, and the rate of information change. Campbell (1988) indicated
that this definition of task complexity was most suitable for analyzing ob-
jective characteristics of tasks.
The framework that Campbell (1988) developed resulted from catego-
rizing the task characteristics identified from the previous work of other re-
searchers as one of the three aspects of complexity identified by Schroder, et
HIERARCHICAL MENU DESIGN

TABLE 1
TASKCHARACTERISTICS
WHICHh l [ A ~AFFECTTHE COMPLEXITY
O F A TASK

Task Characteristics Reference


Uncertainty, ~npredictabilit~,
unfamiliarity Reid, Schingledecker, & Eggemeier
(1981)
Number of commands necessar use of separate Boyd (1983)
storage areas, homogeneity or'edirs,, number of
strategies employed
Magnitude and variation of srimulation, number of Schwab & Cummings (1976)
sensory modalities affected
Path-goal muluplicity Terborg & Miller (1978)
Path-goal multiplicity, decision verifiability, solution Campbell (1984)
multiplicity
Number of cues, cue intercorrelauons, cue validities, Steinmann (1976)
function forms. cue variabilitv
Number of alternatives available, number of dimen- Payne (1976)
sions of information available
Number of dimensions requiring attention, number Schroder, Driver, & Streufert
of alternatives associated with each dimension, (1967)
rate of information exchange
Multiple aths, multiple outcomes, conflictin inrer-
depenLnce among paths, uncertain or protabilis-
tic linkages
Spatial correspondence between user communication Boggs & Simon (1968)
interface and prompt For response
Number of configurations, number of actions Schvaneveldt, Durso, Goldsmith,
Breen, & Cooke (1985)
Procedure-oriented versus choice-oriented rasks and Zhao (1992)
their compatibility with task presentation
Products, required acts, and information cues Wood (1986)

al. (1967). The following describes the four basic characteristics that were
developed (Campbell, 1988):
1. Multiple Paths-"An increase in the number of possible ways to ar-
rive at desired outcome increases information load, and thus increases com-
plexity" (Campbell, 1988, p. 43).
2. Multiple Outcomes-"As the number of desired outcomes of a task
increases, complexity also increases" (Campbell, 1988, p. 43).
3. Conflicting Interdependence Among Paths-"If achieving one desir-
ed outcome confhcts with achieving another desired outcome, complexity
wdl increase" (Campbell, 1988, p. 44).
4. Uncertain or Probabhstic Linkages-"Uncer~aint~ can increase com-
plexity by enlarging the pool of potential paths to a desired outcome"
(Campbell, 1988, p. 44).
In addition, Campbell (1988) recognized that the complexity of a task
should be examined in the context of the interaction between the character-
istics of the task and of the person performing the task. The presentation of
1190 J. A. JACK0 & G . SALVENDY

