binary solo said:
I think that's a little over simplified. You're assuming NPD has no margin for error when it does, because they are still estimating totals based on a sample (probably a wider sample than VGC, but still a sample). So I'd say NPD could have a MoE of maybe 5%. If you factor in both VGC And NPD MoE you might find that the only console that's possibly outside the MoE is 360. Wii might be inside MoE because of the higher absolute number of sales, but perhaps Wii might also be outside MoE; I haven't done all the maths. Taking NPD MoE and VGC MoE into account I think PS3 is pretty much fine along with DS and PSP. You can basically ignore NPD lower margin for error (and VGC upper) when NPDNPD for the purposes of this discussion because with 2 sets of sampling data the true sales number of each console is most likely to be between the 2 estimates, not outside the 2 estimates. VGC also has an additional MoE on the basis that it estimates USA only sales as 83% of total Americas when that % will fluctuate month to month. So the true 360 sales it somewhere between 185K and 233K, leaning closer to 185K. However the matter to consider for VGC adjustments is whether the true sales number is likely to be 10% (arbitrary MoE) less than VGC. How likely is true 360 sales to be <210K? Look at the MoE for NPD (est 5%). NPD MoE suggests true sales for 360 could be up to 203K. That's a MoE gap of 7K, making it reasonably likely that VGC 360 numbers are slightly overtracked, but not in a horrendous game changing sort of way. If the MoE's overlap then it means VGC's estimate is OK provided people know they are an estimate. Hence the statistical tie concept that people like to make fun of. So April NPD 360 & PS3 are statistically tied. |
I explained some of it already. See my previous answers for some clarification.
