[QUOTE="mjarantilla"][QUOTE="bobbo"] People in this forum go on and on about how you can't trust vgchartz numbers. Statements like "they just make them up" ,"they can't be trusted" ,"NPD is God" are all over the place.
Ioi, the founder of vgchartz have taken the time to explain how they get their numbers and also share abit of info about their future plans.
Good read that i highly recommend all you nai sayers and doubthers to read thoroughly. Anyone curious about vgchartz and what they do should read this aswell.
http://www.vgchartz.com/news/news.php?id=629
Enjoy and discuss.
P.s : Comments like "they stink" or simular remarks are just a waste of time. That's how teens and children argue. Adults or reflected mature teens have learned to use arguments... Hint hint..... ;-)
jeffwulf
"Anyone with a background in statistical analysis will know that you don't need a large sample size to get good results, merely a good sample."
Bull****. You need BOTH. That's why NPD has a margin of error of less than 1%: they use both a large sample size and a good sample. And you know what? Anyone with a background in statistical analysis will also know that a margin of error of 15-20% is horrible. This guy is talking out of his ass.
He tries to downplay the fact that their statistics are so terrible, and fails miserably. He basically admits to being an effing pretender and pathetic conman when he says that their margin of error is 15-20%. I mean, consider this: if VGChartz says that the Xbox 360 sold 300,000 units in November, but admitting that their margin of error is 20%, then what that REALLY means is that their ACTUAL estimate for the 360's sales are between 240,000 and 360,000.
VGChartz is trash.
Take Statistics. You might learn something. Like how you don't need a large sample to get accurate data. Polls are regularly done by companies that have only a couple thousand people that represent the entire US with only a standard devition of a single percent or less.
Think it was Mark Twain that once said. There is 3 forms of lies : lies , white lies (lieing with good intentions and statistics.)
The results of the statistics is only as good as the questions and peolpe asked. It depends entirely on the sources and where you get the answers.
So yes if you ask 1000 random people about a simple question like voting republican or democrat , you will get a very accurate statistic.
However if you want to find out how many heavanly swordgames have been sold.You will get different answers depending on the sources used.
Store type and customer base for each type of store comes into play. Units shipped totally and how manywere recieved in the stores you track. Did they get enough? did they sell out? could they easily have sold more if given more?. Is the online stores covered?
Do you cover all retail types ? Know how many copies each store got etc etc.
Tracking software sales is difficult. You will never be 100% correct ,but useing the right sources and estimations you can get close. Tracking hardware is easier, but still much more complicated than tracking who likes bmw's better than audi's.
Basically because these are complex merchandice to track... Again is all retail types and therefore customer demographics covered? How many units was shipped to the market?How many units did the tracked stores get? Did they sell out before the tracked sales period was over etc etc.
So again what i'm trying to say is. Vgchartz is good enough for us gamers to use to check salestrends. Not exact numbers,but they are close enough.
Is vgchartz less accurate than NPD. I don't know and i dare say , neither do you....
Log in to comment