It's not really a variable that can be isolated in any effective way. If you did a massive test across 1000s of sites you might be able to extract some useful data, but even then I think the psychological effect varies quite a bit depending on what kind of site it is to begin with.