2.6 Causal Comparative and Correlational studies

Causal comparison and correlation studies. As discussed in Mertens, causal comparative studies are necessary when there is a need to characteristics which it is either not possible, or unethical to manipulate.  Her examples in the book focused on studies of learning and behavioral differences associated with gender, disability and ethnicity.  In writing studies - or in studies where we want to learn about teaching and learning, it may be necessary to use causal comparative or correlational methods to study groups of writers or certain practices or experiences associated with writing or teaching of writing.

I drew attention to the list of reasons for misperceptions of differences among groups ( p. 154) - but re-cast it as a list of caveats for the design and implementation of comparative studies in general.  

We then walked through design of a causal comparative study to look for correlations between feedback = independent variable (in terms of grades or conferencing/comments) and revision = dependent variable.  We ran into lots of difficulties with too many variables and figured out strategies to eliminate some of them (such as only comparing courses with the same level students + same focus). 

Applying this method.  You then used  a modified/hypothetical version of Merten's "steps"(see below) to design causal comparative/correlational studies for (possible) projects by Tim, Neiha, and Wayne.

1. Identify the research problem.

2. Identify the independent or predictor, dependent or criterion variables, and other relevant factors.

3. Identify participants (groups).

4. Describe methods for collecting data 

5. Reflect on possible sources for bias.
6. Pose (anticipated) competing explanations of (anticipated) findings.

I think I kind of sabotaged these discussions - turning everything into ethnographic/case study research - but then I guess that just goes to show the "methods" bias I carry with me.  I think - as you developed them - there were grounds for doing each study as causal comparative, though you would want to put some more time into the organization + design. So put those ideas into your bag of tricks = and keep them in mind as we work our way through the other methods.  

  

Sommers & Brodkey
Simone's and Musheerah's essays are posted to the right.  While our discussion did not focus on the study design - these two particular studies were selected because they "compared" two groups. In fact, each was a case study, though Sommers' piece comes closest to being causal comparative.

Nancy Sommers:
Using our list above, we might think about Sommers' essay as causal comparative in the following ways.
1. Research problem = how does revision correlate with experience writing?
2. Independent variable = writer's experience, dependent variable = revision process
3. Participants = inexperienced + experienced writers from similar geographic  areas (defined in essay) = (also  similar in terms of first language, socioeconoic, ethnic, gender, class distribution? what other similarities might be important.  In other words, what differences within the groups could confuse the data?)
4. Method: Participants wrote three papers (expressive, explanitory, persuasive),and revised each essay twice.  Each participant "was interviewed three times after the final revision of each essay" Essays were coded to identify process + writers' concerns.
5 & 6. Possible sources for bias + competing explanations: "post hoc" fallacy = "proving" the assumption that writers need to be "taught' to revise (novice writers do what they are "taught") => an alternative explanation might be that the younger writers were "developmentally" at a place where thinking processes might not favor revision (think about Bizzell's essay about Perry); this study is especially vulnerable to that since it does not control for age.  This is a much a study of older v younger writers as it is of experienced v inexperienced writers.  Also - as pointed out in class - we noted that the experienced writers were not "taught" to revise - since they (probably) had similar experiences in school - so the conclusion about changing teaching methods - while it may be true - is not relevant to how the experienced writers learned how to revise, since the study does not really examine the process through which experienced writers learned to revise.

Linda Brodkey .  
Brodkey's study does not fit quite so well as a causal comparative study - though it does look at positioning by identity (teacher-v student) within academic discourse = where teachers control the talk and have the power, and students receive talk and have considerably less power.

The points that most interested me were raised in our discussion of Brodkey's conclusions - that the power differential in the literacy letters exchanged between teachers and returning adult students could be accounted for both in terms of the teacherly discourse -and teachers ( unconsciously) not wanting to give up their power - and class.  In our discussion, we raised questions about whether this dynamic was "only" from class. We raised questions about student agency (especially because these were adult students) in terms of determining/following through with topics, and wondered about alternative explanations for why the exchanges turned out as they did.  We speculated about whether there might have been very different explanations about how/why the letter exchanges went as they did had Brodkey ASKED the participants what happened, or why they responded as they did.

We also noticed that though Brodky presented this as about equalizing power relationships in teacher discourse, she did not make any moves to equalize power within researcher discourse.  That is, she held full power in terms of interpretive authority.  Neither student nor teacher participants were asked for their input or interpretations.

Great class tonight.  Your applications of the methods opened new questions and - hopefully- posed possibilities for you as you approach your research.

For next week:
Read: Mertens, Ch 6: Survey methods; Anderson et al (2006, pdf on Course Blog;  Bartholomae (1985), 523;  Royster (1996), 555; Elbow (1999), 641.

As I mentioned in class, the Anderson piece will illustrate many issues that come up with surveys, and the Bartholomae, Royster and Elbow will take up issues in teaching writing as they overlap with discourse.