basic statistics in python

dsavitsk dsavitsk at e-coli.net
Thu Mar 14 16:46:10 EST 2002


"Louis Luangkesorn" <lluang at northwestern.edu> wrote in message
news:3C910734.9030109 at northwestern.edu...

> I don't recognize bi-serial correlations,

maybe there is a better name? it is used to validate test items. the fmla is
...

x = score on item
y = score on all other items
n = number of people taking the test

n(sigma(xy)) - (sigma(x))(sigma(y))
-----------------------------------
sqrt(
((n)(sigma(xx))-(sigma(x))(sigma(x))((n)(sigma(yy))-(sigma(y))(sigma(y))  )

something like that, there may be a parenthesis or 2 out of place

btw, thanks for all the suggestions.  it still isn't clear what the best
solution is, but i am thinking that dumping data to R or SPSS via COM might
be the least painful.

-d






More information about the Python-list mailing list