DOT/faa/am-99/18 An Investigation of the RelationshipBetween Chronological Age andIndicators of Job Performance forOffice ofAviation Medicine

Washington, D.C. 20591 Incumbent Air Traffic Control Specialists

Michael C. Heil

Civil Aeromedical Institute

Federal Aviation Administration

Oklahoma City, Oklahoma 73125

June 1999

Final Report

Thisdocument is available to the publicthrough the National Technical InformationService, Springfield, Virginia 22161.

OU.S. Departmentof Transportation

Federal AviationAdministration

NOTICE

This document is disseminated under the sponsorship ofthe U.S. Department of Transportation in the interest ofinformation exchange. The United States Government

assumes no liability for the contents or use thereof.

APPENDIX A

PERFORMANCE CATEGORIES FOR PEER AND SUPERVISOR RATING SCALESA. CoordinatingCoordinating with other controllers to minimize traffic problems; coordinating clearances chances in

a.rcraftdestmauons.alt^^manner; presenting the rationale for instructions to pilots or other controllers as necessa^

B. Communicating and InformingUsing clear, concise, accurate language to get message across unambiguously; talking onlywhen necessarvand

appropriate; emptying pr0per phraseology to ensure accurate communications; nofifjjg pi ots/c^n oE/part.es; proving complete and accurate position relief briefings; providing accurate and legible flfght s1mormanon; llscening full d̂ ^ from ^ ^ ^^8 J£*«Pthat they are understood; attending to readbacks and ensuring that they are accurate.

C. Maintaining Attention and VigilanceScanning properly for air traffic events, situations, potential problems, etc.; keeping track of equipment/

weather status; identifying unusual events, improper positioning of aircraft, equipment malfunctions, etc ;recognizing when aircraft have potential for loss ofseparation; verifying visually that control instructions arefollowed; recognizing potential problems in adjacent sectors; remaining vigilant during slow periods.

D. Managing Multiple TasksKeeping track ofalarge number ofaircraft/events at atime; conducting two or more tasks simultaneously

{e.g., issuing instructions while scanning the screen; monitoring pilot communications while writing on strips);remembering and keeping track of aircraft and their positions; remembering what you were doing after aninterruption; returning to what you were doing after an interruption and following through; providing pilotswith additional services as time allows.

E. PrioritizingTaking early or prompt action on air traffic problems rather than waiting or getting behind; knowing what

to do first and which are the most important situations to work on; recognizing that some problems or situationsare less important and can wait; preplanning before busy periods; organizing the board and using flight stripseffectively to keep priorities straight for handling air traffic situations; quickly and decisively determiningappropriate priorities.

F. Technical KnowledgeKnowing theequipment and its capabilities and using iteffectively; knowing aircraft capabilities/limitations

(speed, wake requirements, size, minimums) and using that knowledge; keeping up-to-date on letters ofagreement, changes in procedures, regulations, etc.; keeping up-to-date on seldom used procedures orskills.

G. Maintaining Safe and Efficient Air Traffic Flow

Al

Reacting to and resolving potential conflictions effectively and efficiently; using proper air traffic separationtechniques effectively to ensuresafety; sequencingaircraft effectively for arrival or departure; sequencing aircraftto ensure efficient/timely traffic flow; controlling traffic in a manner that ensures efficient traffic flow;controlling traffic in amanner that minimizes traffic problems (e.g., conflictions, traffic flow problems) forother controllers and pilots.

H. Reacting to StressRemaining calm and cool under stressful situations; handling stressful air traffic conditions in aprofessional

manner.

I. TeamworkWorking smoothly with supervisors and other controllers in the facility; pitching in and helping other

controllers as necessary; accepting and reacting constructively to appropriate criticism from supervisors or peers;avoiding arguments and interpersonal conflicts with other controllers, supervisors, or pilots.

J. Adaptability/FlexibilityReacting effectively to difficult equipment problems, changes in weather, traffic situations, etc., or to

unexpected actions on the part of other controllers or pilots; using contingency or "fall-back" strategieseffectively when unforeseen/unanticipated air traffic problems emerge or if first plan doesn't work; asking forhelpwhen it's needed; developing/executing innovative solutions to air traffic problems; dealingeffectivelywithsituations for which there may not be clearly prescribed procedures, situations which require novel thinking;adapting to equipment updates, new kinds of procedures, etc.

