Hi all,
I'm currently working through a couple of questionnaires (one is a likert-scale based questionnaire - Strongly Agree to Strongly Disagree), and the other is multiple choice (in which there is only one right answer).
I'm just wondering what methods researchers might recommend for setting cut scores on both formats.
Right now I'm exploring ROC curves using SPSS, but any guidance/expertise would be greatly appreciated!
Darko
Hi Darko Giacomini,
Here is a pretty easy way to obtain cut-off scores:
You need to have a great amount of participants. Following the normal distribution law, you can calculate cut-off scores with the Z formula : +/-1,96 represents a p value of 0.05. Then, you can find your X (i.e., the value of your cut-off scores) like this: X = General Mean +/- 1,96 x Standard Deviation.
Note that the cut-off scores you will obtain with this method are not really "correct", because it will depend on your sample. Following this, your sample needs to be the most representative of the general population.
I hope that helps and I would be glad to hear other researchers' opinion.
Best regards,
Julian
Thank you Julian! This has been a great help.
Just one follow-up question: How would you determine whether to add/subtract 1.96 to the general mean?
Thanks again!
You will find these values in the Z scores transformation table (I think). I gave you the different critical values (in the picture added in my previous message): +/-1,65 corresponds to p = 0.10; +/-1,96 corresponds to p = 0.05; and +/-2,58 corresponds to p = 0.01.
1,96 is classically used but you can also use 2,58. However, take also in consideration that following the normal distribution law, if you determine to calculate your cut-off scores with a p value of 0.05 (so, 1,96), it means that you can expect to find in the general population 2,5% of people that would present higher or lower scores than your cut-off scores. If you determine your p value at 0.01, you can expect to find in the general population 0,5% of people that would present higher or lower scores than your cut-off scores.
You're welcome!
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS