By using this site, you agree to our Privacy Policy and our Terms of Use. Close
FightingGameGuy on 09 September 2008
Edit Thread
Grampy said:
I think the sample was random since I took the first ten games listed in alphabetical order and I don't believe a game is affected by alphabetical position.

!0 may or may not be significant given the relatibly small number of reviews available from any one magazine on any one platform. I was almost half way through the aplphabet in some cases but I doubt seriously that a larger sample is going to change things when 90% of the reviews are high. I do a lot of scientific research and while we of course use larger samples, patterns that show up this distinctly at ten random samples seldom cahnge that dramatically.

It was enough of a sample to put IGN within .4 points of perfect average, significantly raising my faith in the reviewers at IGN.

I also was impressed with how fair, even tough the fanboy publications were in comparison to Game Informer. That was a major surprise. My kudo's to their professionalism and balamce.

If I can find the time I will do a more in detail analysis but I'm willing to bet Grandma's virginity that Game Informer won't improve significantly. I do judge Game Informer differently because they are owned and supplied by a huge retailer that uses them as an inducement to get customer loyalty. I think they carry therefore a greater burden to be fair than a completely independant source and certainly a Fanboy magazine; and yet they seem to under perform both.

And what's the beef with the signature exactly? I spent several hours doing it. It was all meant in good fun but if it offends, I will certainly remove it. I've had it for several weeks and no one has complained before.

The issue is that your doing pseudo-statistics, not real statistics, which deprives your results the theoretically proven descriptive capabilities of actual statistics. 

You haven't "confirmed" anything because all your tests are statistically invalid. 

Heck even calling them tests (all you've done is display different trends and never analyzed if the difference between the trends is mathematically significant) isn't correct.