AFHRL-TR-76-9

AIR FORCE BACKGROUND AND INTEREST MEASURES AS PREDICTORS

______,_____ _ OF SUCCESS IN UNDERGRADUATE PILOT TRAINING

IHuByU Nancy Guinn

Bart M. VitolaSandra A. Leisey, AIC, USAF

AN; PPERSONNEL RESEARCH DIVISION

Lackiand Air Force Boe lexas ý3236

May 1976

U•It

SS LABORATORY

AIR FORCE SYSTEMS COMMANDBROOKS AIR FORCE BASETEXAS 78235

NOTICE

When US Government drawings, specifications, or other data are usedfor any purpose other than a definitely related Governmentprocurement operation, the Government thereby incurs noresponsibility nor any obligation whatsoever, and the fact that theGovernment may have formulated, furnished, or in any way suppliedthe said drawings, specifications, or other data is not to be regarded byimplication or otherwise, as in any manner licensing the holder or anyother person or corporation, or conveying any rights or permission tomanufacture, use, or sell any patented invention that may in any waybe related thereto.

This final report was submitted by Petsonnei Research Division, AirForce Human Resources Laboratnry, Lackland Air Force Base, Texas78236, under project 7719, with Hq Air Force Human ResourcesSLaboratory (AFSC), Brooks Air Force Base, Texas 78235.

iThis report has been reviewed and cleared for open publication and/orpublic release by the appropriate Office of Information (01) in

-accrdance 'vith APR 190417 and DoDD 5230.9. There is n: cbjectionto miinhied distribution of ilis report to the public at large, ot byDIDC to tme National Technical Information Service (NTIS).

STbi: technical report has bten r-ekvwed and is approved.

1I•I.AND 1) BROKAW, Ttchitica Director~uuIRoeeareh lMlv~son

Ai Appived for publiii',on.

V)AN D-. FULGHAM. Coloitel, USAF

i'1

*1

$CRIYCLASSIFICATION OF THI4S PAGE Dat~e, 0.1 rif"d)

REPOT DCUMNTATON AGEREAD INSTRUCTIONSREPOT DCUMETATON PGE _______ BEFORE COMPLETING FORM12. GOVT ACCESSION NDOV-CIPIENT'S CATALOG NUMBER

fACKGROUND AND NTEREST ~ASURES ASSREDICTORS ina}-OF SUCCESS IN UNI3RG RADUATE PILOT TRAINING, '7

______ _____________________________________28__ ________-__ De_____I__________

0. CONTRACT OR GRANT NUMBER(s)

It9 PEVF('Rm.'.sGORGANIZATION NAMAE AND ADDRESS 10. PROGRAM ELEMENT. PROVST TSrch DiisionAREA & WORK UNIT NUMBERS

AAir Force Humi Rescuices la boratory 673J la ~~ckland I~ir Fwice Base, Texas 78236 7100

III ONTIOI INCFFICE NAME AND ADDRESS

IHQ Air 'otte H~umana Resources Laboratory (AFSC) // May 076I trooKs Air Force Base, Texas 78235'AE

76 j~77/ Q~~Z' -7/ Unctassified

IS DISTRIQUTION SYA', FUENT (. ht kopwt

Appioved for publicz-4a-distrtbution wiliu~tciI

tmdq*wj~d a~nd iattiluw invaototici-

-'>A sian 14a "'f 593 I-Atot Itilhict in fim () 141I1i4nltt school wete 11wDr~~c Sthe ora Vocatimpl hlem~ilWa.114 7rwJ the Ofimtlkcopqi bt aJ.nd Atiuirdiiu Swvey. 1itwi 11oftwonm in jilto tt~lhSg waos twloolto*CII wOldert to "wýa thwefe~citwoll 4c blof t aj'ik~, mlittdiag,~ aud intewil m ~IIlo pted~leinp ritot lvzlrtlftý cltci:4.Using~ ilCml (out~~S cuntim C ýi1lC kys wete dcultntcd (tomt 0hw %uTVY date. rej'ssh'. u~alyMs wa~s utii~tcrd towkcte Vainous twvtkla of ImVk-104 to IV w~eid as posawbke Owb Sakoion 0wqxWei QI~. kVsinr thte orlpistl ck~offscote (Of PSC~ Modeli I, '46 ;4o of 11w ounimes %gere ~C01'y ideiitlied 'and ml 1 IV.iod-rt (i 1w Rcdullejttoup iflct4teci:ly Ib~ckd L 1xilecuital falluws. For Nkodc 2, IS 14.otwmn the clitoirce C~ps~oil was wkotifed 33

Ni 3ia imAt~ltin along'ith wmi pelcertt of 1he %Umu rd~e.wt od ,A iw fh uro

DO 473 covwuor vOS IS 09OE'0.C1Ke

SECUR I TY CLASSIFICATION OF THIS PAO E(II7he Data Enter"d)

Item 20 Continued:C,,.• -•>were labeled as potential fatlures along with 20 m of the graduate group. The possibility of increasing the

accuracy of prediction by using non-cognitive data and the operational usefulness of the composites are discussed.

oI

Ii.o

*1,, , - ..

I÷ ______________

PREFACE

This work was performed under Project 7719, Ai Force Development of Selection,Classification, Assignment, Performance Evaluation, Retention, and Utilization Devices;Task 771909, Development and Validation of Specialized Procedures to ImprovePersonnel Classification and Assignment.

Speial appreciation is expressed to Mr. Henry Clark and AIC S. Kay Wilson of theComputational Sciences Division, Air Force Human Resources Laboratory, for theirprofessional and technical assistance in the planning, programming, and accomplishmentof desired analyses for this project.

C I0

TABLE OF CONTENTS

Pge

1. Introduction and Background 5...............................5

II. Method .............. ......................................... 6

Ill. Resuls and Discussion ............ .................................. 8

Construction and Cross.Validation of Keys from Biographical/Attitudinal Data .... ...... 8

Validity of SVIB Data and AF SVIB and OBAS Keys against Pilot Training Critciia 8Selection of Optimal Selection Composite ........ ........................ 10Identification of Pilot Training Eliniinces ........ ........................ 11

IV. Condiusions and F.econunendations. ... . ............................ 15

References ........ ... .......................................... 16

Appendix A: Development of Experimental Scores ........ ....................... 17

LIST OF TABLES

TRWl PawI Attritiot• Rates in UPT by Source of Counitssion bor 1971 - 1973 ............... 5

2 Types of'Traning Ebininatiio for Students ui UPT 1971 - 1973 ............... 6

3 Tiaining ()ulcou of Sample Pupulatioo by Trwitiog Kww s.. .. .. .. .... ..... . ....... 74 Con|jksitIa|i of Crtterion Calegorte.s 7

5 Co|nqovwc and Cfr-Validities ofOBAS Keys fot MIT" C'iteria ..................... . 8

6 Cortilatios Iktween SVIB StkosIAF Keys and V-lauiiatitntn Uia " , ..........

7 Cumulmativc Multipk R atil R1 foy 17 SVIU "-S ..ks ...................... 10

8 ,Sautaofy ( ploatwi oalys. . I..............................I I

S|Wcditie Vics A.tuA Outt•ewit ftot 1hre MWidel of i1b PMot Solectiou compoite . . .. 12

