| dc.description.abstract |
<p>This dissertation consists of two separate essays on major choice in higher education.
In the first chapter, I investigate how differences in information affect students'
major choices over time. Since college has such a short time horizon, the amount of
information students have before coming in may play a big role in how well they are
matched to their final major. They may also choose their initial major based on how
uncertain they are about their match with that major, especially since they have the
option to switch in future periods. This paper discusses students' search process
in finding a major, and how information impacts behavior and ultimate outcomes. I
set up a tiered structure where the student must first choose a field (either STEM
or Non-STEM) and then choose a major within that field. This allows for matches within
a particular field to be correlated, thus providing information on non-chosen majors
within the same field. The student makes decisions based on the choices that will
maximize her expected utility over her entire college career. Since her current choices
and information set depend on past decisions, and since there are a finite number
of periods, I can solve the dynamic decision problem using backwards recursion. </p><p>Once
I solve for the student's optimal decision path, I estimate the model using data from
the Campus Life and Learning Survey from Duke University. The CLL data allows me to
observe students' expected majors at multiple points throughout their college career.
I attempt to find the model parameters that best match particular moments in the data.
The first key type of moment involves overall switching patterns, that is, the probability
of choosing a particular field in the initial period, and then the probabilities of
later decisions conditional on the first choice. The second key type of moment I match
captures which students are making which decisions. I look at how academic ability,
as measured by SAT Math scores, and gender affect the choice probabilities in the
data. </p><p>I find that the STEM field has a much lower average match value than
non-STEM, but a higher variance in matches. Thus, students are less certain about
how well they might match with STEM. Students with higher math ability are more likely
to choose STEM in the first period, but the sorting by ability greatly increases in
the later period. It is costly to switch into STEM from non-STEM in the second period,
while the reverse move is virtually costless. All of these results support the theoretical
result that students will choose the field with more uncertainty in the early periods
(given similar expected match values) because of the option to switch later if they
get a bad match. This is especially true when the more uncertain field is also more
costly to switch into in later periods, as in the case of STEM. </p><p>In the second
chapter, co-authored with Thomas Ahn, Peter Arcidiacono, and James Thomas, we estimate
an equilibrium model of grading policies. On the supply side, professors offer courses
with particular grading policies. Professors set both an intercept and a return to
studying and ability in determining their grading policies. They make these decisions,
attempting to maximize their own utility, but taking into account all other professors'
grading policies. On the demand side, students respond by selecting a bundle of courses,
then deciding how much to study in each class conditional on enrolling. We allow men
and women to have different preferences over different departments, how much they
like higher grades, and how costly it is to exert more effort in studying. </p><p>Two
decompositions are performed. First, we separate out how much of the differences in
grading policies across fields is driven by differences in demand for courses in those
fields and how much is due to differences in professor preferences across fields.
Second, we separate out how much differences in female/male course taking across fields
is driven by i) differences in cognitive skills, ii) differences in the valuation
of grades, iii) differences in the cost of studying, and iv) differences in field
preferences. </p><p>We then use the structural parameters to evaluate restrictions
on grading policies. Restrictions on grading policies that equalize grade distributions
across classes result in higher (lower) grades in science (non-science) fields but
more (less) work being required. As women are willing to study more than men, this
restriction on grading policies results in more women pursuing the sciences and more
men pursuing the non-sciences.</p>
|
|
