Normalization of Employee Performance Review Scores

Rating Patterns of Managers in a Performance Review

During a performance review, managers have a tendency to rate differently – some are naturally lenient while some fall on the opposite end of the spectrum. However, more often than not, it is the employees reporting to them that experience the brunt of these variations. This impact becomes crucial in an environment where employees are offered performance based remuneration/ incentives.

Normalization of scores is intended to introduce greater objectivity in the employee performance review software of an organization.

It may be observed from the figure above that Manager ‘A’ has the tendency to rate subordinates at 7 to 8 points on a performance rating scale of 1 to 10. Manager ‘C’, on the other hand, is highly conservative and the best ratings are still in the range of 5-6 points. Thus an average performer with Manager ‘B’ is equated with the best employee under Manager ‘C’ and with the lower average employees of Manager ‘A’.

The training of Managers A, B & C on the rating norms may improve this trend subsequently. However, what should the company do with the evaluations already conducted? How has the company management looked into this problem which has an impact on promotions, compensations and career management of all employees? Some outstanding performers (placed under a harsh reviewer like Manager ‘C’) may quit the organization and some low performers (placed under a lenient reviewer like Manager ‘A’) could become tomorrow’s managers. This vicious cycle tends to boost average performers who cling to their jobs and promote mediocrity in the organization. A performance-driven company need to normalize the rating trends of their managers and weigh employee performance against the proper average rating.

What does normalization mean in a performance review process?

Assume there are ten managers in an organization who are reporting to ten different executives each and they, in turn, report to three different senior managers. In this scenario, there are 13 different appraisers who are reporting on 110 employees in the organization. Among these employees, 100 are at the same level and 10 are at the managerial level. Each of the 13 appraisers has a different rating style meaning that employees reporting to them have a high degree of variability in their performance appraisal scores. The process of balancing this variability is called ‘Normalization’.

Normalization Process in Employee Performance Review

The process comprises the following steps:

• Statistical mean of the rating pattern of all reviewers at the same level in an organization, across various departments, is computed. Let this mean be ‘M’.
  
• Statistical Mean for each of the appraisers at the same level (ie, for all the ten managers in the example given above) is computed. Each manager would have appraised 40 to 50 employees over time. This mean is ‘Mi’ (i = 1 to 10).
  
• A correction Factor (CF) for each of these managers is then computed CF = Mi/M. Its value, for example, will be 1.0 if the rating pattern of a manager is the same as the statistical managerial mean.
  
• Performance score of each individual employee is then divided by the CF for his/her manager to compute the normalized value, which is then utilized for management decisions.

Connected issues of Employee Performance Review Process

1. To implement the normalization process in employee performance review, companies need data from appraisals that a manager has conducted over a period of time. This is so that the statistical mean for each manager could be computed based on 40 – 50 appraisal reports written by him/her. This requires a data base approach.
2. What happens when a manager has not completed the requisite 40 to 50 appraisals? The approach normally adopted, is to compute the mean Mi and CF for the manager based on the number of completed appraisals, but not to apply the CF to the performance scores of subordinates without discreet approval by the management. (Note: Some organizations prefer to use the normalized score awarded by the reviewer).
3. Since the Appraisers tend to refine their rating tendencies automatically as the years roll by, some organizations prefer to use the Moving Mean concept (MMi) for each manager rather than a constant Mi. The moving mean for a Manager is an indicator of the improvement in his/her rating trend. CF = MMi/M is the computing algorithm used in this case.
4. In mid to large organizations, with 4 to 5 different levels of appraisers and a number of departments, functions and employees varying in strength from 200 to 5000, it is not possible to normalize appraisal scores without automation of the entire process through an employee performance review software.

Conclusion
In some organizations, normalization of performance scores is managed by a high level committee and this process is not transparent to the employees. If such committees consider the rating patterns of various managers such as MMi (discussed above), their decisions would be data based and objective.

A customized and configurable performance review software can be used to implement the above discussed normalization features and help in mapping the client needs.