Module 5 discussion

Please begin this discussion early so your fellow discussion group members have time to respond to your question!

Discuss your final paper idea with your discussion group by asking questions of your group about your ideas for your research design and by giving feedback to other group members about their designs.

Questions will be limited in scope to topics raised in chapters 10 and 11.

Questions should be things like: considering different designs for your research proposal with at least one of them being a non-experimental, quasi-experimental or factorial design - or deciding among two or more designs discussed  in chapter 10 or 11, for instance whether to use a within- or between-subjects nonexperimental design, or whether to use a 2x2 Vs. 2x3 factorial design, or about how many independent variables to have in a factorial design.

You must ask at least one question and respond to at least three questions of others for a total of four posts.

Note: posting your question early should mean more opportunities for feedback from others – feedback that can be very helpful for preparing your paper.

Your responses to others questions should show understanding of the strengths and weaknesses of the different types of designs.

Examples:

Question: I want to study the effects of smoking marijuana on academic performance and I am trying to decide whether to do a longitudinal study and look at changes in grades over time in marijuana users versus doing a nonexperimental between-subjects design and comparing grades in a sample of marijuana users (used daily for at least the last six months) and a sample of non-users (haven’t used in the last year and don’t intend to use). Which design do you think would be better, and also, is grades the best way to operationalize academic performance?

Answers: In nonexperimental designs there is a risk of assignment bias; you would have to make sure that the two groups are not very different in areas other than marijuana use.

In the longitudinal design you would have to get a very large sample because even if they are using marijuana at the start of the study marijuana they might use more or less, or use other drugs, or alcohol, that could be confounds.  Maybe you could make sure they don’t use drugs other than marijuana at the start of the study?

Grades would be easiest. But you could look at things like the tests they give in the 10^th grade. It would be more work, but it would be the same across the state. An A at one school might not be the same as an A at a different school, but the state test would mean the same thing for everyone.

Longitudinal research takes too long, do you want your results in 10 years? BTW, have you thought about looking at a variable more specific than grades? I know I’ve read that marijuana affects memory, so maybe you could just test memory.

Question: I was going to do a single factor experiment between subjects treatment and no-treatment control, and now I am thinking about a factorial design. I am studying the effects of a treatment program on autism. I don’t want to make it too complicated, and now I am thinking about age and gender as variables. Boys get autism more than girls, and it is usually better if treatment that begins at a younger age.

Answers: I don’t think you should make your paper too complicated. It is graded on complexity of design so maybe adding a factor is good, but I wouldn’t add more than one. There was an example in the book on dividing them into age groups and using age as a factor to reduce variance. Check out that example at the end of chapter 11.

Is there a good reason to think that the treatment would be better with boys or girls? If there is, then I would use gender and if there isn’t don’t bother – it won’t add anything and will make your study more complicated. And like the last comment, adding in age might be a good idea.

I thought no-treatment control groups are not ethical if there are other treatments that work out there. Maybe you could use within-subject pretest-postest design, or compare two treatments and also use age as a factor so it could be a two factor design with a treatment factor and an age factor.

You get the idea – your comments should attempt to be genuinely helpful to the person asking the question and reflect that you know the material.

And again, please post your questions earlier rather than later!