I am running some content analysis studies on my dataset which has two different classes, and each class has a respective list of the document I am analyzing. I compare the LDA topic model inference and some keyword analysis on the two classes from the documents. There is enough document list on each of the class to be analyzed, but one has relatively larger documents than the other. Will my results on this case be debatable, because of the imbalance of data. But, in a content-based analysis, or LDA study where the sample size of the document is not a variable, will it matter?