I need a little help. I am a Statistician by nature, so sometimes get a little oversensitive when it comes to analysing stats - I get ... wound up? irritated? ... by dodgy stats.
Here's the latest, and you can tell me if I'm right, or if I need to shut up and get on with it.
Last years Y13 did an exit questionnaire, for ALIS. It asked them all about what teaching styles they experienced in each subject that they studied. For example ...
"Presentation of a topic by the teacher" or.../
" Practical Work (Using apparatus or Making things)."
... and so on. There were 22 catagories.
Students rate each one from 1 to 6, on the follwoing measure..
1 = never or almost never
2 = about once a term
3 = about once a month
4 = about once a fortnight
5 = about twice a week
6 = about once a lesson.
Scores for all students in each subject (sample size = 7 in my case, but that's another gripe) were then averaged (arithmetic mean for fellow stats types), and these published and compared with national figures.
Here's the problem(s).
1. My Preparing Essays scores was 1.29 (1+1/7). This means that one student thought we prepared essays about once a term. I haven't prepared an essay since my PGCE in 1996. This is a lack of awareness by the students that sheds doubt on all responses.
2. If 4 students respond "Never" and 3 respond "once a lesson" then the average is 22/7 = 3.14 which appears to be once a month despite the fact that NOT ONE STUDENT said once a month. Surely the mode (most common) response of "Never" is more appropriate?
Please help me out with this one?!
Ah, the joys of student voice. I've recently been giving my students postacrds, which ask what they liked about the lesson on one side and what they didn't like on the other.
ReplyDeleteThe first few times we did this, my Y7 gave some fantastic sounding answers that bore no resemblance to reality: for example one student expounded on the many ways the group work (that we hadn't done) had helped him to learn. I now throw away all the ones that could be filed under 'irrelevant response' - some lessons that doesn't leave much to work with.
Getting back on topic: Personally I'd say that if 4 students respond "Never" and 3 respond "once a lesson" then the results are probably not very meaningful, but if anything I'd go for the mode, which at least bears some resemblance to the actual data.