DOT/FAA"AM-93/4 Contribution of Personality

to the Prediction of Successin Initial Air Traffic Control

Offce of Aviation MedicineWashington, D.C. 20591 Specialist Training

AD-A264 699

David J. Schroeder

Dana BroachWillie C. Young

Civil Aeromcdical InstituteFederal Aviation Administration

Oklahoma City, Oklahoma 73125

April 1993

Final Report 2 99-L

This document is available to the public

LID through the National Fechnical Information

S N Service, Springfield, Virginia 22161.

U.S. Departmentof Transportation

Federal AviationAdministration

NOTICE

This document is disseminated under the sponsorship ofthe U.S. Department of Transportation in the interest of

information exchange. The United States Governmentassumes no liability for the contents or use thereof.

Technical Report Documentation Page

IRprNo2 Government Accession No 13 Recipients Cataiog No

4 Tile ad Sutitl Repon Date

Contributions of personality measures to predicting success of April 1 993

6 Perlorinig Qrqarizarioo Code

7 Author(s) 8 Perlotmong Organ,7alion Reponl No

David J. Schroeder, Ph.D.;Dana Broach, Ph.D.; andWillie C. Youne, M.Ed. _______________

9 Performing Organization Name and Address 10 Works Unit No (TRAISIj

Federal Aviation AdministrationP.O. Box 25082Oklahoma City, OK 73025 11 Contract or Grant No

12 5tr-Snr~nqnn.Agn-ncy narrne ai~d Aacirezis 13 Type of Report ant Period Covered

Office of Aviation MedicineFederal Aviation Administration800 Independence Avenue, S.W.Washington, DC 2059114SosrnAgnyCe

15 Supptementat Notes

16 Abstract

Review,, have cons~stendv concluded that the validity of personality as a predictor of job performance is low (Besco. 199 1: Reilly & Cbhao. 1992:Trenopyr & Oieltjen, 1982). However. Barrick and Mount's (19J91) meta-analysis of studies of personality and job periormance demonsrar

the utility of the "Big Five" model of personality (Digman. 1990J) in personnel selection and training. 1I his stud ' was designed to evaluate theutility of personality in predicting student success in the FAA's Air Trafflic Control Specialist (ATCS) Non radar Screen Program ("the

Screen"). The Screen follows the miniature training, tresting. and evaluation paradigm (Siegel. 1983), in which individuals with no prior

knowledge of the occupation are taught cri~ical aspects of the job and then assessed on a pass/fail basis for their potential to succeed Ascontrollers. The NEO0 Personality Inventory (NEO-PI'; Costa & McCrae. 1985) was administered to 723 men and 30- women at entry into

the 9-week Screen. NEO-PI scale scores arid cognirive iptitude measures were used to predict final composite scores (CCOMP) of students

Men and women air traffic students exhibited lower average scores in Neurotiijm-ir higher averige scores in E-xtro'ersiiOP, ()pen'te~fIr

Experience. and Conscientiousness, and no difference on AgreeableneY5 when compared to normative sample%. Correlations between the

persosnality scales and ClOMP were low for both sexes, ranging from .000 with Imepulsitveness a facet of Neuroficiirm to -. 149 with Feciternent-

se eking, a facet of the Extrarversion dimension. Despite the low zero-order correlations, several of the personality facerts proved useful in a

regression equation, explaining an additional 3% of variance in performance over that explained ye aptitu .de measures. Theseincluded Exciteeeene-seeking, Fantaiy. Atir'iry. and Ideas. While these results were not entirely' consistent with lBarrick and Mount (1991). they

do offet some support for the role of personality variables in th7- pr-cdlctioi, of aic~s~ the ATCS Screen. Continued research ts needed toassess the relationship of theoretically-based measures of personality to success on the job over time.

17 Key Words 18 Distribution Statement

Neuroticism, Extraversion, Openess, Agreeableness, Document is available to the public through theConcientiousness, Personality, Air Traffic Control National Technical Information Service,Specialist, Screen, Domain, Facet Springfield, Virginia 22161.

19 Security Classif (of this reportt) 20 Rrcuotv Claýsi - i.,., 21 No of Pages 22 Pnce

[Unclassified Unclassified 32Form DOT F 1700.7 (8-72) Reproduction of complteted page authorized

ACKNOWLEDGMENTS

The authors gratefully acknowledge the support for data collection provided by the FAAAcademy during the course of this research. The opinions expressed in this report are thoseof the authors and do not necessarily represent official Federal Aviation Administration policy.

A version of this paper was previously presented at the 34th Annual Meeting of theAerospact Medical Association, Miami, FL

iii

CONTRIBUTION OF PERSONALITY 1O TIHE PREDICTION 1 SUCCE-SS IN

INITIAL AIR TRAFFIC CONTROL SPECIALIST TRAINING

The general public and the news media often ascribe controllers were highly similar to those noted by Karsonpersonality characteristics to individuals in certain occu- and (O)'ell. About 15 years latcer, Shrocder and Dollar

pations. For example, the air traffic control specialist (1989) found evidcnce of the same general 16I11 profile(ATCS) occupation is sometimes described in terms of for controllers as Karson and O)'Dell and Roe, et at.

personality traits such as "stress tolerant" and "attentive However, the 1984 applicant group (N = 3,468) also

to detail." This ascription is sometimes extended to reported less insecurity (0), less tension (Q-;), ad greatersuggest that individu.ls with certain personality charac- self-assertiveness (l-. self-discipline (Q3), and emotional

teristics are attracted to specific occupationfi. Soni. re- stability (C) than the 1974 controller group. Similarlysearch findings, such as that of Kassera and Russo (19871, Delonev and Schroeder (19884) found differences be-provide scientific support to this lay notion of conver- tween entry level controllers (,%'= 4,244) enrolled in the

gence between personality and occupation. Personality FAA Academv and individuals in college or other oCcu-is also cited as an explanation for occupational perfor- pational settings, using the lfYers-Rrigy• Type lndicator

mance. For example, in a review of cognitive and (.ABTl; Myers, 1962). There were higher percentages ofnoncognitive factors associated with ATCS performance, controllers classified as Sensing- T(inking-Judginr ( 15 7.Colman (1970, p. 47) noted that the "importance of & E.S17 than among other groups.

personality, interest, and motivation in successful per- On the other hand, Rose, Jenkins. and liurst (19-81formancc of air traffic control work is recognized not found little difference between controller scores (,Vonly by the mental health staff of the agency, but by 391) and population norms on the (a/ifornia Perona/irvworking controllers, supervisors, personnel and general Inventorn( C11[ Gough, 1960). In that study., controllersmedical specialists as well." This view that personality is were lower on 5ocialization and Responibilir, scales thanrelated to occupational choice and performance suggests the normative group. and low average on S.f-tcontrol. but

two basic research questions: (a) do persons who enter were still within normal limits. Air traffic controllers intheATCS occupation differ from the general public with the field, as well as in the Academy classroom, have alsorespect to general personality characteristics; and (b) been consistently reported as having lower levels of traitwhat personality characteristics, if any, predict who is anxiety than individuals in other occupational groupslikely to become a successful controller? (Coollins, Szhroeder, & Nve. 199 1; Nye& Collin,. 1991:

Smith, 1985). In summary, there is some evidence that

PREVIOUS RESEARCH individuals attracted to the ATCS occupation differfrom other, general population groups along at least

Differences Between Controllers and General some personality dimensions.PopulationPrevious research is mixed on the question of personality Personality and Performancedifferences between controllers and the general popula- Studies finding no relationship. Several studies on thetion. On one hand, Karson and O'Dell (1974) compared relationship of personality toATCS training and on-the-controllers (N = 11,074) and the general population job performance are consistent with influential reviewsusing Cattell's 16PF. While most of the comparisons of the selection literature in aviation and other occupa-

were statistically significant, the differences of practical tions which concluded that the validity of personality as ,significance indicated coat controllers were, on the aver- a predictor of iob performance is low, at best (Besco.

