Technical Documentation: 

Attendance and Turnover Test

Criterion‑Related Evidence of Validity for the Attendance and Turnover Test

Criterion‑related validity evidence describes the extent to which the test predicts important job outcomes. We conducted a study in which attendance was assessed by a supervisor's rating and turnover was assessed by having people self-report on the length of time they stayed in their jobs (longevity). Our research showed that the Attendance and Turnover Test predicts both attendance and turnover.

Criterion‑related validity evidence was collected by administering the test to 118 incumbents and by collecting self-report data on longevity and supervisor ratings of attendance. Table 1 presents the relationship (Pearson correlation coefficients) between the Attendance and Turnover Test score and the criteria.

Figure 1 shows the probability of excellent attendance (as rated by supervisors) as a function of the applicant's Attendance and Turnover Test score. Figure 2 shows the probability of staying on the job five or more years as a function of the Attendance and Turnover Test score.

Many employers prefer to determine the validity of a test for their own organizations. We encourage such research and our staff of industrial psychologists can provide any needed assistance.

Table 1. Validity of Test Scores for Predicting Job‑Related Criteria

Note: Sample Size 118.

Probability Not Certainty

The Attendance and Turnover Test can be used to improve the probability of selecting applicants who will become employees with good attendance and who stay on the job for a considerable period of time. 

This test, although very useful in applicant screening, is far from a perfect screening tool. There will be some applicants who are selected on the basis of this test who will be undesirable employees. Likewise, this test will reject some applicants who would have been acceptable employees. This happens in all selection systems. The incorporation of this test into employers’ current selection systems is intended to reduce such errors.

Applicants can be deceptive when taking the test. Applicants can improve their scores by lying or by enhancing their desirable qualities.  The ODD score described below is our attempt to identify such individuals.

If the test suggests that the applicant will have excellent attendance and stay on the job a long time, but the references and background checks reveal that the candidate has recently had poor attendance and changed jobs frequently, we suggest that more credence be given to the background check and reference information because the best predictor of future behavior is past behavior.

Figure 1. Probability of Excellent Attendance

Figure 2. Probability of Staying 5 Years or More On the Job

The ODD score attempts to identify unusually desirable responses. If an applicant obtains an ODD score of 4 or higher, additional information on the applicant should be collected. 

An applicant may obtain a high ODD score by having an unusually large number of desirable qualities. Such applicants may make good employees. However, one may also obtain a high ODD score by exaggerating one’s positive qualities. Such applicants may make less than desirable employees. The dilemma is that it is not readily apparent whether the applicant will be a good or poor employee.  

To assist in making a decision about an applicant with a high ODD score, additional time should be spent with the applicant in interviews and available background and reference information on the applicant should be carefully reviewed.

The Attendance and Turnover test is a composite of an attendance scale and a turnover scale. The attendance scale has an internal consistency reliability of .67. The turnover scale has an internal consistency reliability of .47. The attendance/turnover scale has an internal consistency reliability of .66.

The meaning of an applicant’s test score is increased when it can be compared with a reference group. We suggest that the most meaningful norms are those based on an organization’s current incumbents. The industrial psychologists at Work Skills First, Inc. can assist in preparing a norm table for individual organizations. However, such local norms are not always feasible to collect due to a limited number of incumbents or insufficient resources to test the incumbents.

For those organizations without local norms, we provide norms in Table 2. These norms were obtained by administering the test to 142 nursing assistants and home health aides in one hospital, two home health care facilities, and several nursing homes. We provide norms for the full sample of 142 incumbents and broken down by hospitals, nursing homes, and home health care organizations. The norms are provided separately by type of organization because the test scores varied meaningfully across the three types of employer. On average, scores are highest for the hospital employees and lowest for the home health care employees. The nursing home employees tended to score lower than the hospital employees but higher than the home health care employees.

The percentile scores associated with a given test score is the percentage of the employees in the norm group who scored lower than the test score. For example, for the attendance/turnover scale for the full sample norm group, a test score of 32 places the examinee at the 55‑56^th percentile. This means that 55 to 56% of the norm group scored lower than this score.

Table 2. Test Norms for the Attendance and Turnover Test Score for Nursing Assistants