The issue isn't with the use of a different font, the issue is that there are two series of data points differentiated by font that, according to the study desi

The issue isn't with the use of a different font, the issue is that there are two series of data points differentiated by font that, according to the study design, should be relatively independent, but in fact have improbably similar cumulative distribution functions. This indicates that one of the series was fabricated from the other.