age, brihtcer (B). more conforming to rbc g (G, 199 i ; I-ogi ,V ( ýibb, '2S9; (,mon &I, Tottier. 196"1;

tough minded (4, practical (M), self-disciplined (Q3), Guion & Gibson, 1988; Reilly & Chao, 1982; Tenopyrand less suspicious (W4 than the normative 16PF sample. & )ltjean, 1982). For example, (Colman (1070) de-

Rose, Jenkins, and Hurst (1978) reported that results scribed a study conducted by Karson and O'Dell (1969)•de%

from administration of the 16PF to their sample of 388 that found little evidence ofany significant relationship or

Mat ,/J J . . I U1

S.......A M, o 211

between scores on the 16PF and job performance ratings were consistent, meaningful relationships between par.

in the controller LmploveeApp raisal Record. Research by ticular personality constructs and lob perfoirman Ce Inca-Schroeder (1984) using Barrett's lmpulsiven- aleand sures across tests and acrossor w.ithin oL.cuparions ,MountManning (1984) using Zuckerman's (1979) Sensation & Barrick, 1991). Recent meta-analv,'- of a large num-Seeking Scale found little evidence for relations between ber of personality-oriented validation studies by Barrik

those personality constructs and ATCS performance. and Mount (1991) and Tett. Ja.kson, and RothsteinStudies finding some relationships. On the other (1991 ) demonstrated the utility of the 'Big Five" model

hand, there are also several studies ot controllers suggest- of personality in the prediction of various Job pcrfor-ing that there is a relationship between personality and mance criteria acrosý many occupations. This model ofATCS performance. For example, Colman also cites personality suggests that there are five major factors ofwork by Buckley, O'Connor, and Beebe (1969) in which personality as described by factor and ,ructural analysecs

statistically significant relationships between controller of the domain of trait labels people use to describe

performance on an air traffic control simulation and themselves and others (Digman. 1990: Hofstee. 1984:16PF scores were reported for a very small sample of36 Hogan. 1983: John, Angleitner. & Ostendorf. 1988:controllers. More recently, Collins, Nye, and Schroeder Norman, 1963: Topes & Chri.,tal, 1961) 'Tett. et al.,(1991) found that although entry level controllers re- obtained a corrected estimate of the overall relationported significantly lower levels of anxiety on the State- between "Big Five" personality dimensions and job

Trait Personality Inventorv (STPJ; Speilberger, 1979). performance measures of.24, indicating that personalityself-reported anxiety was still related to success in initial may have utility in the prediction of job performance.ATCS training. Individuals reporting higher levels of This lineof research on the validity and utility ofthe "Big

both state and trait anxiety experienced higher failure Five" model of personality is extended to the ATCSrates than those reporting lower levels of anxiety. In a occupation in this study. Our specific purpoe'. ,,Cre (a)study from the United Kingdom, Nyfield, Kandola, and to investigate the differences, if any, between entrY-levelSaville (1983) obtained 58 significant correlations, rang- controllers and normative samples. and (b to assess theing from .16 to .31 (N= 112), between 32 Occupational incremental validity of a "Big Five" measure of person-

Personalit, Questionnairescores (OPQ: Nyfield. Kandola, ality over cognitive aptitude in the prediction of perfor-& Saville, 1983) and 22 measures of ATC'; job perfor- mance in initial ATCS training at the FAA Academy.malice. Several of the correlations were re!ated to techni-

cal proficiency (e.g., abilityto form a mental picture from METHODflight progress strips only). But a majority of the corre-lations were with assessments of controller relationships Samplewith other personnel (e.g., cooperation from others. Subjects for this study were drawn from students en-doesn't experience difficulty in relation to colleagues and rolled in the FAA Academy ATCS Nonradar Screensupervisors). In Germany, the ATCS selection test bat- program between September, 1990 and May. 1991.terv includes personality dimensions such as Rigidiy, Complete personality, aptitude, and training perfor-Extroversion, and Emotional Stability (Fissfeldt, 1990). mance data were obtained for a total of 1,121 first-timeThe validity (9/ of the German battery, including per- entrants. Sample demographic characteristics are pre-sonalit and cognitive dimensions, with various training sented in Table I in comparison to the population ofperformance criteria ranged from .51 to .61 with sample FAA Academy entrants. The sample was composedsizes of 162 to 196 entry-level controllers. Unfortu- mostly of non-minority men with some college cduca-nately, component standardized regression coefficients tion with an average age of 26 (range from 18 to 32): thewere not reported. majority reported some college education. As shown in

Critique of past research. At the time when most of TFable 1, this sample of entry-level controllers was typicalthe studies just reviewed were conducted, there was little of the population of Academy entrants.

consensus on an acceptable taxonomy for classifyingpersonality traits in such a way as to determine if there

2

Table 1Sample and Population Demographic Characteristics of A TCS Students

Population SampleCharacteristic Category (N = 9,945) (N = 1,09 1)

Sex Men 81.1% (8,065) 64.5% ( 723)Women 18.9% (1,880) 27.4% ( 307)Missing 8.1% ( 91)

Ethnicity White 87.0% (8,650) 73.7% ( 826)Asian 1.1% (113) 2.3% ( 26)Native American 0.6% ( 62) 0.7% ( 8)African American 5.0% (502) 9.5% ( 106)Hispanic Non-white 3.4% (340) 4.6% ( 52)Unknown 2.8% (278) 9.2% ( 103)

Age Mean 26.1 26.3SD 4.9 3.0

Measures grained analysis of persons or groups. l)omain and facetThe NEO Personality Inventory scales descriptions are presented ii "Table 2.

The NEO Personaliq Inventory (NEO-PI; Costa &

McCrae, 1985) is one of the first personality inventories Aptitudedesigned to explicitly assess the five major personality A written aptitude test is administered by the U.S. Office

dimensions identified over the course of more than 50 of Personnel Management (OPM) as the firs: hurdle toyears of factor analytic research. The NEO-PI was devel- entering the ATCS occupation. lie OPM A'ICS writ-

oped through a series of overlapping factor analyses, ten civil service examination batter, is composed of three

longitudinal studies, and peer ratingstudies, usingsamples tests: (a) the Multiplex Controller Aptitude Test (M CAT);of adult men and women rather than just college stu- (b) a test of Abstract Reasoning (ABSR): and (c) an

dents. This research program is described by Costa and OccupationalKnowledge Test(OKT). The general devel-

McCrae (1 985; 1987, 1988a). The normative samples opment, psychometric characteristics, and validity of

consisted of 502 men, ranging in ages from 21 to 93, and this test battery are described bySells, Dailey, and Pickrel

481 women, aged 19 to 93. Estimates for coefficients of (1984). Extensive research conducted by the Civil Acro-

reliability for the scales range from .85 to .93; those for medical Institute indicates that scores on the 6ivil service

test-retest reliability range trom .86 to .91 (Costa & test battery are significantly correlated with student

McCrae, 1985, 1988b).The five primary scalescompris- performance in the FAA Academy ATCS Screen pro-

ing the NEO-PI are: Neuroticism (A); Extraversion (h); gram (Manning, Della Rocco, & Bryant, 1989. Rock.

Openness to experience(0O); Agreeableness(A); and Consci- Dailey, Ozur, Boone, & Pickerel, 1981). Results from

entiousness( (I). Each of the N, L, and Odomain scales are the test battery are combined with any statutory veteran's

composed ofsix subscales assessing specific facets ofeach preference to yield a final civil service rating (RATI N(;).

domain. As a result, meaningful individual differences This rating is used to rank-order ATCS job applicantscan be seen within domains, providing a more fine-

3

within statutory guidelines such (hat hiring is done on PROCEDUREthe basis of merit, as more fully described in Aul (1991).

