• Contact
  • Client Portal

Statistical Compensation Analysis - Step-By-Step

A breakdown of the compensation analysis standards and voluntary analysis guidelines

Multiple-Regression Compensation Analysis - Steps Involved for Analyses
Step 1: Classify Employees into SSEGs (alternately, job groups and/or job titles can be used if SSEGs are not available) The first step for conducting a compensation analysis is to build a database. This database should contain all available incumbents at the employer, or at the specific location under inquiry. All employees subject to inclusion in an AAP should be included in the dataset (including full-time, part-time, and temporary employees as required). Considerations for making SSEG classifications include: similarity in job content, skills and qualifications involved in the job, and responsibility. Consideration should be given to the sample size of SSEGs in order to conduct meaningful statistical analyses.
Step 2: Create Data Set for the Analyses A data set needs to be constructed for the analyses that includes the variables that you believe drive your company's pay decisions. Demographics, pay levels, and other factors are necessary. The recommended initial variables are below:
  • Employee ID
  • SSEG (see above)
  • Race/Ethnicity (White, Black, Hispanic, Asian/Pacific Islander, American Indian/Alaskan Native)
  • Gender
  • Date of last degree earned
  • Highest degree earned (and type and degree area)
  • Date of birth (this information can sometimes be used as a substitute for prior experience outside the company)
  • Time with company or Date of hire
  • Time in current position or date of last change in grade/title
  • Current annual salary or hourly wage
  • Part-time vs. full-time status
  • Exempt vs. non-exempt status
  • Job title
  • Grade level or salary band classification
  • Employee location (if not housed at the facility)
  • Prior experience data (can take a variety of forms)
  • Job performance ratings
Step 3: Review Analysis Feasibility This step is perhaps the most important, and includes two parts. The first step is checking each variable (above) for certain criteria that will allow them to be properly evaluated in a regression analysis. The second step is to evaluate the statistical power of the analysis.
Step 4: Conduct Multiple Regression Analysis

Utilizing the BCG compensation analysis tool, BCG staff will conduct an analysis of all available groups (e.g., SSEGs, job groups, job titles).

LEVEL 1: Conduct regression to determine whether gender and/or minority/non-minority group are significantly correlated with salary (which could indicate disparity).

LEVEL 2: Conduct multiple regression analyses (including job factors) for all job groupings where statistical significance occurred in Level 1.

Step 5: Evaluate Interactions and Interaction Terms (Level 2) If any interactions appear evident, create rational interaction terms by multiplying dummy-coded variables and continuous variables into new variables and loading into the regression equation (built following the process above) as a "Block 2" analysis to see if they add significantly to the regression model beyond their independent inclusion in Block 1 analysis.
Step 6: Analyze Fit of the Regression Model & Evaluate Regression Assumptions
(Level 2)
With final model, evaluate beta weights, significance of slopes (t statistic), R2 value and statistical and practical significance, adjusted R2 value (to assess robustness), and Standard Error of Estimate, homoscedasticity, and make corrections where appropriate.
Step 7: Draw Conclusions and Provide Executive Summary Report Create an executive summary report that describes the methodology followed and interprets the results. Hard copies of the results will be sent to the client and electronic copies will be available as needed. A conference call will then be scheduled to discuss the findings.
Subscribe to Our Newsletter