This past week we went to the lab to learn about how to perform a factor analysis by using Excel and SPSS. We analyzed data collected by the ENVS 220 class in an EcoValues Survey that they administered to their Facebook friends this past fall. There were approximately The whole concept of factor analysis was foreign to me, and it only began sinking in after Jim explained it to me in about six different ways. Basically SPSS takes a bunch of the survey data that uses numerical attributes (such as a Likert scale) and analyzes it for patterns or trends, ultimately creating several different factors from those trends.
I examined how the survey participants perceived themselves by including questions such as how much the participants identified with a certain listed characteristic. After I ran the factor analysis, I found the top three factors had these characteristics:
- having a good time/spoiling yourself
- admiration for yourself
Characteristic, Factor 1, Factor 2, Factor 3
Creativity, , ,.688
Equality, , , .710
Status, , .864,
Surprises, .805, ,
Pleasure, .719, ,
Helping, , , .602
Success, , .847, ,
Adventure, .779, ,
Loyalty, , , .702
Fun, .765, ,
From there, I was left to come up with my own “big words” to encapsulate all these qualities into a single factor. In class we discussed that Big Words generally are conceptual in scope, have cultural relevance, and practical significance. The first factor I called the “Thrill-Seekers,” who seem to run off of excitement and surprises as well as personal experience. The next category I named “Status-seeking,” just in terms of the money and success related attributes of the second factor, and the third factor I labeled as “Justice-seeking” because of the equality and help related aspects of the factor.
I thought this was a really good experience to have, even if the statistical analysis side of big words can be a bit overwhelming. I’m not sure I completely understand how SPSS spits out all these numbers yet, but I thought the way that it can make qualitative data a bit more quantitative was really useful. Instead of making a word cloud or something like that, this tool used statistics and Likert scale data to quantify qualities. Even after it produced the rotated component matrix, it wasn’t easy compiling all of the individual attributes into a single “big word” that encompassed them all, though it did help to have weighted values for each attribute.
Based off of this quick SPSS tutorial we did this week, I can definitely see myself using this tool for future text and survey analysis, especially in the context of people’s sense of place.