Skip to 0 minutes and 1 secondWhen I offer this course on-campus here at Hanyang University, I ask all my students to come up with their original research ideas. They then have to do a simple empirical analysis and write a paper about it. They can choose any research question they like, as long as it is related to crime in some way. In the past, students wrote papers about very interesting and diverse topics such as the effect of the immigrant influx on crime rates and the determinants of in-school juvenile delinquency. During the semester, I have frequent conversations with my students about their research ideas and progress.
Skip to 0 minutes and 43 secondsAnd this is my favorite part of running this course, listening to different ideas and questions people have about crime and helping them to develop their ideas further. Unfortunately, to have all of you start working on your own project and guiding your progress on a one-on-one basis would not be feasible in this online course. But still, I want you to think about your own crime question and share with your classmates here. Hopefully this discussion will help you develop your ideas further and turn it into a solid research question. And you may want to pursue the question further by collecting data and doing the actual analysis by yourself.
Skip to 1 minute and 28 secondsFor those of you who want to pursue your question further, I have the following suggestions. First, choose a question that you are actually interested in. The actual research work can be very time-consuming and tedious. If you are working on a question that you do not find interesting yourself, your research experience can easily become boring and unbearable. Second, be specific. Of course we are more interested in big questions, but actual research questions tend to be very specific. For example, you may want to know more about how providing more and better education can help lower crime. That is your big question, but when you actually work on this topic, you have to choose research question that is more specific.
Skip to 2 minutes and 22 secondsIf you have data on high school dropout in crime, you should probably focus your research on the effect of high school dropout in crime. If you have detailed data on school characteristics across different high schools, you may want to focus your research on how criminal behavior may be influenced by attending certain types of high school. If some government policy substantially changed the schooling experience of some students, you may want to know whether this change had any impact on their subsequent criminal behavior. Note how all three examples here, look at very specific questions which can potentially give you very clear-cut answers. Third, what you can and cannot do is usually bounded by data availability.
Skip to 3 minutes and 13 secondsSo if you choose a research question where a related data is very hard to come by, you probably will not make much progress. For example, whether and how much on influx of illegal immigrants affects local crime rates has been of great policy interest in recent years,
Skip to 3 minutes and 32 secondsbut good data on illegal immigrants are very hard to get for an obvious reason: they are here illegally and they probably want to keep their personal information private. Lastly, think hard about how you can recover a causal relationship. For example, suppose you want to know the causal effect of drinking on crime. Simply comparing the offending rates between individuals who drink a lot and individuals who do not drink at all will probably not give you the causal effect of drinking on crime, because there are probably different in many other aspects.
Skip to 4 minutes and 11 secondsHowever, we may believe that individuals who are just above age 21 would drink a lot more than individuals who are just below 21, when there are not that different in many other aspects. In this case, the legal age cutoff of 21 for drinking provides us an identifying variation that allows us to identify the causal effect of drinking on crime. Think about an existing law, a policy change, or other natural events that you can use as identifying variations for your research question.
Design your own research question
Think about your own research question.
Think about which data and empirical methods you will use to answer that question.
How do you plan to recover a causal relationship from your data? Can you think of a good identifying variation? If you cannot think of a good quasi-experimental identifying variation, describe a hypothetical random experiment which can help you investigate your research question.
After this video, you will be asked to submit your “research proposal”. You will then share your research plan with your classmates and peer-review each other’s proposal.
© Songman Kang, Hanyang University