I aim to assess several correlates of the SPQ in normal controls. Would you set a cut-off value at the lower end of the continuum of SPQ values and how would you set this value (percentile / certain amount of participants scoring the lowest)? It seems to me that it is common practice to use all values, is this true? However, could this practice entail problems?
Hi Lisa,
I don't have any familiarity with the SPQ specifically, but as my background is with mental health screening tools, I'll raise some general questions/cautions that may help you. Using the continuous score is likely the best approach. You lose power when you categorize continuous measures, although this is common practice given that clinically measures tend to be categorized (and if you have strong clinical implications for your paper, then categorizing might be worthwhile to make your paper more relevant).
If you plan to categorize your scores, you need to decide why you are using SPQ scores. Are you using the score an important measure in and of itself, or are you using it because your interest is really in schizotypal PD. If it's the score itself, then any decision could be justified, but you will need to explain why you chose a certain percentile or number of participants.
If, as I suspect, it's because you're most interested in the PD, then you need to be sure the cut-off is valid (in which case percentiles or choosing a certain number scoring the lowest is likely a bad idea, and you should be looking for a cut-off that has high specificity since you are interested in healthy controls). If you can find studies which present sensitivity and specificity of SPQ cut-offs against a diagnosis of Schizotypal PD this would help steer you in the right direction.
Hope this helps.
Michael
Dear Michael,
thank you so much for your contribution. Indeed, I'm rather interested in the PD than the score itself. However, I merely find a paper concerning sensitivity and specificity in the SPQ- brief (which I am not using) in a chinese sample ..
Hi Lisa,
I am unsure how to give a good answer, because in your first post you state that you are looking at schizotypic correlates in "normal controls" and in the second that you are interested in SPD. Furthermore; is schizotypy your IV or your DV?
In the healthy population, I would agree with Michael and try and avoid using cutoffs, unless you specifically want/have to do a group-wise comparison with schizotypy as your IV (e.g., high vs. low schizotypy). In this case I would suggest either dichotomizing or trichotomizing your participants.
If you are, however, interested in SPD, I would use a diagnostic tool for SPD. The SPQ may be used as a screening tool for SPD, but it is not intended for diagnosing SPD.
Hope this helps,
Phillip
Dear Phillip,
Thank you, of course you are right and I got something wrong in my second post. I assess SP in healthy controls (SPQ and german version of the O-LIFE)!
I aim to assess the correlation between the SPQ/O-LIFE scores and other measures, so I must not do a group-wise comparison. I agree with your and Michael that avoiding cutoffs may be the best solution, I just wanted some further input and ideas about this before I start!
best,
Lisa
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India