Patterns of Pay in N.C. State Government Paper

    (

C

PATTERNS OF PA'T IN N.C. STATE GOVERNI'IENT

Thts Study and Rcport were made poss.ible by a grant under the Intergovern-
mental Personnet Act (IPA), Grant Nos. 80-12 and 81-5.

(
T PLAINTIFPT

ofilBlT .
7L
Gmgre s



(

(,

qxEcuTrvE slrl'tMARY

Thts research report examines gay patterns by race and sex erithin
the North Carollna State Government workforce. It conslsts of three
chapters. The first looks at salary paEtern differences among race/sex
groups when one other variable, such as educat.ion or years of aggregate
service, is taken into account. These data are presented in graphic
and tabular form and are intended to give a detailed picture of salary
differences among these groups. Chapter Two involves a Eore elaborate
statistical analysis in which ntore than one relevant influence on salary
can be assessed at one time. The multivariat,e staEistical technique
of multiple regression here will estimate the differences in salaries
among race/sex groups with the effects of number of relevant influences
on salary taken lnto account. The final chapter in this rePort deats
wlth the issue of pay equit,y or whaE has become known as the "comparable
northr. issue. Two samples of jobs are evaluated on the basis of equivalent
pay for equivalent work as measured by two job point evaluation syst.ems
developed to assess the pay strucEure of the state Eovernment workforces
of Idaho and l{ashingt.on. Simple regressions wil l compare the actual
salaries of North Carolina positions with those predicted by the point
value under the Idaho or llashingEon systems. The gap between the figures
rill Ehen be analyzed for any patterns that, emerge in regard to the racial
or sexual composition of the employees. The variety of analytic tools
employed ia this report will provlde different viewpoints from which
to detect pat.terns of Pay in state government.
ln this sumrnary by chapter.

The results are reported

In the tabular analyses in Chapter'One, as a rule, salary Patterns
emerge in which white males are disproportionately represented in the
salary ranges above $13rOOO while white females, black males and black
females are overrepresented at the lower end of the salary scale. Among

the various departments of staEe government t ith over 9OO employees,
the only exception to this rule is the Department of CorrecEion where
all four pay patEerns t.ake on a similar form. The Department of Correction'
tt should also be not.ed, has the lowest percentage of employees in the
$25,OOO and above salary rang,e

I{hen relevant control variables were each assessed'for their effect
on this overall pattern, the tabular analyses showed:

At every education level, white males enjoy a salary advanEage
over their whire female, black male and black female peers.
Only among holdcrs of SraduaLe degrces does any other grouP
(in this case, black males) begin Eo approximate the salary
pattern of white ma1es.

Wlritc males hoLd proporcionately morc jobs that require highcr
educacional attainments than they actually had (19% versus
l2L lor white ft'malcs, 9% for bla<:k nrales, and 1O% for black
fema lcs).(



(

white males also hold proporrionately fewer jobs rhar had
lower educational requirements chan chey actualry possessed
TWrc 31% for white females, 32% for Llack males, and 36%for black females).

rncreasing years of aggregate service pay off more handsome-I'y for white males than any of the other subgroups, especiallyin the uppermost salary ranges. This superiority is compoundedby the advantage white males hold in years of service, thatls while white males make up onry 4r"L of those state employeeswith 1 year or ress of service, they constitute 6g% of all
employees with Io or more years of employment with state
S,overnment.

The patterns controlling for age reflect those seen aboverith exception of youngest age group (age 2s and below) where
all four subgroups exhibit a similar salary disrribution.
rc is not until the 36-45 year o1d caregory that the familiarpicture 

.begins _to emerge with white males dominat.ing the salary
ranges above $15,ooo and the oEher subgroups disproportionately
represented below that mark

One of the primary vehicles for perpetuating salary differences
aIDonB the races and sexes has been occupational segiegation. Some jobs
are known as t'female. or "bIack, jobs and the posilions in which th"r.
SrouPs dominate are for the most part among the lowest paid in most organi-zations. In order Eo concrol for such occupational differences, salarypatterns were examined in occupational categories under two.classification
systems, one developed by the North Carolina Office of St.aEe personnel
(osp) and the other by the Equal Employoent opportunir,y comrqission (EEoc).
The following highlights emerged:

Arnong the clerical and office services classes in the osp
system' a disproportionately high share of white males earn
$13rooo and above despite the tightly clustered range of pay
grades and overwhelming female numbers in this category.

