Convolutions of a Faculty
Salary Equity Study
Michael Tumeo, Ph.D.
John Kalb, Ph.D.
Southern Methodist University
Faculty Compensation Overview
• Faculty compensation while not the sole motivator for
faculty, is an important magnet for attracting and
retaining good faculty as well as and interwoven
component to boosting morale (Shuster, Finkelstein,
2006).
• While faculty salary is an important consideration, other
factors such a job location, benefits, peer interactions,
and non-tangible factors also weigh into the attraction,
retention, and morale of faculty.
• Faculty compensation has many facets, but this study
will focus on faculty salary specifically.
2
Questions and Answers
• Are there Gender inequities regarding faculty salaries at
our institution?
– At the 2007 AIR Forum in Kansas City, Porter, Toutkoushian, &
Moore presented a paper in which they show, using NSOPF
(National Survey of Postsecondary Faculty) data that gender
inequities are pervasive and long-term.
– This then begs the question, “Is the question of gender inequities
the right question to ask?” or has this become the “duh”
question?
• Perhaps the more appropriate questions become,
“Where are the gender inequities? Can they be
explained? What can we do about them?”
3
SMU Solution
• Using a multifaceted approach we attempted to explore
the answers to the first two questions in hopes of finding
a solution to the third.
• We used a graphical analysis, Multiple Regression, and
an “inappropriate” ANOVA
• This presentation will walk you through what we did, why
we did it, and what we found.
• We will also discuss some of the strengths and
weaknesses of each approach and hopefully solicit some
ideas for additional analysis.
4
Graphical Approach
• Does time at the institution, or time since degree impact
salary equity?
• Do tenure status, and discipline of the faculty member
impact salary equity? (only included Tenured and
Tenure-Track faculty in analysis) [Non-tenure track
faculty unnecessarily complicates an already
complicated analysis]
• What is the best way to see the effect of these variables
on salary equity?
• KISS method is important so as to not complicate the
graphic unnecessarily (using Tenure instead of Rank, for
example)
5
F all 2006 A n n u al F acu lty S alary b y Y ears S in ce L ast D egree
y = 9 5 9 .0 8 x + 7 9 5 7 5
H IG H
y = 5 4 5 .2 2 x + 7 2 6 1 7
S alary
M ODERATE
LOW
0
10
20
30
40
Y ears
F em ales
M ales
L inear (M ales)
6
L inear (F em ales)
50
60
F all 2006 A n n u al F acu lty S alary b y Y ears at In stitu tion
H IG H
y = 285x + 95580
y = 2 4 9 .0 3 x + 7 8 6 1 6
S alary
M ODERATE
LOW
0
10
20
30
40
Y ears
F em ales
M ales
L inear (M ales)
7
L inear (F em ales)
50
60
F all 2006 A n n u al F acu lty S alary b y Y ears S in ce L ast D egree
y = 5 7 1 .9 6 x + 9 2 2 0 2
H IG H
y = 6 9 2 .1 3 x + 7 2 8 9 6
y = 2 9 4 .5 7 x + 7 5 9 4 3
y = -8 0 1 .9 8 x + 7 6 7 3 4
S alary
M ODERATE
LOW
0
10
20
30
40
50
Y ears
T en u red Fem ales
Lin ear ( T en u red Fem ales)
T en u re T rack Fem ales
T en u red M ales
T en u re T rack M ales
Lin ear (T en u red M ales)
Lin ear (T en u re T rack Fem ales)
Lin ear (T en u re T rack M ales)
8
60
F all 2006 A n n u al F acu lty S alary b y Y ears at In stitu tion
H IG H
y = -5 1 0 .5 3 x + 1 1 6 9 8 5
y = -5 1 0 .6 5 x + 9 5 2 3 6
y = 1 4 3 5 .5 x + 7 4 5 6 0
y = -2 0 4 4 .2 x + 7 4 7 1 9
S alary
M ODERATE
LOW
0
10
20
30
40
50
Y ears
T en u red Fem ales
Lin ear ( T en u red Fem ales)
T en u re T rack Fem ales
T en u red M ales
T en u re T rack M ales
Lin ear (T en u red M ales)
Lin ear (T en u re T rack Fem ales)
Lin ear (T en u re T rack M ales)
9
60
General Trends Found
• Can clearly see in all graphs “apparent” gender salary
inequity.
