Measuring Turnover: A Review of Traditional Measurement Methods and Development of Measurement Techniques Based on Survival Analysis

Article excerpt

INTRODUCTION

O'Reilly (1991) suggested that research into turnover has entered a "fallow" period and is in need of some stimulus or new direction to more fully address the process of employee withdrawal. Indeed, most research into turnover for the last half century has been based on the March and Simon (1958) model. This model implicitly assumes that the decision to leave is based upon an individual's determinations about the desirability of quitting and the opportunity or ease of doing so. This rational process model has been extended and tested by many researchers over the years. Price (1977) added another dimension by suggesting that the operant factors are dissatisfaction and opportunity to leave. Dissatisfaction suggests the operation of an emotional component to the decision. Mobley (1977) also focused on the connection between job dissatisfaction and turnover.

Mowday, Porter, and Steers (1982) developed a more comprehensive model in which organizational commitment plays a central role. Organizational commitment, it turns out, is a multidimensional construct which includes both a cognitive and an emotional component.

This paper will first review traditional approaches to operationalizing and measuring turnover. Two approaches for measuring turnover based on survival analysis will then be developed and discussed. Measurement is central to both the development of theory and the empirical validation of theory. The more uniform our measures, the more clear will be the theoretical nuances developed to understand the object of our models. Turnover is one area where standard measures have been lacking.

TRADITIONAL TURNOVER MEASURES

The study of employee turnover has produced many options for its measurement. Much of the earlier research considered turnover to be captured in the stayer/leaver dichotomy. Because of this focus, and because cross-sectional tools make data collection and analysis much simpler than longitudinal techniques, measurement was generally based on some variant of the percentage of employees staying or leaving over some period of time--generally one year.

Price (1977) gave a comprehensive description of seven methods in general use. These include:

1. Average length of service. This method sums the length of service for all employees and divides by the total number of employees.

2. Accession rate. This method sums the number of new hires in the period and then divides by the average number of employees over the period.

3. Separation rate. This method sums the number of employees who left during the period and divides by the average number of employees over the period.

4. Stability rate. This method uses the ratio of the number of employees who were employed over the full period by the number employed at the beginning of the period.

5. Instability rate. This method divides the number of employees who left during the period by the number of employees at the beginning of the period.

6. Survival rate. The number of employees who remain through the entire year (as in 4) divided by the number of new employees.

7. Wastage rate. The number of employees who leave during the period (as in 5) divided by the number of new employees.

While each of these methods provides utility for viewing turnover from a particular perspective, research into such an important topic requires a measure with both universal application and a precision not uniformly present in these methods. A more rigorous approach can be developed utilizing the mathematics of survival analysis. This approach is used both in the rigorous and precise statistics of mortality tables and in descriptions of radioactive decay.

SURVIVAL ANALYSIS

An eighth method, not described by Price, is survival analysis. A simple version of survival analysis is used in the life insurance industry to measure the survival of new agents. …