Homework 3
BIOST/STAT 579
Summer 2005
Yanez
Salary Data Description
(Due: Tuesday 12 July 2005)
Background
Differences in salaries between men and women at US colleges and universities are well
documented. Because salary discrimination based on sex is illegal in the US, there is
considerable controversy over whether the differences between male and female faculty
members’ salaries are due to sex bias. Other explanations which have been put forward for
salary disparities between men and women include differences in experience, degree attained,
the field in which one works, administrative responsibilities, and productivity, among others.
Two general approaches are possible for investigating the presence of sex bias. One approach
involves attempting to determine whether individuals have been victims of bias by examining
their particular circumstances. The other approach examines salaries at the group level and
attempts to uncover differences in average salary between groups of men and women within
an institution. This is the general approach that will be used in this analysis.
There are many factors in faculty salaries, and each of these may be confounded with sex
differences. Market forces are important at some universities, i.e., salaries may be determined
by the demand for experts in certain fields and/or by salaries offered outside the university.
Thus, higher values may be placed on some fields within the university and faculty within
those areas will be paid higher salaries.
Experience is another factor that in faculty salaries. A faculty member's salary tends to
increase with the amount of time that person is employed at the university. If the faculty
member is hired after having worked at another institution, then their prior experience may be
used in determining their salary. Sometimes salary increases are given when a faculty
member is promoted from one rank to another, although these increases are not necessarily
based solely on the promotion.
A faculty member's salary is also in by their productivity. Many different productivity measures
can be used, including research grant funding, number of papers published, teaching
performance, and administrative duties, among others.
Questions of Interest
The goal of the analysis is to determine whether sex bias exists and to describe the magnitude
and nature of the effect. Your analysis should revolve around the following specific questions
of interest:
1. Does sex bias exist at the university in the most current year available (1995)?
2. Has sex bias existed in the starting salaries of faculty members (i.e., salaries in the year
hired)?
3. Has sex bias existed in granting promotion from Associate Professor to Full Professor?
4. Overall, how would you answer the question: Is there sex bias in salaries at the university?
What issues are involved in attempting to generalize your results?
In answering these questions be sure you defend the use of each of the variables used in your
analyses. The data set contains a large number of records and this may affect the analysis
approaches and methods used as well as the interpretation of the results.
1
BIOST/STAT 579
Summer 2005
Yanez
The Data
The data to be used in this analysis consist of faculty members' salaries at a single US
university. Data were collected on 1,597 faculty members employed at the university in 1995
(medical school faculty were excluded). Monthly salary was determined for each faculty
member for each year from 1976 through 1995. Other variables available include sex, highest
degree attained, year of highest degree, field, year hired, rank, and administrative duties.
Note that the last two variables may change over time but the others are constant over time.
The file salary.txt is in ASCII format with tabs separating the fields and can be downloaded
from the class web page at (http://faculty.washington.edu/yanez/b579/salary.txt). Each
record in the data file represents the information on salary and the other variables for a
particular year (there are 19,792 records).
The variable names and description are given below:
Id
Identification number for the faculty member
sex
M (male) or F (female)
deg
highest degree attained: PhD, Prof (professional degree, e.g., medicine or law), or
Other (Master's or Bachelor's degree)
yrdeg
year highest degree attained
field
Arts (Arts and Humanities), Prof (professional school, i.e., Business, Law, Engineering
or Public Affairs), or Other
startyr year in which the faculty member was hired (2 digits)
year
year (2 digits)
rank
rank of the faculty member in this year: Assist (Assistant), Assoc (Associate), or Full
(Full) admin indicator of whether the faculty member had administrative duties (e.g.,
department chair) in this year: 1 (yes), or 0 (no)
salary monthly salary of the faculty member in this year in dollars
Admin indicator of whether the faculty member had administrative duties (e.g., department
chair) in this year (1=yes, 0=no).
2
BIOST/STAT 579
Summer 2005
Yanez
Assignment
In this assignment, you will prepare the background and outline the statistical methods that
you will use to investigation the questions of interest. Your report should look like the sections
of a formal manuscript to be submitted to an applied journal. Because a statistical analysis
aims to answer a scientific question, the sections of your report should be organized in the
manner which is customarily used in science. To wit:
Background: Provide a description of the scientific motivation for the analysis. Use
your own words rather than copying the description provided by the client. By
providing your understanding of the problem, the client may be able to correct any
misconceptions that you had about the science. You don't have to go into great detail
here, but do give all the facts that entered into your decision process during the
analysis. The Questions of Interest belongs here. State the specific questions that
have been posed.
Statistical Methods: Describe the methods used for the analysis at a level
appropriate for an applied audience. Include references for non-standard techniques.
You may want to describe the software used, and certainly want to describe the
methods used for assessing the appropriateness of your models. Explain how you
handled common problems like missing data, multiple comparisons, etc. Provide
interpretations for all parameter estimates used to answer relevant scientific
questions. Motivate transformations. Describe the use of P values and confidence
intervals if they play an important role in your analysis. Explain why you didn't use
more common techniques if necessary.
The major theme of the above is to write to the scientific community rather than to a
statistician. In a future assignment, we will proceed to write the results, discussion and
appendices
3