10 DUcskcip4-"% S-arls•sti and 1'etwxu.av (4 WTE Tmainees Atwv and WoluwCutotff P41i4t on M1ot Sekthnt Cnotpxeitt, MODIEL. I by Ttzliing O(utco. .......... 13

I I k•cwiptive Sttslai wld lN-cntage of U1I't Tlnes A6tiv' and WletowCutoff 18tiot kvi iPtlt Sekctiw Conipxslt MODUl 1 by Ti i~inug Outcvuic. .. .. ...... 14

12 Ww!Igiv SAtto vid NievotW (f Atww wd WowCutoff POWntn Wot Sck-f~ioa Cotnpnate M4OD)EL by T#2Wur oulcae . .. .. .. .... 14

3

•BACKGROUND AND INTEREST MEASURES AS PREDICTORS OF SUCCESSIN UNDERGRADUATE PILOT TRAINING

1. INTRODUCTION AND BAM(GROUND

Over the years a great deal of attention has been focused on decreasing training costs by developingtechniques to select individuals foi training programs who are most likely to succeed. Today with thestringent budgetary limitations imposed by Congress on rinlitar spending, more emphasis titan ever hasbeen placed on the armed services to find ways of maximizing the return on their training investment. Themost costly type of training in the Air Force is the training of pilots. Although the cost of undergraduatepilot training varies depending on which costs are included or excluded in the specific estimates, the costper pilot graduate with adjustment for attrition has been placed at S, 1,829.' This estimate does notinclude the addition of costs by Headquarters USAF to cover depot rraintenance and central!y fundedreplenishment spares. The average training cost directly attributable to each elimination fromundeigraduate pilot training (UPT) has been estimated to be approximately S16,082. With costs of thismagnitude, any selection program designed to identify potential elintinees could result in corsiderablesavings.

A review of attrition rates reveals that the winiber of training eliminations in UPT varies by source ofconuilission. As shown in Table I, UPT trainees from Officer Training Schooli (OTS) represent the largestpercentage of elilLnecs while trainees from the Air Force Academy (AFA) have the lowest washout rate. Afurther look at the causes for attrition of OTS pilot trainees indicates that the two main reasons foIeinmination are flying deficiency and self-initiated (i.e., voluntary) elimination (Table 2). An investigationof the factors related to flying deficiency might include, in addition to thle traditional aptitudes tequired forpilot trainees, motivational factors such as the desire to becone a rated pilot anl/or an expressed interest inflying. The wlf.initiated elimination (StE) appears to reprwsent a distinct motivational comiponent since thistype of elintination ii initiated by the student himlself rather than instructor pVot or superior, It wouklwelm that the correlates of belhavior relatwd to thes two mail% causes for elimination should includenon•cognitive/non~inteliectual factors in addition to cognitive abilities, and that the use of non.cognltihedata in tile selection of applicants for pi1ot tIrtaing would make the election ptocvss more cust.efkctiv,

In Navy research, the usofulness of hioglaiAlical Itud interest inwiltories haw also been studioed quiteextelsively (Abrahlius & Neumann. 1971a, Ahrhartus & Netummm, 197.3. Abrahians, Ne-tanoi, & Doun,061), AbIa411uni, Neutrn-mo, & (1t01t"s. 196-S. Wit & Abrcalrit, 1970. 196.i 0, I9()i; N•-wnln. C-'thims,&Alitobsits. 1%)t S'*ks deel tfonk lthe strong Vocaiminal t"terest Wzrik (S'If) Wete foluld to Ihe

vAid in ott Feutf; cig ly rrwiiliatiouAl daicrtruccs fiteii the Naval Aczndenriy uustdsng kthe c.areera runovtlmitlasrwng NRf 'Mt' applic-ntA, anid Iredicting NRO1) offit icermettton (ANohon is & Nounium. 1971h.Abrhtaanm & 'Keutwuit. 1173, Alal , NKimati, & 0311. 169. Abwh.ms, Ncuuntan. & QItwns.W04. t~I serhc ktvn±mntrn asat odtohe sf Ieitot itctive validaty f(Or tho NROWt

seleetloo I4ugranr: howtse,, reeardh usig h aplul ad 00 terview data to idhntil4 4ttnitk-•s5 ftom 11eNaval Acadeiiy indicated il~t initeresi dau t;* frOl e SVWu~ (wa fatv otaddive(i~atn &o Abraluits1970; Nenuman, Gl-ws. & Abcaha's. 1%97).

7~*~ . itigiisHaesIn UP'by ýýoitvu9721971 - 1"'

5:{1 4 12 2 0 ,A S A 9 ,1 1% 1 11, 9 11- 28 .11 %•-

103 22.1% 360%. 14.414 ,,7.,'

(Or1 A~t- (ot i' 73 IwrIt1cA bW t-4 AW-CIA049.Nmm~ 1911k. rhcw co~ics tE', e cs Itt ~

.4

"5~

Table 2. Tyt-es of Training Elimination for Students in UpT 1971 19736

1911972 1973

Type Elimination N NN

Hlying DC fiLiency (F LY) 562 49 484 51 297 51Academic Deticiency (ACM)) 17 2 4 0 t0 2Medical Elimination (MED) 143 12 116 12 539Manifestations of Anxiety (MOA) 1 2) 11 10? 11 53 9Self-initiated Elimination (SIE) 295 25 222 24 159 28Fatality 50 2 030(Abhet 7 1 13 2 7 1

Totl Eliminated 1,158 100 948 100 580 !00

aL)aca turnished by Hq Alt, Training~ Aiulysis Division.

hi previous Air Force research, it was generally concluded that biographical arid interest data wasplositively related to successful muilitary oulcoriws (Muffins, Ewing, & Woods, !962; Taylor, Murray, &lEllison, 1971 - Tupes, 19)52, 0Q57. Usdin & Shenk, 1973). In the study by Tayloir et al. (0971), it wasrecomminended that further research oin the aevelopinent of riotivata19nal assessuient techniqueIs for pilott'alning using biogralphical intormiation be hintiated. When a positiv.: relationship was found betweenbuclkgtoundjintevest mecasures and expressed career intent, L6-din and Shenk (1973) also recommended thatthe study of the overall uitility of background and intterest measures; be extended to. inclule valdating these

intunnsaginst training critezia.

Therefore, tht ptirpose of the curre-nt study was to (a) develop btngrphajici anid interest keys for anAir Vorce sampl~e of 018 pilot traint-cs with tilftleret trulitiog outcouvs;, (b) as~sz the validity ofbioglaplica~l Will interest keys devceloped aptaist thec tlitelion ul eXpressd Coretr ifitet for p4nt tialflingcriteri4, and (,;) vAldate mid asses thc valud of owl-4AVipsivCr dau a PIC4pehein g ut emnloinpdui Ulrainatr

A total tit 54.1 offi~vt3 who entetes) I'll Itt14F1l10 A 04) 1011 1tov~lnW h

samplek puia14ort. Umes offltcets hod rceelwr their colwldm'011n by cv'ili'lkiing t1*0 12-2wcek OTS. I'vog'VoIi(tit colkleg dtzAtei du~rtg, the latter patt 4 1971. %hile 4n the officer usalsog 1xvipat st ddsi~id AP11I,'kesas, fthes sudetU wti &o ndsn dkicd the SVIU am) si Ofti~xt luvq s Wpka a Attitudinal SwuveyfOIJAS).

i1tt SOWi is ;U) ellpý-Uail skoved inventoly ba$44 410 11ttC111s Of bkkes 9.44 44140%s 01*4Al to~sh ..,aoeg %'aiO~l ~eopatosas Th tte Sr 4 tuA--tpalti4131 keys And five oiv~dlenWmtaikey. The

9.0Xk i% o II (tetn invemotoy whlii6 coilains 1414roidan atutudirial Ites ncu 4u41 dintre

resqondent if) inklice the itrnptwan 0( cesurin Cwtrees 41ocds Ud fth extcnt to wl~di ths troods C40 Nefilfi~ld in an Air Vottv job.

