To choose a statistically valid sample size for any serious research, we have to consider 3 factors 1. Population size 2. Margin of error and 3. Confidence level.
1. Population size would refer to the total number of people who fit into our target segment, this could also be a group with something common. This population can also refer to things, as well as people. • An example, All registered voters in Chennai; All players of IPL; All Indian who bought a home in the past year; All daily maximum/Minimum temperatures in July for major Indian cities. To calculate an appropriate sample size, get an estimate of the population size first - the target group should be well defined.
2. The margin of error (also called confidence interval) is the plus-or-minus figure sometimes reported in the media (newspaper/television) or if someone is making claims or you might have seen it in opinion poll results. A margin of error tells you what percentage points your results will be different from the real population value. • An example, if you use a margin of error of 3 and 65% percent of your sample picks an answer “Option -1-Yes” you can be "sure" that if you had asked the question to the entire relevant population between 62% (65-3) and 68% (65+3) would have picked that answer.
3. The confidence level tells you how confident you can be. It is usually seen as 90%, 95%, 99% and represents how often the true percentage of the population who would pick an answer lies within the margin of error. A 99% confidence level means you can be 99% Sure; the 95% confidence level means you can be 95% sure/certain. 95% confidence level is used by most researchers. ….. When you put the confidence level and the margin of error together, For a 95% margin of error with a 4% margin of error means that your statistic will be within 4% points of the real population value 95% of the time. You can also say that you are 95% sure that the true percentage of the population could be 4% higher or lower for any given response. The wider the margin of error you are willing to accept, the more certain you can be that the whole population answers would be within that range.
Below table will help you choose the ideal sample plan, given you have an estimate of the population size and what your desired confidence level and margin of error is.
Sample size required:
There are many online simple size calculators available as well, but understand the three factors well (Population, Margin of error and Confidence level) is important to derive an appropriate.
The bigger your sample size/base, the more sure/certain you can be that their answers truly reflect the population. This also indicates that for a given confidence level, the larger your sample size, the smaller your margin of error. However, the relationship is not directly correlated, that is doubling of the sample size does not halve the margin of error. Apart from the three factors, sometimes certain types of analysis requires a minimum sample size and many times the researcher would stick with the guidelines for that analysis.
Written by Sunil Mukkath (sunilmukkath@elastictree.com)
A senior researcher at Elastic Tree. Let us know what you think about this article. Submit your queries, comments, insights at info@elastictree.com.
Comments