Second Creswell Posting

Creswell, John W. Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. 3^rd ed. Thousand Oaks: Sage, 2009. Print.

(Blog entry based on 129-226.)

I appreciated the pragmatic approach of this book. Creswell’s suggestions and plans for the different research designs made research seem all the more approachable and possible, though he was often quick to point out the amount labor involved. Furthermore, I found the discussions here quite generative. I began thinking ahead to what I want to accomplish with my dissertation and which research designs may help in this.

Research Questions and Hypotheses

As I want to emphasize the procedures for research designs, I only will raise two questions on this chapter. Creswell is seems rather adamant about what to use (research questions or hypotheses) in what types of studies. In his discussion of qualitative studies, he only addresses the use of research questions, stating that hypotheses are not to be used (129). I understand that qualitative studies are exploratory in nature, but does making some assumptions about possible outcomes necessarily constrain the study and limit its ability to be exploratory? Quantitative studies, he says, can use both research questions and hypotheses but only if the hypotheses build from the research questions (133). What strikes me here is that I understood hypotheses as natural outgrowths of a study’s research questions. Are there instances in which the hypotheses are not extensions of the research questions?

Quantitative Procedures

I was glad to see half of the chapter on quantitative procedures dedicated to advice on survey methods. I have been considering a survey for my dissertation, so learning about some issues to consider prompted me to begin considering the forms and questions my survey(s) may take. In terms of population samples, Creswell suggests that random sampling yields more generalizable data (148), but I think I might be more likely to use the purposeful selection common in qualitative studies (178). This will allow me to craft targeted questions/prompts and receive some richer data. As my survey develops, I will undoubtedly be returning to Creswell’s discussions on validity (the possibility to draw reasonable inferences from the data) and reliability (the consistency of the results) of my instruments. At this point, I’m not entirely sure what validity and reliability issues I will have to address, but this section will be a useful reference point for me.

Also helpful in my thinking about research design were the steps on data analysis and interpretation, especially when considering response bias. This refers to the bias created because of non-respondents. Creswell suggests using wave analysis to mitigate response bias—i.e., analyzing returns weekly (or at least frequently) to see how the responses are affecting the outcomes and then using this to determine what the non-responses might have contributed (151-52). Prior to reading this, I expected to deal with non-responses, but I hadn’t considered how it might bias the study or how to respond to it.

Questions on Quantitative Procedures

I am still a little fuzzy about some of the terminology and statistical methods used in quantitative procedures, and given Creswell’s goals here, I didn’t expect him to detail these at length. But I am wondering a little bit about the power analysis model (157). How does one establish values for the three elements (alpha, power, effect size)? This may not play into my work, but some more knowledge of these types of determinations may be of use in my own reading.

Qualitative Procedures

These I am a little more comfortable with—at least in terms of documentary studies. Creswell identifies nine characteristics of qualitative studies. I’ll not cover them all here, but some characteristics include a “natural setting” (i.e., not a lab), the researcher as the tool to interpret data, numerous types of data, and inductive analysis (175-76). These fit quite well with my experiences and strengths regarding research. Again, Creswell does well to provide a series of steps one might use to interpret and analyze the data collected. These are clear and valuable to anyone engaging in this process. Most meaning to me is his discussion on coding data. This is a process I have engaged with only in some unsystematic ways (in terms of how I understood approaching the process at least), so his explanation of the coding process (186), will be something that will be of considerable importance as I move ahead in my dissertation work (and it will help me organize my thoughts as I go along).

Question on Qualitative Procedures

This question has less to do with procedures and perhaps more to do with where Creswell locates some discussions. He discusses IRB considerations in the chapter on qualitative procedures but does not mention them in relation to quantitative procedures. Is there any particular reason for this?

Mixed Methods Procedures

As I stated in my previous posting on Creswell, a mixed methods research approach seemed most appealing and sensible to me, and this has not changed after reading more detailed accounts of all three research designs. To me, exploring my dissertation topic through a combination of survey results, interviews, and documentary research seems to be the most probable and most comprehensive approach.

What I have yet to determine is the type of mixed method model this will take. Creswell identifies six types of mixed methods designs:  

• sequential explanatory: a quantitative study followed by a qualitative study that extends the quantitative data; 
• sequential exploratory: a qualitative study followed by a quantitative one that builds on the qualitative data;
• sequential transformative: one type of study followed by another, both framed by a particular theory; 
• concurrent triangulation: both types conducted at once with comparisons between the data sets; 
• concurrent embedded: both types conducted at once but with one method more prominent than the other;
• concurrent transformative: both types conducted at once with a particular theory guiding the data collection and interpretation (211-16).

Here again, I think Creswell explains the possibilities clearly—I just need to begin thinking about which of these might best suit my needs. My hunch at this point is that I will lean more toward a concurrent model because of time constraints that I may experience.