You send a small list to two programs and receive different quartiles. Before searching for a corrupted file, check the percentile method each program was asked to use.
The NIST guide to percentiles describes several interpolation methods used in practice. When the requested position falls between observations, the rule for locating and interpolating that position can change the answer.
Take the invented sorted list 10, 20, 30, 40. For the 25th percentile, a position rule p(N+1) gives 1.25, producing 12.5 by linear interpolation. The rule 1+p(N−1) gives 1.75, producing 17.5. Here p is 0.25 and N is 4; positions are counted from one.
Record the method with the number
Include the function name, options and software version in an analysis note. Keep the original data and explain how missing values were handled. A label such as “lower quartile” does not preserve all those choices.
These two results are estimates under specified conventions, not proof that the data moved. For a shared report, agree on a convention before comparing outputs and reproduce it with a tiny test list.
Do not choose the method solely because its answer looks more convenient. Consistency and an explicit definition make a percentile useful to someone who needs to reproduce it.
Editorial illustration from this site’s image library; not documentary evidence of the example or object discussed.