A2

An Investigation of the Relationship Between Chronological AgeAnd Indicators of Job Performance for Incumbent

Air Traffic Control Specialists

The relationship between age andjobperformancefor AirTrafficControlSpecialists (ATCSs) isanissuethathas been revisited many times over the past fewdecades (Trites, 1961;Trites & Cobb, 1962; Cobb,1968; VanDeventer & Baxter, 1984). Researchershave consistently found a negative relationship betweencontroller age and performance across studiesthat have used different ATCS options (enroute,terminal), career stages (age at entry into training,current age on the job), and criterion measures (on-the-jobratings, academy performance) (Trites, 1961;Trites & Cobb, 1962; Cobb, 1968; VanDeventer &Baxter, 1984). That research provided support forcongressional action that mandated both hiring agelimitsandretirement age requirements for theATCSjob. As more ATCSs reach retirement age and theFAA prepares for renewed hiring efforts, there is aneed once again to explore this issue. According toSchroeder, Broach, andFarmer (1997), the potentialeffects of aging on cognitive functioning, and theconsequences of these changes on job performanceand future training requirements, are important considerations associated with the aging of the ATCSworkforce. The purpose of this paper is to use asample of incumbentATCSs to explore the relationship between entry-on-duty age, current age, andindicators ofjob performance.

In 1972, in an efforttoaddress safety issues relatedto theage-performance relationship, Congress beganrequiring that applicants for the ATCS jobnot reachtheir thirty-first birthday prior to initial appointment (VanDenventer & Baxter, 1984). Congress alsoset a mandatory retirement age of 56, at which timeATCSs are removed from positions requiring thedirect separation and control of air traffic (Aul,1991).These age restrictions arenot without controversy, as there has been much criticism of thesepolicies by those people who are excluded fromemployment (Aul, 1991). Despite the criticism, it ishard to ignore the consistent results found by researchers who have explored the relationship betweenage and performance.

In his study of air traffic control trainees, Trites(1961) found anegative relationship between age andsupervisor rating, which led to his conclusion thatolder trainees were less likely to beviewed as thebestcontrollers. Using criterion measures that assessedboth FAA Academy training and job performance,Trites and Cobb (1962) found a negative relationship between age at entry into ATC training andsubsequent Academy and job performance. Theseresults led to the recommendation that there be amaximum entry age for air traffic controllers. Theunderlying causes of this negative relationship werenot explored by those researchers.

Cobb (1968) used a sample of journeymen radarenroute ATCSs to provide further evidence of thenegative relationship between age and performance,using job performance rating as the criterion measure. Potential reasons for thelower ratings obtainedby older controllers were physiological aging, lower-levelaptitudes and abilities, and/or lowermotivationlevels (Cobb, 1968). Cobb also suggested that thelower ratings for older controllers could have beendue to biased attitudes toward older people.

Cobb and Mathews (1974) explored the relationship between age, air traffic experience, and experimentalratings ofjob performance forterminal ATCSsat high-density airports. As found in previous studies, the results revealed a negative relationship between age and the experimental job performanceevaluations. The results also demonstrated that controllers over age 40 were rated as less proficient intheir jobs than younger controllers. Cobb andMathews were reluctant to attribute these differencesto physiological aging alone. They speculated thatthe older controllers may not have been as highlymotivated asyoungercontrollers, that the oldercontrollers may not have been among the topcontrollersat anypoint in their career, and that thelow ratingsmay also have beendue to age bias.

VanDeventer and Baxter (1984), administered abiographical questionnaire to trainees at theFAAAcademy. The results revealed anegative linear relationship

between age at entry and pass/fail status after initialATCS screening. These results supported the findingsofresearch fromnearlytwodecadesearlier.VanDeventerand Baxter speculated that this negative relationshipmight havea cognitive basis.

The presentstudy revisited the issue ofATCS ageand performance using incumbent controllers andnewly developed measures ofjob performance. It isimportant to note that ATCSs were not tested ontheir own sectors,sotheir actual levelofperformanceon the job remains unknown. A recent Air Traffic-Selection and Training (AT-SAT) concurrent validation study afforded anopportunity to investigate therelationship between age and performance using criterion measures that did not exist for previous studies. One of these measures, a computer basedperformance measure (CBPM), served as a measureof the technical skills necessary to effectively andefficiently separate traffic on the job. Assessmentratings ofjob performance by peers and supervisorsalso served as a criterion measure for the current

study. Based on previous research, both currentandentry-on-dutyage were hypothesized to benegativelycorrelated with ATCS job performance.

METHOD

ParticipantsA total of 1083 Full-Performance Level (FPL)

enroute ATCSs, supervisors, andstaffparticipated inthe AT-SAT concurrent validation study. A minimum of75 ATCSs from each of 12 enroute air trafficcontrol centers (ARTCCs) volunteered to participate. Personnel employed in supervisory or staffpositions were excluded from the present study,leaving828 FPL enrouteATCSs. All participants identified themselves asbeingFPLATCSs atthe timeoftheconcurrent validation; 141had previouslyheldastaffposition and 30 had previously been supervisors.Demographic information is presented in Table 1.Almost halfof the controllers were between theagesof32and 37 (M=36.4, SJD=5.7), and themajority ofparticipants wereCaucasian males. The average number of years in current position was 8.3, the averageentry-on-duty age was 25-0, andtheaverage numberof years as an FPL ATCS was 7.2. Current age waspositively correlated with many work-related variables, such asnumber of years as an FPLATCS andnumber of years in current position. A positive correlation was found between current age andyears in

Table 1. Demographics

Variable Number Percent

Current Age

31 or Younger 144 17.5

32-37 391 47.6

38-43 197 24.0

44-49 59 7.2

50 or greater 30 3.7

Entrv-on-Dutv Age

23 or Younger 252 33.6

24-25 208 27.9

26-27 127 16.9

28-29 107 14.3

30-31 52 6.9

Gender

Male 690 83.3

Female 138 16.6

Race

American Indian 17 2.1

Asian/ Pacific Islander 5 0.6

African American 36 4.3

Hispanic 33 4.0

White 731 88.1

Other 6 0.7

Education

H.S. or GED 68 9.1

Attended Trade School 3 0.4

Completed Trade School 15 2.0

Attended College less than 2 165 22.1

Attended College 2 or more yrs. 179 23.9

Completed College, 2 yr. Degree 55 7.4

Completed College, 4 yr.Degree 225 30.1

Attended Graduate School 38 5.1

astafforsupervisory position. The duration of theseassignments was quite short, as the average amount oftime spent in these positions wasless than 6 months.As shown in Table 1, only 10.9% ofthe participantswere above theage of43. Although this number isnotproportional to theother age groups, it isconsistentwith skewed age distributions reported by previousresearchers (Cobb, 1968; Cobb & Mathews, 1974).Theage of the workforce ismost likely skewed due tothe firing of 11,000 controllers during the 1981Professional Air Traffic Controllers Association(PATCO) strike.

Procedure

Data collection teams were assembled and trainedto conduct testing at each of the 12 ARTCCs included in the study. Each team was comprised of atest site manager and two to four team members whowere responsible for administration of the predictorbattery and criterion measures. All participants werevolunteers in the AT-SATconcurrent validation studywho were recruited through on-site briefings and aninformational memo. Volunteers were tested over atwo-day period in a room provided by their facility.One day of computer-based testing was devoted tothepredictor test,andonedaywas devoted toadministration of the CBPM. Due to scheduling constraints, somevolunteers took all parts ofthe testovertwoconsecutive days whileothers spread testing outover the course of the on-sitedata collection period.

Each volunteer also identified two peers and twosupervisors tocompleteexperimental jobperformanceassessment ratings. The supervisor and peer ratersparticipated in an orientation and training programtoensure validandaccurate scaling. These raters werethen askedto complete the formsandsubmit them tothe researchers.

Measures

The criterion measures used in the AT-SAT con

currentvalidation study also served as the controllerperformance measures in the current study. Both theCBPMand peer andsupervisor ratingscales served asmeasures of job performance that were comparedwith both entry-on-duty age and current age.

Computer Based Performance Measure. TheCBPM served as a measure of the technical skills

necessary to effectivelyandefficiently separate trafficon the job (Hanson, Borman, Mogilka, & Manning,1998). It is a 38-item, 2-hour test where controllers

are presented a seriesof realisticair traffic scenariosand 2 to 5 multiple choice questions pertaining toeach scenario. Each question has 3 to 5 responseoptions representing different ways in which tie airtraffic problems might be addressed. During administration of the CBPM, controllers were presentedwith scenarios and given up to 60 seconds to revieweach one before itwas presented for scoring purposes.Each scenario lasted no more than 5 minutes. Respondents were thengiven 25seconds toanswer eachquestion. Once a response was chosen, controllerswere unable to return to previous items orscenariosto review information orchange answers (Hanson etal., 1998). Hanson etal. (1998) reported acoefficientalpha of .59 for the CBPM, which was validatedagainst hi-fidelityair traffic simulations.

Peer and Supervisor Rating Scales. The peer andsupervisor ratings are behavior-based scales with 10dimensions and one overall effectiveness scale (seeAppendix A). Rating standards, which describe controller proficiency at different effectiveness levels inan effort tomake ratings moreobjective, are providedbelow each of the 10 dimensions. These rating standards were developed as part of theAT-SATconcurrent validation study (Borman, Hedge, Hanson,Bruskiewicz, Mogilka, Manning, Bunch, & Hogen,1999). Raters were asked to read each category definition andrating standard, thencompare thecurrenteffectiveness of the controller being rated with thatstandard. Ratings were made on aseven-point scaleand werelater combined to produce an overall criterion rating score. Asshown inTable 2,theracing scorewas moderately but significandy correlated with theCBPM score. It should benotedthat these ratings werecompleted independently ofCBPM administration.

AnalysesThe linear relationship between age and job per

formance was analyzed using Pearson's product-moment correlation. Analysisofvariance (ANOVA) andpost-hoc multiple comparison procedures were usedto identify significantly different group means.