the task is clearly a critical component of this interaction. The psychological


relevance of the modes of presentation of information in an interaction be-
tween a human and a computer is widely underestimated. Jorna (1988)
found that the manner in which information is presented plays an integral
role in defining the complexity of a task in an interface between a human
and a computer. An interface between a human and a computer provides
the abhty to change the mode of presentation. For example, MS-DOS com-
mands can be presented in the form of icons and propositions. Different
modes of presentation may affect the complexity of a task (Coury & Pietras,
1989; Jorna, 1988). In dynamic decision-malung environments, graphic and
alphanumeric information interact in a way that renders a display using both
modes of presentation advantageous (Coury & Pietras, 1989).
Design of Hierarchical Menus
Information on a computer screen is commonly presented to the human
in the form of an hierarchical menu. A comprehensive review of literature
pertinent to the design of hierarchical menu systems indicates that the most
critical task characteristics that must be considered in the design of hierar-
chical menu systems are the depth and the breadth of the menu (Jacko,
Salvendy, & Koubek, 1994).
In 1981, depth and breadth were examined formally in a study con-
ducted by Mder. In that study Mdler hypothesized that the structure of a
menu's hierarchy affects the speed and accuracy of target acquisition. It is
known from literature on visual search that positive response times increase
h e a d y with the number of items displayed (Neisser & Beller, 1965); how-
ever, M d e r (1981) found that short-term memory is a limitation of the alter-
native: increased depth of the hierarchy. The objective of Miller's study was
to evaluate the performance of computer users while using four hierarchical
menus with a fixed number of target items. Four structures were examined:
64', 26, 4j, and 82.
The results supported Mder's hypothesis that hierarchical organization
had a profound effect on performance. For systems of moderate size, in-
creased breadth was optimal but not at the expense of display crowding. In
fact, M d e r found that the 82 condition allowed fastest acquisition and few-
est errors. Kiger (1984) extended M~Uer's research by investigating user's
preference for variations of Mder's menu structures. Users' preference was
consistent with performance. Users preferred the 82 structure. Other research-
ers have also shown that menu breadth is preferable to depth (Schultz &
Curran, 1986).
Additional studies following the initial work of Miller (1981) examined
performance in light of users' individual differences (Barnard, Morton, Long,
& Otley, 1977; Hollands & Meukle, 1987; Lee, MacGregor, Lam, & Chao,
HIERARCHICAL MENU DESIGN 1191

1986; Lee, Whalen, McEwen, & Latremou~Ue,1984; Snyder, Happ, Malcus,


Paap, & Lewis, 1985; Vicente, Hayes, & Wihges, 1987), categorical concept
organization (McDonald, Stone, & Liebelt, 1983; Parkinson, Sisson, & Snow-
berry, 1985), error analysis (Snowberry, P a r h s o n , & Sisson, 1985), and de-
velopment of models (Lee & MacGregor, 1985; Paap & Roske-Hofstrand,
1986).
It is apparent from past research that the issue of complexity is an im-
portant consideration. To be regarded as a critical consideration in design it
must be demonstrated that complexity affects performance. Complexity as a
determiner of performance has not been assessed for hierarchical menu re-
trieval. Thus, the first stage of this research must involve examining per-
formance while manipulating relevant characteristics. Frese (1987) indicated
that complexity is determined by the sheer number of decisions that have to
be made and by the relationships among those decisions. In hierarchical
menu retrieval, the essential dimensions of design which have been identi-
fied by previous researchers are the menu's depth and breadth (figer, 1984;
MLler, 1981). Definitions of complexity lrke that of Frese (1987) along with
this study's assertion that complexity may increase as menu depth increases,
indcate that complexity of menu retrieval can be controlled by the manner
in which menus are designed.
The central performance assessment issues are which menu depths yield
the fastest response times and fewest errors and how are these optimal de-
signs perceived by the users in terms of complexity. Specific hypotheses can
be developed to test these performance issues and their implications:
1. Depth is a significant factor influencing time to respond in menu-
based interfaces;
2. Depth is a significant factor influencing accuracy in menu-based in-
terfaces;
3. The interaction of depth and breadth is a significant factor influenc-
ing time to respond in menu-driven interfaces;
4. Depth is a significant factor influencing the users' perception of a
menu's task complexity
The vahdation of Hypotheses 1 through 4 would establish complexity as a
critical design consideration for hierarchical menus. Further, their vahdation
would serve as a tool which identifies the attributes that drive complexity in
retrieval of menus. This study h k e d Campbell's (1988) theory of complexity
to hierarchical menu design by providing an underlying justification for why
the attributes of depth and breadth affect users' perceptions of complexity.
METHOD
Task
The tasks used to test these hypotheses were several menus with varying
1192 J. A. JACK0 & G. SALVENDY