• Time since degree seems to have a larger impact on
salary disparity than does time at the institution.
• Both factors of time have a disproportionate effect
depending on the tenure status of faculty.
• Provides a wonderful display of salary compression for
tenured faculty at an equal rate for both males and
females.
• Does not address the discipline question.
• Discipline is defined by 2-digit CIP Codes.
10
Salaries by Years Since
Degree
HIGH
Communication, Journalism, and Related Programs
MODERATE
LOW
0
HIGH
10
20
30
40
50
10
20
30
40
50
10
20
30
40
50
30
40
50
Education
MODERATE
Discipline Area based
upon 2-digit CIP Code
Classification
LOW
0
HIGH
Engineering
Males
MODERATE
Females
LOW
0
HIGH
Engineering Technologies/Technicians
MODERATE
LOW
NOTE: All charts are
based upon the same
unit scale (original)
0
10
20
Years Since Degree
11
Salaries by Years Since
Degree
HIGH
Psychology
MODERATE
LOW
0
HIGH
10
20
30
40
50
20
30
40
50
20
30
40
50
Social Sciences
MODERATE
Discipline Area based
upon 2-digit CIP Code
Classification
LOW
0
Visual and Performing Arts
HIGH
Males
Females
10
MODERATE
LOW
0
HIGH
10
Business, Management, Marketing, and Related Support Services
MODERATE
LOW
NOTE: All charts are
based upon the same
unit scale (original)
0
10
12
20
30
Years Since Degree
40
50
Salaries by Years at the
Institution
HIGH
Communication, Journalism, and Related Programs
MODERATE
LOW
0
HIGH
Discipline Area based
upon 2-digit CIP Code
Classification
Females
20
30
40
50
10
20
30
40
50
20
30
40
50
30
40
50
Education
MODERATE
LOW
0
Males
10
HIGH
Engineering
MODERATE
LOW
0
HIGH
10
Engineering Technologies/Technicians
MODERATE
LOW
NOTE: All charts are
based upon the same
unit scale (original)
0
10
20
Years at Institution
13
Salaries by Years at the
Institution
Psychology
HIGH
MODERATE
LOW
0
HIGH
10
20
30
40
50
20
30
40
50
20
30
40
50
Social Sciences
MODERATE
Discipline Area based
upon 2-digit CIP Code
Classification
LOW
0
HIGH
Males
Females
10
Visual and Performing Arts
MODERATE
LOW
0
HIGH
10
Business, Management, Marketing, and Related Support Services
MODERATE
LOW
NOTE: All charts are
based upon the same
unit scale (original)
0
10
14
20
Years at Institution
30
40
50
Multiple Regression Analysis
(Enter Method)
• Variables used based upon Luna (2007) and the
previous graphical analysis.
• Rank (Professor, Associate, Assistant)
• Terminal degree (dummy coded Yes)
• Years since degree
• Years at Institution
• Gender (dummy coded Female)
• Market Ratio (account for discipline differences)
• Dependent Variable (Annual Salary)
15
Table of Terminal and Non-terminal Degrees
Degree Type
Terminal (Y or N)
Degree Type
Terminal (Y or N)
AA
N
MBA
N
AMD
Y
MD
Y
AS
N
MED
Y
BA
N
MFA
Y
BBA
N
MLA
N
BFA
N
MMED
N
BJ
N
MM
N
BM
N
MPA
N
BS
N
MPP
N
CERT
N
MS
N
DED
Y
MSA
N
DENG
Y
MSE
N
DM
Y
MT
N
DMA
Y
MTH
N
DME
Y
PHD
Y
DMIN
Y
SJD
Y
DPA
Y
STD
Y
DTH
Y
THD
Y
EDD
Y
JD
Y
LLB
Y
LLM
Y
LTR
N
MA
N
MAST
N
16
Multiple Regression Coefficients and t-scores
Model
Unstandardized Coefficients
B
t
Sig.