Although previous research has found a linearnegative relationship between age and performance,an examination of bivariate scatterplots raised thepossibility of a curvilinear relationship. Therefore,hierarchical polynomial regression analysis was usedto assess the form ofthe relationship betweenage andperformance. For each measure ofjob performance,age was entered into the equation first, followed by

theage-squared (quadratic) term. If therelationshipbetween age andjob performance isprimarily linear,then age alone should explain a significant proportion ofthe criterionvariance, and the quadratictermthat is subsequently added should not account for asignificant change in R2. If the quadratic term explains asignificant proportion of thevariance beyondthe age term, then there is evidence of a curvilinearrelationship between age and job performance.

RESULTS

Results are divided into two main sections basedon age classification. The first section contains results of analyses using entry-on-duty age, whereascurrent age was used in analyses described in thesecond section.

Entry-on-Duty AgeCorrelations. As shown in Table 2, there was little

support for the hypothesized relationship betweenentry-on-duty age and performance. Entry-on-dutyage was negatively correlated with peer and supervisor ratingscore (r= -.07, p_<.01); however, this correlation isvery small.The statisticalsignificance oftherelationship is likely due to the power of the test asinfluenced by the large sample size. Entry-on-dutyage wasnot significantlycorrelated with CBPM score.Table 2 also shows a correlation between the CBPM

and peer andsupervisor ratingscore. Table3 containsthe results of Pearson product-moment correlationsbetween age and variables related to work history.Entry-on-duty age was not significandy correlatedwith any of the work-related variables included inthis table.

Prediction of Performance. The relationship between entry-on-duty ageand criterionmeasurescoreswas explored further using hierarchical polynomialregression to predict performance. Two separate regression analyses were performed with entry-on-dutyage as the predictor and CBPM and rating scoresservingas dependent variables. Results revealed thatentry-on-duty age didnotsignificandy predict eitherCBPM or the experimental rating scores. Enteringthequadratic term into theequation did not result inasignificant change in R2, providing no evidence ofacurvilinear relationship between entry-on-duty ageand the criterion measures.

Group Comparisons. Mean criterion scores fordifferent age groups were compared using one-wayANOVA. Separate ANOVAs were performed for theCBPM and rating score, the results of which arepresented inTable4. No significant differences werefound between entry-on-duty age groups for eitherthe CBPM score or rating score.

Current AgeThe results of analyses using current age as the

independent variable provide partial support for thehypothesized relationship between age and performance. Theresults demonstrate that the relationshipis curvilinear rather than linear.

Correlations. The results of Pearson product-moment correlations between current age and criterion measures are presented in Table 2. Thesecorrelations do not support the hypothesized negative linear relationship between current age and jobperformance. Although the very lowcorrelation between current age and rating score was statisticallysignificant (r=.08, p<.01), this is likely due to thepoweroftheanalyses asinfluenced bysample size. Asshown in Table 3, current age was significandy correlated with entry-on-duty age (r=.47, p<.01), yearsas an FPL ATCS (r=.78, p<.01), years in currentposition (r=.72,p<.01), years in previous job (r=.l6,p-c.01), years in staff position (r=.40, p<.01), andyears as supervisor(r=.27, p<.01).

Prediction of Performance. The relationship between current age and performance on the criterionmeasures was explored further using hierarchicalpolynomial regression. Two regression analyses wereperformed with current age as the predictor, andCBPMand rating score servingasdependentvariables.

As shown in Table 5, age did not significandycontribute to the prediction oftheCBPMscore whenentered into the regression equation. However, thequadratic term, or age-squared, did contribute significantly when added (R2=.04), providing evidenceof a curvilinear relationship. This relationship isdepicted in Figure 1, which reveals that the predictedCBPM score increases gradually until approximatelyage 35, before declining after age 42.

Hierarchical polynomial regression also revealed aquadratic relationship between current age and peerand supervisor rating score. As shown in Table 6,only the quadratic model produced asignificant R2change in predicting the overall rating score (R.2=.02).

Table 2. Correlation of Age and Criterion MeasuresVariable Mean S.D. 1 2 3

1. Current Age 36.4 5.7 1.0

2. Entry-on-Duty Age 25.0 3.2 .47* 1.0

3. CBPM 191.2 14.3 -.04 .02 1.0

4. Peerand Supervisor Rating 5.1 .97 -.08* -.07* .25*

*jx.0i

Table 3. Correlates of age

Mean S.D. 1 2 3 4 5 6 7

1. Current Age 36.4 5.7 1.0

2. Entry-on-Duty Age 25.0 3.2 .47** 1.0

3. Years FPL ATCS 7.2 5.4 .78** -.01 1.0

4. Years in Current Position 8.3 5.9 .72** .00 .81** 1.0

5. Years in Previous Job 3.2 2.7 .16** .05 .18** .06 1.0

6. Years Staff Position .45 1.1 .40** .04 .36** .15** .06 1.0

7. Years Supervisor .13 .69 .27** .04 .21** .05 .07 .35** 1.0

*p<.0S** p<.01

Table 4. Group Means and Results of One-way ANOVA of Criterion Measures by Entry-on-Duty Age Group

