Human Resources Management for Public and Nonprofit Organizations

202 Human Resources Management for Public and Nonprofi t Organizations

There are two types of criterion - related validity: predictive and concurrent.

Both demonstrate a statistical relationship between the test scores and per-

formance measures. What differs is the sample of test takers and the amount

of time between taking the exam and obtaining job performance measures.

The meaning of this will become clearer as the procedures are explained.

Predictive validity. The procedure for predictive validity studies is as follows.

First, a job analysis is conducted to determine the relevant KSAOCs for

the position. Based on the KSAOCs, an exam is developed, and candidates

are tested for the job using the selection instrument. Next, candidates are

selected using some other standard. For example, decisions may be based

on letters of reference, the interview, or previous experience.

After the new employees have been working for the organization for

six months to one year, job performance measures are obtained. These

could be supervisory ratings or performance in a training program. After

these measures have been collected, a statistical analysis is conducted to

evaluate the relationship between the test and the job performance mea-

sure. If the test is to be considered useful for selection, the high scorers on

the test should have higher performance ratings and the low scorers on the

test should have lower performance ratings.

You are probably wondering why the test is not used to select employ-

ees. Why develop an instrument and then ignore it? The test was not

originally used to select candidates because to determine whether the test

could predict performance, both high and low test scores are needed. If

only people who scored high on the test were selected, it would not be clear

whether individuals with low scores on the test could be good employees.

If individuals with low test scores turn out to be the best performers, then

the test is not measuring what it was intended to measure in order to pre-

dict which applicants would be most successful on the job. If it is not yet

known whether the test is valid, making selection decisions on its results

would be premature.

Concurrent validity. Concurrent validity also statistically demonstrates a relation-

ship between the predictor and criterion measure. However, the procedures

are different than they are in a predictive validity study. Instead of applicants,

incumbents take the examination. Because incumbents are already working

for the agency, job performance data can be collected immediately. Then the

predictor and criterion data can be correlated, and a relationship between the

selection instrument and job performance can be determined.

There are some factors to consider when doing a concurrent validity

study. Because job incumbents are being used, it must be recognized that