Wiqvxw data from these wtest measurs where k-eypkincwl wd nt1wta4cerrd to 01V(RS~ fils 014anayvis.A ~h (tom the two hiventories aloj wmith aj4iudo se-oics retrtiewd frots the aoffie UaPC C&S~ omwixpse the

predictor roasur-Utel W in the UnalSis 1421" CA' t11S sn~fevi.Ctijienmn &-tu ftotv UVT was also or nc-4 (tom the 01mee tapk, ftkersaithe by 10m

Cnn~stat 'k-a 'ine ~vsni irVwee Itunalao lkswmsrc Lahoiatey (AI'IIR~L). LIeldand Alt Votee"4se, Tessa. Criteria vuesiIn the analyi43 were wkedett Qfter 2 ltjt WInSl V COf tfe difffient wtr~aing

owconrws in the u'r'1Me prssltirw.~dl 04Cilrirw releCted1 tratM~n1Ci1 A diffentatr~ilAtiti#,M C41 "y txving

wtupitril to t1V Ak'XU group. The flnmnbet 4, C4ses in elac 4tumotiW om irsacpt~ y during the threet2=of Ur awe shown in TaWl 1l. Umd~ wi thms Uae uposale rnst~yss by triaLrSc phaw w=r ri

feuibl6

Table 3. Traninkg Outcome of Sample Population by Training Fbnse

Twuintag PhtaseTraninsn Or-cam.

Claaulllcraaon T-41 T-37 T.32 Toaln

Attrition CategoryAcademic Deficiency (ACAD) 3 1 4MAilitary Training Deficiency (MTD) IIFlying Deficiency (FD) 43 28 33 104Medical Deficiency (ME!))51 11

Sel-Iitite r-imnaton(SIE) 1 566*1Fear of I'lying/Manifestation of Anxiety (MO A) 3 131g

Ot(her (OTHEiR)22Total Number of Ufl Attritlons20

&wca Category*Total Number of UP'1 Graduates (GRAD)36

Total Ufl Sample 543

Final sekuciton us a criteuton was based on the hypothesized rekovance of thle criterion to thec researchobjectives of tile numbler of subjects available in one or inurfe of the attrition categories. As large a nutlberof subjects as possibll in the criterion classfication was desirable to permilt meantingful anllyses ant) toincrease the stability in key &-kveiopurint irid probabilitly of obtainaing significant results. in thlecwnm-apphication anayvrs. Criterion Group I included !ill types of elianinees ant) tluaw who graduated.Criterion Gs5oup 24 was ctnnposed of IJVT flying deficiency elinlinem andi succvssN graduates. Cit ernCloutp I conusalcd those who elected to eliminate theinuelvecs (SIV) and UKPI pduvcws; Ctilteion Gf oup 4includcJ 1Omit. c~lasiid as usniVAtlonal mofloruns (SliP oad MOA) klanig witl% the suco-ss group. All fourcriterion gioupi wowrtw sd in withem Irnialysis phae. trite-ittn Uwtuji I w"as- toed &s Owvzi

ana~sc~. Twk 4dultwots O cosostionof aitetiowt categriteA.

Tubk 4. Cuniposit~ of (ACsicdon CAtcgVcri

NWOkio AtAO MID% VO UfO St riCA 01141' Cki

4466 X X X

Tile ua'ilpk pJndvuaunnI-A waso soeno the stirani.a SYIII tacal &Wt keys dOWLAoped pmvimally by he,Alir Voto (Sihcnk-, 19M3 wegt. spj~tA tQ the irprarce~'sb ty m o~ntsf fth (TAS. tkfoatihinr =ndwcithtt vsed in) gnente tk4w wwos are contaied in Appcndhz A. Two adliliitnul keys, SVW~iknt andDUANIM104 which uwet devclopc4 byý P4b4 aWd S:nk (19731) ue$a.1r3 a ciiitetot of exinesdd CAtcettnteot wttre also itndudc4 to &-2tvnnln tbci* ralty 34ag awntio tvaittiog cliteria.

Iteal " mrerskmtt arwiyts Wert accoNniAiibt to dewtcrp keys% fo pilvoz ttiling crilteia %-nd toawu i fth %Uaeful(A of theion-o6pitlve mearwes aloae ýad in oozbiswsat wa opetth vuiooi aitudeunwzus lin prcdksimg pilot uztmaknv, oulcouts.

VWr item sinysis the Irun swwvy rtmpondcot popikiiion was rsdeda into twot Pamples Ihlf~urn*~

Th (list) and W&= -, RWs 0044a~ thwtn~vtd ft, pwdf ctttflkWvd hyw ob ~ni~ item iutd

7

vaiidity, a composite was built by selecting keyed ittems in a sequence to yield maximum predictiob; and across-application of keys and item sequences was accomplished t,) insure stability from sample to sample.Omlitted responses were considered as a valid alternative. The 01 sionficance key was selected for use in thevalidation and[ cross-applicatioii phases. Items were keyed based onl the validity coefficient for cacti itemah-mative. Alternatives were keyed +1, -1, or 0, 0epending on the significance and sign of the validitycoefficient.

Regression analysis tjlottenberiz & Ward, 1963) was -,Iilized for the model seeking phase t;i slect theoptimal sei of predictofs. Miree miodels- containing those predictor variables foupd to make the most validI and wunhje contribution to predictiton of pilot trrining success were cross-applied to estimate the stabilityof the oredictor composites. An optimum cutoff sco~re in the predicted score range was determined for eachmodel whinch would maximize the identification of potential elintinees to the minininuni exclusion of pilotgraduates. Finally, a distribution of predicted scores !or graouates and elimnwies was accomnpljished todetenoine the number of pilot traince outcomes correctly or incorrectly identified by the selectorcomposites, cutoff Score.

tIll. RF.SUtI S AND t)"SUWSON

Construction and Crovs-Validation of Keys

Foto! separate keys weft develotld fronm lthe rcs-txse data ot the OIIAS. Total Flitination (TlI)-Hlying Ml~elvency (1Z1-) Self.1111tated (SWF)' and Motivational lltkieimcy 011D). As ihown in Table 5, theValidity' to, thc UF key is .37 for the criterion usnin otily seven of the ODlAS lteav:... Cmcst;ýappiyinp the TEk ey ytetds a vahdlty of .13. Vbid the decrease ill the absolute site of the vuhid~y -00'ricimnt was siieabiethe stgoticaixo of the validitv' ellicien mpn ss-appliestaiol tzi reaed the .05 lovel. SAIieav results Wereobtaine t'or the NI) koy. The orttaigvaliVdity wis .40 hilt decitwOes te .14A when ctosa-appketl. Fot thosei~cys destgld to tat, thm ImutivAtional Comptunent of elitmnne h talmoc (511 and MOA, the it-lulls of

ettinu e1Wout-41e, Allhot~lt vab lutes of .33 MA .37 wetfe %lbtained (of the miK0ilkeys eto.app»~gthetse ký-V% rmi~tcd in validity ~owifficionts o1 0Ni fut hokh keys wISKIS Welk, rj-

IAýIoI4, at1 the O 10% lvel ei ttk. these kcyi litioll IV wed adata~ in toCobi~natuln Withaptainhal asid~ Al i tileftc dtsu *Ioc~ they taP amu ki hof lw mid otstiladwda 10fttamwI0r not14motrly itiouded t U1 0$Vrt04lai slecliv.4 te~sti

lotal tlilltmg On 7lUN Iq io611enqt Witlu"ajc41icr jq;

UvtatoitKtial Mkwicicy 00*440(SlIJW)A) Mt.) 5

4 VaW ly qf VIS Do "~.ta AIV S Wa)OIAStys

Lii the 21teinp to build a ptedtior couoltxtml of Clitvaou1en key%. SV1I Ikey. =nd Uecyl doew,01 ilr~mW MI Four~ %Upki it Knm p that ft vatw r"I ttv ixj44 s'rpueae opsW OrM haveto be tciluv,,d to a nuMacAW-2k riufthr. 11W lims mtp to rOhtain ftis (AeICcVe weas tf) tCIcm the Sig SVW~kecys to a Ilmistiuncm oui1lhel Which ~wea ai tifkarity idteb to the ose;ou Cicutiios of 6mimniatln fi10M1,401 Im~ntsý. The coveolcatloim between the 1:itation t4 elitninalives anid the k~eys used 41 thit study(-4'atCseic iP hb&W 6.) weteC OVA hol~ iou. to W)4'ci u&C of ofily m~e of the key% as a $sin 0ýPeedkcln

;uwuftust. Fom the eutite sUro*~ of $93 Pilot ltliioeni the SVW'-JC6Wve score vwith the I4w"~ madltli

Table 6. (orrelatiomi Between SV1 6-Scales/AF Keys and Eliniination from UPT

ScahI)eyCotmmation SciIt/Keiy CorMtation

Occupational Scalej-~SVID Occup~aiional Scales-SVIBD~entfist .0514 Officer Worker f.0487Osteopath .0291 Pur~iaing Alerat - 0557Veterinarian .09)1160 Banker -. 0062IMwsictan .015 b i1arniaL'1st .0162Psychiatrist .0638* Morlician *-.01271\ycho'vpit - 09~ Saks Manager .02 72Uaologist --.0358 Real Estate Salesmanw .0107Architect .0051I Ut"- Insuiance Salesman, .-. 0042ýMiathemnatician .0282 Advertisittg Man .08,34"~

Phsist.0073 LWye r .1106'Chenust .0568 Author-Journalist .-. l)0*0Enguiner .06,20 Presid en1t. Nlt g Ce Ice r -.0042hroduction Managker .105706 C'redit Manager .0388Armny Officer .147900 ('namb11er ot Comnerix vx"eUtive -.07 2Aif Force Officer .1461"' Pliysical Therapist OW)4;

(>uti e.06540 Compite r l1'r~u~rmtm .15~50"6Forest scvfvl.e Man t)t.)0*3 Ilusness Fdjucatium Te-utier .0004

Fame 04712 Conwmwtuliy Rk-CtativnAdjtlnr.Y6Math'stlowx le~cret .0273

No1Iwct 1421,V1INAtinsan .040 SN itk!JIWOt IAtVV -06)

Persom I threeor 06 cu~tn!Iae 00

MUa tiAkIntkr COWistnr .4028 Ac4tk-ak Aduecenunt -,07070

St SA,001

Attot-,04 Pcwtikto S~xwl WWI10

OVA t0aknet 02' ~ WAto e 06SPIat CP k M ky .0114

A~os~tan.0412

SIM4V Of. Al C4.

%ith the L1P1 pawiffall Clalesmn w"t the 1bt1a,1 ci1i (.22 Alt) 'tuh UN? Ait Vrito -Asicet kcy thnwt4 afeburey lo~w tel ,rs~l wthteUTciei .15). it tnay be thal tl~th kiy It mtne I1curjie Pfnoo-uitc-l suvwom In the AM Fooc. Resr~ts r.,f tij~ewon u iay i ttdchtiiy the itinimwum nurnhet of SVIIIw~k w~cik~. w~hen .,xtNswd, at' nmim IreJ~civsse of 'rcces~s &n tiltn4 tial*0tt ioicated dhot 17 "acl s

I lx zmulupkc (malatums of .45 with thg 0cljgioni a1Vea*"t 10 be the 0 WW inaln bel. The WAWcSeexe 1w furthet te"1iao uyatayc atei Umeid ip Table 7.

N om 1f th itMianepossib~iy senis ud~di. in seskirm- m-wauhd, hadt) hown a itpv"A .dioat rýhp wIah bou, came intentkioo (Wifr. & ShcA,~ M07) and -.,iu;4 icu %It 4Nhonk-, 1'0731) W'cij r~~~~~miw w. ýC 4ttsiranly telatod to The J'mJot flaagi~t -,iiiesicm. The hiflav'm and1b~inftuia

keysa d 'do)KfJ co the wuie to pftdkt c.exioaWes caice inumn sh m-d ihnmos Po MA04dtwilp "htheO

"~abte 7. tUWsiELUSU MYvIcEpa

and R' fwI 7 SVIB Wkes

scam !A PI

Nliwcin >85 .0668

CoM1puter Pr~~lograin r .3024 .091")Wte tnsurantv Salesman .32167 WO06

ltyhza Tenpse.3157 .112 7

Public AJ~lweustrator .360 .1346Acemtuitant: .37462 .1415Oftice Worker Xh0) 0151

(urpclttef .4 126 .1702,Lrnuist A41 W .1 7M

(~tiyrh.4.129 .18:,13Advertsing Nn 11W)

CP'AOwrte, .4,35 10.44I 3 "210

UPI' cttIo lOWcývet, tlwt4, thc tit~ta~tcgos of ýatcet ptotentlal howv Wkt of no rtlttmtship to UKPet~un~0~t~gadw tttrtir uwcaelit of al tludictur of panrnutd rcaniin hu'Jt not N# olv-ll~kt4.