Entry-on-Duty Age

Job Performance

Measure

23 and 24-25 26-27 28-29 30-31younger

F

CBPM

Peer and SupervisorRating Score

191.72 191.12 190.47 192.53 193.905.19 5.14 5.01 5.04 5.13

.72

1.83

:p<.05

Table 5. Regression of Current Age on CBPMScoreVariable B B AR*

Current Age 4.68 1.82 .00

Current Age2 -.06 -1.88

Adj

.04**

R2=.04

R2=.04

R=.21**

*p<.0l

240

30 40

Current Age

Figure 1. Regression of Current Age on CBPM

50

O Linear

a Quadratic

60

Table 6. Regression ofCurrentAge on Peer and Supervisor Rating ScoreVariable B P AR*

Current Age .160 1.26 .00

Current Age2 -.002 -1.19

Adj

.02**

R2=.02

R2=.02

R=.15**

*p<01

° Linear

° Quadratic

Current Age

Figure 2. Regression ofCurrent Age on Peer and Supervisor Ratings

Table 7. Group Means and Results ofOne-way ANOVA ofCriterion Measures byCurrentAge Group.

Current AgeJob Performance

Measure31 and 32-37 38-43 44-49 50 or Older

youngerF

CBPM

Peer and SupervisorRating Score

189.10 192.51 193.36 190.54 178.544.98 5.10 5.22 5.44 4.75

7.33**

6.89**

p<.01

This relationship, plotted in Figure 2, suggests thatpeerandsupervisor ratingscores begin to decrease atapproximately age45.

Group Comparisons. A one-way ANOVA wasperformed for each criterion measure to determinedifferences in mean scores based oncurrent age. Theresults of these analyses, which are summarized inTable 7, show significant age differences for each

measure ofcontroller performance. Tukey post-hoccomparisons were performed to determine which oftheage groups differed significantly interms ofscoresand ratings.

Results revealed that the mean CBPM score of178.54 for controllers who were 50 or older wassignificandylower than CBPMscoresfor controllersof theotherage groups (p<.01). Figure 3 shows that

OuCO

mo

CO_c

o

ccd<D

200

190

180

170

31 or Younger 32-37 38-43

Current Age

Figure 3. Mean CBPM Score for Current Age Group

c

OCO

£CDQ.Z3

CO

cco«_

CDCD

Q.

(0CD

44-49

31 and Younger 32-37 38-43

Current Age

Figure 4. Mean Peer and Supervisor Ratings forCurrent Age Group

44-49

50 or Greater

50 or Gi

mean CBPM scores gradually increased with ageuntil peaking for ATCSs in the 38-43 age range.Mean CBPM was lower forcontrollers age 44-49, thendropped significandy for ATCSs age 50orolder.

As shown in Table 7, significant age differencesalso existed for peer and supervisor ratings (F=6.72(4, 744), p<.01). Tukey Post-hoc comparisons revealed that the mean rating score of 5.44 (SD=.65)for controllers age 44-49 was significandy higherthan the ratings of controllers ages 31 and younger(M=4.98, SD=.60), 32-37 (M=5.10, SD=.72), and50 and older (M=4.75, SD=.87). Furthermore, meanpeer andsupervisor ratings for controllers age 38-43(M=5.22, SD=.68) was significantly higher than theratingsgiven tocontrollers age31andyounger (M=4.98,SD=.60) and 50 and older (M=4.75, SD=.87).

Themean peer andsupervisor criterion ratings foreach age group areplotted inFigure 4.Mean peer andsupervisor rating increased foreach agegroupbeforepeaking for ATCSs between the ages of 44-49. Asdiscussed above, the decrease in mean rating forATCSs ages 50 and older isstatistically significant.

DISCUSSION

Theresults ofstatistical analyses, particularlyhierarchical polynomialregression, provide evidence ofacurvilinear relationship between age and ATCS performance. An examination of thelinear reladonshipbetween age and job performance generally yieldednon-significant results. However, the results of linearanalysesdonotprovideacomplete understandingoftherelationship that exists between age and performance.

Entry-on-Duty AgeInclusion of entry-on-duty age in this studypro

vided an opportunity to investigate the existence ofdifferences in performance thatare based on the ageat which current FPLATCSs entered duty. Results ofthe analyses described above demonstrate no relationshipbetween entry-on-dutyageand performanceon the CBPM. A small, but statistically significant,negative correlation was found between entry-on-dutyage andpeerandsupervisor ratings. TheANOVAyielded non-significant results; there were nosignificant differences in CBPM score and rating score forpeople who entereddutyatdifferent ages. Theresultsof regression analyses failed to demonstrate a relationship between entry-on-duty age and CBPM andratingscores. Since the sample ofATCSsusedin this

study was restricted by the hiring-age limit, no assumptions can be made regarding the potential performance levels of people hired over the age of 31.However, it should be noted that earlier researchfound differences in ratings when the age range wentto39years (Cobb, 1968). The current sample isalsorestricted in terms ofATCS performance level, sinceit is comprised of only those people who succeededthrough training; people who failed training are notcurrendy employed as controllers.

Current AgeThere is strong evidence to suggest that the rela

tionship between current age andjobperformance isnot linear. Although previous research revealed anegative linear relationship (Trites, 1961;Trites &Cobb, 1962; Cobb, 1968), such results were notfound in the current study. Analyses reveal that therelationship between current age andperformance is,in fact, curvilinear; performance scores increasedwith age until ATCSsreached their mid 30s. Performance then leveled off until ATCSs reached theirmid 40s, whenscores dropped.

Evidence of a curvilinear reladonship betweencurrent age and performance was provided by theresults of hierarchical polynomial regression, as wellas analysis of variance. The patterns of relationshipthat were found between current age and the twocriterion measures are similar, which is importantgiven that theCBPM represents anobjective measureof performance on air traffic-related tasks and wasadministered independently of the more subjectivepeer and supervisor ratings. The reported R-squaresof .04 and .02 indicate that current age does notexplain a large percentage of the variance in controllerperformance. Theseresults,however, do reveal thetrend of the relationship between these variables.Predicted CBPM score increased with age beforepeaking for ATCSs in their mid-30s through mid40s. At approximately 45 years of age, predictedCBPM score begins to decrease and drop to a levelbelow that predicted for the youngest and newestcontrollers. Asimilar pattern was found when usingcurrent age to predict peerand supervisor rating.

Plotting the mean CBPM and rating scores forcontrollers of different age groups provides anotherillustration of the reladonship between current ageand performance.Mean CBPM scoresincreasedwithageuntil reaching their highest point forcontrollersin the 38-43 year age range. The mean score then

decreased for those age 44-49 before dropping significantly for those 50 or older. The mean CBPMscore for controllers 50 and older was significandylowerthan that ofcontrollers from allofthe other agegroups.Peerand supervisorratings are highestfor theATCSs between the ages of 44 and 49, before dropping significandy for ATCSs age 50 and greater.

The results of the comparisons of current ageandmeasures of ATCS performance provideevidence ofa generaldeclinein performanceforcontrollers oncethey reach their late 40s. The regression lines describedabove indicate that the drop in performancelevel begins to occuraround age 45- The comparisonsof group means clearlydemonstrate that the level ofperformance, asmeasured by theCBPM and experimental rating score, is significantly lower for thosecontrollers age 50 and older. Since the upper agerange is restricteddue to mandatory retirement, theextent to which this downward trend would continue

for older controllers is unknown. One caveat with

these findings is that the number of people includedin this study who were 50 or over is much smallerthan the number of people from other age groups,suggesting that the mean criterionscores for the olderATCSs may not be as robust.

The findingthat, asa group,oldercontrollershavelower ratings ofjob performance is consistent withfindings of previous researchers who have exploredthis issue. However, previousstudies havefound thisto be a negative linear relationship, whereas thecurrent study providesevidence ofa curvilinear relationship. Scores on the current measures ofjob performance actually increased with age until ATCSsentered their late 30s, indicating that job performance increases withairtraffic experience. Thestrongpositive correlation between current age and workexperience indicates that younger controllers had lessexperience than older controllers. Given that moreexperience on the job should result in increasedskilland learning, it is not surprising that people withmore experience would have higher performancescores andratings. Asreportedearlier, however, thereis a point at which scores on these job performancemeasures no longer increase, and actually decrease,with age and despite experience. Inthe current study,performancescores decreased to the point that peopleoverthe age of50 performedata level that was lowerthan that of younger controllers who had much lessjob experience.

Limitations of the StudyThere are limitations to this study that must be

considered when reviewing the results and conclusions that are presented. All of the ATCSs whoparticipated in this study werevolunteers. It is therefore likely that controllers uncomfortable with theirlevel of performance chose not to participate. Thismay be particularly likely for older controllers, sincetheir average level of performance was found to belower thanthatofyounger controllers. Consequendy,theolderATCSs includedin thisstudymayrepresentthebetterperformers from that particular age group,resulting in an under-estimation of the age-relateddecline in job performance.

Although the CBPM used in this study providedvaluable information about ATCS's ability to performsimulatedtasks, it doesnot provideevidenceofactual job performance since the ATCSs were notperforming tasks using the airspace they work withon a daily basis. The experimental peer and supervisor ratings, however, did relateto actual performanceon thejob.Thecross-sectional approach of thisstudyalso creates limitations. Conclusions regardingchanges in ability are based on the assumption thatATCSs possess skill and ability levels that are relatively equal following training. Consequendy, lowerperformance levels among older ATCSs would beattributed to declinein abilityrather than to a lowerlevel ofability throughout their career. Onemitigating factor that must be considered is that the bestperformers are often promoted into supervisory positions. Thismeans that theaverage performance levelof oldercontrollers may be lower because the "best"ATCSs of this age group are no longer activelycontrolling traffic. A longitudinal study of age, sothat individuals aretracked throughout thecourse oftheircareer, isthe onlywayto determinethe extent towhich performance levels change over time,aswell asthemanner in which this change occurs.

