2. if I have 200 respondent for a questionnaire of 25 items included in 4 domains ,on scale score from 0 to 5, so the lowest response will be 0 and the highest will be 125.which scientific way should I follow to make categorizations and why for example into three level (low, moderate or high)or just (weak or strong) or (very low ,low ,moderate, high, very high).
take resilience questionnaire for example highest will be 125
so can we say any score from 0 to 42 he has low level of resilience, from 43 to 84he has medium level of resilience, from 84 to 125 he has low level of resilience
3. that absolutely will be useless if there is categorization in the original questionnaire or not?
Other than in very specific circumstances, you would typically not want to categorize your summary (composite, scale) score for various reasons (most importantly, loss of information about inter-individual differences/between-person variability through categorization). Having continuous (or quasi-continuous) variables is better than ordinal measurement. Therefore, you would typically want to use the scale score with its full range of values. Why would you want to categorize your score?
Christian Geiser Why would I want to categorize my score?
To divide the results of the participants in the research and determine their levels , for example, high, medium or weak skills based on their response and total score they have in the questionnaire.
I am sure it is a question in the basics of research, but I need indications on the best way to determine the levels of participants to match the correct scientific research methods
thanks a lot for your response
Again, from a statistical perspective, it is best not to categorize/form groups and to rather use the variable/score as is, taking advantage of its (quasi-) continuous (metrical) scale level. Otherwise, you lose information. There are plenty of statistical procedures available for continuous variables and those are typically more powerful than procedures for categorical/ordinal variables.
Most categorization methods are also rather ad hoc/arbitrary (e.g., median or quartile split). A model-based approach that is less arbitrary is latent profile analysis (LPA), which allows you to identify latent groups based on continuous variables. Often, these groups come out as low, medium, high. However, LPA requires multiple continuous indicator variables.
- Badges
- Science topic
- Similar topics
- Psychology