Wednesday, 30 October 2013

MAT: Language of Stats

Three points in this post, all related to statistics, aka Data Management. Partly posting it for me, to remember this stuff next year. Which I could do offline, so it's partly for you too - if it's helpful, and because I'm curious to know if anyone else has experienced these effects.


First point. I have been teaching statistics for a number of years. Something I've noticed is that students tend to use the term "survey" almost interchangeably with "sample". For instance, "I will survey the population randomly" or "I will survey by grouping".

In our data collection unit, I often have a word contrast element as part of my quizzes and tests. So last year I put up 'sample' and 'survey' as a pair of terms to compare. It threw some of them off, we discussed, I think things worked out.

This year I kept it on my quiz - and also put it on a quiz to my Grade 12 college level students in their statistics unit. I saw a LOT of this:
Example from 4C class

To be clear, I had defined census previously, so was a bit confused as to where this new "survey"="census" belief came from. Is it that you give a survey to all the people in your sample? Yet I also got the impression that some were simply going for the opposite of 'sample'.

When taking it up, I pointed out that 'survey' is the QUESTIONS, with no direct link to who in the population you end up asking. In general, I find student use of the term "survey" curious. Anyone else run into this?


Second point. When I ask the students to generate a question with bias (eg- it's leading, it's double barreled, it's unclear), the first reaction for a number of them this year has been to offer up what I might call a probing question. For instance, "Do you use drugs?" is offered up as a biased question.

"It's just a question! Answer me!!"
I explain that such a question is not inherently biased. For instance, this could be for a pharmaceutical company which needs that kind of data. If I said "Drugs are bad! Do you use them?", that would be biased. If a guy dressed like a cop give you a suspicious look as he asked the question, that would be biased. The question itself is not biased. And it's not merely the topic, "Do you eat fast food?" was a similar suggestion.

This could, of course, be a misunderstanding as to the nature of bias. But I can't help but wonder if some of this comes from thinking that things which are bad for us have bias by default. Or from the urge we have to dodge questions we don't like!! (Gee, when have I seen that lately? *cough*Stephen Harper*cough*) It concerns me, because if society starts shooting down probing questions by citing "liberal bias" or "conservative bias", we could all be in real trouble.


Third point is more a point of awesome. There's a lot of definitions in this data collection unit, from sampling, to bias, to classification of data and types of studies. One student prepared index cards, with terms on the front and their explanations on the back! Nice! I asked her if I could have them when she was done with the course. She said sure.

Then she simply made me my own set!
There's 25 of these

Today, the day of the test, I also saw two other sets that other students had made. Very nice! Must remember to suggest this next semester, possibly even incorporate them into a review somehow.

Final Unrelated Bonus:
 My Animated GIF Twittereen avatar (adapted from nik_d_maths) isn't displaying properly through twitter. So here it is, as a reward for reading to the end.

It's related to FRAKKED in my web serial

No comments:

Post a Comment