The International Journal of Aviation Psychology

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

This article was downloaded by: [University of Bristol]

On: 25 January 2015, At: 13:48


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK

The International Journal


of Aviation Psychology
Publication details, including instructions
for authors and subscription information:
http://www.tandfonline.com/loi/hiap20

Automation- Induced
"Complacency":
Development of the
Complacency-Potential
Rating Scale
Indramani L. Singh , Robert Molloy & Raja
Parasuraman
Published online: 13 Nov 2009.

To cite this article: Indramani L. Singh , Robert Molloy & Raja Parasuraman
(1993) Automation- Induced "Complacency": Development of the
Complacency-Potential Rating Scale, The International Journal of Aviation
Psychology, 3:2, 111-122, DOI: 10.1207/s15327108ijap0302_2

To link to this article: http://dx.doi.org/10.1207/s15327108ijap0302_2

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of


all the information (the “Content”) contained in the publications
on our platform. However, Taylor & Francis, our agents, and our
licensors make no representations or warranties whatsoever as to the
accuracy, completeness, or suitability for any purpose of the Content.
Any opinions and views expressed in this publication are the opinions
and views of the authors, and are not the views of or endorsed by
Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of
information. Taylor and Francis shall not be liable for any losses,
actions, claims, proceedings, demands, costs, expenses, damages,
and other liabilities whatsoever or howsoever caused arising directly
or indirectly in connection with, in relation to or arising out of the use
of the Content.

This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access
and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [University of Bristol] at 13:48 25 January 2015
THE IMERNATIONAL JOURNAL OF AVIATION PSYCHOUXjY, 3(2), 111-122
Copyright O 1993, Lwrcnce Erlbum Associltes, Inc

Automation-Induced "Complacency":
Development of the
Complacency-Potential Rating Scale
Downloaded by [University of Bristol] at 13:48 25 January 2015

Indrarnani L. Singh, Robert Molloy, and Raja Parasuraman


Cognitive Science Laboratory
The Catholic University of America
Washington,DC

Automation-induced "complacency" has been implicated in accidents involving


commercial aircraft and other high-technology systems. A 20-item scale was devel-
oped for measuring attitudes toward commonly encountered automated devices that
reflect a potential for complacency. Factor analysis of responses (N = 139) to scale
items revealed four factors: Confidence-Related, Reliance-Related, Trust-Related,
and Safety-Related Complacency. It is proposed that these form components of
automation-induced complacency rather than general attitudes toward automation.
The internal consistency (r = 37) and test reliability (r = .90) of the factors and the
scale as a whole were high. Complacency potential is discussed with respect to
interrelations between automation and operator trust in and reliance on automation.

Pilot "complacency" has long been implicated as a possible contributor to aviation


accidents (Hurst & Hurst, 1982). The term has become more prominent with the
advent of automated aircraft, and the possibility of automation-induced compla-
cency has been raised (Billings, buber, Funkhouser, Lyman, & Huff, 1976;
Parasuraman, Molloy, & Singh, 1993; Thackray & Touchstone, 1989; Wickens,
1992; Wiener, 1981). One indication of the importance accorded the concept is the
inclusion of complacency as one of several behavioral coding categories used to
classify aircraft incidents in the coding manual of the Aviation Safety Reporting
System (ASRS; Billings et al., 19'76). which defines complacency as "self-satisfac-
tion which may result in non-vigilance based on an unjustified assumption of
satisfactory system staten @. 23). Wiener (1981) defined complacency as "a psy-

Requestsfor reprintsshouldbe sent to Raja Parasuraman. Cognitive Science Laboratory.TheCatholic


