Nielson has set the blogosphere abuzz with its annoncement that they are replacing pageviews with time spent as their new default sites popularity metric.
Obviously there are problems with the new metric, perhaps more than with the old one, and of course the key to this debate is that advertisers need to look closely at their goals before selecting a metric to base their media descisions on.
But this discussion raises a more interesting point: statistics can lie. And there is probably no are where they lie more frequently than when marketers (liars) sell their wares (you) to other marketers (moe liars).
Creative statistical analysis is nothing new and web metrics are some of the easiest stats to play fast and loose with. If your site is confusing and hard to use your pageviews may be low but your time on site will probably be high. If you publish crappy articles that people mostly skim and never read, break them up across multiple pages and watch your pageviews skyrocket (time on site will probably stay the same).
Good web analysts are (or should be) masters at coercing data to proove any preconcieved point.

BTW I wrote this on my blackberry so it’ll probably need editing. Sue me.