I mentioned here that I would compare my post on online spending trends to Rob Leathern's and see why we came to different conclusions. The answer, unsurprisingly to anyone who's done any work with this type of industry data, is that we use different numbers.
I used TNS for ad spend data. Rob used both TNS and IAB numbers. TNS and the IAB have wildly different estimates for online ad spending. For example, US online spend in billions:
TNS has consistently had smaller numbers for online spend than the IAB. I'm not going to pass judgement on whose numbers are better: I do not know. I merely noted the difference and decided that if I was going to compare online to other media, I would use spend numbers from a single source. This, I think, gives a better comparison, if not necessarily a better absolute number.