« PreviousContinue »
test the proposed standard for its applicability and impact on GAO audit
Since the proposed "Evaluator" standard covers substantive (i.e., nonsupervisory) work and most of the GS-13 and 14's are supervisors it is possible that analysis on a factor by factor basis would obscure differences based on supervisory responsibilities exercised. To cover this
contingency, the "whole job" was ranked independent from the standard by *-. •& group of classifiers and a group of Gao managers..--The classifiers par- . : titioned the questionnaires into stacks of 20 based upon an evaluation of
• the whole job and ordered from the strongest stack of jobs to the weakest stack of jobs. The classifiers then went through each stack and through a method of round-robin comparisons of sets of 5 jobs came up with a whole · job ranking for each questionnairc. A group of GAO managers also ranked
the 20 questionnaires in each ståck using the round-robin, whole job :. evaluation method. RESULTS .
The mean number of points for each factor and the mean total points by racial and sex subgroup appear in Tables 2 and 3 respectively. The significance of the differences between the means was tested using a one-way analysis of variance.
Examination of the data in Table 2 revcals that although nonwhites generally received fewer points on most of the factors, the differences were not statistically significant. Likewise the data in Table 3 shows a similar pattern of results with females tending to be credited with an insignificantly lower number of points.
· While no great disparities were found in the factor points carned by race or sex, some additional analyses were needed before a firm conclusion could be drawn. The FES system calls for grades to be determined by a conversion of the points into grade equivalents. Therefore while the total number of points separating the nonwhites and whites (73) and females and males (63) are not great enough to reach statistical significance, they
may be practically significant if they cause a disproportionate number of: __nonwhites or females to miss the cutpoints for inclusion in the next... higher grade level.
* Tables 4 and 5 show the distribution of completed grade levels by race and sex respectively. While the data shows a tendency for white males to have. somewhat higher computed grade levels than their minority and female counterparts, these differences again did not reach statistically signi-... ficant levels. .
. . . . . . . . . . Because the sample included a large number of supervisory personnel . and the classification standard covers only non-supervisory work, the ex
amination of computed grade levels versus actual grade levels has little ... · real meaning. However, to the extent that supervisory duties are equitably
distributed, it would be expected that there would be no differences among the subgroups on grades which are supportable solely on the basis of nonsupervisory duties. Among the racial subgroup sample, 23 of the nonwhites (26.7%) and 29 of their white male matches:(33.7%) wound up with computed grade levels which were at or above their actual grade levels. In the sex subgroup sample, 22 of the females (25.6%) and 23 of their white male matches' (26.7%) grades were supportable without receiving credit for