Those response rates are fairly awful with two groups that are markedly different. Seems very likely that they’d self-select on the face of it especially if they knew what the research question was.