University of America, Washington, M3 20064.
1 12 SINGH, MOLLOY, PARASURAMAN

chological state characterized by a low index of suspicion" @. 117). As these


varying definitions indicate, although aviation researchers agree that complacency
may be a potential problem in automated aircraft, there is considerable uncertainty
regarding the dimensionsof complacency. Nevertheless, the frequency with which
the concept is encountered in analyses of aviation accidents suggests that attempts
should be made to understand and validate the concept.
To begin, one can ask whether complacency is engendered by particular aspects
of the flying environment. Although a number of potential candidates might be
identified, one feature of the modem aviation environment seems pruticularly
salient: automation. Because of its reliability, automation may engender excessive
Downloaded by [University of Bristol] at 13:48 25 January 2015

trust and overreliance in pilots (Wickens, 1992).It is perhaps not a coincidence that
many of the ASRS incident reports mentioning complacency involve automated
aimaft. The following case from the ASRS database provides one example:
The aircraft was at FL370 in Washington center airspace, with the f i t officer (F/O)
flying, using the autopilot. Air-traffic control (ATC)gave aclearanceto cross 20 miles
west of DQO at FL240.At the top of the &scent point, the aircraft began a power-off
descent To comply with the crossing restriction ATC requested an increase in cruise
speed to 320 knots when level at FL240. The captain used the Flight Management
Computer (FMC) cruise page to re-establish a cruise altitude of FL240 and a new
cruise speed of 320 knots. This action eliminated the earlier altitude crossing restric-
tion. Since the F/0 was on the progress page and the captain was on the cruise page,
the loss of the restriction went unnoticed (the legs page displays altitude restriktions).
The aircraft reduced its descent and slowly added power, causing the aircraft to cross
the restriction fix 1,000 feet high. The problem was noticed just prior to the fix, too
late for any action. In filing this report, the captain felt that the crew's confidence in
the aircraft to make crossing restrictions that are programmed into the FMC caused
the crew to become moderately complacent. (Billings et al., 1976)

Pilot attitudes of overconfidence in and overreliance on automation may not be


sufficient in themselves to lead to complacent behavior, but may only indicate a
potential for complacency. Complacent behavior may occur only when compla-
cency potential coexists with other conditions such as (a) pilot inexperience with
equipment; (b) high workload brought about by poor weather, heavy traffic, or
equipment trouble; (c) fatigue due to poor sleep or long flights; and (d) poor
communicationbetween ground and crew or among crew members. The combina-
tion of the crew's attitude toward automation (e.g., overconfidence)and aparticular
situation (e.g., high workload) may lead to complacent behavior.
This approach suggests a distinction between complacency potential, which
may be revealed in attitudes towards automation such as overconfidence, and
complacent behavior. A number of performance measures of the latter are possi-
ble. P i w a s m a n et al. (1993) operationally defined complacent behavior as
inaccuracy and/or delay in detecting a failure in the automated control of a
simulated engine-monitoring task. They found that subjects failed to detect
malfunctions in the automation after about 20 min into the simulation when the
reliability of the automation routine was invariant over time and when subjects
had to perform other manual flight tasks simultaneously (tracking and fuel man-
agement). Behavioral measures based on the performance of other flight-related
tasksare also possible.
The aim of the present study was to assess complacency potential. It should be
possible to measure a person's potential for exhibiting complacency by evaluating
his or her attitudes towards automation. According to this view, complacency
reflects attitudes that arise in certain contexts and that may subsequently (e.g., in
a different context) affect performance adversely. For example, a pilot may skip
certain items on a takeoff checklist without untoward incident for years, hence
fostering an attitude of complacency and a potential for complacent behavior. This
has a consequence for performance only when a particular context arises-for
Downloaded by [University of Bristol] at 13:48 25 January 2015

example, when there is a failure associated with the unchecked item and some of
thepreviously identified situational variables occur (e.g., high workload). Langer's
(1989) concept of premature cognitive commitment is relevant to this view of
complacency as an attitude that arises from initial experience with automation.
Premature cognitive commitment develops when a person f i t encounters a
stimulus, device, or event in a particular context; the attitude is then reinforced
when the stimulus is reencountered in the same way. According to Langer,
when we aocept an impression or piece of information at face value, with no reason
to think critically about it, perhaps because it seems irrelevant, that impression settles
unobtrusively into our minds until a similar signal from the outside world-such as
a sight or sound-calls it up again. At that next time it may no longer be irrelevant,
but most of us don't reconsider what we mindlessly accepted earlier. @. 22)

