Subtracting Group Medians Did Not Reproduce the Median Paired THC Difference
A median difference calculated within people is not necessarily the difference between two group medians. The capillary–venous THC table illustrates why preserving paired calculations matters.
The published summaries can look counterintuitive
At thirty minutes in occasional flower users, the reported capillary and venous medians were 2.4 and 2.1 nanograms per milliliter. Yet the median within-person difference was negative, approximately −0.4. Simply subtracting the two group medians would produce a positive value instead.
The operations summarize different information
Each separate median identifies the middle value within its own ranked set. The paired analysis first subtracts the corresponding specimens from each person and then summarizes those differences. Ranking and subtraction need not yield the same result when performed in the opposite order.
Use the statistic that answers the question
A question about paired matrix bias calls for the paired observations and analysis, rather than an improvised subtraction of marginal summaries. The numerical contrast alone is not proof of a reporting error. It also should not become a conversion formula for an individual sample. Retaining the calculation helps avoid reversing the apparent direction of a matrix comparison, while leaving collection conditions, uncertainty and the distinction from impairment assessment intact.
Related analysis: the study’s broader findings and limitations.
