RANGE RESTRICTION PROBLEMS IN USE OF SELF-SELECTED GROUPS FOR TEST VALIDATION

DISCUSSES PROBLEMS IN CORRECTING SAMPLE STATISTICS FOR RESTRICTION OF RANGE. THE EMPHASIS IS ON CASES OF RANGE RESTRICTION RESULTING FROM SELF-SELECTION WHERE THE EXPLICIT SELECTION VARIABLE IS EITHER UNKNOWN OR UNMEASURABLE. AN EXAMPLE ILLUSTRATES THE EFFECTS OF SUBSTITUTING AN AVAILABLE IMPLICIT SELECTION VARIABLE FOR AN UNAVAILABLE EXPLICIT ONE IN THE STANDARD CORRECTION FORMULA FOR VARIOUS CORRELATIONS BETWEEN THE 2 AND FOR VARIOUS SELECTION RATIOS. THE CALCULATIONS INDICATE THAT THE SUBSTITUTION RESULTS IN AN UNDERCORRECTION IF THERE IS A SUBSTANTIAL POSITIVE CORRELATION BETWEEN THE IMPLICIT AND EXPLICIT SELECTION VARIABLES; HOWEVER, IT RESULTS IN AN OVERCORRECTION FOR SMALL OR NEGATIVE CORRELATIONS BETWEEN THE 2. (11 REF.) (PsycINFO Database Record (c) 2006 APA, all rights reserved). © 1968 American Psychological Association.