Friday, October 2, 2009

UCASS Questions


Good morning all!

Alrighty here's a good one - one that I'm hoping someone can help me understand or maybe we can obtain some clarification from the Education folks :)

UCASS - is a population study - so essentially a Census - gathers information from all Universities and Colleges in Canada. So far so good. Now my researcher has asked the Division to obtain a mean and a standard error for a number of continuous or scale variables. (Aside: yes she's paying for each round of calculations and custom tabulations.)

The response they gave her yesterday was they cannot calculate a standard error for any values since it's a population study. Hmmm... ok if I think about this - yes it's a census so in essence this is the only population of this sort available and working on this notion maybe my researcher should have asked for a standard deviation. However... this information was not suggested and my researcher was also told that because this is a population study they are unable to provide her with any variation measures. This is where I get confused.

Can anyone help me understand this? Why is it NOT possible to obtain any variation measures from UCASS?


I did get some clarification from Education Division on this, and it seems to support what Susan explains below:

"We cannot provide any measures of variance (i.e. standard errors or coefficients of variation) because UCASS is a census of its targeted population. Measures of variance measure how much an estimate calculated from a sample may be different from an estimate calculated from the target population. However, since estimates coming from UCASS are calculated from the population, there is no need to figure out how much it differs from the population since it *is* the population."

I hope that helps.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.