Well this is a lazy one but I think it’s probably worth saying …
Details #
Just as n is the usual abbreviation for the number of participants in a study (or substudy, or contributing to dataset or just an intersection cell of a study!) so k is the usual abbreviation for the number of items in a multi-item measure. Why does k matter? Because the more items you have in a measure, the more aspects of what you want to measure you can tap and, generally, the more you can separate what you want to measure from “noise” or “error”. That’s the same as saying the better your internal reliability.
In the 21st Century there has been a huge shortening of the length of measures. I have huge sympathy with getting away from the mid-20th Century tendency to use measures with many items such as the 90 item SCL (Symptom Check List) or the many measures that were over 100 items and sometimes up into many hundreds of items, for example, the MMPI-2 (Minnesota Multiphasic Personality Inventory) has 567 items. However, I have real reservations about using measures with k < 10 but that’s too big a story to fit here.
Try also #
- Cronbach’s alpha
- Internal reliability
- Macdonald’s omega
- Reliability
Chapters #
Not covered specifically in the OMbook.
Online resources #
None currently.
Dates #
First created 19.iv.25.