Langer identified a number of antecedent conditions that produce this attitude,


including routine, repetition, and extremes of workload. Each of these conditions
is possible in the automated cockpit.
In the present study, a scale was developed for measuring attitudes (favorable
or unfavorable) toward different aspects of automation that reflect a potential for
complacency. An altitude can be defined as a personal disposition common to
individuals, but possessed by them to different degrees, which impels them to react
to objects or situationsin favorable or unfavorable ways (Ajzen & Fishbein, 1980;
Fishbein & Ajzen, 1975). The proposed Complacency-Potentid Rating Scale
(CPRS)sought to establish a significant relationship between favorable and unfa-
vorable automation attitudes and various dimensions of complacency concerning
automated devices.

METHOD

The development of the CPRS consisted of three stages. The f i t stage involved
the generation of a set of questions (scale items) to probe attitudes--favorable or
unfavorable-toward various aspects of automation. The interitem correlation
matrix was then analyzed using factor analysis to reveal the underlying statistical
1 14 SINGH, M0UX)Y. PARASURAMAN

factor structure of the items. The reliability estimates of the extracted factors were
also computed. The second stage was an initial validation study that investigated
the network of relationships of the complacency scale developed in the first part
of the study with selected preexisting variables such as age and sex. Finally, the
test-retest reliability of the instrument was evaluated by administering the scale to
the initial sample for a second time.

Subjects
A total of 139 undergraduate students (50 men, 89 women) from the Catholic
University of America volunteered to participate in the study. The students majored
Downloaded by [University of Bristol] at 13:48 25 January 2015

in a variety of subjects within the arts and sciences. Subjects ranged from 17 to 38
years of age, with most subjects being close to the mean age of 20 y m . Subjects
received credits towards a course requirement for their participation.

Measures
Initially, 100 general statements about the benefits and potential costs of
automation-items related specifically to complacency potential-were devel-
oped. Items were generated by examination of ASRS reports, perusal of com-
puter-attitude scales (e.g., Igbaria & Parasuman, 1991), and discussion with
engineers and psychologists cognizant of the problem of complacency. These
statements were examined by four readers knowledgeable about automation-
and complacency-related ASRS reports. The readers selected 20 statements on
the basis of face validity in assessing several dimensions of automation-related
complacency. The 20 items included 4 "bogus" or "filler" items, which were
used as a check on the readers' consistency in rating items conscientiously. The
16 complacency-related items included parallel positive and negative state-
ments on various aspects of attitudes toward automation (i.e., rating order was
counterbalanced). A pilot-test of the 20-item scale was administered to 10
volunteer subjects, who were asked to indicate their agreement or disagreement
with each of the statements. A 5-point Likert-type scale was provided, with
response anchors ranging from strongly disagree (1) to strongly agree (5). Thus,
with the 16 test items, the range of possible scores on the CPRS (after allowing
for the counterbalanced rating order) varied from 16 (low complacency poten-
tial) to 80 (high complacency potential). The final version of the 20-statement
questionnaire was administered to the 139 subjects in several small groups. The
respondents also provided demographic and other personal information, includ-
ing their education and their experience in using microcomputers.

RESULTS

Scoring and Item Analysis


All favorable statements were scored on a scale ranging from strongly disagree (1)
to strongly agree (5). Unfavorable statements were scored on a scale ranging from
COMPLACENCY-POTENTIALRATING SCALE 1 15

strongly agree (1) to strongly disagree (5). The total score for each respondent was
obtained by adding scores of all items except for the bogus or filler items. The mean
CPRS score was 57.69 (N = 139,SD = 6.09). and scores ranged from 40 to 75. Item
analysis was carried out to ascertain the best statements for use in a scale measuring
complacency potential. In general, item analysis provides an index of difficulty as
well as an index of validity (Guilford, 1954). The index of validity refers to how
well the item measures or discriminatesthe construct in agreement with the rest of
the test items, or to how well it predicts some external criterion. Product-moment
correlation coefficients were computed between the score of each item of the
questionnaireand the total score. Results indicated that all statements were highly
intexcorrelated(r > .98), indicating high internal consistency among the items. The
Downloaded by [University of Bristol] at 13:48 25 January 2015

mults of the item discrimination analysis also showed extremely low (r < .22)
correlations, thus favoring the retention of all statements.

Factor Structure
The interitem correlation matrix was computed for the 20 scale items. Factor
analysis using the SPSS-XO (Statistical Package for the Social Sciences) proce-
dure (SPSS, Inc., 1983) was performed on the correlation matrix. The principal-
components method with varimax rotation was used to extract the underlying
factors. The complacency-potential factors are listed in Table 1. Examination of
the results in conjunction with the scree test (Catell, 1966) revealed the presence
of five factors with eigenvalues greater than 1. These five factors accounted for
over 53% of the explained variance (see Table 2). The factor loadings and their
variances are presented as per their order of importance, determined by the magnitude
of their eigenvalues. The criterion used was that an item should have a factor loading
of .3 or greater to be considered to contribute significantly to that factor.
Factor 1, which accounted for 19.1%of the variance, had 4 items with loadings
greater than the criterion of .3. These items appeared to be related to general
attitudes concerning automation.For example,Item 3 stated that "People save time
by using automatic machines (ATMs) rather than a bank teller for banking trans-
actions," and Item 12 stated that "I often use automated devices." Factor 1 was
accordingly labeled General Automation.
Factor 2 consisted of 4 items that reflected various aspects of confidence
regarding automation, such as "If I need to have a tumor in my body removed, I
would choose to undergo computer-aided surgery using laser technology because
it is more reliable and safer than manual surgery." High scores on this factor were
interpreted as reflecting overconfidence in automation, and therefore indicative
of a potential for complacency. Factor 2 was accordingly named Confidence-Re-
luted Complacency.
Factor 3 consisted of 3 items reflecting an individual's perception of the
reliability of automation and consequent reliance on automation, such as Item 10,
which stated that "ATMs provide a safeguard against the inappropriate use of an
individual's bank account by dishonest people." This factor was labeled Reliance-
Related Complacency.
TABLE 1
Complacency-Potential Factors and Associated Items

Item
Factor Factor Item Loading
Confidence
1. I think that automated devices used in medicine, such as CT
scans and ultrasound, provide very reliable medical
diagnosis.
2. Automated devices in medicine save time and money in the
diagnosis and treatment of disease.
3. If I need to have a tumor in my body removed, I would
choose to undergo computer-aided surgery using laser
technology because it is more reliable and safer than
Downloaded by [University of Bristol] at 13:48 25 January 2015

manual surgery.
4. Automated systems used in modern aircraft, such as the
automatic landing system, have made air journeys safer.
Reliance
1. ATMs provide a safeguard against the inappropriate use of
an individual's bank account by dishonest people.
2. Automated devices used in aviation and banking have made
work easier for both employees and customers.
3. Even though the automatic cruise control in my car is set at
a speed below the speed limit, I worry when I pass a police
radar speed trap in case the automatic control is not
working properly.
Trust
1. Manually sorting through card catalogues is more reliable
than computer-aided searches for finding items in a library.
2. I would rather purchase an item using a computer than have
to deal with a sales representative on the phone because my
order is more likely to be correct using the computer.
3. Bank transactions have become safer with the introduction
of computer technology for the transfer of funds.
Safety
1. I feel safer depositing my money at an ATM than with a
human teller.
2. I have to tape an important TV program for a class
assignment. To ensure that the correct program is recorded,
I would use the automatic programming facility on my VCR
rather than manual taping.
- -

