If selection methods are invalid, employee selection decisions are no more accurate than decisions based on a toss of a coin. Validity is the degree to which a measure accurately predicts job performance. Selection methods are valid to the extent that predictors measure or are significantly related to work behavior, job products, or outcomes. The process of demonstrating that a predictor is significantly related to a measure of work behavior, job products, or outcomes is validation.
The validation process demonstrates that a significant statistical relationship exists between a predictor and a criterion measure of successful performance on a job. A predictor is any piece of information that can be used to screen applicants. Predictors include information from application blanks (education level, experience, and so on) and reference checks; scores on tests of skill, ability, or aptitude; data from interest and personality inventories; and interviewer ratings of an applicant. Criterion measures are any measures of work behavior, job products, or outcomes that have value to an employer. Job success is an abstract concept that means different things to different employers.
Major Types of Validation
There are three major types of validation used to validate predictors. They are (1) criterion-related validity; (2) construct validity; and (3) content validity.
Criterion-related validity. A predictor has criterion-related validity if a statistically significant relationship can be demonstrated between the predictor and some measure of work behavior or performance. Examples of performance measures are production rates, error rates, tardiness, absences, length of service, and supervisor’s ratings. Suppose a department store uses as a predictor for its sales personnel one year of sales experience. To validate this predictor, the employer would have to demonstrate that a statistically significant relationship exists between one year of sales experience and some measure or measures of work behavior or job products, perhaps number of sales and/or low percentage of errors in ringing up purchases.
Construct validity. Instead of directly testing or using other information to predict job success, some selection methods seek to measure the degree to which an applicant possesses psychological traits called constructs. Constructs include intelligence, leadership ability, verbal ability, mechanical ability, manual dexterity, etc.
Constructs deemed necessary for successful performance of jobs are inferred from job behaviors and activities as summarized in job descriptions. They are the job specifications part of job descriptions. Construct validity requires demonstrating that a statistically significant relationship exists between a selection procedure or test and the job construct it seeks to measure. For example, does a reading comprehension test reliably measure how well people can read and understand what they read?
Content validity. A selection procedure has content validity if it representatively samples significant parts of a job, such as a filing test for a file clerk or a test of cash register operation for a grocery checker. Selection tests that approximate significant aspects of a job are called job sample tests. Job sample tests require applicants to perform certain aspects of a job’s major activities, thus demonstrating competence at tasks which are an actual and important part of the job. Significant aspects of a job are determined through job analysis and set forth in job descriptions of jobs. Job sample tests should approximate aspects of the job as closely as possible, since this increases content validity.