Friday, May 05, 2006
Does the sophomore economics tutorial a student takes influence his academic and/or career interests?
Generally, college students are not randomly placed into their courses; rather, they have some freedom to choose which courses they want to take. Consequently, it can be difficult for researchers to determine whether taking a particular course alters a student’s academic or career preferences or whether it is a student’s preferences that cause him to choose that course.
For the economics sophomore tutorial at Harvard, students rank their preferences and are then placed into a course through a lottery system. Because some students are placed in their first choice course while others are not, we can form an experimental group and a control group accordingly. For instance, there are a number of students who ranked “Everybody’s Doin’ it: Social Interactions and Economics” as their first choice. Among these students, those who got into Bryce’s class would serve as the experimental group, while those who were placed in a different tutorial would serve as the control group.
The idea is to determine whether there are any consistent differences between the academic and career preferences of the experimental group versus the control group. If so, because there is no reason to suspect bias, then we can attribute these differences to how the given tutorial is affecting the students who take it. My tentative hypothesis is as follows: “Students who are placed in their first choice tutorial will have different academic and career interests than students with the same first choice who were placed in a different tutorial.”
I will be gathering data using a survey I have created. Depending on how many responses I get, I may need to divide the tutorials into various categories, such as Behavioral, Legal, Finance, Healthcare, etc. If that is the case then my hypothesis will argue that students who are placed in a different type of tutorial than their first choice will have different academic and career interests than students who were placed in the same type of tutorial as their first choice.
The questions on the survey include: gender, house affiliation, race/ethnicity, occupation of mother and father, likeliness of attending graduate school or professional school, summer internship information, and likeliness to take certain courses while at Harvard. Most importantly, the survey asks the students to list their tutorial rankings as they did for the official lottery system and to state which tutorial they actually took. Accordingly, I will have many variables to work with, and I should be able to control for various factors when taking the regressions.
What my regression equations will look like depends on what kind of data I obtain from the surveys. At this point, it is likely that one of the equations will regress getting into one’s first choice tutorial on wanting to go to graduate school. Another equation would regress getting into one’s first choice tutorial against the likelihood of taking a particular class. After I get the data, the regression process will be much more concrete.
Concerns and Limitations
The biggest limitations have to do with the data collection process. First, because of the limited amount of time I will have to collect data, it is likely that the sample size will be relatively small. Secondly, I am collecting information about students’ tutorial choices based on their own memories of their rankings. While most students will probably remember their first and second choice, some students (particularly the seniors who took the course two years ago) will perhaps not remember their lower choices correctly.
As for the experimental design, a potential problem I foresee is that students who do not get into their first choice tutorial may be affected by the tutorial in which they are placed. If there is no bias in terms of the type of tutorial in which such students are placed, then the data will be virtually unaffected. If, for some reason, students with the same first choice often had similar types of tutorials for their lower choices, then this could present a problem.
Despite these limitations, this study should be able to tell us a great deal about how students are affected by the sophomore economics tutorials they take.
This all looks good. You are correct in that your main problem is going to be with sample size. The small number of observations you will obtain (from an already small sample) is likely to make observing statistically significant results difficult. That is ok for this project though.
Y = a + b(in tutorial X) + c*(ranking of tutorial X) +[controls]'*d + e
Subscribe to Post Comments [Atom]
Subscribe to Posts [Atom]