High rates of attrition exist among college students in science, technology, engineering and math (STEM) fields, especially among women and minorities. This paper seeks to examine factors that influence the decision to persist in STEM majors. I use data

1
Persistence in Science, Technology, Engineering
and Math Fields Do Instructors Matter? Joshua
Price Cornell University
High rates of attrition exist among college
students in science, technology, engineering and
math (STEM) fields, especially among women and
minorities. This paper seeks to examine factors
that influence the decision to persist in STEM
majors. I use data from the 2000 cohort at all
public 4-year universities in the State of Ohio
to match students to the instructor and
classmates of each course. The objective is to
test the hypothesis that sharing racial and
gender characteristics with instructors of STEM
courses increases the probability of persisting
towards a STEM degree. Results indicate that
Black students are significantly more likely to
persist in a STEM major when they have more black
instructors in STEM courses. There is no
significant effect on persistence for Female
students who have more female instructors.
Effect of female instructors on
female students - Increase in grades in intro
courses (Carrell et al, 2009) -
Identification - Random assignment - Small
increase in grades in first semester (Hoffman
Oreopoulos, 2007) - Identification
Exogenous course selection in 1st semester -
Mixed Results for additional credit hours taken
(Bettinger Long, 2005) - Identification
- Instrument with deviation from steady state of
faculty gender composition Effect of
minority instructors on minority students - No
studies for college students - Positive effect
on academic performance in elementary school
(Dee, 2005)

• Having instructors with similar racial and gender
  characteristics increases persistence in STEM
  fields
• All Public 4-year Universities in Ohio (13 in
  total)
• 2000 entering cohort, tracked for 6 years
• Transcript, Enrollment, and Faculty Files
• Match students and faculty to each course

Abstract
Hypothesis
Methods
Persistence in STEM Major by Race and Gender

• Estimate Linear Probability Model
• Y 1 if persist in STEM field after 1st Year
• X is vector of student characteristics (race,
  gender, ACT score)
• F is vector of faculty characteristics of first
  term courses (race, gender, rank, tenure-status)
• C is the number of STEM and non-STEM courses
  taken the first term
• XF is the racial and gender match between the
  student and instructor in STEM and Non-
• STEM courses
• ? is institutional fixed effects
• How to overcome self selection into courses
• - Rely on exogeneity of 1st semester (Similar
  to Hoffman Oreopoulos, 2009)

Fraction who persist in STEM major after Year 1
Data
Science, Technology, Engineering, and Math (STEM)
Pipeline in College
Results
Linear Probability Estimation of Persistence
Persistence in STEM major After 1st Year (Race)
Persistence in STEM major After 1st Year (Gender)
STEM occupation or graduate degree in STEM field
Enter the pipeline prior to college
Persistence by Initial Major
Initial STEM Majors
Initial Non-STEM Majors
Outcome After 1st Year
Outcome After 1st Year
Over ½ of initial STEM majors do not receive a
degree in a STEM field
Literature Review
Outcome After 2nd Year
Outcome After 2nd Year
Faculty Characteristics
Conclusion