I'm planning to analyze sleep quality using Pittsburgh Sleep Questionnaire Index (PSQI) score and assessing insomnia severity using Insomnia Severity Index (ISI).
I've seen at least three dozen publications using mean and SD to describe both. The question is: - is it actually possible to describe PSQI and ISI in a population using mean and SD?
For what I mean, lets take example from NIH Stroke Scale for stroke;
NIHSS = 4 means "mild stroke", but it's not only can be interpreted as "alert patient with partial gaze palsy with drifting left arm and left leg" but also "alert patient with limb ataxia and gaze palsy". Does this apply for PSQI and ISI?
Anyone can help?
you want to say normal range or severity of insomnia from PSQI and ISI score if am not wrong?
Yes. Because they both yield a continuous total score you can calculate and report mean and SD. In your example of the stroke scale, the outcome is ordinal rather than continuous.
Maitrey shaileshbhai Pandya well, not really. But if there are any studies which convert those two scoring systems into binomial outcome, please let me know.
Susan R Wilbanks thank you! this is actually interesting, because many studies are using mean for such ordinal system (NIHSS, UPDRS, CHAD-Vasc, sICH, and the list goes on). I wonder if there is actually a way to represent those ordinal numbers without relying on nominal re-grouping or falling into non-parametric tests.
Correct. When you use the basic assumptions of statistics you cannot technically calculate mean and SD from ordinal variables, but in practice, composite scores from those measures are used as continuous variables.
- Badges
- Science topic