Mass significance – information criterion

#1
In search for the ’best’ linear regression model with 10 potential predictors I calculated the Schwarz Bayesian Criterion (SBC) for every possible combination of predictors (POWER(2;10) = 1024). Do I have to take these 1024 models into account when adjusting my results for false positives? I can’t see that either fitting a line with the least squares method or calculation of the SBC is a statistical test, and therefore I don’t understand why the 1024 models may contribute to mass significance. On the other hand, by considering all models and selecting a few for further analysis I make comparisons and this could be interpreted as ‘testing’.

Thanks for your consideration!

I posted this question on 03-06-2012 under the heading “Does fitting of a regression line as such contribute to mass significance?” in the forum ‘Statistical Research \ Regression Analysis’ but did not get any reply. Maybe I chose the wrong forum or header or I wasn’t sufficiently precise about my problem. Therefore I make a new start here.