You are right: searching for deviant cases and amending your analysis as necessary is one remedy for anecdotalism. Analysing sequences of interview talk helps – don’t just pull a line or two out of an interview because the positioning of talk is very important.
More generally, I worry that treating what interviewees say as a transparent guide to their perceptions and experiences is commonsensical. Discourse and conversation analysis as well as narrative analysis offer sounder, theory-grounded approaches.
Hope this helps.