depths and breadths. Hierarchical menus were chosen for the experimental
task for two reasons. First, they had not been examined in the past within
the context of complexity. Second, manipulation of the attributes which
could drive complexity, namely, depth and breadth, can be directly manipu-
lated by the experimenter. The menus were constructed utilizing batch
programs which were written in MS-DOS, Version 4.01. The computer
which ran the MS-DOS and was used for the experiment was a Bondwell
IBM-compatible laptop computer. It had an 80386SX microprocessor and
operated at 25 MHz. The display was a VGA-compatible backht liquid crys-
tal display. The six different menu structures generated on this machine
were:
1. Two choices at each of two levels (2*),
2. Two choices at each of three levels (2'1,
3. Two choices at each of six levels (z6),
4. Eight choices at each of two levels (82),
5 . Eight choices at each of three levels (8'1,
6. Eight choices at each of six levels (86).
The and g2 structures were the same as those Kiger (1984) used. The re-
maining structures were variations of I g e r ' s (1984) structures. These struc-
tures were used because they were slight variations of the menus that had
been used by both Mdler (1981) and Kiger (1984). Thus, they could serve
well for replication purposes.
Each subject was required to uthze one of three menus to satisfy a goal
presented to them by the researcher. Presentation of the goals provoked tar-
get acquisition. An example of a goal used in this experiment is:
You are interested in obtaining secretarial work in the near future. Utilize the menu to obtain
information about this area of interest.

The selections that the subjects made at each level of each menu were
entered into the computer's 81-key keyboard.
Design and Procedure
The subjects were 12 students who had prior experience with menu-
driven interfaces and whose mean age was 25 yr. The group was composed
of seven men and five women. The subjects were recruited through adver-
tisement and paid $5.OO/subject for their time.
The dependent variables were response time and accuracy. The inde-
pendent variables were depth and breadth. The experimental design was a
nested factorial design.
At the beginning of the experimental session, the subject was seated in
front of the computer. The subject was given a brief description of what the
experiment would entail. A practice session was conducted prior to any ex-
HIERARCHICAL MENU DESIGN 1193

perimentation wherein the subject was presented with a goal and then allow-
ed to traverse a menu to satisfy the goal. The menu and goal used during
practice were not used during experimentation. Each subject encountered
each of three menu structures six times during experimentation. A unique
goal was provided to the subject every time a menu was encountered. The
order of presentation of the menus and their associated goals were random-
ized.
Subsequent to the practice session, the subject was presented the first
goal and asked to traverse the menu to satisfy the goal. The subjects tra-
versed from a superorhate menu panel to a menu panel below it in the hi-
erarchy by entering the number of their selection into the computer through
use of the computer's keyboard. The experimenter manually recorded the
time it took the subject to make the response. To achieve high accuracy, the
subjects were required to maintain finger position on a "home" key when
they were not engaged in physically executing a keystroke to make a menu
selection. In this way, the experimenter was able to capture accurately the
time subjects took to process information on the computer screen. The time
involved in physically executing the keystroke was not of interest and, there-
fore, was not recorded. Errors were counted by the experimenter as they oc-
curred. This process continued for each subject until six trials for each of
the three randomly presented menu structures were completed. The first
group of six subjects were presented with the 22, z3, and Z6 menu structures.
The second group of six subjects were presented with the 82, 83, and g6
menu structures.
At the close of each experimentation session, the subject was asked to
complete a questionnaire. The purpose of the questionnaire was to assess the
user's perception of complexity of the three menu structures utilized. The
questionnaire required the subjects to rank each of the three menus on a
scale of 1 (low complexity) to 7 (high complexity). There was a total of six
items on the questionnaire that required the subjects to rank complexity.

Influence of Depth on Response Time


Past literature indicated that users of menu-based interfaces reahze en-
hanced performance when the dimension, depth, is minimized. In fact, er-
rors have been prevalent in the deeper hierarchical menus (Snowberry, et al.,
1985). Time to respond has also been degraded as a result of deeper hierar-
chical menus (Miller, 1981). Both phenomena were observed in this study.
The first hypothesis contends that depth has a significant influence on
response time in menu-based interfaces. A 2 x 3 analysis of variance was
conducted to test the first hypothesis. The depth and breadth factors were
fixed, and the subjects factor was random. The subjects factor was nested
1194 J. A. JACK0 & G. SALVENDY

within the breadth factor. An expected mean squares analysis (Hicks, 1982)
indicated that the breadth term be tested with the subjects term, the sub-
jects term and the interaction between subjects and depth be tested with the
errors term, and the depth term and the interaction between depth and
breadth be tested with the interaction between subjects and depth.
The results corresponding to response time in Table 2 indicate that
depth had a significant effect on time to respond (F2,20=88.04, p<.0001).
Furthermore, a Newman-Keuls range test indicated that response time dif-
fered significantly for depths two, three, and six. It is interesting to note that
the breadth factor also had a significant effect on time to respond (F,,,,=
15.08, p < .003). It is known from the literature on visual search that positive
response times increase linearly with the number of items displayed (Neisser
& Beller, 1965). A Newman-Keuls range test indicated that there was a sig-
nificant difference when comparing the breadth at level two and the breadth
at level eight. Additionally, the analysis of variance for time to respond
shows a significant subjects' effect (F,,,,,, = 2.43, p < .01).
TABLE 2
SUMMARY OF VARIANCES
OF ANALYSES FORRESPONSE
TIME,ACCURACY,
AND PERCENED
COMPLEXITY

Variable df SS F
Response Time
Breadth
Depth
Subjects
Breadth x Depth
Depth x Subjects
Errors
Total
Accuracy
Breadth
Depth
Subjects
Breadth x Depth
Depth x Subjects
Errors
Total
Perceived Complexity
Depth
Subjects
Errors
Total

Influence of Depth on Accuracy


The second hypothesis contends that depth is a significant factor influ-
encing accuracy in menu-driven interfaces. The results corresponding to ac-
HIERARCHICAL MENU DESIGN 1195

curacy in Table 2 are the outcome of a 2 x 3 analysis of variance designed


to test the effect of depth on accuracy. The depth and breadth factors were
fixed, and the subjects factor was random. The subjects factor was nested
within the breadth factor. An expected mean squares analysis indicated that
the terms of the model analysis of variance be tested in the same manner as
utiized for the response time variable. Depth had a significant effect on ac-
curacy (F2,2,=11.33, p< .0005). Furthermore, a Newman-Keuls range test in-
dicated that accuracy differed significantly for depths of two, three, and six.
Practically speaking, the significant effect of depth on both dependent
variables is a reflection of the fact that increased depth involves additional
visual search, decision-making, response selection, and greater uncertainty as
to the location of target items due to the increased number of menu frames
with which the subject must contend.
As part of the analysis of response time and accuracy, a regression anal-
ysis was conducted. This h e a r regression analysis yielded a Pearson product-
moment correlation coefficient of .70 ( p < .0001).
Influence of an Interaction of Depth and Breadth on Response Time
The third hypothesis contends that the interaction between depth and
breadth is a significant influence on time to respond in a menu-driven inter-
face. The results in Table 2 corresponding to response time indicate that the
interaction of depth and breadth had a significant effect on time to respond
(F2,,=3.86, p<.04). A plot of the interaction between breadth and depth is
shown in Fig. 1. The significance of this interaction is an indication that the
increased visual search associated with broad menus and the additional vi-
sual search, decision-making, response selection, and uncertainty associated
with deeper menus interact co increase the users' time to respond.
Two t tests were conducted on the extreme points shown in the plot of
the interaction of depth and breadth. The t tests indicated significant differ-
ences between the mean times to respond for the & and 22 menu structures
(t,,,, = 10.18, p < .OOOl) and significant differences between the mean times to
respond for the 82 and 26 menu structures (t,,,,=6.00, p< .0001). Also, a
simple main effects analysis was conducted to examine further the effect of
depth on each level of breadth. The analysis showed there was a significant
difference in the effect of depth on response time when breadth was at level
two (F,,,,, =54.68, p< ,0001) and when breadth was at level eight ( F,, =
55.10, p < .0001).
Influence of Depth on Users' Perceptions of Complexity
The fourth hypothesis asserts that depth has a significant influence over
the users' perceptions of the complexity of a computer menu. The interrela-
tionship between a menu's depth and the complexity of the menu has not
J. A. JACK0 & G. SALVENDY

0 1 2 3 4 5 6 7 8
Depth

FIG.1. The interactive effect of depth and breadth on mean response time

been explicitly explored in past research. This interrelationship was exam-


ined through the use of a questionnaire which the subjects completed subse-
quent to data collection. A 1-way analysis of variance was used to assess the
influence of depth on users' perceptions of complexity. In this model, the
depth term was fixed, and the subjects term was random. Both terms were
tested with the interaction between depth and subjects. The results in Table
2 correspondmg to perceived complexity indicate that depth had a signifi-
cant effect on the users' perceptions of the complexity of a menu (F,,,,=
53.45, p < .0001). A Newman-Keuls range test indicated that the depths two,
three, and six were all significantly different from each other. A plot of
mean complexity ratings for all six menu structures is shown in Fig. 2.
These results indicate that with the increased depth comes additional visual
search, decision-makmg, selection of responses, and greater uncertainty as to
the location of target items. This translates to an increase in perceived com-
plexity as depth increases.
The questionnaire's internal consistency, using Cronbach coefficient al-
pha (1951), was .67. Although the reliability of the questionnaire is not as
high as would be desirable, the robustness of the relationships tested indi-
cates that the results support Hypothesis 4. It is clear that the subjects con-
sidered the tasks involving the deeper menus to be more complex than both
the tasks involving the menus with moderate depths and the menus with low
depths.
HIERARCHICAL MENU DESIGN

High 7
cornplerityj

Depth

FIG. 2 . The effects of depth and breadth on the perceived complexity of hierarchical
menus

Conclusions
This study demonstrated that as the depth of a computerized, hierarchi-
cal menu increased, users perceived that the complexity of the computerized
task increased also. The theoretical justification of this observed phenome-
non is provided by Campbell's (1988) comprehensive framework for the
complexity of tasks.
Campbell (1988) identified multiple paths, multiple outcomes, conflict-
ing interdependence among paths and uncertain h k a g e s as four characteris-
tics of a complex task. A concurrent examination of Campbell's (1988)
framework and a task involving hierarchical menu retrieval suggests that
these four characteristics are present as depth increases, and their presence
explains the simultaneous increase in complexity of the task as depth in-
creases. In a task involving menu retrieval, a path can be considered the suc-
cessive choices made while traversing a menu. Outcomes can be considered
the number of terminal options at the lowest level of the hierarchy. Linkages
in a task of menu retrieval correspond to the progression from one option to
the option directly above or below it in the hierarchy.
Table 3 shows the quantitative demonstration of Campbell's (1988) qual-
itative characteristics in hierarchical menus. The descriptive statistics show
that both response time and total number of errors increase as depth is in-
creased. As a consequence, users perceive that complexity increases as depth
increases. These quantitative results correspond to Campbell's (1988) quah-
1198 J. A. J A C K 0 & G. SALVENDY

tative characteristics of complex tasks. Also shown in Table 3 are two of


Campbell's characteristics to indicate how both increase in depth increases,
thereby increasing perceived complexity. In summary, as menu depth in-
creases, the four task characteristics (Campbell, 1988) become more sahent,
and this in turn drives users' perceptions of task complexity.

TABLE 3
QUANTITAT~VE
DEMONSTRATION (1988) QUALITATIVE
OF CAMPBELL'S
CHARACTERISTICS
I N HIERARCHICAL
MENUS

Measure Menu Structure


Breadth: 2 8
Depth: 2 3 6 2 3 6
Response Time (sec.)
M 7.15 15.57 40.21 16.48 24.24 64.82
SD 2.34 7.02 19.89 12.33 153 33.44
Tordl Number of Errors
nd 0.03 0.50 0.64 0.22 0.31 0.64
SD 0.16 0.87 0.85 0.53 0.78 0.85
Perceived Complexity* 1.83 3.58 5.50 1.91 4.24 6.89
Number of Paths to Goal: 4 8 64 64 512 262,144
Number of Outcomest 4 8 64 64 512 262.144
'Rated on a scale of 1 to 7; 1 =low complexity, 7 =high complexity.
t A characteristic of complex tasks (Campbell. 1988).

It is asserted that the underlying cognitive process responsible for the


effects of the four characteristics on complexity of a task is short-term mem-
ory load (Jacko, Salvendy, & Koubek, 1995). As the number of paths, the
number of outcomes, che ~nterdependenceamong paths, and the uncertainty
of linkages increase, the mforrnation load on the user increases also. This ad-
ditional information load challenges the users' short-term memory as they
must attend to more items. There are several practical implications of this re-
search. It has been widely recognized that performance degradation accom-
panies increased menu depth. However, prior to this study, complexity had
not been examined for computerized hierarchical tasks involving menu re-
trieval. This study identdiedincreased depth of a menu as an attribute that
significantly affected users' perceptions of the complexity of the task. Fur-
ther, this assertion was consistent with the Campbell (1988) framework for
complexity. Finally, this phenomenon was explained on a cognitive level by
suggesting that the underlying basis for degradation of performance associ-
ated with increased depth and the resulting complexity of the task is a height-
ened challenge to short-term memory.
T o expand generalizabhty of the results, further work should focus on
additional hierarchical structures and should involve a larger subject pool.
The analysis of variance for time to respond showed a significant subjects ef-
HIERARCHICAL MENU DESIGN 1199

fect (Fl,,l,, = 2.43, p < .01) which may indicate that uncontrolled factors hke
practice and learning along with individual factors like personality and moti-
vation could influence the outcomes of such investigations.
Despite the small sample of participants in this study, the results over-
whelmingly indicate that software designers should consider the saliency of
Campbell's (1988) four characteristics of complex tasks in computer inter-
faces. For example, in hierarchical menu systems, mininlizing the depth di-
mension reduces the number of paths to a goal, the number of outcomes,
the interdependence among paths, and the uncertainty of linkages. The con-
sequences of this minimization is a decline in perceived complexity.
REFERENCES
BARNARD, I?, MORTON,I., LONG,J., &OTTLEY,E. (1977) Planning menus for display: some ef-
fects of their structure and content on user performance. In Dicplays for man-machine sys-
tems, Institute of Electrical Engineers, conference publication 150, 1977.
Boccs, D., & SIMON,R. (1968) Differential effect of noise on tasks of varying complexity.
Journal of Applied Psychology, 52, 148-153.
BOYD.S. I? (1983) Assessing the validity of SWAT as a workload measurement instrument. In
Proceedings of the Hziman Factors Society 27th Annual Meeting, Norfolk, Virginia, October
10-14. Santa Monica, CA: The Human Factors Society. Pp. 124-128.
CAMPBELL, D. J. (1984) The effects of goal-contingent payment on the performance of a com-
plex task. Personnel Psychology, 37, 23-40.
CAMPBELL, D. J. (1988) Task complexity: a review and analysis. Academy of Management Re-
view, 13, 40-52.
COURY,B. G., & PIETRAS, C. M. (1989) Alphanumeric and graphic displays for dynamic pro-
cess monitoring and control. Ergonomics, 32, 1373-1389.
CRONBACH. L. J . (1951) Coefficient alpha and the internal structure of tests. Psycbometrika, 16,
297-334.
EGGEMEIER, F. T., MCGHEE,J. Z., &REID,G. B. (1983) The effects of variations in task loading
on subjective workload rating scales. Proceedings of the IEEE National Aerospace and
Electronics Conference CNAECON), Dayton, Ohio, May 17-19. Pp. 1099-1105.
FLEISHMAN, E. A., & QUAINTANCE, M. K. (1984) Taxonomy of human perfoonnance: the descrip-
tion of hzrman tasks. Orlando, FL: Academic Press.
FRESE,hl 11987) A theory of control and complexitv: implications for software design and in-
rcgr.lrlon of computer systems into the work In. M. Frese, E. Ulich, & W . Dzida
(Cdb ), Psychological issrres of human-computer interaction in the workplace. Amsterdam:
Elsev~erScience. Pp. 313-337.
HICKS,C. R. (1982) Fzindamental concepts in the design of experiments. New York: Holt, Rine-
hart & Winston. Pp. 56-59.
HOLIANDS.J., & MERIKLE,I? (1987) Menu organization and user expertise in information
search tasks. Hzdrnan Factors, 29, 577-586.
JACKO, 1.. A., SALVENDY, G., & KOUBEK.R. J . (1994) An experimental study for menu dch~gn
g u ~ d e h e sfor menu selection. Proceedings of the 12th Triennial Congress of the 1ntcmn-
tional Ergonomics Association, 4, 351-353.
JACKO, J . A.. SALVENDY, G., &KOUBEK, R. J. (1995) Modeling of menu design in computerized
work. Interacting with Computers, 7, 304-330.
JORNA,R. (1988) A comparison of presentation and representation: linguistic and pictorial. In
G. C. V. Veer & G. Mulder (Eds.), Hziman-comnpziter interaction: psychonomic aspects.
New York: Springer-Verlag. Pp. 172-185.
KIGER,J. I. (1984) The depthhreadth tradeoff in the design of menu-driven interfaces. Inter-
national Iozirnal of Man-Machitre Studies, 20, 201-213.
LEE, E., &MACGREGOR, J. (1985) Minimizing user search time in menu retrieval systems. Hu-
man Factors, 27, 157-162.
1200 J. A. J A C K 0 & G. SALVENDY

