COMPILER Research

Eight weighted factors combine 35 behavioral trait scores across 253,717 eligible historical profiles. Tail cutoffs depend on the number of selected traits: 1 trait: 3.5%; 2 traits: 18.5%; 3 traits: 32.5%; 4 traits: 43.5%. The top or bottom tail is applied to every selected dimension using fixed cutoffs from the full eligible population, with tied scores kept together. These settings aim for a typical music-filled sample near 5,000 profiles, not an exact size for every combination. Selection counts without a separate tail sample reuse the existing global median splits. Unselected dimensions are unrestricted. The full eight-letter type remains the default. Sample sizes vary by combination and category.

Distinctive entries require 70 usable profiles in the type, 200 mentions overall, 10 in the type, 50 outside it, at least three in each deterministic half, a lift of at least 1.25 and a gap of at least 0.25 percentage points. The direction must recur in both halves. Missing mentions are not dislikes. These exploratory display rules were relaxed after inspecting the initial results; they are not independent validation.

Most-listed results show up to 20 entries with at least ten mentions. If no distinctive entries qualify, the explorer shows most-listed results with a notice; this does not mean the type has no distinctive preferences. Music shows the existing 695-band/group catalog without genres; solo artists are not exhaustively covered. Genre counts remain in the underlying research. Books and television contain a mixture of works, authors, series, channels and genres. Religion and politics use only the previously reviewed label groups. Education describes recorded concentrations, not degree completion.

The within-Religiosity view compares profiles on the same Religiosity pole using the current search's cutoff system. It is not full statistical adjustment. The historical data span 2008–2011. The new questionnaire is experimental and uses different, provisional midpoints.