Std. Error
(Constant)
FEMALE
TERMINAL DEGREE
YEARS SINCE DEG
YEARS AT INSTITUTION
MARKET RATIO
STUDY RANK
-45418.277
6651.084
-6.829
.000
-5702.960
2543.721
-2.242
.025
11373.917
5004.147
2.273
.024
568.848
180.677
3.148
.002
-1082.334
152.975
-7.075
.000
86554.912
4521.985
19.141
.000
22630.020
1959.562
11.549
.000
a Dependent Variable: Annual Salary
17
Studentized Residual Plots
Studentized Residuals Against Years Since Degree
6.00000
5.00000
4.00000
Studentized Residual
3.00000
2.00000
1.00000
0.00000
-1.00000
-2.00000
-3.00000
-4.00000
0
10
20
30
Years
18
40
50
60
Studentized Residual Plots
Studentized Residuals Against Years at Institution
6.00000
5.00000
4.00000
Studentized Residual
3.00000
2.00000
1.00000
0.00000
-1.00000
-2.00000
-3.00000
-4.00000
0
10
20
30
Years
19
40
50
60
Influence and Leverage Plot
Measure of Data Point Influence and Data Point Leverage
(Data in Upper Right Corner are High Influence and High Leverage)
0.10000
0.09000
0.08000
Cook's D (Influence)
0.07000
0.06000
0.05000
0.04000
0.03000
0.02000
0.01000
0.00000
0.00000
0.01000
0.02000
0.03000
0.04000
0.05000
0.06000
Centered Leverage Value
20
0.07000
0.08000
0.09000
0.10000
Multiple Regression Analysis
(Stepwise Method)
• Same variables used in the previous
analysis
• Interested in model selection
• Most parsimonious model selected using
change in R2 rule
• y = -41,625.651 + 89,844.209 * Market
Ratio + 26,581.145 * Rank + (-711.610 *
Years at Institution).
21
Stepwise Data Table
Change Statistics
Mode
l
R
R Square
Adjusted
R Square
Std. Error of
the Estimate
R Square
Change
F Change
df1
df2
Sig. F Change
1
.640(a)
.410
.408
$29,157.236
.410
312.984
1
451
.000
2
.777(b)
.604
.602
$23,916.244
.194
220.322
1
450
.000
3
.799(c)
.639
.637
$22,846.032
.035
44.148
1
449
.000
4
.803(d)
.644
.641
$22,713.234
.005
6.266
1
448
.013
5
.806(e)
.649
.645
$22,572.063
.005
6.621
1
447
.010
6
.808(f)
.653
.649
$22,467.606
.004
5.166
1
446
.024
a Predictors: (Constant), MARKET_RATIO
b Predictors: (Constant), MARKET_RATIO, RANK
c Predictors: (Constant), MARKET_RATIO, RANK, YEARS_AT_INST
d Predictors: (Constant), MARKET_RATIO, RANK, YEARS_AT_INST, FEMALE
e Predictors: (Constant), MARKET_RATIO, RANK, YEARS_AT_INST, FEMALE, YEARS_SINCE_DEG
f Predictors: (Constant), MARKET_RATIO, RANK, YEARS_AT_INST, FEMALE, YEARS_SINCE_DEG, TERM_DEGREE
22
Model Validation
• Condition Index of the Collinearity Diagnostics
table yielded a value of 11.6
– General Rule (values of 15 or higher = moderate risk
of mulitcollinearity while 30 or higher is a serious risk).
• Two additional Multiple Regressions were run
(Forward and Backward) to ensure the Stepwise
Regression was not a mathematical artifact.
• Did not do a split sample validation or a cross
sample validation, but the model is not being
used for predictive purposes so further validation
procedures were deemed unnecessary at this
time.
