Thalidamide was also before modern biostatistical techniques for evaluating and approving drugs, but yes, there have been high profile screwups since then too. Might be fun subjecting clinical trials higher certainty standards but... I think the industry would buckle under the financial weight. Maybe that wouldn't be so bad though?
Anyway, we're not in (complete) disagreement. It's absolutely valid to say that if these 2200 households were sampled from the US, they are not representative of Mexico, nor China, nor Wyoming. The first two because they were not a part of the population sampled, the last because it's a stratum within the population. But it absolutely would be representative of the US as a whole --
provided proper sampling techniques were used, the data collected was of good quality, etc. etc. There are about 100 disclaimers that we really ought to print on any statistics like these, but most people would fall asleep after the first one. Regardless, this would indeed
not say a damn thing about San Fran or the Amish, but
only about the group proportion for the nation as whole.
And of course using the US to extrapolate to the whole world is utter folly. If someone could manage the logistics of sampling 2200 households from the whole world (hint: not. financially. possible.) and looking at the proportions, I'd be comfortable with those saying exactly that -- the rate of participation in gaming among the global population, which
again would
not say anything at all about any given stratum -- first-world, third-world, northern, southern, temperate, equatorial, much less the constituent countries -- only that the sum of the parts has such a participation rate. Remember, statistical theory is derived based on finite samples and infinite population sizes -- 2200 of 300m vs 2200 of 7b doesn't make an ounce of difference for the validity of the point estimate nor its confidence interval.
This is also why I hate any time an article uses statistics -- they never explain enough about the methods, nor do they caution the readers about interpretation
s.