I read two of your papers describing the generalization of the Ferguson's formula for tests with multiple response options. The rationale and methodology of computing delta is amply clear. I find this method very useful for those interested in psychometric evaluation of tests/scales in general and clinical tools in particular. However, the manual calculation even through electronic spreadsheets may be time consuming. Is there any SPSS macro, syntax or Excel macro or any other software (other than R codes as I am not well versed with R) for computing delta?
Article How discriminating are discriminative instruments?
Dear Rakesh, the ML estimation of Delta needs numeric resolution of nonlinear equations, therefore we built the application 'Delta' (for Windows plattform). Nevertheless, as we have received more suggestions in this way, we will update the application for the use of raw data with the capability of reading Excel files.
Delta is available for downloading in http://www.ugr.es/~bioest/Delta
Best regards
I have attached an Excel spreadsheet to calculate Ferguson's Delta for multiple response options. The instructions are (I think) quite straightforward - just generate the frequency table in SPSS, then copy the frequencies into column A, update the number of items, and the number of responses.
Thanks Dr. Hankins for providing the Excel spreadsheet to calculate Ferguson's Delta.
Dr. Femia, I am also thankful to you for providing the link to your software 'Delta'. Is this software calculate Ferguson's Delta? From the description available on the webpage it appears that this software helps to identify random response pattern or guessing on multiple choice tests with one correct answer and not for multiple point scales where there is no correct or incorrect response. Further, it can also be used to compute inter-rater agreement.
I find this software very useful for the said purpose, though, my interest at present was in computing the Ferguson's Delta to determine the extent to which the test discriminate among the test takers in terms of the construct measured by it.
Thanks to all for the valuable support and help.
Dear Dr. Hankins I calculated the delta for a 12 item questionnaire with 7-point rating scale with sample size 420. When I pasted the frequencies for first item the n was correctly displayed (i.e., 420) but as went on pasting the frequencies of next items the n was changing with a factor of 420. Is it correct or have I made some error in pasting the frequencies. My idea was that the n will remain constant i.e., 420.
For your convenience I am attaching the Excel sheet with calculations.
Stata module to calculate delta from http://ideas.repec.org/c/boc/bocode/s456851.html
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software