Jobs in the rnstitutional services and Human Services cate-gories have the two most homogeneous pay structures of the
12 osP classifications examined. AII the rest show some variant
of the dominant. salary patt.ern where white males are dispro-portionately represented at the top end of the pay st,ructure.

The officials and Administrators EEoc classification shows
a distinct separation by sex but not by race. Black males
in this category parallel the salary pattern of white males
but, the gap between male and female managers remains.

(

The Paraprofessional occupations are the most
of any of the 8 EEOC classification. AI"t the resr
the Tcchnicians and SkiIlcd Craft classes, show
advantagc for whit.e males.

homogeneous
, especial ly
a dist inct

- 1L -



(

The final exarninat.ion of jobs in the tabular section involves
a comparison of race and sex segregaEed posiEions. Those Jobs above

95% white or 557" black are defined as race segregated. Those jobs where
either gender makes up more. Ehan 7O"L of the incumbents are defined as

sex segregated.

Under these definirions, 11r259 or 22.57. of a1l scate employees
sork tn race segregated Jobs. An examination of each subgroup within
rrrrhiterr and t'black" job classes reaches a similar conclusion in each

case: each group does bet.ter {n I'white'r than 'rblack" Job classes.

Arnazingly, 36,O51 or 72.17" of all state employees met our deflnition
of working in a sex segregated Job. The salary Patterns of the four
subgroups show more differences than in race segregated Jobs. Iltrite
aales actually are better represented in the upper satary brackets in
rrfemale" job classes than in'rmal.e'r ones. For white females the opposite
holds true they are better paid in jobs where men predominate. Black
females also show this pattern but to a lesser degree. In sex segregated

Jobs of either kind black males do poorly; they show a few representatives
at the upper ends of both the "male" or "female" job classes.

These tabular analyses presenE a great deal of information but
are limited to an examination of one "conErol variablerr at, a time. In
Chapter Two, a more sophisticated approach is taken enabling a number

of relevant variables to be statistically controlled. Using a muttiple
regression analysis, even when the factors of education' years of aggregate
service with staEe government,, age, and proportion of supervisory personnel
are aIl simultaneously statistically account.ed for, salary penalties
of 92529.L9 for white females, $22L2.62 for black malesr and $3271-O8
for black females emerge.

As eartier analyses indicated, occupational category is a great
influence on salary. Performing this same regression analysis itithin
six occupational groupings reveals both similaritles and differences
among them. In all six regressions white males reEain their salary advantage
over the other three subgroups. However, the returns to education among

officials and professionals are far greater than to any other of the
other groups. Additionallyr among officials and adminisErators' years
of aggregate service had no significant impact on salary. As far as

absolute salary costs 80r black females are under the greatest dollar
disadvantage in all job streams except among technicians and clerical
workers, where black males do the poorest. FinalLy, the equations for
officials and administrators and for professionals explain far less of
the overall differences in salaries than do the ones for the other four
subgroups; unmeasured influences on salary missing from these analyses
have stronger effecLs in these tr.ro occuPational areas than in the others
examined.

Separating employees along racc/sex lines and predicting salaries
reinforces the findings already presented. White males enjoy a higher
rate of rcEurn to educat ion, years of aggregate scrvice r suPervisory
posit-ion, and age over thc othcr three subgroups

- lll -

(

(

Adding controls for



(

occuPational cateEory produces a significant improvement in the amount
of variance in salaries explained for all four race/sex groups.

Slnce both different returns Eo, and different levels of education,
aggregat,e service, ag€, and supervisory position are involved in the
salary differences among race/sex groupsr sorn€ artificiaI manipulat.ions
are used to assign the relative weight to these two sets of differences.
lJhen the levels of education, aggregate service, etc. of white males
are subscituted into the equations for the other three groups, black
atales benefit more from this change than either white females or black
females. However, when white male rates of ret.urn to these variables
are substituced for those of Ehe other Broups, both female categories
benefit more than do black males.

The above regression results provide absolute dollar values to
atEach to influences on salary but their statistics are neither directly
comparable nor explicitly ordered. A path analysis provides a theoretically
based explanatory model and standardized scatistics that allow direct
comparison of effects. The overall path analysis shows that while education
has Ehe single most influential impact on salary the effects of race
and sex are also significanE. The considerable direct effects of race
and sex on salary (that is, those noE transmltted through differences
ln educational levelsr )€ars of aggregate service, occupational placement,
or supervisory placemenE) indicate that. otherr perhaps, illegitimate,
sources of salary disparities are present.

