Mike, Not quite: Table 2.5 still breaks it down to average values per distractor, so still not what I'm looking for. I entered the options & key using IREFER into winsteps, and so most tables use the data to give the answers per option chosen. Unlike this, Table 2.15 just has the average for those who got it right and wrong (missing combined with incorrect responses), and its a table that has these numbers that I'm looking for.
The reason why I want the figures is so that I can set some kind of yardstick for revising test items. I want to say things like "in this test, X number of items had the average measure of those getting it incorrect within X pts of those getting it correct, and so should be revised".
Of course, I could get these figures by revising the input file so that it is composed entirely of '0's and '1's, but then I would lose the useful distractor analyses that winsteps gives.
Oh yes, another question!
Is it possible to compare the performance of different sections of the same test? My test has 28 questions composed of 4 sections of 5-10 items each, and I want to see how each section contributes to the overall reliability of the test. At present I can do this by making four different versions of the same control file, and copying the results of each into an excel file, but this is rather labor intensive. Maybe there is an easier way?
Thanks, as always, for your time and patience. David
David, Winsteps does support summarizing item statistics by sections of the test, and also doing DPF (differential person functioning) by sections of the test. But comparing different sections of a test for statistics such as "test" reliability requires a separate analysis of each section. This is easily done in Winsteps by putting a section code in the item labels, and then doing multiple runs of Winsteps using ISELECT= from the "Extra Specifications" prompt.
Hi Mike, I just want to check something related to deleting items from the analysis.
This time I'm trying to remove three poorly performing items, so I rerun winsteps and at the extra specifications menu I type 'IDELETE= 4,11,22' (following the format given on the help page). However it comes up with an error message asking if ';' is missing.
So, is it correct to say that multiple single items cannot be specified at the 'Extra specifications' line before running the analysis?
I have found that it can only be run from the 'specification' menu after an analysis has been run. However, if I do it this way winsteps will not provide the Cronbach's Alpha figures, which I need to include.
Please let me know if there is a way to specify multiple single items to delete and still have winsteps report Cronbach's Alpha (otherwise it requires the laborious process of creating a new control file).
David, you may want to 1) reboot your computer 2) uninstall Winsteps 3) reboot your computer 4) install Winsteps (be sure to do it in Administrator mode: right click on the install.exe and "run as administrator".)