Sorry you're having troubles with this. CVA should automatically compute and report results across all different questionnaire versions for you. I cannot think right now how/why you'd be getting the reports broken out for you by questionnaire version, unless you selected the version# as a "banner point" within the SMRT market simulator.
If you are doing paper-and-pencil interviewing, then just make sure to put all the responses for all respondents and all questionnaire versions within the same .csv data file. Respondent number is followed by version# and then respondent answers within the .csv file. Then, computation happens all together and all the data are within the same data file in your project.
If that's not the issue, please call our technical support line and see if they can ask the right questions and come to a solution for your issue.