Internship

It is the official ending of the summer internship program and we completed an evaluation form. But I guess I have more to say, so I will just put it here.

Application

Before applied to this internship, I was in great trouble finding an internship that suits my interest. There were tons of career fairs, recruitment opportunities around, unfortunately, most of them on campus were for students interested in banking and finance. As an economics student, most of my classmates take economics as a close substitute for business and they all want investment banking or consulting jobs upon graduation. But I knew, that's not what I want.

Starting January this year, I was in this economics honors seminar which introduces the Panel Study of Income Dynamics, and I thought, hmm this is interesting, maybe I should give it a try. And I was so glad to find out that the SRC does have an internship opportunity every summer. So I got the application material in the internship fair and started application right away. (My friend was complaining to me that she was late for the fair and there were so many people at the SRC booth, so she could not get a copy of the application material)

The application was quite user friendly. A statement of purpose, which I believe anyone who has been through the application period should have one in hand, a cover sheet, transcript and two recommendations. It took me less than an afternoon to finish the application.

I guess I was lucky that I got to know the honors seminar professor well, who later become my research sponsor this summer. So even before I started applying, I talked to him about things that I would be expected in the internship program. At the same time, I kept busy with other internship applications.

The application turned out not to be surprising, I was not replied by most of the investment banks--those who replied said "No". To my great relief, I got the offer from SRC. It did not take me any time to reply and say "Yes" (I guess I was a bit desperate for something to do in the summer).

Paperwork

I admit that I hate paperwork, but I was really lucky that Anita helped me for most of the tedious stuff. And in the end it turned to be less trouble than what I have anticipated. I guess this is another benefit of SRC internship, since it is on campus job, it is easier for international students.

Choice of research topics

I spent the first one week of the internship re-studying the PSID. I went through the tutorials again in details, and paid particular attention to the way that the PSID data were sedigned. I was particularly fond of the use of Family Identification Mapping System(FIMS) and cross generational analysis. My office mate Noura, who was also an SRC intern, helped me a lot with the design of the Child Development Supplement. =)

Starting the second week, I began to play with the PSID. I had some ideas left from the honors seminar and I talked to my sponsor about them. I am now still amazed by his ability to see a problem even before I started the analysis, and the ability to see the result and comment "hmm this is interesting"... Maybe it's the 40 years of experience with the PSID? I don't know, but I absolutely hope that I can do it one day.

Research is tough tough tough

Despite having some exposure to the PSID and research on PSID before this internship, I still found it hard. There are just so many problems. For example, at first, I was using the parents' reported spending on child in PSID core as a predictor variable, but when I compare them to the total of a detailed break down spending in the Child Development Supplement, about 1/3 of the parents reported more in PSID core.

It was like a parent reported "I spent 50000 on my child last year" and "I spent 1000 on the child's education, 1000 on child's clothes and 2000 on child's food and nothing more" at the same time. I was in such a panic that I did not know which predictor variable to use.

When I ran to my sponsor for help, he looked at the data and said "hmmm interesting" and we discussed several causes--when reporting in aggregate, parents tend to over-estimate; parents' tend to "please" the interviewer by over reporting; parents tend to report at whole intervals say 100, 1000 rather than being specific and etc. And he showed me common practice how economists deal with similar problems--using instrumental variables.

In the first few weeks, I was so easily getting into panic, and slowly, I learned to solve the problems myself. And this was the first time that some of the statistics knowledge that I learned in class started to make sense--I did well in most of my stats classes, and I thought I understand the concepts. But I only started to appreciate those concepts once I saw how they changed my analysis.

Summer Institute

That was the most tiring four weeks of the internship. I work from 9am to 2pm, and take class from 2pm to 5pm, and I still have to do the reading and assignment at night. I took a course in analysis of complex survey data and it just went at full gear.

That was an intensive training in both the concepts and analysis with various softwares. We used STATA, SAS, IVEware, and learned to interpret results from SUDDAN and etc. However, I think this is the best part of the internship. I can make use of whatever I leanred in class the previous day in the work next morning and with the knowledge of complex survey sample, I was able to appreciate the design of the PSID even more.

The only setback was in the last week of the course, everyone was exhausted and I had to start preparing for the symposium. So I only completed the coursework in the last week, without takinga further look into it. And I though, if I can have one more week before the symposium, I would be able to use model based analysis for the data too. But anyway, since I will continue working, I will have more chances to practice that soon.

Other interns and activities

I still remember in the first day of the orientation, everyone looked so tensed up. But we got to know each other pretty fast, thanks to all the outings and dinners. It is interesting to listen to others talking about their research topics. Economists see things differently from sociologists, statisticians or demographers, so when we talk we can actually see how different disciplines interact. I remember in the meeting with Bob Groves, he mentioned that social science research will generally go to the following directions: more quantitative, global and interdisciplinary...

Symposium

I was really stressed up before that. Especially after the dry run, when I saw everyone so well prepared, I was nervous. I practiced so much that I dreamt about it almost every night before it. But I was surprised in the morning of the symposium, when I went to see my sponsor and present in front of him, I found I was well prepared too. =) and to my greatest relief, he raised several questions that I found I could give a satisfactory answer. Maybe just as Noura has told me, since I have been working on it for the whole summer, I really know where every single bits come from.

Statistical software

I learned to use R, SPSS, STATA in classes, and SAS in CSCAR workshops. Before this internship, I mainly used R and STATA. But in my work, I found some work can be so easily handled in one but not others, and I had this problem of switching between softwares. Sometimes it took me ages to write a program to do something in R but it is just a build in function in SAS, sometimes STATA kept giving outrageous results and when I read the mannual, I realized that was because how they handle missing data and etc...

One lesson that I learned was, read the mannual, know how they handle certain analysis before you proceed with any software. Otherwise, it would be a waste of time. I am glad that I learned it--no matter how unpleasant the process was.

================================

Well, that's all. I will be leaving for Chicago tomorrow and be back next week. =) I am excited..

Labels: eeks ECON, Sexy Stats

posted by panda xuan @ 8:21 PM