Yes, the best way is to create two separate files for parents and students for a perfect match in the final data sets. Generate unique id for each household so that they can be identified with no duplication in the sample. For instance, for a parent from a household no. 1 in a town/village no. 1 in a district no. 1 in a state no. 1, the unique id will be 1111 (state, district, town/village, household, parent) in the parent file. For the same household the student id will also be 1111 in the student file, in case there is only one student interviewed from that household belonging to the parent id 1111. If you have two students from the same parents living in the same household, the second student id will be 1112. But notice that the household id for both the students will remain same. Use this id as key to match the each student and parent to create your final data set for analyses. Think about another way of identifying households, parents and students if the parents have students living in two separate households. I am not aware if you have such respondents in your sample.

If you just want to investigate the differences in opinion between parents and students, another way would be to use a single database and create a binary variable of the role, parent/student (coded as 1 or 2). You can then fill in the database by inputting all the questionnaires and then run hypotheses tests to check whether their answers are statistically different or related according to their role (parent or student). for example independent samples t-test( or the equivalent non-parametric test) between the binary variable of their role and their opinion on various questions.

Yes, using the final single data set one can create many such required variables and conduct various tests of hypothesis. But before arriving at the single final file, two separate files for parents and students, respectively, may help the researcher always in checking if there is anything wrong with the data inputs.

Thank you for the suggestions. Yes, parent and student survey responses can be done using a binary variable. But however will SPSS read certain variables as missing if I put it in for one and not the other? I have demographic data that I received from parents. Is there a way to let SPSS know that certain information was only taken from parents but not for students so it doesn't code it as missing?

I currently have two separate excel spreadsheets for parent and student data information that I am using to create the SPSS data set.

Thank you again.