<< Click to Display Table of Contents >> The end? |
|
Here we are, time's up.
What have we talked about?
Items
cognitive and affective
Item pools
we got a bit into the deep end
may have been boring
Marbles and the binomial
having "all our marbles" is usually what we want
the binomial is neat
Samples and confidence intervals
bigger samples and better items reduce the size of confidence intervals
Reliability
we want to have it
it's really important
coefficient alpha is commonly used as the reliability figure for a test or survey
alpha is good when it's 0.80 or greater
low alpha may not always be the end of the world
we can forget about the total score and concentrate on results at the item level
but while this is true, good alpha is what everyone usually wants
Measurement error
we don't want it
but we can't get rid of it
bigger samples (longer tests) help to minimize it
Validity
several ways to measure it
many of them look at correlations with other tests or surveys
Coffee and donuts
Happiness
I thank you very much for your attention, and for the questions you asked (if you asked questions).