Comparison to norms

Criterion Three analytic procedures were employed to exploreThe FAA Academy Nonradar Screen. The FAA Acad- possible differences between persons entering the ATCSemy Nonradar Screen ("the Screen") was established in occupation and the general population. First, meanresponse to recommendations made by the U.S. 94th scores for the sample of entry-level controllers on theCongress House Committee on Government Opera- main and facet scales of the NEO-PI were comparedtions (U.S. Congress, 1976) to reduce field training with the published norms.The null hypothesis that thereattrition rates. The most recent version of the Screen was were no differences between entry-level controllers andimplemented in October 1985, supplanting predeces- normative groups was tested by t test computed on the

sor, 6ption-specific (Terminal and En Route) programs pooled variances of the groups. The overall risk of Typein place from 1976 through 1985. This Screen was I errors across the multiple comparisons was minimizedrevised again in 1986, shortening the course from 60 to through the use of an unordered Bonferonni procedure42 working days; the program remained relatively stable (Rosenthal & Rubin, 1984). The corrected criterion ctin content and process until it was terminated in March, for any given t test was .004, based on 12 comparisons1992. The Screen was based upon a miniaturized train- for each sex. Second, mean differences were translateding-testing-evaluation personnel selection model (Siegel, into Common Language Effect Sizes (CL; McGraw &1978, 1983; Siegel & Bergman, 1975) in which indi- Wong, 1992) as an aid to clarifying the practical signifi-viduals with no prior knowledge of the occupation could cance of any differences. The CL metric for effect size isbe assessed for their potential to succeed in air traffic the number of times out of 100 that a randomly sampledcontrol. entry level controller (group 1) will have a higher score

on a given personality scale than a randomly sampledPerformance measures. The Screen was developed person from the general (normative) population (group

with a clear emphasis on the assessment ofdevelopmen- 2). For example, if CL = 68 with respect to the Extraver-

tal performance using multiple methods (Boone, 1984). sion domain scale of the NEO-Pt, we would expect that,As a result, thirteen assessments of performance were for any random pairing of a controller with a member ofmade during the Screen (Della Rocco, Manning. & the general normative population, the controller wouldWing, 1990). These measures were derived from tests have the higher score 2 out of3 times. A CLof50 suggesis

administered in the classroom, observed performance that compared to a normative sample controllers wouldduring laboratory simulations of non-radar air traffic be no more likely to have higher scores on a scale thancontrol, and a final written examination. The measures might be expected by chance alone. The third analyticwere summed and weighted to create a final composite procedure illustrates these mean differences and effectScreen score (COMP), ranging from a theoretical mini- sizes in terms of the degree or percent of overlap in themum of 12 to a maximum of 100. A minimum score of distribution ofscale scores for the controllers and norma-70 was required to pass the Screen. Failure in the Screen tive groups. The greater the amount of overlap, asresulted in the removal of the student from the ATCS indicated by Tilton's 0(1937) statistic, the less effective

occupational series at the very least, and often in termi- a scale is in separating or discriminating between the twonation from employment with the FAA. The final corn- distributions. Taken together, these three analytic proce-posite score (COMP) was the training performance dures provided information about potential differences

criterion of interest in this study. between entry-level controllers and members of thenormative populations.

4

Table 2

NEO-PI Domain and Facet Scale Interpretations

Low Scores Scale High Scores

Calm, stable, relaxed, NEUROTICISM (N) Worried, tense, unstable,secure, deliberate nervous, impulsive

Relaxed, calm Anxiety (NJ) Fearful, worriedAmiable, even-tempered Hostility (N2) Angry, easily-frustratedHopeful, feels worthwhile Depression (N3) Guilty, hopelessSecure, comfortable Self-consciousness (N4) Ashamed, easily

embarrassedSelf-controlled Impulsiveness (N5) Impulsive, unable to resistResilient, hardy Vulnerability (N6) Intolerant of stress

Reserved, aloof, quiet, EXTRA VERSION (E) Outgoing, gregarious,reticent talkative, energetic

Cold, formal Warmth (El) Talkative, affectionateSolitary, self-contained Gregariousness (E2) ConvivialUnassuming, retiring Assertiveness (E3) Dominant, forcefulSlow, deliberate Activity (E4) Energetic, vigorousCautious, staid Excitement-seeking (ES) Flashy, takes risksUnenthusiastic, serious Positive Emotions (E6) Cheerful, high-spirited

Unlearned, realistic, OPENNESS TO EXPERIENCE (0) Inquiring, analytical,pragmatic, dogmatic tolerant, curious

Realistic, practical Fantasy (01) ImaginativeInsensitive to art and beauty Aesthetics (02) Moved by art and beautyNarrow range of emotions Feelings (03) Empathic, values feelingsPrefers familiar, routine Actions (04) Prefers new, novelPragmatic, factual Ideas (05) Curious, analyticalDogmatic, conservative Values (06) Tolerant, non-conforming

Cynical, rude, ruthless, AGREEABLENESS (A) Trusting, helpful, forgiving,uncooperative gullible

Unreliable, disorganized, CONSCIENTIOUSNESS (C) Organized, reliable,negligent punctual, obedient

5

- - 00

0 0 0 \ (a

r0 _ 1 it ON 00

',I;I

""o b

00 - cl .J: 09

zf 0\ 000a

LOjN'X

C-9 od CD '7

(t4- -I-

Nn ON E 0t

- L-0.l

a- 0 0 0 0 )

zo *

I "6

Prediction of performance RESULTSZero-order correlations berween the final OPM rating athire (RATING), personality facet and domain scale Comparison to normsscores, and final Screen composite score (COMP) were Mencomputed. These correlations were not corrected for Domain scales. Means and standard deviations for thedirect and incidental restriction in range due to selection NEO-Pl domair, scale scores for the men ATCS stu-of subjects on the OPM rating at hire. Therefore, the dents, adult men, and college men are presented in Tableestimates from this analysis represent lower-bound esti- 3. While the-e appeared to be some degree of overlapmates of the relationships between aptitude, personality, between ettrv-letel controllers and both adult and col-

and perikormance in the general population. Blockwise lege men normative distributions (Table 4), the t testsmultiple regression analysis was used to assess the incre- and CIl. effect sizes, using a criterion of CL < 25 or >_ 75.mental validity of the personality scales over that of the indicated important differences betwe. - the groups inaptitude measure in the prediction of performance in the the ove:alI profiles. ATCSs differed significantly fromFAA Academy ArCS Screen program. The OPM rating adult men on 4 of the 5 domain scales (all except

at hire was entered into the regression equation in the A, eeableness) and from college men on each of thefirst block in order to identify the proportion of variance domains. As a group, ATCSs reported significantlyin Screen performance accounted by cognitive aptitude. lower (p < .004) neuroticism than the adult or collegeThe personality N, E, and 0 facet sca'es and A and C men. The C1 suggest that 9 or fewer entry-level control-domain scales were examined in the second block using lers might be expected to have higher N scores thanstepwise regression. Use of this procedure provided an normative adult or college men in 100 random pairings.assessmentoftheadditionalsignificantvarianceinScreen ATCSs were also more extroverted and conscientiousperformance accounted for by personality beyond that than either adult men or college men, with far morealready accounted for by the genera! cognitive aptitude controllers expected to have higher Eand Cscores thanmeasure. The null hypothesis was that personality mea- normative men in 100 random pairings.sures would not account foradditional variance in Screen On the Openness to lxperiencedimension, ATCSs hadperformance beyond that accounted for OPM rating. higher scores than adult men (p:< .004) but lower than

Table 4

Percentage Overlap (Tilton's 0) in Controller and NormativeDistributions for Men on NEO-PI Domain Scales

Overlap of ATCS Men withAdult Men College Men

Domain Scale (N = 502) (N = 250)

