Janice M. Morse
32
DEVELOPING THEORY USING META-ANALYSIS
“I am paying for this microphone!”
—Ronald Regan (when the moderator threated to cut off his mike at a debate he paid for)
Brainstorming: Sorting Out Meta-Analysis
Morse: | But studies overlap in different ways. They overlap because they are about the same phenomenon or the same concept, or they overlap because they are about the same disease entity, or they overlap because they are about something else. And you have to make your decision about what it is you are collecting studies on. And people go around collecting studies about illness. Now when Joy [Johnson] and I put together all of those studies on illness, we were criticized because people said that Norris’s study was about abortion, and “abortion wasn’t illness.” This is ridiculous. And when I told Judy [Norris], she said, “My study wasn’t about abortion; it wasn’t about illness—it was about soul ache.” So what our theory is about is human responses to a threat to the self, or whatever. And if you look at it that way, then Judy’s study fits perfectly. If we create these artificial boundaries and don’t let these studies, as Glaser says, “earn their way into data,” then you get into a whole lot of trouble, and you have a tremendous threat to validity—if you want to use that bad V word. So that’s why I think you have to start with the most abstract concept and work down. |
May: | That’s a nice bit of intellectual footwork! But 99% of folks, who attempt to do that, can’t do that. Even in my field if you took all the quantitative work on at-risk childbearing—there are not very many, but you could pile them up on the table. But then to push through and find out what the study really was about. And to think: this one is about “technological management of conditions.” And this one is about “suffering and anticipatory grief.” Do you see it’s a level problem? You’ve got apples and Tuesdays. And the source data don’t necessarily tell you [what it’s about]. Jan [Morse] had to get inside those studies and then say, “But what did they really do?” |
Morse: | But it is false for us to go and collect studies on heart attacks and strokes. |
Sandelowski: | But it’s the way you start. In order to get to the point where you are—that’s something that’s invented by the analysis. And in order to get there, you have to go through this process. A naturally occurring way to start is, “Gee, I am interested in people’s responses to heart attack,” and then to say the essence of this thing is not the heart attack, per se. The main part of the analytic work is getting there—you don’t get there first. |
Morse: | You are totally right. But I don’t know if you have to do the coding, or if you have to be theory-smart—the problem is, if you start off with the data, you get bogged down. You get into not seeing the forest for the trees thing—you can’t get there. To do a meta-analysis, you have to decide on your goal before you start. So you are doing this analytic stuff and working almost deductively, organizing your theory. |
—Wuest and Merritt-Gray (1997)
META-ANALYSIS
Throughout this book we have discussed various ways to develop solid, clear, and appropriate concepts. In this chapter we will discuss how to create a mid-range theory from multiple interrelated studies that resulted in lower level theories. Qualitative meta-analysis has a different purpose, and therefore a different outcome, than qualitative meta-synthesis techniques. Recall that meta-synthesis “smoothes” categories across attributes, looking for commonalities, for the “most similar,” for common agreement, to make the original concepts or theory stronger. When doing meta-synthesis, researchers use studies that address a similar area or processes, and the final labels used in the theory may or may not be the same as (or similar to) those used in the original study. It makes for a stronger theory about a certain topic.
On the other hand, qualitative meta-analysis is a technique that develops a higher level theory from a cluster of studies from different contexts and different populations, but studies that address the same concept and the same phenomenon. Rather than meta-synthesis that produces a concept about the same topic, meta-analysis produces a higher level concept. For instance, in Chapter 33 we use grounded theories that address all sorts of illness problems to develop a meta-analysis of the experience of illness.
Qualitative meta-analysis uses data and analyzes the internal structures of mid-range theories to create new, higher level, formal theory (Estabrooks, Field, & Morse, 1994; Schreiber, Crooks, & Stern, 1997). This formal theory is different and more insightful than the studies from which it has been derived; it is at a higher level of abstraction, and the study represents the domain of the higher level concept. By conducting a meta-analysis, the researchers are working with a different level of data and manipulating it differently, than when using the meta-synthesis. Researchers are working with the concepts and theories developed from other studies.
For instance, in this example, grounded theory studies that have developed theory about women adjusting to heart attacks, having a hysterectomy, discharge from a psychiatric hospital, mothers involved in their daughters’ abortions, and husbands’ experiences during their wives’ chemotherapy are used to create a higher level theory of illness. The theory of illness is much broader in scope than any of the single theories from which it is composed, and most importantly, is applicable to conditions that were not included in the original study—and to the experience of people who are yet to experience illness. It is a way to make the resulting theory more generalizable, more useful. If done well, the theory should be applicable to future instances of the phenomenon. That is, the theory developed in Chapter 33 should be also applicable to cases or illnesses that are not included in the original analysis. The level of abstraction is not particular to any specific illness.
In the literature there is some crossover with the terms “meta-synthesis” and “meta-ethnography” (Sandelowski, 2003). Therefore meta-analysis may be found as meta-ethnography (Noblit & Hare, 1988) or qualitative meta-data-analysis (Paterson, Thorne, Canam, & Jillings, 2001). Note that this process of meta-analysis and the outcomes are quite different from meta-synthesis.
DOING META-ANALYSIS
Data for meta-analysis are derived from a number of studies that address the same concept, but in different topics. Does that make sense?
To Select Your Studies
Supposing you are interested in a concept, such as posttraumatic stress disorder (PTSD); you may identify studies describing returned soldiers, people who have had serious accidents, or even women who experienced difficult childbirth. All of these studies do not have a context in common; they do not have a population in common, but they address the same concept. And as with meta-synthesis, your task will be easier if they all use the same methods that will produce the same theoretical form. In the Chapter 33, the studies are all about various aspects of different health crises, using Glaserian grounded theory (Glaser, 1978).
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