LEE,E., MACGRECOR, J . , LAM, N., &CHAO,G. (1986) Keyword menu retrieval: an effective al-
ternative to menu indexes. Ergonomics, 29, 115-130.
LEE, E., WHALEN,T., MCEWEN.S.. &LATREMOUILLE, S. (1984) Optimizing the design of menu
pages for information retrieval. Ergonomics, 27, 1051-1069.
LOCKE,E. A., SHAW,K. N., SAARI,L. M., &LATHAM,G. P. (1981) Goal setting and task per-
formance: 1969-1980. Psychological Bz,lletin, 90, 125-152.
LYSAUGHT, R., HILL, S., DICK,A,, PIAMONDON, B., LINTON.P., WIERWLLE, W., ZAKLAD,A,,
BITTNER, A.. &WHERRY, R. (1989) 0 erator workload: comprehensive review and evalua-
tion of operator workload methodoyogies. Technical Report 851 of the United States
Army Rcscarch Institute for the Behavioral and Social Sciences, Army Project Number
2Q262785t1790, 7-33.
MCDOIJALI). .
J STONE.J., &LIEBELT, L. (1983) Searchin for items in menus: the effects of or-
ganization and type of target. In Proceedings of t f e Hurnarr Factors Society 27th A n n z ~ l
Meeting. Santa Monica, CA: The Human Factors Society. Pp. 834-837.
MILLER, D. (1981) The depthhreadth tradeoff in hierarchical computer menus. In Proceedirrgr
of the Htrtnan Factors Society 25th Atrrrz~alMeeting. Santa Monica, CA: The Human Fac-
tors Society. Pp. 296-300.
NEISSER,U., & BELLER, H . (1965) Searching through word lists. British Journal of Psychology,
56. 349-358.
PAN, K. R., &ROSKE-HOFSTRAND, R. J. (1986) The optimal number of menu options per panel.
Human Factors, 28, 377-385.
PARKINSON, S., SISSON, N., &SNOWBERRY, K. (1985) Organization of broad computer menu dis-
plays. International Journal of MUP PI-Machine S~zldies,23, 689-697.
PAYNE,J . W. (1976) Task complexity and contingent processing in decision making: an infor-
mation search and protocol analysis. Organizational Behavior and Human Performance, 16,
366-387.
REID, G. B., SHINGLEDECKER, C. A,, &EGCEMEIER, F. T. (1981) A plication of conjoint measure-
ment to workload scale development. In Proceedings of l i e Hzrman Factors Society 2Tth
Annual Meeting. Santa Monica, CA: The Human Factors Society. Pp. 522-526.
SCHRODER, H. M., DRIVER.M. J., &STREUFERT, S. (1967) Hurnan information processing. New
York: Holt, Rinehart, & Winston.
SCHULTZ. E. E., &CURRAN, P. S. (1986) Menu structure and ordering of menu selections: inde-
pendent or interactive effects? SIGCHI Bulletin, 18, 69-71.
SCHVANEVELDT, R. W.. DURSO,F. T., GOLDSMITH, T. E.. BREEN.T. J., & COOKE,N. M. (1985)
Measuring the structure of expertise. International Journal of Man-Machine Siridies, 23,
699-728.
SCWAB. D. I?, &CUMMINGS, L. L. (1976) A theoretical analysis of the impact of task scope on
employee performance. Academy ofMariogetnent Review, 1 , 23-35.
SHAW,M. E. (1976) Gmzrp dynamics: the p~vcholog~ of anall group behavior. New York: Mc-
Graw-Hill. Pp. 309-321.
SHERIDAN, T. B., &SIMPSON, R. W. (1979) Toward the definition and measurement of the men-
tal workload of transport pilots. FTL Report R79-4, Contract DOT-0s-70055, Depart-
ment of Aeronautics and Astronautics, Flight Transportation Laboratory, Cambridge,
MA. P p . 1-72.
SNOWBERRY, K., PARKINSON, S. R., & SISSON,N. (1985) Effects of help fields on navigating
through hierarchical menu structures. International Journal of Man-Machzne Studies, 22,
479-491.
SNYDER, K.. HAPP.A., MALCUS,L., PAAP,K.. &LEWIS,1. (1985) Using co nitive models to cre-
ate menus. In Proceedings of the Hzrrnar~Factors Society 29th ~ n n z r a f ~ e e t i Santa
n ~ . Mon-
ica, CA: The Human Factors Society. Pp. 655-658.
STEINMANN. D. 0. (1976) The effects of cognitive feedback and task complexity in multi-
ple-cue probability learning. Organizational Behavior and Hurnan Performartce, 15, 168-
179.
TERBORG, J. R., &MILLER,H. E. (1978) Motivation, behavior, and performance: a closer exam-
ination of goal setting and monetary incentives. Jozrrnal of Applied Psychology, 63, 29-39.
VICENTE,K., HAYES, B.. & WILLICES, R. (1987) Assaying and isolating individual differences in
searching a hierarchical file system. Hzrmarz Factorr, 29, 349-359.
HIERARCHICAL MENU DESIGN 1201

WOOD,R. E. (1986) Task complexity: definition of the construct. Organizational Behavior and
Human Dension Making, 37, 60-82.
ZHAO,
8.(1992) A s~ructuredanal sis and quantitative measurement of task corn Iexity in hu-
man-computer Inreractlon. d p u b l i s h e d Ph.D. Dissertation, School of ~nzustrial Engi-
neering, Purdue Univer., West Lafayette, IN, December, 1992.

Accepted April 9, 1996.

You might also like