If lic, SVIRAIn pte 1npuut paancj~bailtv kcyý atr I'nund inl wtcaiu eMo sane wrelwitoxrp w.s11 savtt covt%ýx4kwui U~ that siniwn titt l1Nv~mh rovavh itt ~Ic54i" catee intern. ttw,,t keys CýMW't W O'Seeuly

M41jncitt w~~tt will Os paoows (it M5ktctm to piltut rtahlttgýj wtial wickcuon wnutd boe bwas Ocnia0y4Iposttdl" adtt olther n sovx,;we 40:s pr~t-i~ew u! jttot tritltlIK %vxwCfl It the akeultdjn w~io of l toeqwU(hfyag tnt Owl-C useuuzei iý (vtwoa*, fina tckciti bu w-4 txi itWuhgat6 of tvw-ar*"~y co44 be

A ~~~w-c4 to ttItitttt INe Mt Iomcetetmumi wt A kxm*4 1ritzutg ttnwn0wit.

in j de tllrgut b- 4+Cltxn cutlty jot 00atit4 WWin wt.c n mt~in A 'w r4KWN Ou 1 th p4otei dw

a~dtu&'nt4 ";on itt gon aM natcy tosn~tetn 1 I ~ ~ ~ e tccsn kyitcsn, lhe uae of ItN IWillandJrn WJAS fit lddititn 1to tre AU Virqs lkc {Tiryiw tM (ARKM tnw* in tut wouM rrnit +4

ofta ~m Ow- ojv.&tvtwnA tItiu~ pftv~,xn Uwvnw. ionic aIecnt~ot *"Jamtlttvl t(VMIIIIA rrfrbintig

The rim ovp lo, tkvelofrs; Ut olnvi" CAhtwtt4ic ý4 pecdkIVs fat wc toi ,celtgcrio t pjlkt traincrat to etervxac wisehet nd ptci~ t pltYp of preictwom twiad 2 oriAfjcwt 1tottfliunr tn

poitdrcing d~uauomwehntutwn fliam piwltitakluvng A Mwsutity 4 IfvrertuiOn anahvs is pmtntet inTable A

R4141%t "I IrqtIeatVis VWnbktmi I Wrd ttubIC the lI 17 WVIl scakt aMd (fl elftwfnc key% Mnth a4-fc Nt w'tbuttnn ip the pttdctint of Ow Olcitciier. L; 0,6bulmkns 3il wa &ntd that the WVI 0101

Key and (UPS Ley% pt'dqtay ckwi"pcd oXI a czitclion 4i e~4wtmdc4cAttjt itnetu writ noti OpsicnilKityedied the Ow IIN paC!falcoitwi. CornhsnAl4cut of tOe SVIO * "-ak ~nfc WAS keswith Nir ACVITl ptIotc~nnnut .rcalt stjpifkzt e, the .01 kw)4tn'?n 4, S% and 6). ta ntoairnpa ; the ofc~c~nuo h

non coV.ltwv 4au wiAh Owe AIK)t MWm Conpote (1vowfrun 7. S. ar4 9), it was fowid thAt Or NMIlV.Mk WO CUAisc ktp*. ak"IC and Wn Ol~tltrJon4¶ ttmAe a uniq e W gafstc'ntun .Itcifre 4ccntiy n~vc and akWv thre AV(XSI Nt" Vrnwa tc.( On the mes h-144 reu Of veo

4 psnJX~bun 11 irtLC4ic tht ti4A onnOI~At aiwO MAM a si'ifai SWiibW 600M an4n tnoncnWAlfl data

W' ý7- 7

Table 8. Summary of Regression Analyses

UPTFull Model Restricted Model

Problem Criterion R I Predictors R 2 Predictors df, df, F Ratio

I P'ass /Ia i .2009 17 SYIit scales .0000 - - 17 278 4.l11i'2 PasslFailI .1674 4 OBAS Eliniinee keys

(TE. FD, Sir , MD)) .0000 - - 4 291 14.62 *i Pass/Fail .00q9 Career intent keys

(5 OitAS, I SVIB P'ilot Ke y) .0000 - -6 289 *24fls

4 l'ass/Fail .30311 17 SVIB scales, 4 OBAS Elint

composite21460000 F22 273 5.41*

5 ~ ~ ~ ~ ~ ~ ~ 1 SVs,1tIi scalesSI caes&AFQPltComposite &00 4 - mky 18 273 4520'

CýoOTPioiteCmo~i .2000 17SIsae 5 273 14.06"7 Pass/ Fail .3038 17 SVI B Scalts, 4 0OBAS

Flint keys & AFOQT Pilot 1 VBsaeComposite .1674 4OLA ilm keys 1 273 2.97"

80 .'as/Fail .3038 1'7SVIB scales, 4 OBASHim keys & AFOQT PilotAOQPitComsposite .00396 Compositles 21 273 4.93"

9S Pfiass/Fad .3038 7level.ls 40 A

two additional maue.38 Fo this pupscaes, the moel ofpeitrvaibeAeeSomdnterltv

SVIB ~ ~ ~ ~ i scales for&A es n h AFOQT Pilot Compsite Usigllo2peitost ialmlil

AFOQT Pilot Composite.Tee1 rdcospoue utpe R09 Cmofit .46 .11 oncosvldto.9Mode

3Bascmrsed of thee fosurts itA kpeyrsan thet aFding Pilo CmoieFor bohoho-onthise moelsthes mtiplheobtaie Piot opst n the validation safplelwastr44nwithw.14doncross-validationvTe. obtained, correlationsingth

vlatie otnta aendficrof-aldidtionsamlscwreeig stmeatiticlyinificant atul oedrbeyond towr he pralevel withtherexcepsttionofthea rosignfiancei ofo Modweleto 2. cdr hc ol nld teamnsrto foeo

Itfiscrealiees thatatileomagnitudesofethengultiple casaresed.tMondereainslWhenl the redit componente is

used5wa onbfturained;s anRowever, itnhudas entdta the correlpli ation ohs.Mde obindthied on the scaplesn

compatedonmprirstied population. Fromey an the QPioCmpsete reslts Modes Imaddeapea toe monstiate soRobtasefuness in the selecation af potential 4 pilts. .4o rs-aiain heotie orltosi h

dentxcaption of theo Týrainingalidiy orMoel s

Based on fthre regrspim analysis, it whudalso beonoted that the corrinationi oftatitude on thersand/oe

biogrphical dotal inretito d a pot pulation. C rompoite res)culds fMoctis as ad3 screeain toemonstrate tomiettusrfuiness wnho mighctio be potential failurs.

Ihudetfcto be aPplodto Tasficieng ltn numer fnwtaneeneigplttriigsoeesn h iBaf e oilo thmieei relretiveysn l andlssi waus tothed thateruie the combtinanadtion l of aptitu e, inest, wando

decrapicedat ~asss ithe usfuNeSeofcthree Coposie (lct composfuites otial scrpeeving tho~ugh poritionofthes thota sine had bepoentuied failr teskydvlpeto.h mdlseigwds

~' {. . ~The overall statistica, significance of a predictor composite ofteni do,-% not reflect the actual practicalvalu ;fa scef"i~g tecluniqtw. A more useful assessmnwlt of. the PSC models canl be made by a compi~prsoi

vau fasceS)

of the nuniber of pilot trainees whose outcomes are correctly identitied (i.e.. iits) versus tile ni:.inbel ofpilot trainees incorrectly classified (i.e., false positives and misses). ibis inclh.1o all personnel identified asgradutes who have successfully completed pilot training and tl,,r idc,itificti as potential ePr iaee.: wvho

' did attrit from training. False positives include those successful graduates who were identified as poten!ialelinlinees, misses include elimiices classified as successful graduates.

