Factor Analysis in Dissertation & Thesis Research
In some dissertation and thesis research designs, you may want to break a large set of variables down into smaller sets of related data. Factor analysis is the statistic used to determine if any of the independent variables comprise common underlying dimensions called "factors." Through factor analysis, you can find variables that are correlated with each other, but relatively independent of other sets of data.
Factor Analysis Uses
After employing factor analysis, you may believe that your subjects' responses to the many different questions are driven by just a few of these underlying factors. For example, your paper may hypothesize that there are two major factors that underlie mechanisms for coping with depression. One assumed factor could be locus of control (internal versus external) and another factor could be presence of pleasant events (seeking them out versus withdrawing from them).
Factor analysis is particularly helpful in developing and testing theories. For example, your research may find that women use the two coping mechanisms hypothesized in your dissertation, while men in your sample use a totally different factor structure to cope with depression.
One of the biggest uses, though, of factor analysis in dissertation and thesis research is the development of objective tests for psychological measurement. If your goal is to develop a new measure of personality, for instance, you may write a very large number of items to include on your personality instrument. This large instrument then is given to a random group of subjects. Once the data has been entered and a factor analysis has been run, individual items on the instrument may be added or deleted, depending on how they "group" together into factors. This second instrument then is administered to a different group of randomly-selected subjects. Your data collection and analysis would continue like this until you had an instrument with several items comprised of factors you believe represent the construct you are trying to measure. In this example, that construct would be personality.
Factor Analysis Steps
The steps involved in performing a factor analysis for a dissertation or thesis include choosing and measuring a set of variables, running a correlation matrix, pulling out a set of factors from that correlation matrix, determining the number of factors observed in the correlation matrix, possibly rotating the factors (see a good statistics book), and then, finally, interpreting the results of the factor analysis. Sounds complicated, right? Actually, most statistical software packages make this process very easy. You simply select options from a pull-down menu and fill in boxes using the variables in your data base.
You will know if your factor analysis is a good one because the factors will make sense. A good factor analysis is one in which the named factors are consistent with the meaning of the combinations of variables that make up each factor, and these variables are highly correlated with one another. Furthermore, a good factor will have several highly-correlated variables that do NOT correlate with other variables that "load" onto another factor.
Factor Analysis Types
There are two major types of dissertation factor analysis. One is exploratory factor analysis. Exploratory factor analysis usually is used in the beginning stages of analysis, as it can be a tool for condensing variables and generating hypotheses about the underlying dimensions of your paper. The goal of exploratory factor analysis is to summarize and/or describe the data by grouping variables that are correlated with each other.
There are two types of exploratory factor analysis. One is principal components factor analysis and the other is principal factors analysis. For most dissertations and theses, it is customary to use a principle components factor analysis, as this analysis technique allows for the extraction of as many significant factors as possible from your data set. Unless you have clear information from an advisor, for example, that principle factors analysis is better, use a principle components factor analysis for your work.
The second type of factor analysis used in dissertation and thesis research is confirmatory factor analysis. This statistic is much more sophisticated than factor analysis and is used when the research is in an advanced stage. The confirmatory factor analysis is used to test a theory of an underlying process. In confirmatory factor analysis, the variables are chosen carefully and specifically to illustrate the underlying process of the construct. Most dissertation research involves exploratory factor analysis, so we won't go into a more detailed description of confirmatory factor analysis in dissertations here.