TABLE 2
Factor Structure and Internal Consistency of the
Complacency-Potential Rating Scale
Factor
Criteria General Confidence Reliance Trust Safety
Eigenvalue 3.05 1.76 1.44 1.21 1.05
Cumulative percent
of variance 19.1 30.1 39.1 46.7 53.2
Internal consistency
(alpha) .97 .82 .85 .89 .95
Three items achieved loadings higher than .3 for Factor 4, reflecting the
individual's trust in automation. For example, Item 17 stated that "I would rather
purchase an item using a computer than have to deal with a sales representativeon
the phone because my order is more likely to be c m t using the computer." Factor
4 was labeled Trust-Related Complacency.
l b o items achieved loadings on the fifth and final factor. These items appeared
to reflect the individual's perception of the safety afforded by automation. For
example, Item 6 stated that "I feel safer depositing my money at an ATM than with
a human teller." Factor 5 was labeled Safety-Related Complacency.
The development of this scale involved testing the internal consistency reliabil-
ity of the items used to interpret each of the five factors. Coefficient alpha
Downloaded by [University of Bristol] at 13:48 25 January 2015

(Cmnbach, 1951) is the basic formula for determining reliability based on internal
consistency, and is the most widely used index of internal consistency reliability
(Cmnbach & Meehl, 1981; Kerlinger, 1986; Nunnally, 1978). Cronbach's alpha
has been found to be a lower bound to the true reliability (i.e., alpha is aconservative
estimate of the reliability of a scale; Armor, 1974; Carmines & Zeller, 1979). The
alpha coefficients for the five factors ranged from .82 to .97. The high alphas
obtained for the five factors provided further confiidion of the homogeneity of
the items comprising them, and indicate an acceptable level of reliability. The
overall reliability for the questionnaire as a whole was .90.

Correlations With Selected Antecedent Variables

The second part of the study attempted to investigate one aspect of the construct
validity of the CPRS by examining the relationships among individual subject
variables with scores on the CPRS. The pattern of product-moment correlations
revealed that age and education were negatively correlated, whereas computer
experience, computer use, and mouse use were positively correlated with compla-
cency potential, although the correlation coefficients were genenlly not signifi-
cant. The negative relations with complacency of demographic variables such as
age and individual variables such as computer experience and computer use are
similar to those reported in studies of computer anxiety and attitudes towards
microcomputer use (Igbaria & Parasman, 1991). The low correlationsbetween
these antecedent variables and complacency scores are perhaps due to the low
variability on the individual variables within the sample studied.

Test-Retest Reliability

Test-retest reliability (or stability) coefficients were computed for the CPRS. The
coefficients were based on the scores of 44 (10 men, 34 women) of the original
sample of 139 subjects who returned mail-in copies of the questionnaire 3 months
after the initial administration. The reliability coefficients for internal consistency
(a = .90)and for stability (a = .87) were high and statistically significant.
118 SINGH, MOLLOY, PARASURAMAN

Discriminant Validity

Discriminant validity was preliminarily assessed by computing the correlation


coefficient between scores of the CPRS and a scale measuring attitudes toward
microcomputer use for decision making and planning activities (Igbaria &
Parasuraman, 1991).The obtained value was near zero (r = .05) and nonsignificant,
showing very low association between the two measures. A further assessment was
made by comparing the five within-factor interitem correlations with the between-
factors interitem correlations. The majority of the correlationsobtained were higher
within groups than between groups, indicating good discriminant validity.
Downloaded by [University of Bristol] at 13:48 25 January 2015

