Now Playing: Dekho Maine Dekha Hai (Love Story)
Topic: Factor Analysis
Well, I've been reading the paper John sent me on "Teacher's beliefs and intentions concerning teaching in higher education" (first author is Norton, L.). I only have the author's proof mind you ... so the document might change ... but so far very interesting document ... I am taking my time reading through it since I believe I may use similar methodology in my work. I've only read the first 9 pages of the 35 page document!!!!! Oh well, I took a short break to read up on factor analysis in Andy Field's "Discovering Statistics" - it is much easier read than Howell's "Statistical Methods for Psychology" ... besides Howell doesn't have factor analysis. Had to do that before I proceed further into the paper - because I think I would have been lost!
I think I've finally understand factor analysis ... (well ... at least to understand the words when I read the article) ... for so many years when I read those stupid papers on their questionnaires and their mention of factor loadings and varimax used to get me so confused ... and what they meant by a factor ... I finally think I know ... Could you imagine, these factors that these people supposingly discover, are just a figment of their imagination ... well, not literally, but they are taking variables that cluster together ... and then they 'dream up' some word that will describe what these variables are describing ... like if age correlated with sex (M/F) and those two correlated with hair colour ... then they will say that factor is personal characteristics ... but whose to say the factor is actually personal characteristics but not describing something different ... you got to know good vocabulary to come up with these factor words ... to know what they could be related to ... you should take an English course in word association for factor analysis in my opinion. As for varimax ... its just simply a way of understanding the data better by transforming it... and the transformation is accomplished by just rotating the data about the factor axes - yup the factors - whatever name you decide to call them, can be considered on their own right as a kind of variable, and have relationships with one another, and hence can be plotted as a graph, with each having an axis!! If the variable (a real variable from the questionnaire) seems to be more incorporated towards one factor then it has a higher factor loading (basically a larger value on that factor's axis). But yeah that reminds me ... these variables from questionnaires ... what exactly are they?? I mean Field gave me the impression that the variables can be a question from the questionnaire ... but reading other people's papers ... I got the impression that variables are pre-determined and questions are asked that relate to that variable, and the means from these questions are used to represent the score of the variable. Now ... who's to say that those questions accurately measure that variable?? Won't a separate factor analysis will be needed on these set of questions, to see if they are measuring one thing i.e. the variable - then from a factor analysis, only one factor should be given ... not sure if this is sense to anyone out there ... but it is making perfect sense to me!
John's paper by the way is looking at teaching beliefs and teaching intentions of teachers in higher education institutes ... cool isn't it? They used a questionnaire based on some guys called Gow and Kember ... can't find a reason why they used it -probably in the paper somewhere and I am missing it. Anyway, this will be cool to check if teaching beliefs or intentions are related to the use of software and its types ... don't you think? Well ... first got to develop a questionnaire like Gow and Kember for classifying the software into computational and learning tool (unless I can find one ... somebody help me please!!!)... and maybe then I can find some factors ... and throw in the words varimax and factor loading wherever I feel like!