The third multiple regression was computed using the Wherry technique with the 

 C pool of LHQ variables reflecting changes during childhood and youth. Five 

 variables produced a multiple R of .54 accounting for 29% of the criterion 

 variance. The variables resulting in this correlation were: changes in 

 quality of clothing, changes in type of residence, changes in health, changes 

 in fathers employment, and changes in mother's employment. 



The crucial point for this discussion is that, using three conceptually inde- 

 pendent sets of predictors derived from LHQ, variables from each set can inde- 

 pendently account for a meaningful percentage of the variance recorded on the 

 criterion variable. These results provide considerable evidence that the LHQ 

 is a highly sensitive instrument. 



Cross-validation of LHQ Predictors 



The results from multiple regression with LHQ variables as predictors are so 

 strong as to raise questions about whether they might be accounted for by 

 peculiarities in the relatively small sample of scientists employed in the 

 analysis. As a first cross-validation of the instrument, one of us (Radloff) 

 obtained complete LHQ responses from a class of U.S. Navy enlisted men attend- 

 ing 2nd Class Diver's School (n = 38). As a criterion measure for the LHQ of 

 this population, we used completion of school defined as = failure, 1 = com- 

 pletion of SCUBA training, 2 = completion of SCUBA and Hard Hat Diving courses. 

 This group provided a difficult test sample for cross-validation. Not only 

 was a highly restricted criterion employed, but the two populations were dra- 

 matically different on almost every possible dimension. In addition to the 

 fact that the mean age of the Navy sample was 10 years younger than that of the 

 TEKTITE aquanauts (21 vs. 31), the two groups differed significantly (£ <.05) 

 on 16 of the 39 variables derived from the LHQ. To note a few of the major 

 differences between the samples, the TEKTITE scientists came from larger home- 

 towns, received more parental praise and physical affection, performed better 

 in school, and had more changes in place of residence, type of school, and 

 changes in school performance. The Navy divers received more physical punish- 

 ment, came from more homogeneous communities, went out at night and dated more, 

 and had more clashes with authority. These differences were predictable and 

 clearly reflect, among other things, social class differences between the two 

 samples. 



The cross-validation was conducted by using the same predictors and the same 

 Beta-weights employed in the TEKTITE sample to predict the 3-point school com- 

 pletion criterion. The results are extremely encouraging. The multiple R' s 

 obtained were: E predictors, R = .37; I^ predictors, R = .33; C predictors, 

 R = .46. Given the great differences in the nature of the two populations, the 

 fact that the same variables weighted in the same way predict so well in both 

 groups attests to the overwhelming importance of a limited number of demographic 

 variables. Predictors derived through consideration of the requirements of the 

 Navy Diving task and the characteristics of the available population of poten- 

 tial divers should be far more effective in accounting for variance on the school 

 completion criterion. This research needs to be undertaken. 



In conclusion, the research conducted during TEKTITE II has provided an oppor- 

 tunity to validate a sophisticated technique of demographic prediction against 

 a highly reliable and objective criterior of behavior. The results obtained 



VIII- 55 



