The JP seems to have no love for consoles anymore. It's portable or nothing.

That explains why japanese hardware sales are rising... Oh wait, this study is completely invalid.

The Japanese game market is expanding, albeit at a rate slower than expansion in he UK or Americas.

2007-2008 was the only year in which sales dropped, because 2007 was an incredibly high year for japan

They surveryed 1000 people

It looks like the users on this thread don't understand statistical inference.

In most statistical polls 1000 people is the standard amount of people to do it on. Fewer means the results won't be as accurate, more takes more time and money.

Sorry, do you mind if I explain a thing called 'the Scientific method' to you.

We have 3 seperate experiments, which all show that that the japanese industry is growing. They all agree with an assumption of about 10% uncertainty. These experiments are called 'VGChartz numbers, Famitsu numbers and Media Create Numbers.

We have a 4th experiment, which tries to present the exact opposite of what these three are saying. however, it is a much less convincing experiment, as it involves much smaller sample sizes, and it involves asking people if they are gamers, rather than infering this from the number of games they buy.

Clearly, 3 good experiments say one thing, and 1 bad experiment says the opposite.

it is our job as rational minded people to try to explain what the people who ran this survey did wrong.

I suggest as one thing, they took it over too short a time scale for it to be scientifically valid

Secondly, 1000 people is not enough, 10 000 would be a much more sensible number, but even then the uncertainties would be huge.

Anyone else got any ideas as to how the survey could have been improved?

I do mind as I was commenting on the 1000 people thing, not the article as a whole or the results, this is normal practice for 'poll' takers, 1000 people is standard number for these types of thing.

But whether this data is correct or even accurate is a different matter entirely which I wasn't commenting on.

Not one of those "experiments" is taking part in the scientific method. They could engage in the scientific method if they chose to, but it's alot more comprehensive than that, and your analysis would certainly not be sufficient. This poll has little to do with the scientific method, it's a simple poll with all the limitations polls usually have.

The alpha level of this poll would actually be better than the gallup polls and all the other polls in the US presidential election. In fact they could have chosen a smaller sample and still had a reasonable alpha level.