Oh dear. A Tauranga resident is claiming a Tauranga poll about a \$21 million waterfront museum plan is claiming a poll of 1,000 people is too small. The City Council has refuted this saying even 400 would be valid.

The resident, Ken Evans, is pointing towards his petition-online surveys as a better way to gauge reaction. Umm, no they’re not. A self-selecting sample is meaningless in terms of being representative of the whole population.

Now the size of the population of the area being polled has very little effect on the maximum margin of error of a poll.

A 1,000 person random poll has a margin of error of 3.16% for an unlimited population. For a population of 100,000 like Tauranga the moe is 3.13% – almost the same.

Even a 400 person poll would have an moe of 5.01% for an unlimited population and 4.99% for a population of 100,000.

Only when the population is so small as to be approaching within a magnitude of the sample size, does the population have much effect on the margin of error. The only times one would want a sample of much more than 1,000 is when you want to break the responses down by gender, age or area.