Hi, I have a dataset from a questionnaire with 498 observations. There is item nonresponse present, where some variables have 56% missing values. Should I use multiple imputation when there is as many as 56% missing, or should I just use the non-missing values? If I get rid of the missing values, I am only left with around 230 respondants, which is substantially small. Therefore I would like to avoid this.

I also have a question regarding the use of scale in surveys and how I can solve this in Stata. For instance, I have a variable named "employees" that I want to construct based on two items where the respodants were asked to rate - on a scale from -10 to 10, where -6 to -10 indicates disagree, 0 to -5 indicated neutral, 1 to 5 indicates agree and 6-10 is irrelevant - how they make use of the excess capacity of their employees in the firm. I want to change the scale from 1-21 and sum the two items from the survey where respondants were asked to evaluate: 1) if they use the excess capacity to train employees and 2) if they use the excess capacity to solve problems that they normally didn`t have time to do under normal circumstances. When I sum these two items, I have a scale ranging from 2-42. How can I code this in Stata?

Thank you in advance.
Best, Danijela.