This is going to be vague but I recall a news item on Radio 4 where someone was studying the disparity between actual crime rates and perception/fear of crime. The hypothesis was that that crime in the UK had become more visible but highly localised to particular places and times - i.e. nearly everyone was aware of centre of town binge-drinking and resulting high level of associated crime on Friday and Saturday nights even though those areas were at all other times remarkably peaceful. This led to a higher perception of crime because, I guess (though I don't think that was explicitly stated), people extrapolated out from the Friday/Saturday night problem to a perception that the crime rate was constant across the week. I'm afraid I can't remember any names or publication venues for the research...