Resources

Research and Statistical Tips

Choosing Random Samples, Part 2
The most common question about random samples for pastoral planning is "how many do I need?" This question is considerably more difficult than one might think because of the formula to calculate sample size for a finite population. Not only is the formula complex but the results are somewhat counterintuitive. For example, let's presume that we are surveying each household in a 3000 household parish. For a certain level of precision 1017 households need to be randomly selected -- about 33% of the parish. Now what if we are surveying a parish of 300 households: how many households are needed? One might be tempted to guess about 33% or 100 households. But the formula gives a result of 251 households, or about 84%.Fortunately the Hartford Institute for Religion Research has provided practitioners with some good guidelines that avoid formulas. They recommend that parishes of less than 300 households select everyone, those with 300 to 350 should randomly select 75% of parishioners, those with 350 to 700 should select 50%, those with 700 to 1000 should select 33%, and those with more than 1000 households should randomly select 25% of parishioners. These guidelines specify parish size in households when one is surveying the household, i.e. one questionnaire per household. When one is interested in surveying individual parishioners, the same guidelines can be used but substitute the word "parishioners" for "households." Next: Choosing Random Samples, Part 3: How to select the sample. Archived Tips