Methodology

The Partnership for Public Service and the Institute for the Study of Public Policy Implementation created the Best Places to Work rankings to provide a comprehensive and authoritative rating of employee engagement across federal agencies and their subcomponents.

The data used to develop these rankings was collected by the Office of Personnel Management (OPM) in its most recent Federal Human Capital Survey, completed in the summer of 2006. This survey was distributed to a stratified random sample of more than 436,000 executive branch full-time permanent employees, making it the largest survey of federal employees ever undertaken. The survey achieved a response rate of 57 percent, resulting in a final sample of more than 221,000 employees. The agencies included in the study account for 97 percent of the executive branch workforce.

Working with the consulting firm Hay Group, the Partnership for Public Service and the Institute for the Study of Public Policy Implementation created a statistical model to transform this raw survey data into specific measures of workplace engagement. We created the overall Best Places to Work index plus composite scores for each of ten work environment categories. We then used this information to calculate the results for each agency and subcomponent, as well as demographic groups in these organizations. Finally, we performed statistical analysis for each of the 283 federal agencies and subcomponents to identify the relationship between the 10 workplace engagement dimensions and the overall Best Places to Work index score for each organization.

We rated and ranked all agencies and subcomponents with 100 or more full-time permanent employees. Agencies with at least 2,000 full-time permanent employees are included in the "large agency" category.

The snapshot data for each agency or subcomponent was gathered from OPM's FedScope database or directly from the individual agency or subcomponent. All current workforce information used to build the agency snapshots is from fiscal year 2006, except for demographic data, which is from 2004.

In a few cases, survey results were available for the questions we used to calculate the employee engagement index, but not for the demographic groups or the workplace environment categories. We assigned "no data" to those areas where data were not available.