actually it's will be statistically significant (you can already see the bimodal distribution). The problem is with sampling bias, and the fact that the data is a self report measure (unknown reliability). Could adjust using recruiter supplied stats in an attempt to correct the self report. The self selection however is insurmountable. Personally I find these things interesting because I assume the data is at least somewhat reliable for the sample that choose to reply to the poll.

Looks about right to me. Thing is with stats you want to tell a story and ideally create testable hypothesis based on it.It might be a wrong story but mouthing off "Correlation does not imply causation or the plural of anecdote is not data"(statistical tourettes anyone?) adds no useful insights and qualifies as not even wrong(adds no new information and is intrinsically untestable) . We see price stickiness around the £50 mark with some very high paying jobs for some clients with deep pockets who aren't so price sensitive (London finance most probably),even if we assume half of the +100 people are lying it still represents an interesting view of the market.

Im in the 50 an hour band myself,would be interested in scaling up a bit.