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!