The article is focused on the validity of 19 selections procedures for predicting job performance (and also future training performance) and the validity of paired combinations of GMA (General Mental ability) with other methods available, where the higher multivariate validity and utility for job performance is using the GMA plus a work sample, the GMA plus an integrity test or the GMA plus some structured interview. Both integrity test and structured interview have an advantage over the work sample as there’s the possibility of using them for entry level (without any experience) selection and for experienced employees. The (direct) determinants of Practical Value (utility) of selection methods, besides validity, can be seen via other measures as variability of job performance (measured by dollar value of output of output as percentage of mean output) and selection ratio. Regarding the selection ratio, no hiring procedure has any practical value if you hire all applicants (ratio equals to one) and there are practical value gains extremely large if the company has the luxury of hiring only the top scoring 1%. In the other hand, when the variability of job performance is zero, all applicants will have the same level later if hired so the utility of all selection procedures is also zero as it doesn’t matter who is hired; while if it’s very large it’s an extreme case but that appears to be the reality for most jobs and would be even larger if all applicants were hired or selected randomly from an applicant pool, the variability operates do determine practical value as one is selecting new employees from the applicant pool.
The Validity of Personnel Assessment Methods over 85 years of research findings headed toward a Theory of the Determinants of Job Performance but many of those (and this) studies have several limitations that don’t let us hypothesize the whole picture. Sometimes, organizations use more than only two selection methods and so incremental validity from adding a 3rd predictor would be useful. Saying this, they didn’t examine combinations without GMA due to inexistence of estimates of predictor intercorrelation and, nevertheless, even within the context, some intercorrelations are not as precise as we would like them to be (for example, the job tryout is similar to extended job sample test so, in the absence of data estimating the job tryout correlation, use correlation of job sample). Gender or minority subgroups have been left beyond the scope of the study but differential validity says that general finding is that validities don’t differ appreciably for different subgroups and that subgroups differences are rare or nonexistent. The most relevant finding was the lack of predictive bias; even when subgroups differ in mean score, selection procedure score appear to have the same implications for later performance in all, hence same predictive interpretation of scores.
Indeed, employers can choose which methods to use but different methods and combinations obviously lead to different validities for predicting future job performance. Interests and amount of education have low validity while graphology has no validity and would be almost the same as hiring randomly. GMA and work sample as already mentioned has high validity but there’s more practical combinations with high compositive validity, GMA plus integrity test or structured interview, that can be used with entry level (no experience) and experienced and both predict performance on the job and in job training quite well; also, are less expensive than other combinations and in particular cases there are reasons that other combinations are better even with high but lower validity.
The cumulative research findings over the last century make it possible to create and test several theories of job performance with less psychological processes and a lot more basic understanding of human competence in the workplace. The validity of the personnel selection measure is directly proportional to the practical value of method: there are gains associated to the increase of the validity, big losses in reduced production if using low validity methods. While observing real world companies, researchers found that many employers are currently using suboptimal selection methods and so, they are unnecessarily creating a competitive disadvantage in a competitive world, even if they have better alternatives they could use.