FUTURE SELECTION OF AQUANAUTS 193 



be considered semiobjective, since they were made by hard-headed, experienced men using 

 quantitative scales. 



Using, then, the five harder or better criterion variables and the five predictor variables — 

 eliminating education, since the civilian-Navy differences would tend to restrict correlations — 

 we have a 5 x 5 matrix. Within this 5x5 matrix there are 10 out of a possible 25 significant 

 correlations, with three more of borderline significance. By chance, we would expect only one 

 significant correlation from this matrix. The demographic and criterion variables have an 

 extremely high number of correlations. 



Finally, there is the fact of internal consistency. Looking down the columns of Table 7 it 

 can be seen that the signs of all significant correlations within each column are the same. 

 That is, age is always positively correlated with the criteria when the correlations are signifi- 

 cant, and size of home town is always negatively correlated, and so on. 



We can conclude from Table 7 that the more successful aquanaut in Sealab II was older, 

 had more diving experience, was more likely to have been later born, and was raised in a 

 smaller-sized town than was the less successful aquanaut. By success is meant that he went 

 on a greater number of sorties, spent more time in the water, was chosen as a teammate more 

 often by his peers, his performance was rated higher by his leader, and he appears to have 

 been more satisfied with life in Sealab, as indicated by the fact that he made fewer outside 

 telephone calls. A word of caution is in order in considering these results. These correla- 

 tions are group tendencies only. They cannot be applied to individual aquanauts, excluding 

 other considerations. There may well have been first-born, young men, with relatively little 

 diving experience, not raised in small towns, who were among the best divers in Sealab. If 

 this is true, then what do these results mean? Just how these predictors should enter into a 

 selection process will be discussed after additional results are presented. 



Correlations Between Allport-Vernon-Lindzey Values and Criteria 



The Allport-Vernon-Lindzey Study of Values has six value scales. Ten criterion variables 

 were used. In the 6 x 10 matrix of values by criteria there was only one significant correlation. 

 Since the appearance of one significant correlation could be a chance occurrence, this correla- 

 tion probably does not merit further consideration. 



The lack of significant correlations for the group as a whole does not, however, mean that 

 values have no relation to performance in Sealab. There were several types of men in Sealab. 

 Two well-defined subgroups are the Navy and civilian divers. It is reasonable to assume that 

 the men in these two subgroups entered Sealab with somewhat different goals. These differ- 

 ences in goals may in turn have been reflected in basic values, as measured by the "Study of 

 Values." Thus, while one value may have correlated with performance positively for say the 

 civilian subgroup, the correlation on the same value may have shown a negative correlation for 

 the Navy subgroup. This was in fact the case. Before presenting the results illustrating this 

 point, it is necessary to discuss first the criterion factor scores on which the analysis was 

 based. 



Criterion Factor Scores 



In attempting to evaluate performance and adjustment in a complex situation such as Sea- 

 lab, it is necessary to use multiple criteria. Ten have been used in the present analysis, and 

 more are available for future analyses. The use of multiple criteria has both advantages and 

 drawbacks. The principal advantage is that multiple criteria provide a more complete picture 

 of behavior than would a single criterion. The principal disadvantage is that the picture is 

 fractionated and diffuse, since some criteria correlate with some predictors but not with oth- 

 ers. Factor analysis provides at least a partial answer to the choice between a simplistic few 

 or a confusing multitude of variables. 



Three factor scores were derived from an analysis of the ten criterion variables. One of 

 the factor scores is an unrotated or general factor. For the general factor, each of the ten 



