The jobless survey is taken from 60,000 households. It only takes 600 households to report a job for a resulting 1% decrease in the statistics. A group of 100k people could inform their members if surveyed to respond one way or another and it would be enough for a 1% swing. We need a more accurate measure.