I have a very simple question regarding the methodology section. My question is that for example, suppose we have collected 687 questionnaire from the respondents. Among those some of the respondents demographics information are missing. So shall I consider those questionnaire missing in providing personal information for analysis? Even though our research objective has nothing to do with demographics
Thanks all in advance
If the research question has nothing to do with those demographics, you could include them. You might also do the analysis on the entire data set and the subset which has the full demographics available, and see if the additional data points make a difference.
Hi Muhammad,
it is not exactly clear what you mean
a) What percentage of gender and age is missing?
b) do you want to exclude the entire case that is missing?
Eliminating complete cases is never recommended. If for instance gender is missing, reporting %females/males may be misleading IF there is a systmatic dropout in gender (this happens when respondents of a specific gender had not participated in the study). This is non-random missing. If the dropout is non-systemtic, the percentages will be correct. This is missing completely at random (MCAR).
If gender is an important control variable, include it and apply either full information maximum likelihood (e.g., in SEM) or multiple importation. This avoids elmininitating that which at best results in low efficiency/power or worse, systematic bias.
Best,
Holger
the missing data must included in data analysis and it can take percentage form total number of data collection
- Badges
- Science method
- Similar topics
- Methodology