What am I doing today? I’m re-interpreting Q-method output for my dykeland study because of a late discovery about how PQMethod identifies ‘defining sorts’. Q-method uses statement-sorting (or photo-sorting, viz Milcu et al. ) to understand the public discourses that exist around a given issue. I seem to be doing quite a lot of it of late with students and colleagues, but this is the first time that I’ve been the analyst. Near-ubiquitous freeware program PQMethod does an outstanding job of providing statistical output that is easily interpreted, but it is important to dig into the manuals to understand the steps it takes along the way. Once factor analysis identifies various discourse ‘types’ based on sorting on a forced-normal distribution, PQMethod helpfully identifies ‘factor-defining’ sorts, which you can use to characterize each one. These are the individuals who sorted similarly, driving that particular ‘archetype’. Using the demographics of these defining sorts to be reflective of a discourse is particularly useful when you have a lot of sorts, which is a new use of Q-method which is not entirely consistent with the rationale behind its design. PQMethod identifies as ‘defining’ those sorts where: (1) The factor explains at least half of the common variance, that is, the factor loading for that particular respondent on a particular factor must be at least half of the variance explained by all factors pulled out of the model for that respondent; and, (2) the loading must be significant at p < .05. However, it also includes in that mix those for which the correlation (the loading) is negative, that is, the complete opposite. Perhaps for a qualitative interpretation of the factors this would be irrelevant, but I designed my concourse of statements such that scores could be derived to summarize perspectives on a range of themes. As folks like me push the method and the software to places it was not intended to go, it behooves us to be careful that we fully understand the tools we are using.