Update post exam, 26th April:
We had to answer 5 questions (no options) in 2 hours. First impressions: this is going to be much easier than the stats exam. But surprisingly, I found myself running out of time, and actually failed to find the time to answer a part of one question, and not quite 100% sure of my answers to some others. I think this is the price I paid for being too overconfident. However, I am confident of at least passing it. A few colleagues remarked that they’ve scored full marks on both stats and SPSS exams, hence they’ve got at least 65% of the QA course marks in the bag, meaning they can skip the research methods module altogether!
For this module, there are 300 slides to go through in 5 days.
By the end of the 2nd day, we were already at the 178th slide. I must add here that it’s rather embarrassing to the lecturer and INTAN in general regarding the state of the computers in the huge lab. The PC she used popped up a window every few minutes with the big bright words: “Warning, infected with virus”. Quite a few of the computers are infected with malware, and some did not function at all. At least one had a problematic SPSS installation – it would quit without warning. It was so bad that she had to declare that the exam would be paper based instead of computer based. And the INTAN director’s office is just a short walk down the block!
Back to the module. My initial thought was that this module was going to be easy – all computer based – let it do all the number crunching – but as we went on, there are actually so many new things to be learnt with regards to arcane statistical concepts. We didn’t actually go through the background of the names, there was simply no time for that as the lecturer zoomed through the slides like there’s no tomorrow.
For example, in checking on the responses to your questionnaires for your research, you’d have to ensure that it has what’s called reliability of scales, meaning the scales you use consistently measures whatever it’s supposed to measure. Scales in this case can be of the form: 1-strongly disagree, 2-disagree,…, 6-agree, 7-strongly agree. For that, you’d have to first compute the “Cronbach Alpha”. If its value is more than 0.6, then you can rest assured that your questionnaire is OK.
But if not, you’d have to fix it by doing the next step called the “Factor Analysis”, where you’d select the “Rotation method: Varimax with Kaiser Normalization”. Then you’d display the “Scree plot” and extract “Eigenvalues” that’s over the value of 1.
We’d encounter dialog boxes like this
The resulting scree plot
Another crazy dialog box
Hair-tearing stuff indeed, especially when it’s a hot afternoon in a room containing more than 70 computers with less than adequate airconditioning.
Apart from the Scree plot (chart), I also encountered at least 3 other charts which I would never have imagined the usefulness of had I not embarked on this course. They are the mysteriously named “stem & leaf plot”, “box plot” and “Q-Q plot”.
A box plot
A Q-Q plot
I also encountered colourful sentences in the notes like the following: “Need to test formally using Kolmogorov-Smirnov or Shapiro-Wilk statistics to check for departure from normality.” No need for that, we’re confirmed to be way departed from normality…
And we have not even started on hypothesis testing yet. That should cause us to let off more steam from our heads.
The exam is next Tuesday.