You go with the data that you have, not the data you wish you could get. It would be nice to redo the study with a large population that is randomly selected.

You go with the data that you have, not the data you wish you could get. It would be nice to redo the study with a large population that is randomly selected. However there is also no reason to believe that the results will be significantly different in direction. Therefore it would also be unwise to discount this result.