CONCLUSION

Overthe years, researchers have speculated aboutthe possible causes ofthe decline intraining andjobperformance scores with age (Trites, 1961; Trites &Cobb, 1962; Cobb, 1968). Themost likely cause ofthe decline in job performance measures is an age-related decline incognitive ability. Such adecline has

10

been found in other studies, including those thathave focused on cognitive decline in older pilots(Hardy & Parasuraman, 1997). Cobb (1968) speculated that lower performance ratings could have beentheresult of,among other interpretations, age bias onthe part of the rater. The current study does notsupportthatoption: Controller'sscores on theCBPM,which is an objective criterion measure, were positively correlated with peer and supervisor criterionassessment ratings. The pattern of relationship between current age and performance was relativelyconsistent across all criterion measures. Such a consistent pattern wouldnot be expected ifeithertheCBPMor the ratings were inaccurate orbiased.

As the ATCS workforce ages and reaches retirement age, the actual job performance ofolder controllers becomes amore important issue. The presentstudy provides experimental evidence that controllers over theage of 50 perform, onaverage, atalevelthatislower than thatofyoung controllers withlitdeexperience. Theyalso perform, onaverage, ata levelthat issignificantly lower than ATCSs who are just afew years younger. Although only asmall percentageof thecontrollers in this studywere age 44orolder,it is only a matter oftime before the bulk ofcontrollers reach this age. Almost one-halfof the ATCSsincluded in this study were between the ages of 32-37. In 10years rime, the majorityof these controllerswill begin entering the age range atwhich predictedjob performance level begins to decline.

Since work experience and current age are highlycorrelated (as shown inTable3), it is not possible todisconnect the effects of these two variables. Thecurvilinear nature ofthe relationship between ATCSage and performance scores suggests that the youngestcontrollers are notnecessarily the best. Theregression lines shown in Figures 1and 2suggest thatthereis a ten-year span of time, between the ages of approximately 35 and 45, in which ATCSs are at theirpeaklevel ofperformance. It seems thatATCSs needto have a certain amount of work experience, or"seasoning," before theyreach this peak performancelevel. The extent to which older controller's performance would have declined had theynot had at least10 years work experience, is unknown. It is likely,however, that they would not have peaked at alevelas high as other controllers before experiencing age-related decline.

The mandatory ATCS retirement age means thatthe FAA loses its mostexperienced controllers whentheymayno longer bethebestperformers. However,this does not necessarily mean that they are no longercapable controllers. This raises questions about thelevel ofATCS performanceconsidered"good enough"to continue separating traffic. Thus far, there is noapparent answer to this question. ATCSs work in anenvironment where there are sector differences andworkload variability. Since changes in sector orworkload may affect controller performance level,additional information is needed to better understand how they operate in sectors with differentworkload demands and conditions. It isimportant tounderstand whether ornotexpertise helps older controllers cope with novel or demanding situations asthey control traffic. Another question that ariseswhen considering the mandatory retirement age iswhether or not modifications to the ATC work environment would compensate for possible age-relateddeclines in performance. It should also be noted thatnot all older ATCSs exhibit a lower performancelevel; in fact many over the age of 45 achievedperformance scores higher than those attained byyounger controllers. Whenviewed as a group, olderATCSs are found tohave lower levels ofperformance.

Since ATCS work in asafety-related occupation,the importance of understanding the relationshipbetween current age and job performance cannot beover-emphasized. As has been the case with earlierstudies, thecurrent study cannot provide definitivereasons for the decline in performance scores thatoccurs for older controllers. To have a better understanding of this dynamic, it is important that thisissue be studied in a more comprehensive manner,where age and performance are the specific researchissues under investigation. Anopportunitytoconductalongitudinal study ofage and performance exists nowthat the FAA plans toincrease its hiring efforts over thenext several years. It is recommendedthat new ATCSsbe tracked throughout their careers, with job performance being measured periodically tostudythemannerinwhich performance changes withage.

11

REFERENCES

Aul, J.C., (1991). Employing air traffic controllers. InH. Wing &C.A. Manning (Eds.). Selection ofairtraffic controllers: Complexity, requirements, andpublic interest (pp. 7-12). (DOT/FAA/AM-91/9).Washington, D.C. Federal Aviation Administration Office of Aviation Medicine. Available

from: National Technical Information Service,Springfield, VA 22161; order # ADA238267.

Borman, W.C., Hedge, J.W., Hanson, M.A.,Bruskiewicz, Mogilka, H., Manning, C, Bunch,L.B., & Horgen, K.E. (1999). Development ofCriterion Measures ofAir Traffic Controller Performance. (DOT/FAA/OAM Technical Report).Manuscript under review.

Cobb, B.B. (1968). Relationships among chronological age, length ofexperience, and job performanceratings ofair route traffic control specialists. Aerospace Medicine, 39, 119-24.

