The questions in the book are to get you thinking. There are no right questions. The idea is to explore the data. So if you think that being a single parent might lead to more behaviour problems, figure out some way to use the data to address this. If you think there might be a relation between income and emotional problems in girls, figure out a way to ask it. ANY question is OK, as long as you can arrange the data to answer it. In their papers, Offord's team examined some questions. There are many others.
As to displaying the data. It's not an issue of GRAPHING the data. Basically, just about all the data are in the form of dichotomous variables, so has to be displayed using contingency tables. I showed lots of examples in the class. Generally you put the risk factor (e.g. welfare status) on the left, labelling the rows, and the outcome (behaviour problem) on the top, labelling the columns.
I talked about different ways to characterize the associations (odds ratio, chi-square, phi coefficient, relative risk, attributable risk). They give different information, and may lead to different interpretations. DON"T think that you are required to do all possible ways I just listed. Try a couple and see what information you get out of it.
What seems to be hard to grasp, is that we don't have a hidden agenda. All along, we've been trying to get the point across that the goal is simply to learn how to address resea4ch questions with data. And the nature of research is that no one gives you the questions -- you have to dream them up yourself. In this data base in particular, there are no "hidden" questions. There is no fancy theory guiding the research. Conversely there are dozens (literally 12x11/2 = 66) of possible associations. So any reasonable association can be explored. The choice really is yours. The important issue is simply to use the computer to calculate the association (generally in a pivot table), then use your head to figure out what the table is telling you. That's all.