[Edu-sig] programming skills and IQ tests..

[Jeremy Gray]

Hi Jurgis,

being edu-sig, I'll send this to the list and not just you personally,
despite it having somewhat low actual python content.

I teach at grad and undergrad level, and do research into individual
differences in self-control, and sometimes use IQ tests (see
http://www.yale.edu/scan/) python is pretty useful in the lab, especially
PsychoPy.

I'd like to test my students' general ability to undersnand examples.
> Because I am confused, is it my fault or their laziness, or sth else,
> that some of them (nearly 30%) understnand nearly nothig what they were
> taught.
>

rather than to get into assessing and interpreting IQ-ish scores, I think
you'd be much better off asking other teachers / instructors for advice on
how to get through to those 30%. from what you are saying, the real problem
you want to solve is how to get through to those students, right? maybe a
mentor teacher can sit in and observe you and the class and offer
suggestions? or ask someone who taught the course previously what examples
they used, and how the course went? maybe you could ask students for
anonymous feedback after each class one what was really clear and what was
really unclear. (how useful this might depend on the age group.) you might
ask about whether students can see well enough, hear well enough, and so on.

it sounds like you are hoping to tease apart motivation from ability, which
is not a trivial thing to do. for this, giving students problems to do in
class is unlikely to be informative: its the same class (with the same
social setting including distractions, history of expectations, etc), the
same time of day (fatigue, etc). its not just as simple as giving a page of
problems out, and getting a score. once it become "IQ" there's an implicit
threat of revealing that a given student is either lazy or incompetent, so
some will become nervous and do poorly as a result. and there's what's
called stereotype threat, which is the fear of confirming a negative
stereotype about one's racial or ethnic group. its basically a way of
eliciting anxiety that interferes with performance, and everyone is
susceptible to it (including white males in the context of being compared to
Asian males on a math-y test). so IQ: its better not to go there if you
don't have to. and the real problem is getting through to that 30%, not
quantifying their ability.

if you still want to go the IQ route, out of personal curiosity or whatever,
you would be better off if you can obtain student scores on a standardized
test that someone else has already administered, on which the student was
personally invested in doing well, and took under optimal conditions (low
distraction, etc). if you are in a school setting, maybe there's a way to
work with the administration to get standardized test scores? if these are
college level, SAT scores work surprisingly well as a measure of general
cognitive ability (you can convert one into the other with pretty high
accuracy, for individual people).

or perhaps down the road we could collaborate on a research project :o) I'm
only half-kidding. there would have to be a scientific question, not just an
assessment of students in a class.


> Main question of email:
> Does anyone know freely distributable DB of such IQ quiz questions --
> preferrably mostly graphical, like
> http://iq-test.co.uk/iq-test/ or so
>

sorry, I don't know of any but I've toyed with the idea of developing a
battery of mini-tasks for this. there are a number of pretty creative, and
non-threatening tasks out there that correlate to some degree with IQ.
reaction time does, for example. if we got enough of such tasks, implemented
as mini-games in python, ... could be interesting.


> I think it could come in handy for others (mathematics here as well)...
> I also hope, one can  train the attention/concentration with IQ tests
> -- which is verty important skill in understanding examples.
>

yes, its possible to train attention / concentration to some degree, hard,
but possible. people in my field talk about "working memory training" (for
which the wikipedia page is pretty bad and out of date). the most
interesting study I've seen is by Jaeggi et al, available free,
http://www.ncbi.nlm.nih.gov/pubmed/18443283  (or pdf:
http://www.pnas.org/content/105/19/6791.full.pdf+html)

best regards,

--Jeremy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/edu-sig/attachments/20100506/7c62b4c2/attachment.html>