Hi..
I want to use "Narcissm" variable. There are standard instrument for this variables consist of 40 statements (A and B option). My question is, can we reduce the question to 5 or 10 question, so my respondent (busy people) have time and still focus to fill all my variables instrument? If it can, how?
Hello I Gede,
You may be able to shorten an existing measure by dropping items, but doing so will most likely affect both the reliability of the resultant scores (lowering, typically) and validity of the scores for your application.
If item-level correlations for the 40-item version are available, you could evaluate:
(a) the factor structure of the measure, with the notion that any items not loading on a unitary factor are not salient for your construct of narcissism;
(b) a successive dropping of items, starting with those having lowest item-total correlation (so as minimize reducing internal consistency reliability of scores);
(c) choosing only items that have strongest correlation with some criterion variable (e.g., score on another, verified measure of narcissism);
Note that doing either requires having data from a sample (relevant to your target population), responding to the original, full-length version of the measure. For purposes of noting whether any deleterious impact on score validity occurs as a result of method (a), (b), or (c), scores on some external criterion variable would be very helpful.
Good luck with your work!
Sounds like your aim is to focus on shorter unidimensional scale in which case your concern if saturation on a primary factor.
Hope this helps,
Matt
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress