世論 What Japan Thinks

The one complaint I most often hear about polls is from people who, because they weren’t asked their own opinion, believe the poll to be fatally flawed. However, statistics tells us that small sample sizes can give reasonably accurate results even for very large populations; therefore, the usual sample size of 1,000 or 2,000 for predicting the opinion of the whole country of Japan is sufficient, assuming the 2,000 are picked in a reasonably random fashion.

The second most common complaint is that internet polls are fatally flawed: many of the polls I translate here are carried out “on the internet”, but the most common type of internet poll I translate are the ones from closed monitor groups: these ads you see (probably in this story too!) for “Get Paid to Take Surveys” are from companies inviting you to join monitor groups, which you can usually do for free anyway, so please don’t part with any money in reply to these ads!

However, open internet surveys are usually hopelessly inaccurate, but I try to avoid these whereever possible, or at least make clear to the reader that there is a high degree of probability that the figures are totally inaccurate.