Digital trace data from social media provide large amounts of information on individuals, their behavior, and their interactions with each other. Social media data have been employed to study personality, social networks, and other phenomena. However, employing social media data for research causes some issues: for example, data have to be transformed to fit analytical methods, and data may have been shaped by the social media information systems through which they were produced. In turn, the ways in which these issues are accounted for significantly affects research results. This study contributes to the methods used to analyze social media data by proposing a method to compute frequency measures on users' preferences (formally comparable to survey items) from answers to multiple-choice questions in online reviews that are repeatedly given by users over time. I evaluate the method by computing travel motivations from online travel reviews and comparing my results to findings on travel motivations obtained through classic surveys. Since both results are very similar, I conclude that my approach is appropriate and should be tested for other domains and datasets. I discuss the limitations of the method and the evaluation and these issues can be alleviated in further research.