This situation that requires an imputation method that you must pick from. Dealing with missing data is a sensitive matter. Depending on which way you choice to handle missing observation will impact your project and in this case (with observations less than 100) could make your analysis poor/basis.
Here are some popular options:
- Impute the value of the missing data
- Remove a variable which has a lot of missing data and use other variables which measure similar aspects of the characteristics being studied.
- You could create a regression model to impute missing values.
- You can use the average (caution when using this method as there may be other variables in your data set that may be dependent).
- Duplicate the previous or nearest value.
Here is a website on some tips: http://www.real-statistics.com/descriptive-statistics/missing-data/
Here is a working paper on the subject: liberalarts.utexas.edu/prc/_files/cs/Missing-Data.pdf
Missing values is delicate subject. Use caution and always, always, always DOCUMENT WHAT YOU ARE DOING, because others want to know and it makes your work so much better.