Predicted scores based oil the weighted linear combination of the variables in the thre- nodels wereobtained for all 593 subjects. Cutoff scores for the three models were computed wh.ch differentiatedbetween pilot trainees who would not be eliminated it some phase of UPT fron those who would havedifficulty in completing this training program. Table 9 shows the possible utility of t.e PSC" models.

Table 9. Predicted Versus Actual Outcomes for Three Models'of the Pilot Selection Composite

Predicted Outcoj•ne

SuccC•s Fahure Total GroupModel Actual uPT

Number Outcome N % N N %

Success N 34') 71 37 32 386 65)0 10 to0

(t is) ,(false p sitives)

F:ailure N I 2) 27 78 t(1 207 35(/2 38 100

(ilmsses) (hits)

Total 478 I A) 11, 100 593 10091 100

2Success N 300 68 26 40 316 659". 7 (10

(hits) (false tIIiives)

Failure N 169 32 38 60 207 35,2 18 IO

(nisses) (i5ts)

• otal N 5 i 1)i '. W 100 593 100)7; ' 14' 11100

Succes's N .10 73 6t 45 .186 08") 20 100

I +" (Off..) (f(alse !ItlXWOM)

hlidcl N 113 . 9.1 55 207 35

,owld N 423 100) 170 100 543 100

7I "it) I )

I I, lI I I f41 i.$ I-~k¢l . tl• Mi lt 'lPl ,, 1l "4l<

I.d. I I 4 Vi1i1s 4•'' . lit ASI sl A I4T x.11' h l.oIt p. tr ,

<,2

1;I,11kki 4,1 , ý nItko

tS( A' odei I. Thiry-eight percent of the eliminees were identitfed by Model I as high-risk for

attrition and only ten percent of the UPT graduates were incorrectly classified by the PSC Model I cutoff

score of .41). With a favorable selection nitio, the misclassification of tea percent of a groap of potentially

successful pilot trainees could be tolerated

The base rate of' successful completion of UPT in the population sample was 65 percent (i.e.. 386 of

the total 593 were successful). Using the predictor composite, the rate of correct classilfations of UfT

outcomes (i.s., hits) is increased to 72 percent.

Table 10 shows ihe number of predicted/actual UPT outcome classiftications by specific type ot

efiminee. Nineteen percent of the total sample was labeled as high risk by this composite cutoff. Of thie 19

' percent identified as potential failures, 68 percent did, in fact, eliminate for one reason or another. Among

the various reasons for e!inunation, over 40 Perc,,nt of the flying deficiency and self-intiated eliniinces were

identified by the PSC Model I cutoff score. Only 18 percent of the fear of flying category was identified as

high risk. Obviously, the primary behavioral chiaracteristic related to this type of failure were not tapped

by the measures included in the proposed composite.

"N'- Table 10. Descrptive Statistics and Percentage of UPT Trainees Above and.Beow Cutoff Point on Pilot Seection Composite MODEL I by Training Outcome

Training Outcome

Type ECtitnc

Nj ACAD.UPT MED Total

GRAD O`V4,R4 FLY SE MOA T Al. Group

Cutofe Score N Co.% N Cot% N Co % N CoW% N Co % N Co. % N Col%

PSC .49 37 10 7 30 42 40 20 41 3 18 78 38 115 19

tP(r&dcnad Fatilue) Row Y ~ 32 6 ?8 3 3 68 1 00

PSC-..49 449 90 16 70 62 60 37 59 14 82 129 60 47" 81

tPicted Succweil Qow% 73 3 13 8 3 27 Ic0

TowIN 3%- 100 73 100 104 100 63 100 17 100 207 ¶00 593 tO0

Row, 65 4 17 11 3 35 IOU

Mon.7%27 56791 %A94 "0 510) .5349 .6468

SO AIN .1"I1 .180 .2018 ,182 .1906 .low

The I•ams and stan.1ard dcvtulionls of iUSC Mo-del I skcres are aulso showo in Ta3)k, 10. Restults of

t-lets iktwonl diw "walls of the graduate gnup n.and the Iosi groups indicate that inll •1 omlpariolls the

meian dif weelts wowl•p'fikwi at oz ty0yood the .01 level,

KVC ihxleI 2,h t hUng the optimal cultoff sMite oft 35, nllY 1$ Ivttt of the clin•nmes and We0e1

Iwtcent tIel3l ' gmaduate categoty wot identified is Potential fhlr While classIfication of 18 r(of tif ellnhti:ces would doease cost rlated to attritilon. tills lmdel does nit appear ito lie W; effectivc ut

the idetlitication of i1kential failuies as ilte other two models, Oi t1lW othr ht nldl, tilts loodol Iim h the

lowest rate of oicasliztrnKt of g ttrially sucrssii4 jýiot ttailnes. IlsWnle thig auodel. till. tale oft vourctl-11casflatiostt of till oukeolls itcsoetM (toln thle buzlate nof n( O to 68 1vttX41t*

As hlnownt ill Toble II .1I Ifermect of 0lW total s"annple wa. li.te~ti asi hligh-isk fo• 1pgot trailirtg ,us1intile .35 cutoff icure. ()f tire I I per0ctt., 5O p.eVrmt werme elinltkted frmttl ;;aRtnlng, 11,i, mol•dl api;'Tars toidalitiv api)oxil"YJtey the ,aute ifrccnta% of failums int tile diil•fealt elbiince cAtej•.ios.

Result, oI-tests otwActWll tie incawts of tilre ghadftaic gwup and the h, ý itwuf 11tellttl rit 1 able t I.4 dl0to th1 11 i all omlparisons eXcel tone tile ltiak diffetcms wele siifica,3nt at or beynld thie .05 let'e.

Th H1411e 1an co0ijadsnl•l belwee- UPI' gAttstcs antd elititilnee due to feia of tlyitg wa4s nm o gtigl•lt.k

Indicated pxriously, dtis mo1da apjxwtc an be tiW least effotiv as a scsceaslng :tevitc fiw pil4 tvtiltcs.

'13

:'" 13

Table 1. Oescriptive Statistics and Percentage of UMT Trainees Above andBelow Cutoff Point on Pilot Selection Composte MODEL 2 by Training Outcome

Training Outcome

Type Ellminee

ACAD,UPT MED Total

GRAD OTHER FLY SIE MOA Total Group

'ýutotf Scura N Col % N Col % N Col % N Cot% N Col% N cal% N Col%

P Z0.,3: 26 7 5 22 19 18 11 17 3 18 38 18 64 11(Predicted Failure) Row% 41 8 29 17 5 59 100

PSC < 35 360 93 18 78 85 82 52 83 14 82 169 82 529 89(Predicted Success) Row % 68 3 16 10 3 32 100

Tota' 386 100 23 100 104 100 63 100 17 100 207 100 593 100Row % 65 4 17 11 3 35 100

Mean .6645 .5570 .5377 .4940 .5324 .5261 .6162

SD .2140 .2349 .2086 .1833 .2251 .2076 .2217

.PSC tMocde 3. The Model 3 optimnal cutoff score of .51 identified 45 percent of tOI failures and 20

percent of thei graduates as high-risk for attrition, The nusclassihication of 20 percent of potentially

successful pilot trainees is rather xcv.sive for opational use. Using PSC Model 3, the rate of correct

classifications of UPT outcolh.s is in, teased .om the base rate of 65 percent to 67 percent.