Neuroticism 82% 59%

Extravcrsion 56% 88%

Openness 91% 80%

Agreeableness 99% 83%

Conscientiousness 84% 60%

7

00 *0

00 C\ m 00 1

M 00

C)l

0-- C14 QN )

~7-

<~0 00 kn V)(D--00

LP~~ CL ~ '

~~Cl

co00

(2( >4 - C) 4 O00 U JO EN ) ~2 E

HitC%Cz \O 0 z z

Cl O ON Cl8

Table 6

Percentage Overlap (Tilton's O) in Controllerand Normative Distributions for Men on NEO-PI Facet Scales

Overlap of Men ATCS Students withAdult Men College Men

FACET (N = 502) (N = 250)

NEUROTICISM

Anxiety 96% 76%Hostility 88% 73%Depression 82% 62%Self-Consciousness 90% 77%Impulsiveness 91% 66%Vulnerability 71% 60%

EXTRAVERSION

Warvith 86% 86%Gregariousness 77 % 99%A ssertiveness 81% 79%Activity 73% 87%Excitement-Seeking 46% 99%Positive Emotions 68% 100%

OPENNESS

Fantasy 94% 70%Aesthetics 93% 74%Feelings 88% 83%Actions 95% 68%Ideas 80% 89%Values 97% 94%

collegc men (p _ .004). Average ATCS scores on the A middle, and men controllers had lower average scores. As

scale did not differ from those of adult men but were shown in Table 6, there was less overlap between entrN-higher than those of college men (p < .004). level controller and college men distributions than with

Facetscales. ComparisonsoftheavL...geATCS men's adult men. Means, standard deviations, results of the tscores on the 6 N facet scales with those of the two tests and CL estimates for the three men's groups on thenormative groups appear in Table 5, with overlap in 6 Efacet subscales are presented in'Table 7. Mean scores

distributions presented in Table 6. The pattern of scores for men controllers were significantly higher than forfor the three groups was consistent on each of the 6 facets, adult men on each of the six Efacets (p_ .004). However,except for Anxiety. That was the only facet on which controllers differed from their college counterparts on

scores for ATCS men and adult men did no-, significantly only the Warmth, Assertiveness, and Activity facets, withdiffer. Otherwise, college men had rPnerally higher controllers presenting higher mean scores. Overall, theaverage scores on the N facets, adult men were in the effect sizes in Table 7 and overlap indices pres'-ited in

9

Cj

b" , C4 6-

a) Ci m r'-

ci - - cic

ON "0 .t N ON\C0 00 N 00 ON 00

ON -.1-00 0ON r- m 0) It -

ON r'n \0 kn I- \0

C14 00 1- o 0 N7"~

(N)I

cn oz~

M. t-- 00 0 . iýE-

E. 0) LA

w Ei

10

2:I

-4 09 C> (ON-

61 co tn

C)I

000 \.D 00

-r:

09 -

00 0 n te ) ON N N

R d

LU

00

E~ u' m.)

C,, 09 -- -rU &)I C> 'o

CU~~ a O ''

ui. 00 00 C

=1I

W)~ 00 C) OIN

let O N V V1

cz*

0 N

06 ký C' 0 0('

cn -d -n<' 0'

u oI

oc

r--

C14 0 00 -. I

- 6 r- W; to

0 co

000 -t 00W-0X

Qj 0 0

cn u

E uo

00z 0 -

01

Table 10

Percentage Overlap (Tilton's 0) in Controllerand Normative Distributions for Women on NEO-PI Domain Scales

Overlap of Women ATCS Students withDomain Scale Adult Women College Women

(N = 481) (N = 276)

Neuroticism 74% 49%

Extraversion 57% 78%

Openness 93% 87%

Agreeableness 97% 89%

Conscientiousness 77% 57%

Table 6 suggested that entry-level controllers were more 5 dimensions. Women entry controllers did not differsimilar to college men in the E domain than to adult from adult women on the Agreeableness and Opennessmen. Means, standard deviations, t tests, and CL esti- domain scales. In comparison to adult women, ATCSsmates for the 0 facet subscales are presented in Table 8. reported less neuroticism and greater extraversion, open-Average scores for ATCS men were significantly higher ness, and conscientiousness. They also appeared to have

than those of adult men on the Feelings (p < .004) and less neuroticism and openness than college women.Ideas (p < .004) facets. Men controllers differed signifi- Women ATCSs were more extraverted and exhibitedcandy from college men on each of the 0 facets except higher agreeableness and conscientiousness scores thanActionsand Values. Compared to theATCS men, college did their college counterparts. Overall, the pattern ofmen were higher on Fantasy, Aesthetics, and Feelings, but effect sizes and degree of overlap between distributionslower on Ideas. Overall, the pattern of CL effect size (Table 10) suggested that entry-level women controllersestimates and overlap in distributions suggested that were more similar to adult women than they were toentry-level controllers were more similar to adult men in college-aged women.terms of their openness to experience than they were to Facet scales. Table 11 presents the means, standard

college men. deviations and ttest comparisons for the three groups onthe 6 Neuroticism facet subscales. ATCS women had

Women significantly lower scores (p < .004) than either adult

Domain scales. Comparisons of the means and standard women or college women on 5 of the 6 Nfacets; womendeviations forwomenATCSs, adult women, and college controllers and adult women did not differ on thewomen on the 5 NEO-PI domains appear in Table 9. Impulsivefacet. The pattern of scores for the three groups

Estimates of the degree of overlap between controller was similar to that noted for men, ATCS women had theand normative score distributions are presented in Table lowest scores, adult women were in between, and college10 foi the domain scales. Scores for ATCS women women had the highest scores. Overall, the pattern ofdiffered significantly from those of adult women on 3 of overlap between distributions (Table 12) and CL effectthe 5 dimensions and from college women on each of the sizes suggested that entry-level women controllers were

13

0 C) 00 0't 2.10

C14.

00 V") ýo 00 4

rn ONt 0-ON~'

o 000

2* ý 0

W) Ný N* q 0 -N 1

=W 00

CE~0 CU .0

0~~~ -4 E' t ~

co~~U -U r-0

C ~ ~- 00)75 II

4-.c

14

Table 12

Percentage Overlap (Tilton's 0) in Controllerand Normative Distributions for Women on NEO-Pt Facet Scales

Overlap of ATCS Women withAdult Women College Women

Facet Scale (N = 481) (N = 276)

NEUROTICISM

Anxiety 82% 59%Hostility 89% 68%Depression 71% 55%Self-Consciousness 81% 64%Impulsiveness 91% 84%Vulnerability 63% 51%

EXTRAVERSION

Warmth 92% 78%Gregariousness 82% 98%Assertiveness 76% 71%Activity 75% 79%Excitement-Seeking 41% 99%Positive Emotions 72% 88%

OPENNESS

Fantasy 98% 72%Aesthetics 86% 77%Feelings 93% 82%Actions 88% 92%Ideas 78% 86%Values 87% 93%

more similar to adult women in terms of the Ndomain pairings. In general, the pattern of overlap betweenand its facets than theywere t- college women. Compari- controller and normative sample distributions on the Esons of the means and standard deviations for the three facet scales (Table 12) suggested that women controllerswomen groups on the Extraversion facet subscales are were mote similar to college women than adult womenpresented in Table 13. Average scores for women ATCSs in terms of their extraversion. Finally, facet scores for thewere significantly higher than those of the adult women 0 domain subscales are compared in Table 14. Womenon 5 of the 6 E facet subscales and higher than those of controllers had significantly higher scores than adultcollege women on 4 of the 6 subscales. The largest mean women on the Ideasand Values facets. The CLeffect sizedifference occurred on the Excitement-seeking (E5) estimates indicated that, on the average, women control-subscale (20.3 for ATCSs versus 12.9 for adult women); lers would be expected to have higher scores on thesewomen controllers would be expected to have higher E5 facets than would adult women. On the other hand,scores than adult women in 99 out of 100 random women controllers would be expected to have lower

15

00 ýT r- 00 - Nr- - 00 r-- Vr

tn 00 c) 0 "D W

00 0(nO t

N- 00 00 ON 0

c:x

w 0C- \D 00 00 CN ON

o t to C14 00EU

0 --5~ - * .. cis

000 '0 00LO

S-.S

-. c

Ch C Nl m~ \C 0 a C

C.) C ... ..