DISCUSSION

It was uncertain at the beginning of this research whether a construct such as


"complacencyn toward computer automation was psychometrically measurable.
The results obtained in this fust study suggest that the potential for complacency
can be evaluated using attitude ratings of items related to everyday automation
technology. Psychometricians suggest that in the early stages of research on a
hypothesized constructmeasure, reliabilitiesof .7 or higher are adequate(Nu~ally,
1978), whereas for widely used scales, reliabilities should not be less than .8
(Carmines & Zeller, 1979). By these criteria, the 20-item CPRS developed in this
study has satisfactoryreliability typical of that of more highly developed and tested
scales. Considering the relatively small sample size and restricted population
tested, the values obtained for internal consistency and test-retest reliability of the
CPRS are encouraging.
Factor analysis of the CPRS revealed five independent factors, one of which
represented general attitudes toward automation. The other four factors were
related to what we propose are components of automation-induced compla-
cency-namely, confidence-related, reliance-related, trust-related, and safety-
related complacency. The internal-consistency reliabilities of these four
components of complacency potential were found to be higher than the pre-
viously mentioned criteria for reliability. The internal-consistency and test-re-
test reliability coefficients of each of the factors and the scale as a whole were
found to be high. Together, these results satisfy the reliability criteria for a
psychometrically sound inventory.
Although the CPRS was found to be reliable and internally consistent,additional
work is necessary to c o n f i i its construct and discriminant validity. Evidence of
criterion-relatedor predictive validity (Schwab, 1980) must await further research,
because behavioral or performance measures of complacency (e.g., inefficiencyin
detecting automation failures), although currently under investigation (Pamura-
man et al., 1993), are not yet fully developed. Analysis of correlations with the
computer-attitude questionnairedeveloped by Igbaria and Parasuraman (199 1) and
comparison of the within-factor and between-factors interitem correlations suggest
good discriminant validity.
COMPLACENCY-POTENTIALRATING SCALE 119

The obtainedpattern of relationships of the CPRS, particularly the internalcon-


sistency reliability coefficients and test-retest reliability coefficients substan-
tially fulfill the requirements of a psychometrically sound scale. However,
because the majority of subjects were young undergraduates with little
computer or work experience, the results of this study may not generalize to
the population at large or, more specifically, to pilots or other users of
automated systems. Thus, further research should be conducted using sam-
ples of employed adults in different occupations or organizations to deter-
mine the generalizability of the CPRS for use in a variety of settings, as well
as its utility and predictive efficacy in explaining task performance with
highly automated systems or in optimizing equipment design. Although the
Downloaded by [University of Bristol] at 13:48 25 January 2015

results of this study provide evidence of the content validity and certain
aspects of the construct validity of the instrument, the convergent and dis-
criminant validity of the CPRS need to be evaluated through the use of the
multitrait-multimethod approach (Campbell & Fiske, 1959). Additional re-
search is also needed to investigate the relations of automation-related com-
placency with a broader set of individual variables, including contextual,
task-related, and key-outcome variables.
The present results are pertinent to recent conceptualizations of the psycho-
logical impact of automation. Riley (1989) proposed that three cognitive
components of the human operator-reliance, trust and confidence-act as
mediating variables in determining the interaction of human operators with
automation. Riley proposed that the trust that pilots have in an automation
system may determine their choice to rely on or ovemde the automation. In
addition, their level of trust will depend on their evaluation of the abilities of
the automation to do what it is supposed to do. According to Riley's model, if
confidence is higher than trust, then pilots will shift their preference from
automation to manual control, especially in risky situations. Regardless of the
level of trust, however, a pilot may tend to rely on automation under high
workload conditions, and this may lead to complacency. Taylor (1990) also
emphasized the importance of trust in highly automated situations, particularly
when functions and tasks are distributed among the crew. He interviewed 50
operational aircrews using 12 tactical decision-making scenarios. Six of the
scenarios were related to navigator decision making, and 6 were related to pilot
decision making. Of the 6 scenarios in each group, 3 were described as
conditions engendering high trust, and 3 were described as engendering low
trust. Principal-coordinates analysis of the pilot responses indicated that demand
for trust was related to the perception of risk, whereas supply of trust was related
to the level of judgement/awareness and uncertaintyldoubt. Demand for trust
generally exceeded supply. However, Taylor did not examine the relationship
between levels of trust and automation, and this would seem a worthwhile topic
for further investigation.
There has been little empirical research on the mediating influence of psycho-
logicalconstructs on human interaction with automation. For example, trust, which
appears to be an important concept on the basis of this study and others (Moray &
120 SINGH. MOLLOY. PARASURAMAN

