Showing papers in "Personnel Psychology in 1974"

Predicting tenure with biographical data: exhuming buried evidence

Abstract: RESULTS of efforts to predict tenure with biographical data have frequently been reported. Schuh (1967) compiled results of 21 studies investigating this relationship. After noting that all but two found a t least one significantly predictive item, Schuh concluded, “it appears that some items in an applicant’s personal history can be found to relate to tenure in most jobs (p. 145) .” Unfortunately this conclusion can be questioned on two counts. First only 14 of the reviewed studies were cross-validated (ix., reported a validity coefficient obtained from applying the predictor weights to a sample separate from that used to determine the weights). The possibility of thus capitalizing on chance relationships is substantial since studies using biographical data typically begin with a large number of variables. At a more general level, estimating the probability of obtaining significant results in any one study from the proportion of published studies reporting significant results is risky. This is true because (a ) investigators tend to “bury negative results (Dunnette, 1966, p. 347) ” and (b) journal editors disproportionately reject studies that do not report significant findings (Bakan, 1966). Consequently, the information to be gained from insignificant results and failures to replicate is suppressed (Walster and Cleary, 1970).