OFFICER SELECTION SYSTEM

1
(No Transcript)
2

General Directorate Belgian Defence
H
H
R
R
Pensacola 4 - 6 Nov 2003 I M T A
3
VALIDATION OF THE BELGIAN MILITARY PILOT TEST
BATTERY

• Yves DEVRIENDT
• Psychologist
• Belgian Defence
• General Directorate HR
• Research Technology

4
Plan
Problem definition Participants Variables Interna
l validation External validation Conclusions Dis
cussion
what methods criteria results
5
Problem definition

• Ø Pilot selection
• Ø Belgian context

Importance
Psycho technical Technical Practical
Problem areas
Considerations
6
Participants

• Ø Pilot applicants / trainees
• Flemish / Walloon
• Male / female
• Selection procedures

7
Variables 1

• Ø IV

Academic
Physical
Interview
Psy tests
Maths Physics Languages English Sciences
Swimming Running Shuttle run Pull ups Sit
ups Jumping Throwing
Psy Prof
PM PS
8
Variables 2

• Ø IV or predictors

PM
PS
C1 Coord T1 Coord C2 Coord T2 Coord G1Discr G2
Discr TR Discr RI Discr (RM Discr) RS 506
G Reaction test Reaction time Reaction test J
ABL304 J ABL 17 J ABL 152 S1 J ABL 152 S2 J ABL
152 S3 J ABL 152 S4
J Cubes test Digit recall 1 Digit recall 2 J
Arithmetic Manikin Spiral time
9
Variables 3

• Ø DV or criterion

Time between GFD1-GFD12 Passed GFD12
(yes/no) Number of cards (green or blue)
Revisions DO1 (ordinal) Supplementary hours
Higher Brevet (yes/no)
Number hours GFD12 Number hours Fail GFD
12 Number hours BFT Number hours fail BFT
10
Internal validation PCA PM / postulants

11
External validation

• What
• Methods
• - Regression
• - Correlation

12
Criterion

• ? What
• - GFD 12 Continuous

13
Criterion

• - GFD Dichotomous
• Ø Value of the criteria
• - Statistical level
• - Content level

14
Multiple regression PM

• Ø Linear model
• Statistical not significant
• Ø

15
Multiple regression PM

• Ø Non linear model
• Statistical not significant

16
Correlation matrix PM Number green and blue
cards GFD / incorporated

• ------------------------------------------------
  -------------------------------
• STAT. Correlations, Casewise MD deletion,
  N119
• BASIC
  
• STATS
  
• -------------------------------------------
  ---------------------------
• C1_COOT1_COOC2_COOT2_COOG1_DISG2_
  DISTR_DISRI_DISRM_DIS
• Variable R R R R CR
  CR CR CR CR RS_506
• -------------------------------------------
  ---------------------------
• C1_COOR 1,00 -,20 ,78 -,12 -,20
  ,04 ,11 ,09 ,18 -,11
• T1_COOR 1,00 -,13 ,60 ,00
  ,00 ,02 -,01 -,00 -,12
• C2_COOR 1,00 -,10 -,17
  -,01 ,07 ,02 ,16 -,09
• T2_COOR 1,00 -,10
  -,03 -,05 ,09 ,10 -,06
• G1_DISCR 1,00
  ,15 -,30 -,44 -,97 ,16
• G2_DISCR
  1,00 -,22 -,17 -,40 ,32
• TR_DISCR
  1,00 ,30 ,33 -,03
• RI_DISCR
  1,00 ,45 -,06
• RM_DISCR
  1,00 -,23
• RS_506
  1,00
• -------------------------------------------
  ---------------------------

17
Multiple regression PSNumber green and blue
cards GFD

• ------------------------------------------------
  ---------------------------
• STAT. Regression Summary for Dependent
  Variable TOTGFD
• MULTIPLE R ,40631838 R² ,16509463 Adjusted
  R² ,11827751
• REGRESS. F(6,107)3,5264 plt,00318 Std.Error of
  estimate 2,3907
• ---------------------------------------------
  -------------------------
• St. Err. St.
  Err.
• N114 BETA of BETA B
  of B t(107) p-level
• ---------------------------------------------
  -------------------------
• Intercpt 5,634058
  1,898140 2,96820 ,003697
• J_S4 ,542463 ,214415 ,186963
  ,073899 2,52996 ,012864
• J_ARITH -,161389 ,090289 -,139459
  ,078020 -1,78748 ,076691
• CIJFER1 -,212888 ,094029 -,501147
  ,221348 -2,26407 ,025586
• MANIKIN ,140490 ,089596 ,132641
  ,084590 1,56804 ,119824
• J_304 ,167000 ,096018 ,091116
  ,052388 1,73926 ,084865
• J_S2 -,334114 ,213209 -,189577
  ,120975 -1,56707 ,120051
• ---------------------------------------------
  -------------------------

