Appendix A - Part 1

Technical notes

a. Use of average salary instead of average grade

Comparing the average salary between groups of employees appears to be the best method of measuring the extent of the inequitable distribution of a particular group. Average grade, which is often used for this purpose, can only show that one group is higher than another, but it does not provide an adequate measure of the difference. The fact that one group has an average grade of 12.8 while another group's average is 13.2 may or may not indicate a significant disparity. What is .4 of a GS grade? Or, in comparing this difference to that for another bureau or office where these groups of employees had average grade levels of 10.9 and 11.6, for example, it would be difficult to evaluate which office shows the largest disparity. A given difference in GS grades would have different meaning at different grade levels. In addition, because GS levels are not really additive -- a GS 6 is not two times as high as a GS 3 -- the level of the average yielded by aggregating these GS grades becomes rather meaningless. Using average salary instead, avoids these problems.

b. Aggregating the education data

Because of the manner in which the sample for the education survey was selected, the distributions and averages shown in Chart 16, p. and Table 8, p., had to be derived in addition to those shown in the series of tables containing the data on educational attainment, Tables 2-12, pp. - ). In order to include an adequate number of all employees in the sample, it was stratified according to 6 groups -- three race groups each of which was divided among males and females. (This is discussed more fully in Part 2 of this appendix.) Because of the stratification in the sample, each of the 6 groups is represented differently in the sample than they are represented in the Department's workforce. For example, blacks comprise 37 percent of all professionals in the sample, but they constitute only 14 percent of all professionals in the Department's workforce. As a result, blacks would be reflected disproportionately in an average of the racial groups in the sample.

To compensate for the distortion noted above, the distributions and averages in Chart 16 and Table 8, p., were calculated by assigning a weight to the professional category in each of the six stratified groups. Each of the professional categories in the six groups were assigned a weight corresponding to the segment that they comprise of the professional workforce. For example black male professionals who comprise 29 percent of all professionals in the sample were given a weight of .3 so that they would reflect only the proportion that they represent in the DOL workforce, 9 percent.

The use index reflects the representation and distribution of minorities and females in the National Office and Field units of an

adminstration. For each of these units, the minority or female representation as a percent of their goal is multiplied times the percent which their average salary comprises of that for all professionals. In addition, both the representation as a percent of the goal, and the average salary as a percent of total average salary are constrained to be no greater than 1.0. This was done to prevent a good representation to compensate for an inequitable distribution, or vice versa. For example, consider a Field unit of an administration where the representation of blacks is 9.6 percent and their average salary is 90 percent of that for all professionals. In terms of representation, 9.6 percent constitutes .80 of their 12 percent goal. Multiplying this .80 times the .90 which the average salary of blacks comprises of that for all professionals gives a use index of .72. A similar use index is calculated for the Field unit of the administration. The indices for the National Office and Field units are then combined according to the percent of professionals in that administration located in the National Office or the Field.

r

The use index for the Department is an average of the use indices for each of the administrations. These indices are combined according to the percent of the Department's professional workforce that is found in each administration.

d.

Classification of professionals and nonprofessionals

The time-in-grade data show that there are a significant number of professionals with over 12 months of time-in-grade and in grades GS 6, 8, and 10.. This contradicts the Office of Personnel's definition of professionals as "employees moving at two grade intervals every year up to GS 11." As a result the Task Force examined the professional-nonprofessional classification system used by personnel. This investigation covered the classification system in effect at the time of the report, and the slightly modified system that will be used when the IPPBS becomes operational. The examination which used the Personnel Staffing Report and the Handbook of Occupational groups, and series of classes revealed several professional series, even in the new system, where individuals doing nonprofessional work were listed as professionals. A list of questionable series follows:

In addition, a spotcheck of the individuals listed as professionals, who were not advancing like professionals should, disclosed that some were indeed doing professional work.

e.

Calculating the number of years that it would take to achieve parity or an equitable distribution

This note explains the calculations used to derive Charts 17 and 18. The first step is to subtract the percent of all professionals at a particular grade with less than a year of time in grade from the percent of the minority group professionals (blacks, for example) with time in grade of less than a year. If the difference is positive it is multiplied times the percent blacks comprise of the grade. This gives the percentagepoint increase in the proportion that blacks will comprise of all professionals at that grade next year based on last year's rate of entry for blacks compared to the rate for all professionals at this grade. The percentage-point increase is then divided into the difference between the goal for the representation of blacks and the proportion they comprise of the grade. This provides the percent of their goal that blacks would attain over a one year period. Dividing the percent into 1.0 gives the number of years that it would take to achieve the goal.

Appendix A - Part 2

Description of the Survey of Attained Education

Purpose

The survey was conducted to collect information on the educational attainment of DOL employees for inclusion in the Task Force's evaluation of the status of minorities and women. This information seemed crucial to our study, but, unfortunately it was not contained in the Department's computerized personnel records. A survey of a sample of DOL employees appeared to be most efficient means of gathering the data on education. 1/

The information from the survey will be included in a regular quantitative analysis as well as in a more sophisticated statistical study. The quantitative analysis will measure and evaluate the difference or similarity in education among racial and sexual groups within the professional and nonprofessional classifications. It will also compare the status of employees with equal levels of education across racial and sexual groups. The sophisticated statistical analysis will estimate the effect of education along with other explanatory factors in accounting for the lower position of minorities.

Sample design

The ideal is to select the smallest sample which will provide a specified level of certainty in generalizing for the population. Without advance knowledge of the characteristics of the information being sampled, it is difficult to determine the sample size needed to achieve a required level of certainty. As a result of the absence of data on the educational attainment of either DOL or Federal employees, we were unable to derive a sample size that would assure a specified degree of certainty. Only a general estimate of the required size was possible. This was based mainly on the advice of statisticians who had substantial experience in sampling. The size of the sample finally selected was determined by this advice together with our evaluation of the limitations on what size sample could be collected within a certain time period. A sample of 1.000 employees was chosen.

Having decided on a sample size, the second major consideration was to select the sample in a manner which would insure that it adequately represented all the racial and sexual groups for which separate information was desired. This is especially important since these groups are represented in different numbers in the Department's workforce and a completely random sample of the whole population might not include enough employees of a particular group nonblack minority females, for example

1/ This decision was reached during discussions with Chuck Roberts and Bill Dedham.