Cobb, B.B., &Mathews, J.J. (1974). The relationshipsbetween age, ATC experience, and job ratings ofterminal area traffic controllers. Aerospace Medicine, 45, 56-60.

Hardy, D.J., & Parasuraman, R. (1997). Cognitionand flight performance in older pilots. Journal ofExperimental Psychology: Applied, 3(4), 313-48.

Hanson,MA, Borman,W.C., Mogilka, H.J., &Manning, C. (1998). Computerized assessment ofskillfor a highly technicaljob. Unpublished manuscript.

Schroeder, D.J., Broach, D., & Farmer, W.L. (1997).Current FAA Controller Workforce Demographics,Future Requirements, and Research Questions. InR.S. Jensen & L.A. Rakovan (Eds.) ProceedingsofThe Ninth Annual International Aviation Psychology Symposium, (pp. 135-41). Columbus,Ohio: The Ohio State University.

Trites.D.K. (1961). Problems in airtraffic management:I. Longitudinal prediction of effectiveness of airtraffic controllers. (DOT/FAA/AM-61/1). Oklahoma City, OK: FAA Civil Aeromedical Research Institute. Available from: National Tech

nical Information Service, Springfield, VA 22161;order #AD268954.

Trites, D.K. & Cobb, B.B. (1962). Problems in airtraffic management: III. Implications of age fortraining andjobperformance ofairtraffic controllers. (DOT/FAA/AM-62/3). OklahomaCity, OK:FAA Civil Aeromedical Research Institute. Available from: National Technical Information Ser

vice, Springfield,VA 22161; order# N62-10353.

VanDeventer, A.D., & Baxter, N.E. (1984). Age andperformance in air traffic control specialist training. In VanDeventer, A.D., Collins,W.E., Manning,C.A, Taylor,D.K., & Baxter, N.E. Studiesof poststrike air traffic control specialist trainees:I. Age, biographical factors, and selection testperformance related toacademy training success.(DOT/FAA/AM-84/6). Washington, D.C. Federal Aviation Administration Office of AviationMedicine. Available from: National TechnicalInformation Service, Springfield, VA 22161; order # ADA147892.

12

1. Report No.

DOT/FAA/AM- 99/18

4. Title and Subtitle

An Investigation of the Relationship Between Chronological Age andIndicators ofJob Performance for Incumbent Air Traffic ControlSpecialists

7. Authors)

Heil, M.C.

9. Performing Organization Name and Address

FAA Civil Aeromedical Institute

P.O. Box 25082

Oklahoma City, OK 7312512. Sponsoring Agency name and Address

Office ofAviation Medicine

Federal Aviation Administration

800 Independence Ave., S.W.Washington, DC 20591IS. Supplemental Notes

16. Abstract

2. Government Accession No.

Technical Report Documentation Page3. Recipient's Catalog No.

5. Report Date

June 1999

6. PerformingOrganization Code

8. PerformingOrganization Report No.

10. Work Unit No. (THAIS)

11. Contract or Grant No.

13. Type of Report and Period Covered

14. Sponsoring Agency Code

Over the last few decades, researchers have consistendy found a negative relationship between theage ofAirTrafficControl Specialists (ATCSs) and both training success and ratings ofjob performance (Trites, 1961; Trites & Cobb,1962; Cobb, 1967; VanDeventer & Baxter, 1984). As more ATCSs reach retirement age and the FAAprepares forrenewed hiring efforts, there isaneedto onceagain explore this issue. According to Schroeder, Broach, and Farmer(1997), the potential effects of aging on cognitive functioning, andthe consequences ofthese changes on jobperformance and future training requirements, are important considerations associated with the aging ofthe ATCSworkforce. The present studyrevisited the issue of ATCS age and performance using incumbent controllers andnewlydeveloped measures ofjob performance. A recent AirTraffic-Selection and Training (AT-SAT) concurrent validationstudyafforded anopportunity to investigate the relationship between age and performance using criterion measures thatdid not exist for previous studies. One ofthese measures, acomputer based performance measure (CBPM), served as ameasure ofthe technical skills necessary to effectively and efficiendy separate traffic on the job. Assessment ratings ofjobperformance by peers andsupervisors also served as acriterion measure for the current study. Results ofANOVA andregression analysis revealed that, on average, older ATCSs received lower scores on measures ofjob performance.

17. Keywords

AirTraffic Control Specialist, Job Performance, Age,Personnel Selection

18. Distribution Statement

Document is available to the publicthrough theNational Technical Information Service,Springfield, Virginia 22161

19. Security Classil. (of this report)

Unclassified20. Security Classif. (of this page)

Unclassified21. No. of Pages

17

22. Price

Form DOT F 1700.7 (8-72) Reproduction ofcompletedpageauthorized

i 3

r

• • •". VOLPE NAT2CNAL'i SYSTEKS r ,

DGE. HAi" . "142

NCE CEN