Lee, 1990; Riley, 1989; Taylor, 1990). has largely been unexplored in the
psychological or ergonomics literature. Trust has behavioral consequences
(Taylor, 1990)and is probably a multidimensional construct with both cognitive
and emotive components (Riley, 1989). Hence trust can be influenced by
individual differences (Scott, 1980). and it may also be developed under
conditions of high workload (Hake & Schmid, 1981). Muir (1988) suggests that
a person's trust or distrust of a machine will affect his or her allocation of
functions in the system, and this in turn may affect the stability of his or her
trust or distrust. Furthermore, when trust is violated, it may not be regained
easily (Deutsch, 1958; Muir, 1988). Nevertheless, pilots may rely on the
Downloaded by [University of Bristol] at 13:48 25 January 2015

automation more when subjected to high workload demands even if their trust
in an electronic crewmember or other automated system is low. Further research
needs to be done to investigate the impact of variations in trust and related
psychological constructs on performance with automated systems under differ-
ing levels of workload.
The importance of variables such as reliance and trust in the development of
automation-related complacency is borne out by the results of a recent large-
sample survey of British pilots and their attitudes toward cockpit automation
(James, McClumpha, Green, Wilson, & Belyavin, 1991). The study revealed
that many in the sample believed that pilots of automated aircraft rely too
heavily on automation. Techniques to combat pilot overreliance on automation,
particularly under high workload conditions, have also been a concern in the
Airbus 310 and 320, which are the most highly automated commercial aircraft
currently flying (Speyer, Monteil, Blomberg, & Fouillot, 1990).
In conclusion, these results suggest that (a) attitudes specific to complacency
may be revealed by the CPRS, (b) the potential for complacency can be
distinguished from general attitudes towards automation, and (c) components
of complacency related to such dimensions as trust and confidence can be
identified. Given the preliminary nature of this study, pilots were not tested,
and subjects were queried only about commonly encountered automated de-
vices. To examine the generality of automation-induced complacency, further
studies are required in which other subject populations are questioned about a
wider range of automated systems. Additional investigation is also required on
how complacency potential manifests in performance with automated devices
(e.g., Parasuraman et al., 1993).

ACKNOWLEDGMENTS

This research was supported by Research Grant NAG-1-1296 from the NASA
Langley Research Center, Hampton, VA, to the Catholic University of America.
Alan Pope was the technical monitor.
The views presented in this article are those of the authors and do not necessarily
represent the views of NASA.
COMPLACENCY-POTENTIALRATING SCALE 121

REFERENCES

Aj- I., & Fishbein. M. (1980). Understading attitudes andpredicting social behavior. Englewood
Cliffs, NJ: hatice-Hall.
Armor, D. (1974). Theta reliability and factor scaling. In H. L Castner (Ed.), Sociological melhadolo~y
(pp. 17-50). SM Francisco: JOISCY-Bass.
Billings, C. E., Later. J. K.Funkbourer. H.. Lyman, G., & Huff, E. M. (1976). NASA aviation s&ty
reporting system Mcma No. TM-X-3445). Moffeu Field, CA: NASA Ames Research Center.
Campbell. D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait
multimethod matrix. Psychdogical Bulletin, 56.81-105.
Cumines, E. G., & Zeller, R. A. (1979). Reliability and validity assessment. Beverly Hills. CA: Sage.
Cattell. R. B. (1966). ?he scree test for the number of factors. Multivariate Behavioral Research, 1,
Downloaded by [University of Bristol] at 13:48 25 January 2015