NQ

0 0

E E00~0 U)r -

Cl C

16

vi

17ý Cl! I

0 .4

C'~' (7N l -

CL "T 0 0 \C0

00

Q CV

ClE 00

0 MU,

0 ~ 0 0 0

5 o~ o17

Table 15

NEO-PI Domain and Facet Scale Scores for Controllers by Sex

Male ATCS Female ATCS(N = 723) (N = 307)

DOMAIN/Facet Scale M SD M SD t CL

NEUROTICISM 64.6 19.1 66.3 20.2 -1.28 39N1)Anxiety 12.5 4.5 13.0 4.5 -1.63 43N2)Hostility 9.4 4.4 9.1 4.2 1.01 54N3)Depression 9.3 4.5 9.2 4.9 0.32 51N4)Self-consciousness 12.6 4.4 12.4 4.4 0.67 53N5)Impuliuve 14.0 4.3 15.1 4.9 -3.60- 36N6)Vulnerability 6.9 3.6 7.5 3.6 -2.45 41

EXTRAVERSION 121.7 16.8 124.5 17.0 -2.44 33El)Warmth 23.6 4.1 24.5 3.9 -2.44 38E2)Gregariousness 17.3 4.5 17.9 4.6 -1.94 42E3)Assertiveness 18.8 4.1 19.0 4.7 -0.68 47E4)Activity 19.0 3.9 19.8 4.3 -2.92 39E5)Excitement-seeking 21.6 4.3 20.3 4.6 4.34"" 67E6)Positive Emotions 21.5 4.2 23.1 4.0 -5.67"" 29

OPENNESS 113.0 15.8 118.2 15.8 -4.83"* 18Ol)Fantasy 17.3 4.7 17.2 5.0 0.31 5102)Aesthetics 15.8 5.7 17.8 5.2 -5.28"" 2803)Feelngs 20.7 4.0 22.5 3.8 -6.70"** 2604)Actions 16.1 3.5 17.2 3.7 -4.53- 3405)Ideas 22.0 4.6 21.3 4.7 2.22 4106)Values 21.2 4.4 22.3 3.8 -3.82"" 35

AGREEABLENESS 48.3 7.0 51.0 6.6 -5.76"" 23

CONSCIENTIOUSNESS 52.9 8.0 55.2 7.6 -4.28"'" 28

NOTES: CL = Common Language Effect Size, ***p < .004or number of times that men would have higherscores than women in 100 random pairings

18

scores on the Aesthetics facet than would adult women. In Prediction of Performancecontrast, women controllers differed significantly from Correlationscollege women on 4 of the 6 0 facet scales. Women Zero order correlations between the mxsure of cognitivecontrollers were more likely to have lower scores than aptitude (RATING) for the ATCS occupation, NEO-111college women on the Fantasy, Aesthetics, and Feelings facet and domain scale scores, and final composite scorefacets, and higher scores on the Ideas facet. Overall, the (COMP) in the screen for both men (N= 529) and womenpattern of overlap and effect sizes suggested that women (N = 193) cases with complete valid data are presented incontrollers were more similar to adult women on the first TFables 16 and 17. Correlations between aptitude (IRlAT-three facets (Fantasy, Aesthetics, and Feelings) and more ING) and personality scores for men ranged from a low ofsimilar to college women on the Actions, Ideas, and .000 with Impulsive (N5) to an absolute maximum of .098Values facets. with the Ideas(05) facet. Correlations between aptitude and

personality scores for women were similarly low, rangingComparison of Men and Women Controllers from .002 with Fantasy (0 1) to a high of. 169 with ValuesMeans and standard deviations for the NEO-PI domain (06). Correlations between the personality measures andand facet scales for men and women ATCS students are Academy screen score (COMP) for men were also low,presented in Table 15. While there was some degree of rangingfrom -.005 forAesthetics(02) to-. 148 with the facetoverlap between the sexes, the t tests and CL effect sizes Excitement-seeking(E5). The pattern ofcorrelations betweenrevealed that men and women differed significantly and personality and performance for women was very similar,practically on 3 of the 5 domain scales. Women ATCSs with generally low correlations ranging from -.005 withhad higher scores than men on the 0, A, and Cdomain Assertiveness (E3) to .178 with Ideas (05).scales (p < .004). The higher 0 scores for women ATCSstudents were consistent with sex differences reported by RegressionCosta and McCrae (1989), Averages for women were Resultsof the regression analysis to determine if the person-also higher than men on the Nand Edomain scales, but ality variables contributed to the prediction of Academydid not meet the practical cutoff of 25 _< CL > 75. None Screen performance above that contributed by cognitiveof the differences noted on the facet scales between men aptitude are presented in Table 18 by sex. The resultsand women met the stipulated cutofffor practical signifi- suggested that the null hypothesis should be rejected for bothcance, although there were several statistically significant men and women controllers. There was a small, but signifi-differences of interest. For example, men were less im- cant increase in incremental validity for men when person-pulsive (N5) and reported less positive emotions (E6) ality variables entered the prediction equation (AR2 =.033,than women controllers (p < .004), but sought more AF (4,524) = 3.84, p < .05). The significant personalityexcitement (E5). Women had higher scores, on the factors included the Fantasy (01), Excitement-seeking(ES),average, than men controllers on 4 of the 6 Opennessfacet and Activity (E4) facets. However, the total percentage ofscales: Aesthetics (02), Feelings (03), Actions (04), and explained variance that those three variables added was quiteValues (06). Overall, men and women entry-level con- small (3%). Just one facet, Ideas (05), entered the stepwisetrollers appeared to have quite similar profiles, with regression solution for women, after entering aptitude intowomen generally having just slightly elevated scores over the prediction equation (AR' = .028, AF(2,190) = 5.53, pmen. <5.05). As with men, personality accounted for a bare 3% of

additional explained variance in performance above thatalready explained by cognitive aptitude.

19

ri.- 00I

rl r)

-00 -j 00

¶ 0(? r '.0

'- 00 0 r I

C-ci ri <.

c-i 0- -It i fn- tn 0

r! *, r- o c - ri

c---~-c o-- ~ W > 1!

~- 0

w 0 v) Tr-

1-C C ~ 0 r!- c'0 004-L

0 q- j,

Ccu

It rl r - - a, a,00 en c-.i -1i-en ~cn 0 0 0 c-- -ý It W

-l 'IT MZL

ý--0t 000c a O ý0 rý Z

IIJC

00 W) C4.4l )0 * - ! t

it'.CI- '.OO - - -- C I0r rl -C l r0 *- z

............................................... t o'.E

00 C ~ itt~ l ir0 CýiO rlc O 0.0 ;00 Qz

Q = UJ00 r- W -ri 'IT It --i 0 W) 00 CA' O\i r- mt m~ 0 a, 00 -0-t 1

C!~~~-r 0 .i~ 0- - 0tI .~ .' . .i 4 ZW )00 0 00 0 00 0 -tOC t l-

m m n ýtm Vm It r,,ýc r1 00t*-0It .

