Learn how you can use statistical sampling for greater insight into your data—quickly and cost-effectively. Read about the proven tools, techniques and tips you can use to better manage litigation costs and drive more efficient strategies.
The Statistical Sampling feature in LexisNexis® Early Data Analyzer 1.6 provides a powerful new tool for forecasting review effort and quality control of searches and filters on top of the existing import, culling and filtering workflow.
By reviewing only a couple of hundred documents via the Web interface of LexisNexis® Early Data Analyzer, the number of responsive documents within a particular set can be determined with high statistical accuracy. This analysis can be performed on specific keyword searches to determine the accuracy of the search terms, or on the entire set of filtered or unfiltered documents within the case to fine-tune filtering.
Statistical sampling is “the selection of a subset of individuals from within a statistical population to estimate characteristics of the whole population.”1 Most simply put, choose a subset of the entire population, make an observation and then estimate how many of the total population have that same characteristic.
A common example is a presidential election. Polling organizations like Gallup® do not need to ask every person in the United States who they’re going to vote for. Instead, they randomly select a portion of the population and project the results of the sample to the entire population.
Typically, organizations like Gallup only poll about 1,000 citizens, which is about 95% (confidence level) ± 4% (confidence interval). Confidence level and interval appear to be really complex, but in the simplest of terms: if you conducted the test 100 times, 95 of those samples will yield a result within our confidence interval of ± 4%.