Table 12 indicate" tha, .- ) petcent of the total population would be identified by Model 3 as high-risk

lesOlt,.l T11s T.1C'el ih6etitles a larger number of petionnel as high-risk than the other two models. Of the

20 ir-cent identified as high.fisk. 55 pet(-,it were elimtinated from training. Forty.seven percent of tlte

flying deficiency category and 4$ :xcelt of the SIl| were identified by tile PSC Model 3 cutoff score. As

with ihe other modeL, the fear of' flying :alcgt -.y it Itss accurately pricr.cctk allhough NSC Model 3

idwitlliks 24 .'lre'lt of this category as compared 1,# 18 pircnt idertfijed by PSC Models I and 2.

*abk" 12D l Iriipilivt Sta:titis .intd P-ct.iittge of U'11 lrf-ines Above W id-Lt.w 'CujofI PNtiit ni Nilot Sekelion (Con"ml1t MUUEL 3 by Ttrlai Olcoume

VOWe*# 00g 60

TVgW rotA4404

SUPY MtI '1I

01•AU ,iY'4BJ I+t,.V ' Kf I • Thl~el Qtl~pp

Cu4loo scot's N CN ft Cot% N C+ % N C.o% N C04% N Ci% N Cot %

7t 0 47 30 43 4 2 94 46 175) 2'4IIidPht4 5w~ 2P 18 2 56 too

PS' .51 I.0 6012 Q ' 5 5 ,,3 67 13 78 113 \,3 423 71(tdttwJ Sucg*WJ Pow % 73 3 13 Ii 3 V 100

ToJ• I ,00 23 1WO0 t'L4 100 63 W00 ,' 100 20u 7 ,c 693 100.ow % 65 4 ?7 1I 3 38 100

666 .6010( -Sb SA7 ZIA00 M7IS .6147

SO ,19-6 , I 4 2012 .21W ,IIM .2033 .2W2

i

•14

T-tests between means of the graduate group and the loss groups presented in Table 12 indicate thatall comparisons between giaduates and elirninees were significant at or beyond the .05 level.

The possible monetary savings from implementation of a selection procedure such as one of the PSCmodels can be demonstrated by using the sample population as an example. A total of 386 traineesgraduated from pilot traifing in the OTS sample representing 65 percent of the total group. Using Model 1,the total number of trainees who achieved a predicted score of .49 or higher was 478; 349, or 73 percent ofthis group, were successful graduates. Assuning that 386 graduates were reqcired to meet operationalneeds, using the 73 percent success rate of the Model I PSC cutoff, a total input of 529 would have beenrequired instead of 593. Instead of the origiial 207 eliminees, only 143 would not have completed pilottraining. The original number of eliminees (207), at an average cost of $16,082 (see Footnote 1) pereliminee, repre-ents an approximate cost of $3,328,974. The lowered number of elirninees (143) representsa cost of $2,299,726, or a savings of $1,029,248.

Although the required input and number of eli ninees change using PSC Model 3, the possible savingsto be accrued are the same. This finding would tend to indicate that the additional costs associated with theadministration of the SVIB arc not warranted. From a cost standpoint, the PSC Model 3 is more effectivethan the other two models. The use of Model 3 would avoid the purchase and administrative time of acivilian test ior screening. The only drawback in using PSC Model 3 is the high (20%) missdassification rateof potential UPI' graduates. The selection ratio would have to be highly favorable before such a screeningprocedure could be used.

Using Model 2, a total input of 568 would be required. The lowered number of elimtinees (182) usingthis model represents a cost of $2,92-6,924 or a savings of $402,050 front the costs associated with a totalof 207 eliminees.

"Tlhe savings based on this sample using the three PSW models are given for illustrative purposes only,Although the saiuigs to be accrued front tile total Ulq" input would le sizeable, the cost of programadministration and rvcruitmnent to obtain a suffident number of high (luolity trainets" was not taken intoaccount. It is also wrcogtiiel that the savings calculated ont OTS input would not accurately reprsent thecosts associated with input from AFROTC and AFA.

IV. C•.MN1LUSItNS AND IU.ECOMkNt)A11`0NS

Tlle correlations ubtaiced for the INA' twidels ajlilar to indicalte tihat a scr•ixhig device toingnonlXo•intive data h•s practical value li tile slection of pilot trailtees. Obviouisly, sucd a iyStel setfectivene1s would bV related to t11h avalability of qualk),1'd jilot applican1ts. If the selection ratiO in thepilot apilicant iosol bhc1mes e0,4 favota•hk, the •rc•cetage of lfalw os•itives for the KC todels, nmay bhe too

tigh to wartant their Use in the seleCtion p1ro-s.

If th oilunlter Of potential p111lot appichants OUtlnumGdhet the projected ned lt t tilolt trtiteeS, theCon111rn about eclusion ittf false positives wowld it bhe a lilltitg factor in •w•llonuwdihg t1e use of IVC

doelts I or 3 in ail o•rational settillg,

Ilasd (In 11tW results of this p•h•l••ni y study, tile following eGnmuse of action awe fwconlntolaed:

u1. rthr research be acconlplished (4 a neow sainple of lilot Itinees to tevalidate t~le 01AS andSVIIJ. 11e recent dcange in the 111 "PT cuiculum to pjerit tiie 1'41 -aircraft ilraae of lralining of ti"lto •esactmonlplished dJuing Officer Tr•tisig Sdcool neotsWstates further lave slitgtion. ".e present study did notattempt to pctdict the Pass/fall Crileriol foir any one 1la.t of tralikig. An evalualtio of (' th IV'Lc%)opo~sites kill Sj.jCtc4X l Cto111mpleetion of the 1-41 phise apitears appromptite followed tly reseafci on theusefulless of these m1eaures in tIl lae2 lhaeas of UI.P' Similar studii s isould aho be asxmpsliehd kmAPIHOTC and AFA pilot t•,itec inpu.11.

2. in the next revislon of the APOOT, itemn toutwld significant in the develollnlert if the IiBASeli•nination kteys CaZ be cunsidered fur i•nicmpatimr ibto the lillt Hiuglsldrcal i•wnluty portion of theoperational test.

3. ll1 value of these screlcing tednliques Should Ib fuirtlihr evuated by Coliwdering the (a)feasibility and cost of adninixite~ialg the additianal tests, (b) the poitential saing, to he aceued by avoldhingthe training eClemlies Of trainces ws o C.urlol conmllete OPT1, Wid (c) the impact l " tite IoW 1W of poletially

,.:. 15

REFERENCES

Abrahamns, N.M., & Neumann, 1. Evaluation of the Strong Vocational Interest Blank as a screening devicefi)r NES'P applicants. Research Memorandum SRM 72-6. San Diego, CA: Naval Personnel andTraining Research Laboratory, Bureau of Naval Personnel, October 1971. (a)

Abrahans, N.M., & Neumann, 1. The assessment of career motivation among NROTC applicants with theStrong Vocational Interest Blank. Research Report SSR 72.9. San Diego, CA: Naval Personnel andTraining Research Laboratory, Bureau of Naval Personnel, October 1971. (b)

Abrahams, N.M., & Neumann, 1. The validation of tihe Strong Vocational Interest Blank for predictingNaval Academy disenrollment and military aptitude. Technical Bulletin STB 73-3. San Diego, CA:Naval Personnel and Training Research Laboratory, February 1973.