.ýR - , -.Ci 0,r-f-0r

ZZZknLLLTO--m00000<O

m zw 0Ttrl 20

0l "t-

tol-'1 %n o r-

* Cl

FA (- .1 Cl

-r at .

r4 0. l- . r- 4S- CO

Orl m l.-0 C0 rl

cl' o 0- 0l~rl

0 -

c-C' rl0 00 -'0 Ir)cnC- C ! lfC-! l (l ý W)w mE

CkA

U~~r 000'C -'-o ý' C'i

fý) ~ ~ r- rl*&m rlý rOql s -0 Al- o(S - 0 0

It~~~~~ ~ ~ ~ (S0 r4Mý IAý - 0

00 _ C

C,-CIAr- r-00 CIA C IA r4 0 m 00C-

r- lO0I( o ýo ri r- '*-kn S-Cl 0 CI 0n cl- 0 000 't(

q fn Cl-0 - 0 (0CjO Q CA '0

C1k fn -- =S =l 0- m 8

*t 0 M 0 -01 O ItM0 l ClC' fn M S0 8 ný.

t-O ICI ~, 0 . .- . .- cl 06 Z'LZC

C:

z 0 Cl( ( 0, SnO'A0.- . 00 m r-CAl .;

W) 10 m r A m 0 Cl -O lCJS. %n r w 00 ;Zt(S*tSC--O I-0- tf7l ri C

C- < <

zzz ww 00- 0- < I u-ZW

21

Table 18

Regression Anaiysis of Incremental Valicity of PersonalityOver Aptitude in Prediction of A TCS Screen Performance by Sex

FStep Var 3 R R Adj-R2 AR2 AF (d()

Men

RATING' .230 .230 .053 .051 29.34-"(1,527)

2 RATING .220E52 -. 132 .265 .070 .067 .017 9.86" 19.85""

(2,526)

3 RATING .215E5 -. 14701 .094 .281 .079 .074 .009 4.92" 14.97"'"

(3,525)

4 RATING .214E5 -. 17301 .101E4 .086 .292 .086 .079 .007 3.84' 12.25"'"

(4,524)

Women

RATING .148 .148 .022 .017 4.25'(1,191)

2 RATING .13305 .167 .222 .049 .039 .028 5.53' 4.94"

(2,190)

NOTES: 'RATING was entered 'p < .05 ") < .01 -p < .001"7E5 Excitement-seeking;

01 = Fantasy;05 = Ideas;E4 = Activity

22

D~ISCUSSION however. they would st ill experienie a normal aniou tH ,,

disappointment and anger. The AT(l :s ~p a s a•hole.Differences Between Controllers and General had stores in the upper-range of the L'.xtraj-r.'rzoo do-

Population main. Individuals with these scores are cheerful, gener-

Practical versus statistical significance. Overall, most of ally satisfied with lifte, and pref'er excitement and

the comparisons between entry-level controllers of both stimulation along with thc company of others most of

sexes to their respective adult and college normative the time. The Openness to tE•sperrence scores for these

samples were statisticall" different. Htowever, the rela- entry-level controllers suggested that thes-could be shar-tively' large sample sizes provided a great deal of stat istical acierized as having broad interests, knowledgeable aboutpower. such that even very small, practically insignificant many topics, and as being intellectually' curious or inves,differences could be reliably detected. In ihis study, tigative. The distribution of scores in the AgreeabkeneB

Tiltor's C) ( 93-) and the Common Language L'ffect domain hinted that some AT'CSs present themselves as

Size (CL: McGraw & \Wong, 1992) were employed to generally warm, trusting. agreeable. and sympathetic to

evaluate the practical magnitude of any detected statisti- others while others present themselv,.s as more tough-cal differences. O)f the two, C/. appeared to provide a minded, skeptical. and ompetitive. ATCS group scoresmore parsimonious description ofdifferences bydescrib- in the ]ornscienttousnesr domain were also in the mid-

ing the number of occasions certain differences w,,ere range. Such scores suggest that the entry-level controllers

likely to be observed in a random pairing of 1 00 eases exhibit an average level of determination, reliability, andf'rom each sample. In general. the patter ns ofttest resulks self-direction Overall, the pattern of scores for men and

combined with the overlap and ef'fect size estimates women controllers suggested a certain level of intensity.

indicated that women entry-level controllers were lower energy-, ambition or purpose. gregariousness, and gener-in neurouicism and higher on extraversion and conscien- ally good adjustment within this sample of controllers.

tiousness than either the adult or college women norma- Wtomen entry-level controllers, in particular, appearedtive groups, and that these differences were practically as to oe more dominant and forceful than their normativewell as statistically significant. Differences between men peers.

entry-level controller, and adult men were generallysimilar to those noted tor women. Men controllers w,,ere CONCLUSIONS ABOUT DIFFERF.NCES%expected to have higher Cscores than adult men in 78 of BETWEEN CONTROLLERS AND GENERAl.

100 random pairings; in comparison, controllers were POPULATIONexpected to have higher E scores in 99 of 100 pairings Persons who enter the FAA Academy ATCS Nonradarcompared to a higher N score in only 9 of 100 such Screen program do differ from college students and

pairings. Entry-level men controllers were lower in adults on anumber of personality,'dimensions as assessedneuroticism, openness to experience, and cotisciert~iuus- by' the NEO-PJ. Differences were found generally onness than college men. For example, men ATCSs were domain scales, with those differences being traced in

expected to have higher conscientiousness than college some cases to specific facets within a domain. Thosemen in just 2 of 100 pairings, indicating that the statis- differences were similar for men and women controllers;

tical difference between men controllers and college men controllers were more outgoing and had higher-levels ofon the C domain was practically significant, excitement-seeking, expressed more positive emotions,

and were more conscientious than the normative samples.

Profile interpretation Some of the differences noted above, however, may be

(;iven these apparent differences between controllers attributable to the tendency, for job applicants and

and normative samples, how would men and women employeesundergoingevaluations to present themselvescontrollers be described on the basis of the NEO-PI in a positive light. Previous research with the 16PF and

profile? Overall, ATCSs scored in the low-average range ATCS applicants and job incumbents supports theof'the Neuroticism domain. Similar individuals might be presence of positive self-presentation (Karson & O'Dell,seen as being generally calm and able to deal with stress; 1974; Schroeder & Dollar, 1989). However, the

23

differences between ATCS studeats and the normative The impact of that gain of 30 developmental ,ontrolgroups were not evident across each of the dimensions or lcrs from a sample of about IO00 entrv.level controller,facets of the N"O-PI that one would expect to be may not seem significant until placed in the contex ofsensitive to a positive test-taking attitude (e.g.. the relative costs to the agency. The FAA Ac.adcm,Agreeableness), AI[CS Nonradar Screen cost approximately S 10,000 per

student to administer (AerospaLe Siences, Inc., I1))I1).Utility of Personality Measures With about 2,000 candidates entering the FAA Acad-

Selected aspects of personality also demonstrated incre- cmv each year, reduction of the failure rate by 3(0mental validity over cognitive measures in the prediction controllers per 1000 candidates would have resulted in aof performance in the FAA Academy ATCS Nonradar savings of S600,000 per year in terms of wasted trainingScreen program. Two of three facets that contributed to resources. The total sav ings over the S full years of thethe final regression equation were drawn from the Open- ATCS Nonradar Screen's operational life would Itave

nessto Eperiencedomain. That finding is consistentwith been on the order of S3 Million in saved trainingBarrick and Mount's (1991) conclusion that scores from resources. In other words, implementation ,-f an AICS

the Odomain were likely to be related to performance in selection test procedure with just marginal improve-training. ments in validity Would have resulted in a signifikant

While the absolute amount of additional variance savings to the agency.