Separate path analyses within the six occupational streams show
that race and sex retain their influence in each case but the explanat.ory
Patterns fa11 into trro distinct patterns. Officials and bdministrators,
professionals, and (to a lesser extent) skilled craft workers share one
causal PacEern where education dominates and race and sex are relatively
minor influences on salary. On the oEher hand, technicians, office and
clerical workers, and service and maint.enance employees all show aggregat,e
service as the clearly dominant factor in salary determinaEion. Race
and sex also have a far greater impact on salary than ln the first group
of occuPations. The sum total of the findings indicate rhat the ability
of state Eovernment to close the salary gap between white males and other
SrouPs ought Co be greater in t.he latter set of occupational groups where
intraorganizaEional experi,ence plays a greater role in salary determinaEion.

The overall picture obtained from the regression and path analyses
in Chapter Two is one that features the advancages of white males over
other race/sex groups. Even with the effects of different amounts of
educationr aggregate service, age, supervisory and occupational placement
statistically controlled, white males conEinue to oyrn a salary advantage
over all other groups. This pattern holds both when separate regressions
are run within occupaEional groupings and within race/sex groups.

Thc final chapter of this report involves a study of somc selected
North Carolina jobs classcs on the basis of tr"ro job poinc evaluation
systems dcvclopcd by Hay Associatcs for the Stace of Idaho and Wiltis
Associates for the State of Waslrington. Since it was not. possible to
have a job evaluation done for thc North Carolina Scatc Government workforcc,
thc next besc alternntive was sc.lected. Job descriptions from these

-lv-



tlro states were maEched to N.C. job classes
sysEems are compared with Ehe . presenE,
st,ructure.

and then these salary assignment
lnternally developed N.C. pay

(
The analyses presented ln this final chapEer cannot be taken

as a full evaluation of t.he N.c. salary system. only a portion of all
Jobs are included in these samples and they are not representatlve in
their selection. The only criteria for membership in each sample was
that a suitable job class be found among the Hay (raaho) or l{illis
(Washington) evaluations. This does not mean the results obtained are
lnvalid, only that they cannot be generalized to the entire N.C. State
Government workforce.

SinpIe regressions of salary onto evaluation points showed that
female dominated (7O'L or more) N.C. jobs received only $25.71 per Hay
evaluation point while male dominated Jobs returned an average of $33.75
Per Hay point. Similar resulcs are obtained using the WilIis system
and also in substitut,ing hiring rates for average salary.

Another ruethod of analyzing these salary discrepanies is to examine
the race,/sex composition of job classes with salaries farthest array from
their point evaluation rat,ing. Jobs in these two samples that are paid
significantly above their rating are dominaced by white males almost
213 of the job classes over I standard deviation above their Hay rating
had no ltomen or blacks in their workforce. In jobs that paid I standard
deviatlon less than their Hay poinE rating, t.he situation is reversed;
here only 73'L of. the job classes failed to have any rromen or blacks.

Another ltay of analyzing the salary differences between male
and female dominated occupations is to examine only those job classes
of equal Hay point value. In this sample of 77 Jobs, under the present
N.C. salary structure, the mean salary of the female dominated positions
Iras 78.8L of the mean salary of the male dominated positions. Howeverr
when the Hay system hras used to predict salaries t.he average salary of
female dominated positions rose to 92.4% of the mean salary of male dominated
pos it ions .

Finally' the regression analyses of positions perforrned earller
produced cost estimates for implementing each of these tlro systems.
The Hay system would require over $17rOOO,OOO to implement while the
Willis system would require over $39rOoO,OoO. It must be remembered
both the regressions and cost esEimates are influenced by the samples
from which Ehey are taken. These $rere not. randomly drawn and are not
rePresentative of atl of state government. Actual cost figures for all
of statc Sovernrncnt would be different and of course subject to external,
pol.itical and budgetary constraints.

The overall impression of this study has to be one that emphasizes
the Patterning of salaries by racc and sex. Although the pattern varies
under certain conditions, it is clear that. the three dif[erenc analytic
techniques used in this rcport all. point to a salary advantage for white
males ttndcr molit circumstances. The ultimate sources and further details
of this advantaB(! remil in ob ject.s of f urther study.

-v-