Abraha•ns, N.M., Neunmnn, i., & Damn, J.E. Use of the Strong Vocational Interest Blank in identifyingNaval Academy early motivational disenrollees. Research Report, SRR 704. San Diego, CA: NavalPersonnel and Training Research Laboratory, Bureau of Naval Personnel, October 1969.

Abralhms, NKM., Neumann, I., & Githens, W.H. 77le Strong Vocational Interest Blank in predicting NROTCofficer retention: Part L Validity and reliability. Technical Bulletin STB 68-7. San Diego, CA: U.S.Naval Personnel Research Activity, January 1968.

Bottenberg, R.A., & Ward, J.IL, Jr. Applied nidtiple linear regression. PRL-TDR-63-6, AD-413 128.Lackland AFB, TX: 6570th Personnel Research Laboratory, Aerospace Medical Division, March1963.

Dann, J.E., & Abrahlnis, N.M. Use of bgrah&kit ,,d interview information in predicting Naval Academydisenroliment. Research Report SRR 71-7. San Diego, CA: Naval Personnel and Training ResearchLaboratory, September 1970.

LaGuipa, J.. Development of officer cndidlate biographical information blank. L Validation of someexpedr•nestal inventories. Teduical Bulletin 60-9. Waqdington, D.C.: Bureau of Naval Personnel, June1960.

Mufliu, CJ., Ewing, F., & Woods, F.E. The use of biographical quwstioninaire.•-IW in prmdicting OCSsuctvs& Technical Memorandum PR.TM.62.27. t•ckland AF8, TX: 6570th Personnel ResearchLaboratory, Aerospace Medical D~ivlion, Dewimber 1962.

NeIuwan, I., Githeos, W.It, & Abraha s, N.M. 77we deelQpnient of the US Navy backgroundquesrhmnalre fJir NROTCy(Rgular) selection. Reswarch Report SRR 68-3, Sam Diego, CA: US Naval

NiwnlRosearch Activity, August 1967.SI" . F. kvdelopment aN validlatm of Pwres to pretdict ofifcer career status, AFIIRL.TR-73.1, AD)-60

540, lUckland AFOU, T'X: Pewou w) Resoarcd Division. Air Forea tlnuman R msourc Laboratory,Murdi 1973.

laytor C.W. Murray. S.L, ElisAn, R,L, & "aisay, M.S. lvdelopnwnt of motivatltot asseswsmntiet-t'hniqewx ftw Air l-'orc, offfitr tralintg and education programs: Motfvatll!n fow ielor tralning,AI:ItUTR.71.21, AI),7S1 497, Brooks AFB, TX: lhotWoWal Edication Divisions, Air Foicelwuman Rtcowees Laboratory, July 1971.

Tuz'upm E.C. A oPx, m for aw officer effectiventss seketlo, battey bserd oan mcasurs- obtaNabkl duringBasic and Adv•acvd AIRO . APIiRC-TN-S7T87, AD-134 207. LUckwd AItB, TX: Air ForcePorowwwl and Tr•ahnig Resarach Center, J wic 1957.

Thplu^r E.C. Vic validity of commerdal interstt Inantorks for pmdictiron of suactss il USAF offiscercasndldate schmal. Reseamch Nnte PERS: $2.50. Laklmid AFB. TX: Ituan Reso•wu RwsardCenter, Air Ttinig Cuouuamd, •e•oeter 1952.

LUdia, E., & Shea-, F. VaIldty of bae•kw)ind and interest tests fiw officer stattc car-er intentAI4IRIATR.72.58, A0).75% 818. ia'ckldal Al"B, TX: Pe1rsanel Riearcd Divislo, Air Foxv }mumnResurce, Labwatoty. ibruay 1473.

16

APPENDIX A: DEVELOPMENT OF EXPERIMENTAL SCORES

Various experimental scores were generated for responses to the Job Importance-Job PossibilityScale (Shenk, 1973). To obtain these experimental scores the following weighted values were first assignedto each alternative:

Alternative Value hnportance Scale Response Possibility Scale Response

a 1 Not important at all No possibility at all

b 2 Somewhat below average in importance Less than average possibility

C 3 Of average importance Average possibility

d 4 Somewhat above average in importance Better than average possibility

e 5 Extremely important Very good possibility

Using these weights, the following experimental scores were generated for each subject:

Total Score Algebraic sum (incduding both plus and minus values) for the weightapplied to the item on the Possibility Scale subtracted from the weightapplied to the same item on the Importance Scale.

Zero Frequency The number of zero differences when the weight applied to the item onthe Possibility Scale is subtracted from the weight applied to the sameitem on the Importance Scale; i.e., the number of tinws the importanceweight was the same as the possibility weight.

Positive Frequency T1e number of positive differences when the weight applied to the itemon the Possibility Scale is subtracted fiom the weight applied to the sameitem on tle Importance Scale; i.e., the number of times the importanceweight was greater than tie possibility weight.

Positive Score Tlhe sum of all positive differences when the weight applied to the item onthe Possibility Scale is subtracted from the weight applied to the sameitem on the Importance Sale; i.e., the iinportanoe weight was greater thanthe posswbLdqy weight.

Total Powssbility Stum of all tie weighted responses for items on the Possbility Scale.Score

*US1 GO• SJI•NI Ul• O iff0 l47t-r,1.-t.1i i'l 17

SLJPLEMENTAR.""q.21

'-I

..x

.4.

A

INO»S.~~I.

AIR FORCE I ICIMAN RESOURCES L~ABOR AT( RYBrook., kir Force Base.,rTexas 7,8235

Errata

Firstmu Icr utIhor Title

F I I 1 1.-TH - _3 - 111 (D -7 7 1) Gll l Effect of an AII-Volumiteer Fore mi thniptiJt into (heSchool of Militarv Sciences. Officvr Traiiroimgprograim

~ ~'L~FIllI-TII7(m~D (I) 81 1)8I Guinn Backgrotand and Interest Mnicstires as P'redictors of- ~Success in Undergradtiatm Pilot rruh iinmg

i.di II i .-THR-.7.-:u ' D-Al 1.2 6.89) % alsgmiuew Nav'igaitir-O.( wrver Seleetion Resoearch: Develop-men~mt of New %ir Forrm* Offiocer Qtialifyinig TestNavigator-Tecinical Comp1 osite

~~~FI ~ %) %0LIB-83 (~ .I58 118) limiuter Pilot Selectioni Sysitemi I ev'elopitteal

Div lipe toworitig error,. hirb Ii. ivrv founid ill ltie data riles of ltie Air Pore. Officer Qtialificatinmi lestl -

F, m!. L. M. and N. all aeialym-es uivig aptitude sc-ores t~vived from the-se test forms which are cowlaiine( inldiii *ubhjec tulumuivaI report- ails.u a are consiudered erroneous.

%laau;oe.er &amml Per.sonikt IVi% i!-ioo