explahied was small (2%), such marginal gains havesignificant utility impacts in large-scale selection systems Future Research Needssuch as that for controllers. For example, the increase in The results presented in this paper provide some empiri -the proportion ofemployees considered satisfactory from cal counterpoints to the claim by Bcsco (1991) thatthe use of aptitude plus personality measures can be personality measurement in selection programs for pro-estimated using the Taylor-Russell tables (Taylor & fessional performance lack both scientific and practicalRussell, 1939). Overall, about 60% of persons entering value. Additional research is needed to provide furtherthe Screen were successful. The current selection ratio, validation of these outcomes. Differences between en-based on the cognitive aptitudc ratings, is about .10. The try-level controllers and the general population norms

validity coefficient of the cognitive aptitude rating by were found, with controllers being overall less neuroti.,

itself for this sample of men was about .23, and for more extraverted and conscientious. Entry-level control-women, .15. The validity coefficient for the combined lers reported few neurotic symptoms, appcared to cx-group was .23. About 74% of the combined sample of hibit better adjustment, and tended to be outgoing

men and women entry-level controllers would have been individuals with higher levels of excitement-seeking andpredicted to succeed in the Screen, based on interpola- more expressive of positive emotions. This provides a

tion of the Taylor-Russell tables. The validity coefficient generally positive picture of those who apply and arefor the cognitive aptitude rating plus personality facets selected to become operational controllers, and alsowas .29 for men and .22 for women, or about .27 for the suggests that they are well suited for handling the de-

combined sample. The interpolated predicted success mands of a highly responsible job. These differencesrate for men and women would have been 77 % if the were in keeping with lay perceptions of the "controllerNEO-PI scores had been used with cognitive aptitude personality," and supportive of research that suggestsscores in selecting persons to attend the FAA Academy, that certain occupations may attract individuals withOverall, use of a combined personality-aptitude selec- different personality characteristics (Kassera & Russo.

tion procedure would have resulted in net gain of about 1987). Additional research might investigate the rela-3% in predictive efficiency, or approximately 30 addi- tionshipý hetween personality and biodemographic fac-tional developmental controllers (out of every 1000) tors, such as reasons for occupational choice and cateerbeing available to the FAA for field training, expectations in the controller occupatinn.

24

Such research could provide the foundation for ATCS RETE-LENCLScareer guidance tools for use by aviation educators andagency recruiters. Aerospace Sciences, Inc. (1991). Air traffic control spe-

The findings reported in this paper demonstrate that cialist Pre-Training Screen preliminary validation:personality, as represented by scores on a theoretically- Final report. (Final report delivered to FAA underbased and psychometrically sound instrument, explained contract I)FAO -90-Y+O 1034). Washington, I)(:additional variance in a technical performance measure Federal Aviation Administration Office of thebeyond that explained by cognitive aptitude. It must be Deputy Administrator.

noted that this performance measure appeared to be Aul, J. C. (1991). Employing air traffic controllers. Inuncontaminated by evaluative biases noted by Besco as H. Wing, & C. A. Manning (Eds.). Selection ofairfatal flaws in personality-oriented research. However, traffic controllers: Complexity, requirements, andfurther research is needed to identi•v which aspecrsofthe public interest. (DOT/FAA/AM-91/9). Washing-Screen program are most influenced by personality, such ton, DC: Federal Aviation Administration Office

as the Instructor Assessment of student potential. While of Aviation Medicine.

the observed relationships were small, it must be noted Barrick, M. R., & Mount, M. K. (1991). The Big Fivethat our sample was a highly select group. Selection on personality dimensions and job performance: Athe cognitive aptitude rating likely resulted in incidental meta-analysis. Personnel Psychology, 44, 1-26.restriction in the range of personality scores. It is prob- Besco, R. 0. (1991). The myths ofpilot personalitv stereo-able that if the group were not initially selected on the types. Paper presented at the 6th Internationalbasis of cognitive abilities, we would see a higher rela- Symposium on Aviation Psychology, Columbus,tionship between the NEO-Pl scores and performance OH.in the FAA Academy ATCS Nonradar Screen. In addi- Boone, J. 0. (1979). Toward the development of a newtion, further research is warranted on the interaction selection battery for air traffic control specialists.between personality and aptitude in performance to testthe hypothesis that certain personality attributes may eral Aviation Administration Office of Aviation

enhance or detract from performance for persons with Medicine. [NTIS AD A080 065].high or low aptitude for the occupation. Finally, longi-tudinal research assessing the utility of personality inca- comparative analysis of individuial and system perfor-sures in predicting performance across time is required.

mance indicesfor tie air traffic co ,.-rol system. At lan-It is likely that the relationships between performance tic City, NJ: Federal Aviation Administrationand stable personality traits change over time. The initial National Aviation Facility Engineering Centerpredictive power of personality dimensions may be low (NAFEC)due to two factors: (a) variance in initial performance Cascio. W. F. (1984). Costirg human resources. Themay be accounted for by ability and prior experience; financial impact of behavir in organizations. Bos-and (b) as the "honeymoon", characterized by initially ton, MA: Kent Ptfblishaiig Co.high commitment and effort, ends in a new job, thenovelty and challenge in a job may fade and dispositional Carrell, R. B_, Eber, H. W., & Tatsuoka, M. M. (1970).fact(,,s may become increasingly important determi- Handbook for the 16PF. Champaign, IL: Institute

nants of performance (Helmreich, Sawin, & Carsrud, for Personality and Aptitude Testing (IPAT).1986). Bothofthese factors may account, at least in part, Collins, W. E., Nye, L. G.. & Manning, C. A. (1990).for the seemingly low observed personality-performance Studies ofpoststrike air traffic control specialist train-relations. Long-term follow-up studies will enable us to ees: IIl. Changes in demographic characteristics ofinvestigate changes in the personality-performance rela- Academy entrants and biodemographic predictors oftions within the controller occupation, success in air traffic controller selection and/Academry

screening. (DFOT/FAA/AM-90/4). Washington,[)C: Federal Aviation Administration Office ofAviation Medicine.

25

Collins, W. E., Schroeder, D. J., & Nye, L. G. (1989). Deloney, J. R., & Schroeder, D. J. (1983). Ways OfRelationships of anxiety scores to Academy and field perceiving the world and the relationship to successtraining performance ofair traffic control specialists. among air traffic controllers. Presented at the 54ith(DOT/FAA/AM-89/7). Washington, DC: Fed- Annual Scienitific Meeting of the Aerospace Medi-eral Aviation Administration Office of Aviation cal Association, Houston, TX.Medicine. Eissfeldt, H. (1990). The DLR selection of air traffic

Colman, J. G. (1970), Review and evaluation of present controlapplicants. German Aerospace Research Es-system for selection of air traffic controllers. (Phase 1, tablishment DLR, Hamburg, Germany.Task I Report submitted under Contract No. ;ough, H (. (1960). Manualfor the California Psycho-DOT-FA70WA-2371.) Washington, DC: Edu- logia H.vento60. (Rev. ted Palo At C to-

cati o a nd P ubl c A ff i rsI n c .l ogi a l l niven to ry . (R ev . E d .). P a lo A lto C A : ( ,.o n -cation and Public Affairs, Inc. suiting Psychologists Press.

Costa, P. T. Jr,, & McCrae, R. R. (1980). Still stable Guion, R, M., & (libsn, W. M. (1988). Personnelafter all these years: Personality as a key to someissues in adulthood and old age. In P. Baltes & 0. elt, and p mA-R oPc

G. Brim, Jr. (Eds.). Lifespan, development and ogy, .39, 349-374.behavior. New York: Academic Press. (,uion, R. M. & Gottier, R. F. (1965). Validity of

personality measures in personnel selection. Per-Costa, P. T. Jr., & McCrae, R. R. (1985). The NEO sonnelPsychology, 18, 349-374.

Personality Inventory rranual Odessa, FL: Psycho-

logical Assessment Resources. Helmreich, R. L., Sawin, L. L., & Carsrud, A. L. (1986).The honeymoon effect in job perfo~rmance: Vein-

Costa, P. T. Jr., & McCrae, R. R. (1988a). Personality poral increases in the predictive power ofoachieve-

in adulthood: A six-year longitudinal study of self-

reports and spouse ratings on the NEO Personal- ment motivation. Journal of Applied Psychologv,

ity Inventory. Journai of Personality and Social 71, 185-188.

