Editing: Editing is a process that tries to ensure the accuracy, completeness, and consistency of survey data. Efforts are made at all phases of collection, processing, and tabulation to minimize reporting, keying, and processing errors.
Although some edits are built into the Internet data collection instrument and the data entry programs, the majority of the edits are performed post collection. Edits consist primarily of two types: (1) consistency edits and (2) a ratio edit.
The consistency edits check the logical relationships of data items reported on the form. For example, if a value exists for employees for a function then a value must exist for payroll also. If part-time employees and payroll exist then part-time hours must exist and vice versa.
For each function reported for the employees, the ratio edits compare data for the number of employees and the average salary between reporting years. If data fall outside of acceptable tolerance levels, the item is flagged for review. Additional checks are made comparing data from the Annual Finance Survey to data reported on the Survey of Public Employment & Payroll to verify that if employees are reported on the Survey of Public Employment & Payroll at a particular function the government also reported a corresponding expenditure on the Annual Finance Survey.
For ratio edits and consistency edits, the edit results are reviewed by analysts and adjusted as needed. When the analyst is unable to resolve or accept the edit failure, contact is made with the respondent to verify or correct the reported data.
Imputation: Not all respondents answer every item on the questionnaire. There are also questionnaires that are not returned despite efforts to gain a response. Imputation is the process of filling in missing or invalid data with reasonable values in order to have a complete data set for estimating state and national totals.
For nonresponding general purpose governments, dependent and independent school districts, and for special district governments, the imputations were based on recent historical data from either a prior year annual survey or the 2007 Census of Governments: Employment Component, if available. These data were adjusted by a growth rate that was determined by the growth of responding units that were similar (in size, geography, and type of government) to the nonrespondent. If there were no recent historical data available, the imputations were based on the data from a randomly selected responding donor that was similar (based on the same criteria) to the nonrespondent. For general purpose governments, and for dependent and independent school districts, the selected donor's data were adjusted by dividing each data item by the population (or enrollment) of the donor and multiplying the result by the nonrespondent's population (or enrollment).
Because of the merging of dependent and independent schools in Maine, this state had to be imputed by itself. We also had to use a crosswalk so that the proper prior year data would be used for imputing the nonrespondents.
Note: Between years 2002 through 2006, individual government imputed data were not released to the public. Beginning with 2007, the imputed data are available on the Individual Government Data file. Data flags are available on the Individual Government Data file to denote the imputed data.
Estimation: Estimation is the process by which sample data are used to project the value of an unknown quantity in a population. In the publications for employment statistics, total full-time employment, total full-time payroll, total full-time equivalent, total part-time employment, total part-time payroll, total part-time hours, and their coefficients of variation are published. Estimates of these major totals are made using a model-assisted approach called Decision-based Estimation. Papers on this methodology are included in the “For Further Information” section of the Survey of Public Employment & Payroll Methodology. A composite estimate for each state by function code variable can be obtained from the sample data and known 2007 Census of Governments: Employment estimates.
To obtain separate estimates for each state by function “cell” (e.g. Corrections for Minnesota), we use small area estimation. There are two straightforward methods to make the estimates, and better results are obtained overall by combining the two methods. The Horvitz-Thompson or HT estimator is a weighted sum of the sample data. Intuitively, each unit in the sample represents itself and possibly many other units. To calculate the HT estimator, each data point in the sample is multiplied by the number of units it represents, and then the units are summed. The synthetic estimator assumes that employment in 2011 is proportional to employment in 2007 for the same state and item.
These two methods have different tradeoffs. The HT estimator has no bias (the expected value equals the true value), but it can be sensitive to units with high weights. The synthetic estimator can be biased, but often has lower variance than the HT. We can improve the estimate by taking a weighted estimate of the two, called a composite estimate. Modified direct estimates replace the HT estimator in the first component of the composite estimate.
Sampling Variability: The data that are provided come from a sample rather than a census of all possible units. The particular sample that was selected is one of a large number of possible samples of the same size and sample design that could have been selected. Each sample would have yielded different estimates. The estimated coefficients of variation, which are provided for each estimate, are an estimate of this sampling variability. In this tabulation the coefficients of variation are expressed as percentages. The coefficient of variation (CV) is the ratio of the standard error to the expectation of the estimate. We used a Taylor series method to estimate the standard error.
State government employment and payroll data are not subject to sampling error. Consequently, state and local government aggregates for individual states are more reliable statistically than the local government only estimates.