1. **"Random"** means each individual element of the population being
surveyed has an equal chance of being chosen. It does not have the same
meaning as "arbitrary", in fact, just the opposite. A random sample must be
drawn sytematically, rather than by convenience.

2. The needed **sample size** is not related to the size of the population.
This is counterintuitive, but easier to accept if you consider that national
polls Gallop poles, and exit poles, for example, a very small portion of the
population. Sample size should be calculated using a formula which takes into
account the desired precision ("confidence interval") and confidence
"p value"), as well as the variability within the population.

3. A **cluster survey** is less accurate than a systematic random sample.
It's for that reason that the formula for cluster sample size often doubles
the sample size (a 2 for "design effect") to compensate for the shortcoming.
A cluster sample, however, is often the only feasible sample design, when the
population is quite large, or when there is no complete list
("sampling frame")of the population to be surveyed.

4. Survey results must report both **confidence, and precision**. Without
that information, it is impossible to say how close the estimates may be.
When estimating proportions, the confidence interval (precision)is narrower
when a proportion is vary high, or very low; and wider as proportions approach
50%.