Psychology, 54, 853-863. Hofstee, W. (1984). What's in a trait: Reflections about

Costa, P. T. Jr., & McCrae, R. R. (1988b). NEO-PI/FFI the inevitability of traits, their measurement, andManual supplement, Odessa, FL: PsychologicalAs- taxonomy. In H. Bonarius, G. Van Heck, & N.sessment Resources. Smtid, (Eds.). Personalitypsychology in Europe: Theo-

retical and empirical developments. Lisse. the Neth-

Costa, P. T., Jr., McCrae, R. R., & Holland, J. L. erlands: Swets & Zeitlinger.(1984). Personality and vocational interests in an Hogan, R. T. (1983), A socioanalvtic theory of person-adult sample, Journal of Applied Psychology, 69, H anR. T. ( .Pa oi t t. N e or of peuson-390-400. aliry. In M. Page, (Ed.). Nebraska Symposium on

Motivation, 55-89. Lincoln, NE: University ofDella Rocco, P. S., Manning, C. A., & Wing, H. Nebraska Press.John, 0. P.. Angleitner. A.. &

(1990). Selection of air traffic controllers for auto- Ostendorf, F. (1988). The lexical approach tomated systems: Applications from current research. personality: A historical review of trait taxonomic(DOT/FAA/AM-90/13). Washington, DC: Fed- research. European Journal of Personality. 2. 17 1-eral Aviation Administration Office of Aviation 203.Medicine. [NTIS AD A238 2671. Karson, S., & O'Dell, J. W. (1969). A criterion factor

Dolgin, D. L., & Gibb, G. D. (1989). Personality analysis of performance ratings and personalit, flac-assessment in aviator selection. In R. S. Jensen tors in radar controllers. Paper presented at the(Ed). Aviation psychology, p. 288-323. Brookfield, Annual Scientific Meeting of the Aerospace Medi-VT: Gower. cal Association, San Francisco. CA.

Kassera, W., & Russo, T. (1987). Factor analysis ofpersonality preferences and vocational interest.Psychological Reports. 60, 63-66.

26

Manning, C. A., Della Rocco, P. S., & Bryant, K. I). Reilly, R. R., & Chao, G. T. 1982). Validity and(1989). Prediction of success in FAA air traffic co11- fairness of some alternative employee selectiontrol field training as a function of selection and procedures. Personnel Psychology, 35, 1-62.screening test performance. (DOT/FAA/AM-89/6). Rock, 1). B., Dailey, J. T., Ozur, H., Boone, J. 0., &Washington, DC: Federal Aviation Administra- Pickerel, E. W. (1982). Selection of applicants fortion Office ofAviation Medicine. [NTIS AD A209 therelc cWntro lection o TsfAA3217/16XAB). the ai, traý c controller ,ccupation. (DOT'I'FA tA1

7 A AM-82/1 I). Washington, DC: Federal Aviation

Manning, C. A., Kegg, P. S., & Collins, W. E. (1989). Administration Office of Aviation Medicine.Selection and screening programs for air traffic [NTIS AD A122 795/8].control. In R. S. Jensen (Ed.), Aviation psychology, Rosenthal, R., & Rubin, D. B. (1984). Multiple con-321-341. BrookfielW, MA: Gower Technical.

trasts and ordered Bonferonni procedures. journalManning, C. A., VanDeventer, A. D., & Baxter, N. E. of iducational Psychology, 76, 1028-1034.

(1984). Sensation seeking and performance in air Schroeder, D.J, (1984). Cognitive style andA FCSAcad-traffic control specialist training. Presented at the erny pass/fail status. Presented at the 55th Annual55th Annual Scientific Meeting of the Aerospace Scientific Meeting of the Aerospace Medical Asso-Medical Association, San Diego, CA, ciation, San Diego, CA.

McCrae, R. R., & Costa, P. T. Jr. (1987). Validation of Schroeder, D. J., & Dollar, C. S. (1989). Personalitythe five-factor model of personality across instru- characteristics of pre/post-strike air traffic controlments and observers. Journal of Personality and applicants. Presented at the 60th Annual ScientificSocial Psychology, 52, 8 1-90. Meeting of the Aerospace Medical Association,

McGraw, K. 0., & Wong, S. P. (1992). A common Washington, DC.language effect size statistic. Psychological Bulletin, Seigel, A. 1. (1978). Miniature job training and evalua-111 , 361-365. tion as a selection/ classification device. Human

Myers, 1. B. (1962). The Myers-Briggs Type Indicator. Factors, 20, 189-200.Palo Alto, CA: Consulting Psychologists Press. Seigel, A. I. (1983). The miniature job training and

Naylor, J. C., & Shine, L. C. (1965). A table for evaluation approach: Additional findings. Person-determining the increase in mean criterion score nel Psychology, 36, 41-56.obtained using a selection device. JournalofIndus- Seigel, A. I., & Bergman, B. A. (1975). A job learningtrial Psychology, 3, 33-42. approach to perfbrmance prediction. Personnel

Norman, W. ( 963). Toward an adequate taxonomy of Psychology, 28, 325-339.personality attributes. Journal of Abnormal and Sells, S. B., Dailey, J. T., & Pickrel, E. W. (Eds.) (1984).SocialPsychology, 66, 574-583. Selection of air traffic controllers. (DOT/FAA/

Nye, L. G., & Collins, W. E. (1991). Some personality AM-84/2). Washington, DC: Federal Aviationcharacteristics of air traffic control specialist trainees: Administration Office of Aviation Medicine.Interactions of personality and aptitude test scores [NTIS AD A147 765].with FAA Academy success and career expectations. Smith, R. C. (1985). Stress, anxiety and the air traffic(DOT/FAA/AM-91/8). Washington, DC: Fed-eral Aviation Administration Office of Aviation control specialist: Some surprisingconclusions from

Medicine. a decade of research. In C. D. Spielberger & 1. G.Sarason (Eds.). Stress and Anxiety. (Vol. 9, pp. 95-

Nyfield, G. R., Kandola, R. S., & Saville, P. F. (1983). 100). New York: Hemisphere.The Selection of air traffic controllers: Concurrent Spielberger, D. C. (1979). Preliminary manual for thevalidity study. London, England: Saville and SpebgeDC.(9).Plinaym ulfothHoldsworth Ltd. State-Trait Personality Inventory. Tampa, FL: Uni-

versity of South Florida, Human Resources Insti-tute. SPSS, Inc. (1990). SPSS reference guide.Chicago, IL: Author.

27

Taylor, H. C., & Russell, J. T. (1939). The relationship Trites, 1). K., Kurek, A., & Cobb, B. (1967). Personalityof validity coefficients to the practical effectiveness and achievement of air traffic controllers. Aero-of tests in selection: Discussion and tables. Journal space Medicine, .38, 1145-1150.ofApp!ied Psychology, 23. 565-578. Tupcs, E. & Chlisail, R. 0t%l;. Rec'ui,,• l ,oil ,,

Tett, R. P., Jackson, D. N., & Rothstein, M. (1991). factors based on trait ratings. (Technical ReportPersonality measures as predictors of job perfor- No. 61-97). Lackland AFB, TX: U.S. Air Forcemance: A meta-analytic review. Personnel Psychot- Aeronautical Systems Division.ogy, 44, 703-742. United States Congress. (January 20, 1976). House Com-

Tenopyr, M. L., & Oeltjen, P. D. (1982). Personnel mittee on Government Operations recommendationsselection and classification. Annual Review of Psy- on air traffic control training. Washington, DC:chology, 33, 581-618. Author.

Tilton, J. W. (1937). The measurement of overlapping. Zuckerman, M. (1979). Sensation seeking. Beyond theJournal of Educational Psychology, 28, 656-662. optimal level of arousal. H illsdale, NJ: ErIbaum.

*U.S. GOVERNMENT PRINTING OFFICE: l9qI-761-O0,,'60060

28