245-276.
Cronbach. W. (1951). Coefficient alpha and the internal structure of a test. Psychometrika,16,297-334.
Cmbach. W., k Meehl. P. E. (1981). Construct validity in psychological testing. In R. 0. Mason &
B. E. Swanson (Eds.), Measurement for management decision (pp. 335-359). Cambridge, MA:
Addison- Wesley.
Deutsch, M. (1958). Trust and suspicion. Journal of CorJlicrResolution,2.265-279.
Fishbein, M.. & Ajzen. 1. (1975). Belief, atiitude, intention and behavior:An introduction to theory and
nseamh. Cambridge. MA: Addison-Wesley.
Guilford. J. P. (1954). Psychometric methodp. New York: McGraw-Hill.
Hake, D., 8 Schmid, T. (1981). Acquisition and maintenance of trusting behavior. Journal 4 the
Experimental Analysis of Behavior, 35, 109-124.
Hunt. K.. k Hurst, L (1982). Pilot error The humanfactors. New Yoric: Aronson.
Igbaria, M., & Parasuraman. S. (1991). Attitudes toward microcomputers: Development and construct
validation of a measure. /nternarionalJournalof Madachine StudiPs, 35.353-373.
James, M.. McClumpha. A.. Green. R., Wilson. P.. & Belyavin, A. (1991, March 14). Pilot attitudes to
flight deck automution. Paper presented at the meeting of the Royal Aeronautical Society, London.
Kerlinger, F. N. (1986). Foundations of behavioral research. New York: Holt, Rinehart & Winston.
Langer. E. (1989). Mim#ulness. Reading, MA: Addison-Wesley.
Moray, N.. & Lee,J. (1990). nurfandallocation offunction in the control of automaticsystems (EPRL
Tech. Rep. N a 05). Uha-Champaign: University d Illinois. Department of Mechanical and
Industrial Engineering.
Muir. B. M. (1988). Trust @ween humans and machines. and the design of decision aids. In E.
Hollnagel G. Mancini, & D. D. Woods (Eds.). Cognitive engineering in complex dynamic worlds
@p. 71-83). London: Academic.
Nunnally, J. C. (1978). Psychometric theory. New York: McGraw-Hill.
Parasuraman, R.. Mollcry, R., & Sigh. I. L (1993). Performance consequences of automation-induced
"complacency." The /nternational Journal of Aviation Psychology, 3, 1-23.
Riley, V. (1989). A general model of mixed-initiative human-nlachine systen~s.In Proceedings ofthe 33nl
Annual Meeting offhe Human Factors Society @p. 124-128). Denver. CO: Human Factors Society.
Schwab. D.P. (1980). Construct validity in organizational behavior. In B. M. Staw & L L Cummings
(Eds.). Research in organizational behavior @p. 3-43). Greenwich, CT:JAI.
Sco& C. (1980). Interpersonal trust: A comparison of attitudinal and situational factors. Human
Relations. 33.805-812.
Spcyer. J. J.. Monteil. C., Blomberg. R. D., & Fouillot, J. P. (1990). Impact of new technology on
operational interface: From design aims to flight evaluation and measurement. In Aircrajl trajecto-
ries: Computation-prediction-control. Volume I (AGARDograph No. 301). Neuilly sur Seine.
Fiance: Nonth Atlantic T n t y Organization.
SPSS, Inc. (1983). SPS-X urerf grridp. New York McGraw Hill.
Taylor. R. M. (1990). Trust and awareness in human-electronic crew teamwork. In Proceedings ofthe
122 SINGH, MOLLOY. PARASURAMAN

CorJcnncr on the Human-Electronic Crew: Is the Team Motwing? @p. 99-103). Ingolstadt,
Germany.
'hackmy, R. ,.I & Touchstone. R. M. (1989). Detection efficiency on an air MICcontrol monitoring
task with and without compter aiding. Aviation. Space, and Environmental Medicine, 60.7444'48.
Wickens. C. D. (1992). Engineeringpsychology and human performance. New Yo*: HarperCoIlins.
Wienu, E L (1981). Gnnplacarcy: Is the term useful for air safety? In Proceedings cf the 26th
Corpomtc Aviation Safety Seminor @p. 116-125). Denver, CO: Flight Safety Foundation, Inc.
Downloaded by [University of Bristol] at 13:48 25 January 2015

You might also like