Raosoft Calculator for TAR
When preparing to conduct Technology Assisted Review or predictive coding, use the Raosoft calculator in order to determine how large your sample size should be. See the calculator on this site: http://www.raosoft.com/samplesize.html . The calculator requires that you enter a margin of error, a confidence level, a population size, and a response distribution. The confidence level is the indication of how certain the confidence internal is - how likely it is that the value that is returned by the calculator falls within the confidence interval. So if your estimate is between 40 - 80,000 documents, the confidence level would be how certain you are that the true required sample size falls within that range. The margin of error is the percentage that the actual value deviates from the best estimate. So if the best estimate is 60,000, and the actual value is 55,000, you have a margin of error of about 8%. You set the response distribution to 50% because we can't have any advance knowledge of whether or not the total ESI population was adjusted to be more responsive or not. The Raosoft calculator is recommended by Catalyst in its TAR for Smart People guide previously discussed on this blog. The Raosoft calculator states that results don't vary significantly for population sizes greater than 20,000.
"If 50% of all the people in a population of 20000 people drink coffee in the morning, and if you were repeat the survey of 377 people ("Did you drink coffee this morning?") many times, then 95% of the time, your survey would find that between 45% and 55% of the people in your sample answered 'Yes'."