23
ANOVA
The Final Frontier
• Wanted to explore possible interactions between
gender and other factors related to salary equity
(finally getting back to the original question)
• Market Ratio was categorized into Market Value
(based on Luna 2007, paper)
• 3-way ANOVA with Gender (Female, Male),
Market Value (Below Average, Average, Above
Average), and Rank (Assistant, Associate, Full)
with Dependent Variable (Salary)
24
ANOVA Cautionary Notes
• Violated several fundamental rules for an
ANOVA, but this was exploratory, so tread lightly.
• ANOVA done on a population, not a sample (All
faculty were included because of sample size
concerns).
• Not really a true “experimental” design.
• Groups size differences at more refined levels
are a concern because of variance differences.
• Interpretation of results and generalizations are
very tentative because of these caveats.
25
Average Salary for General Groups of Faculty
Gender; F = 1.524, p = .218; Not
Significant
Females
Males
Rank; F = 39.342, p < .001; Significant
Assistant
Tukey HSD results show all pairwise
comparisons are significantly different.
Associate
Full
Market Value; F = 107.331, p < .001; Significant
Below Average
Tukey HSD results show all pairwise
comparisons are significantly different.
Average
Above Average
Lower Salary
Higher Salary
26
Mean Annual Salaries for Female and Male Faculty by Rank
Higher Salary
Gender x Rank Interaction
F = 3.429, p < .05
Lower Salary
Difference = $16,902
Difference = $7,050
Difference = $4,276
Male
Female
Assistant
Associate
Rank
27
Full
Average Salary for Market Value based upon Gender and Rank
Lower Salary
Higher Salary
Gender x Rank x Market Value Interaction
F = 1.960, p = .100
Below Average
Average
Above Average
Market Value
Female Assistant
Male Assistant
Female Associate
28
Male Associate
Female Full
Male Full
Conclusions
• The simple answer to the question of gender
salary inequity at SMU is “YES” (a simple
question deserves a simple answer after all,
right?).
• As you can see the “real” answer is quite a bit
more complicated than, simply “Yes”.
• Factors like rank and discipline complicate the
picture considerably.
• Complications regarding sampling, and group
size differences additionally complicate finding a
clear statistical answer.
29
Added Factors not Considered
• Additional information regarding faculty standing
would be critical to gaining a fuller picture of any
potential gender inequities.
– Time in rank
– Performance measures (publications, class and
supervisor evaluations, service, etc)
– Outside job offers
– Changing market demands
– Etc.
30
Lessons Learned and Next Steps
• Discipline specific evaluations may be needed
instead of University level evaluations
• Better data about performance measures
needed
• Need to explore ways to counter salary
compression for both genders
• Need to look more closely at the disparities at
the higher ranks to determine the reality of those
disparities or if other factors are influencing the
apparent salary disparities
31
References
•
•
•
•
•
•
•
•
Barbezat, D. A. (2003). From here to seniority: The effect of experience and job
tenure on faculty salaries. New Directions for Institutional Research, 117, 2147.
Bellas, M. L. (1997). Disciplinary differences in faculty salaries: Does gender bias
play a role? The Journal of Higher Education, 68 (3), 299-321.
Boudreau, N., Sullivan, J., Balzer, W., Ryan, A. M., Yonker, R., Thorsteinson, T., &
Hutchinson. (1997). Should faculty rank be included as a predictor variable in
studies of gender equity in university faculty salaries? Research in Higher
Education, 38 (3), 297-312.
Luna, A. L. (2006). Faculty salary equity cases: combining statistics with the law. The
Journal of Higher Education, 77 (2), 193-224.
Luna, A. L. (2007). Using market ratio factor in faculty salary equity studies. AIR
Professional File, 103, 1-16.
Schuster, J. H., & Finkelstein, M. J. (2006). The American Faculty: The restructuring
of Academic Work and Careers. Baltimore, MD: The Johns Hopkins University
Press.
Porter, S. R., Toutkoushian, R. K., & Moore, J. V. (2007) Gender differences in salary
for recently-hired faculty, 1998-2004. Scholarly Paper, Presented at the 2007
AIR Forum in Kansas City MO.
Webster, A. L. (1995). Demographic factors affecting faculty salary. Educational and
Psychological Measurement, 55 (5), 728-735.
32
Descargar

Faculty Salary Analysis and Gender Equity