week 8 Flashcards
What is the idea behind classical test theory?
Observed score = true score + measurement error
X=T+E.
if scores on the same test correlate, even throughout different times, raters etc, then reliability is good.
What is standard error of measurement?
SEM indicates the level of measurement error associated with (an individual’s) test score. It is simply the SD of measurement error.
What are the five stages of test development?
Test conceptualisation, test construction, test-tryout, analysis, revision
What is a dichotomous measure
Only two options, like yes/no, t/f
What is a polytomous measure
2 or more alternative answers
What is a forced choice format
When theres an even number of questons, with no middle ground (i.e. no ‘neither agree nor disagree.’
What does item difficulty mean in this context?
Applies to tests of achievement or ability, the % of people who get them correct. good tests have a large spread (HD, D, C, fail etc)
What is Discriminability?
When you takr the top and bottom 33% of your group and see which ones they got right and wrong. Also known as a point biserial method. Correlation between performance no particular items (dichotomus) and performance on whole test (continuous)
What is the item characteristic curve?
A fucking irritating statistical thing.
So on a plot, the X axis is theta. This is the thing you are trying to measure. Y axis is probability of response to ‘Yes’ for item.
So, we look at this curve, and at p=.05, thats where people go from saying no to yes. the sharper the curve the more abrupt the transition.
‘probability of endorsement’
criterion-referenced tests
performance measured against pre-determined critera (driving tests)
norm-referenced tests
performance measured relative to others (like intelligence)