Academic journal article Public Personnel Management

Racial Differences on Job Analysis Questionnaires: An Empirical Study

Academic journal article Public Personnel Management

Racial Differences on Job Analysis Questionnaires: An Empirical Study

Article excerpt

Selection measures that are content-validated may be biased if the job content domain, as defined by a job analysis, is dependent on the characteristics of the people who hold the job or who complete the job analysis ratings. The purpose of this study was to determine if a relationship may exist between race and job analysis ratings. Clerical workers employed by state government agencies completed job analysis questionnaires on their own jobs and their co-workers' jobs. The workers evaluated job tasks which belongs to three content domains. lob content for the clerical job was related to race, such that the relative ratings of the three job domains for the black incumbents who had black coworkers differed from those for the other three racial groups - black incumbents who had white coworkers, white incumbents who had white coworkers, and white incumbents who had black coworkers. It was argued that these differences represented actual job differences. Under these conditions, a job analyst should not define different job titles for different racial groups, but attempt to understand why the differences exist. Through such analyses and evaluation of results, bias in application of the job analysis data, such as defining the content domain of selection measures, can be avoided.

Guion (1978) noted that tests developed from content domain samples of jobs are likely to be viewed as unbaised. An assumption underlying this belief is that "The job content domain is independent of the characteristics of the people who hold the job" (Guion, 1978, p. 502). He suggested two conditions under which this assumption may be violated. (a) The tasks and duties associated with a job may differ for various groups of job incumbents. For example, task differences might occur as a result of an affirmative action program. In order to be in compliance with an affirmative action program, an employer may have minority and majority group members occupy jobs with the same job title, but with different job duties. (b) Job duties may not differ for distinct groups of incumbents, but the stylistic approaches the groups take to perform the job may differ. If members of a minority group are socially isolated on a job, they might develop a stylistic approach that differs from the one used by the majority group members. When stylistic approaches to a job differ across groups, bias might be minimized by using task-oriented rather than worker-oriented job analysis to define job content because task-oriented job analysis focuses on the work performed on the job, the end products of these activities, and the equipment and/or reference materials used to accomplish the activities rather than on the workers who perform it (Miller, 1962).

Regardless of how the job content-incumbent independence assumption is violated, if a job analysis assesses the aspects of the jobs on which the groups differ and only one group is sampled, the test based on this job analysis will be biased. It should be noted that bias occurs when an application of a technique results in some negative consequence to a group of individuals. Consequently, if the independence assumption is violated, it is not the job analysis that should be considered bias, but the applications based on the job analysis, such as using a test for selection purposes.

From the above discussion, a corollary to Guion's assumption is proposed: a test may be biased if the perceived job content domain on which it is based is dependent on the characteristics of the people who assess the job, even if all incumbents perform the job identically. In other words, if two groups of incumbents complete a job analysis questionnaire differently (regardless of the content of the jobs held by the two groups), a test based solely on one group's job analysis data will be biased. The degree of test bias would be directly related to the degree that the groups differ on the job analysis questionnaire and the extent that any one group is overrepresented or underrepresented in the job-analysis sample. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.