Use the mean score to replace the missing data to handle the issue. It will least affect the results

Hi Megan,

so you have N = 118 and for 2 of the 10 variables were not answered by 15 respondents?

At any rate, of you want to do a regression analysis, go for multiple imputation which is implemented even in SPSS.

Alternatively, run the regression in lavaan/R and use the FIML (full information maximum likelihood) estimator. The approaches assume that the causes of the missings are not correlated with your dependent variables.

Best,

--Holger

I think it is not missing data but it is incomplete data collection. I suggest collect the data; if harder then use conventional means like a phone call, email, fax, or online questionnaire. your data is not missing at random which could be imputed by some statistical method.

However, there are methods for the missing data called "Nonresponse" but that would be harder to help in your case.

Three possibilities:

(1) Just ignore the missing values (i.e., delete the missing observations). This means that you assume that the pattern leading the missing is random. Maybe it is not a bad guess. If you know that there were not collected by any particular reason, but just a mistake or so, it could be a good strategy.

(2) With some of the other variables carry out your analysis and repeat it excluding these individuals. Try to see if your results with the rest of variables are sensitive to the exclusion of those individuals with missing values. If not, just ignore the missing.

(3) In the case your results are sensitive or you are convinced that the missing values respond to a non-random process, try an imputation using the rest of the variables. I would not use, at least in principle, multiple imputation from the beginning. You always have time to experiment with it later.

Thank you all!