TO: Students in CJ 605
FROM: R. B. Taylor
DATE: 4/12/04
RE: Comments on your intro and methods sections

Below are some generic comments, based on reviewing the sections you all have submitted. I will be setting up appointments to talk with each of you individually about your papers next week.

Although there are many good starts here, just about everybody has a lot of work to do.

INTRO

There are three general sections to an intro.

First, draw the reader in; in just a page or so get him/her excited about your dependent variable, and the questions you are going to ask that no one else has answered. What is your outcome, why should we care, and what are the cool questions to look at here? This should be short - could be half a page, no more than two pages.

Second, take the reader carefully through the previous research. For every predictor that is going to show up in your model, at both L1 and L2, you want to tell the reader about what previous research has actually said. This is where you organize for the reader the previous work. Do not just go through study after study in a serial fashion. Provide an organizing framework. Subheadings are wonderful. At the end of this review, you leave the reader with a set of questions - what SPECIFIC things don't we know yet.

If when you are going through these previous studies, you have bones to pick with them, either in terms of indicator selection, or measurement issues, or analytic approach, or something else, it is not a bad idea to put all those criticisms together.

Then you end with your specific statements about hypotheses and rationales. This is where you tell the reader what your analytic model is. You want to provide a hypothesis and accompanying rationale for each of your L1 and L2 hypotheses. All bundled up, this is your conceptual model.

METHODS

There are two purposes to the methods section: to help the reader "get close" to your data, so he/she understands what is going on; and to provide enough detail so that someone else trying to do the same thing could replicate it. Here are some things you want to pay special close attention to.

1. Usually methods sections have subheadings like Context, sample, and the like. Subheadings are a good thing.
2. You will want to either summarize in your own words, or quote directly from the PHMC source material, a description of their data collection procedures. People need to be able to understand: what was the sampling frame?; what was the sampling strategy?; what was the response rate?; how was the survey conducted -- telephone or in person for example?; what kinds of sections were there in the survey?
3. Be sure to explicitly describe your outcome variable, and each of your predictor variables. For the outcome, be sure the reader understands its specific distribution. For each of the other predictors, at the minimum you want a table with: mean, median, min, max, and sd, and n if there are varying Ns because of missing cases. Remember the ASC guidelines: each table goes on a separate page. You will have predictor variables to describe at L2 as well as L1. You will need to tell the reader about the L2 units: what were they?; how were they constructed? Take a look at the PHMC documentation to see if it can help you. Each predictor should have its own subhead.
4. If there are any special analytical things you did before you got started with the analyses, either in terms of missing values, or special recoding, tell the reader about that.
5. If you have developed an index, tell the reader about each item that went into each index.
6. If you are using the crime data, be absolutely clear about the time period covered, the offense in question, and the rate.
7. Your table needs to be a formatted, word processed document, not just a bunch of patched together spss printout. Each table needs to be totally self sufficient - each variable is clearly and fully explained.
8. Be sure you can clearly explain the weighting variable; you also might want to report on the range of weights applied.
9. You want to provide the reader with a rationale for the analyses you use - what are the reasons that hlm is being used here, and why is it better than another approach - do not need a lot here, just a short para.