18
Association J-ABL152 and criterion
19
Association J-ABL152 and criterion
20
Multiple regression Academic tests Number green
and blue cards GFD / incorporated

• ------------------------------------------------
  ---------------------------
• STAT. Regression Summary for Dependent
  Variable TOTGFD
• MULTIPLE R ,61403803 R² ,37704271 Adjusted
  R² ,33810787
• REGRESS. F(2,32)9,6839 plt,00051 Std.Error of
  estimate 2,4706
• ---------------------------------------------
  -------------------------
• St. Err. St.
  Err.
• N35 BETA of BETA B
  of B t(32) p-level
• ---------------------------------------------
  -------------------------
• Intercpt 9,171350
  4,441459 2,06494 ,047110
• PHYSIQUE ,547729 ,139909 ,816862
  ,208655 3,91490 ,000445
• LG -,239984 ,139909 -,117712
  ,068625 -1,71529 ,095962
• ---------------------------------------------
  -------------------------

21
Multiple regression Physical tests Number
green and blue cards GFD / incorporated

• ------------------------------------------------
  ---------------------------
• STAT. Regression Summary for Dependent
  Variable TOTGFD
• MULTIPLE R ,16075744 R² ,02584295 Adjusted
  R² ,01765676
• REGRESS. F(1,119)3,1569 plt,07816 Std.Error of
  estimate 2,4749
• ---------------------------------------------
  -------------------------
• St. Err. St.
  Err.
• N121 BETA of BETA B
  of B t(119) p-level
• ---------------------------------------------
  -------------------------
• Intercpt 2,233246
  3,541190 ,630649 ,529480
• COURIR ,160757 ,090478 ,319503
  ,179823 1,776765 ,078162
• ---------------------------------------------
  -------------------------

22
Logistic regression Interview 1 Passed GFD
(Yes/No) / incorporated

• VARIABLES
• 11 PSY -9999
• 12 MOTIVATI -9999
• 171 SAT_GFD1 -9999
• Dependent variable SAT_GFD1
• ------------------------------------------------
  -----------------------------------
• STAT. Model Logistic
  regression (logit) N of 0's20 1's101
• NONLIN. Dep. var SAT_GFD1 Loss
  Max likelihood (MS-err. scaled to 1)
• ESTIMAT. Final loss 51,493433979
  Chi²(2)5,5108 p,06360
• -----------------------------------------------
  ----------------------------------

23
Logistic regression Interview 2 Passed GFD
(Yes/No)

• ------------------------------------------------
  -----------------------------------
• STAT. Model Logistic
  regression (logit) N of 0's20 1's101
• NONLIN. Dep. var SAT_GFD1 Loss
  Max likelihood (MS-err. scaled to 1)
• ESTIMAT. Final loss 52,076876738
  Chi²(1)4,3439 p,03715
• ------------------------------------------------
  ----------------------------------
• 
  
• N121 PSY
  
• ------------------------------------------------
  ------

• ------------------------------------------------
  -----------------------------------
• STAT. Model Logistic
  regression (logit) N of 0's20 1's101
• NONLIN. Dep. var SAT_GFD1 Loss
  Max likelihood (MS-err. scaled to 1)
• ESTIMAT. Final loss 54,094375991
  Chi²(1),30892 p,57834
• ------------------------------------------------
  ----------------------------------
• 
  
• N121 MOTIVATIE
  
• ------------------------------------------------
  ------

24
Conclusions

• Ø Internal structure

PM
PS
factor structure
factor structure?
25
Conclusions

• Ø External validity

PM
PS
No predictivity
predictivity
But!!!
26
Conclusions

• Ø Interview
• - separate
• - together

Professional
Psy
No predictivity
predictivity
27
Conclusions

• Ø Academical tests

Physics 1 language
predictivity
28
Conclusions

• Ø Physical tests

No predictivity
29
Discussion 1

• Ø General
• ? Per test battery
• - PM
• - PS good tests
• - Interview
• - Academic tests
• - Sport
• ?
• ? Academisch
• ? Sport

ABL 152 Digit recall 1 Arithmetic ABL 304 Manikin
30
Discussion 2

• Ø Range restriction
• ? Criterion
• ? Recent developments

31

• ? ? ?