Seems obvious that larger sample size will give more close results to what happens in realty. Cool, if it's easy to select large elements to represent the sample, then we should do. Usually a sample of size 30 to 40 elements is great to have.
We should increase the sample size whenever we:
- Can easily handle/select sample items (randomly).
- Perform the measurement with little effort/time consumption.
Notice these are highly dependent on the problem or population nature and situations. Another point to keep is to have accurate measurements for the sample as possible.
In conclusion, two important factors should be regarded:
- Good sample size as large as possible. This guarantees more information diversity is included in our sample. Thus it's more close to describe the whole population facts.
- Using accurate measuring tools/equipment when measuring the required property of the sample because these measurements will judge next steps. This requirement becomes more critical in small sample sizes.