If data is not representative it's just a metric. Its worse than anecdotes because an anecdote has a lot of information and nuance that can be obtained from it.
You can't just say "the data COULD be unrepresentative", then just use some random guys anecdotes on the internet as more valid than national statistics.