Books and COMPILER Personality Types

Patterns with substantial differences

These examples require at least 25 mentions in a type of at least 300 profiles, in addition to the common screening rules. They are examples with larger cells, not a claim of independent replication.

The catalog is derived from common book-field entries, with obvious title, series and author aliases combined. Results mix titles, authors and genres, which are labeled as entries rather than all being described as individual books. Generic placeholders were not treated as preferences.

Explore the 256 profiles

Reading the eight letters

COMPILER uses the original proposed pole names below. These are provisional shorthand for broader dimensions, not clean logical opposites, abilities, diagnoses or judgments of worth.

In particular, Autonomy does not mean an inability to care; Frugality is broader shorthand here than thrift alone; Leadership includes control/confrontation preferences rather than demonstrated skill.

What the comparison can establish

The catalog contains 302 entries. Distinctive rankings require at least 70 usable profiles in a type, 200 mentions overall, 10 in the type, 50 outside it, and at least three in each deterministic half. Listing-rate lift must be at least 1.25, the absolute gap at least 0.25 percentage points, and the direction positive in both halves.

Lift compares the selected type with everyone outside it in the same usable category cohort. Lists can contain fewer than twenty entries. The split-half rule screens consistency; it is not external validation or a multiple-testing guarantee. Larger ratios based on ten or twelve mentions remain uncertain.

Each person counts at most once per matched entry, but can match several entries. Common whole-field dislike/negation phrases are conservatively screened out of preference fields. Education concentrations are not dislike-screened. The matching catalogs are not exhaustive and may miss languages, aliases or prose descriptions.

The factor model combines 35 original trait scores. Each of its eight global cutoffs splits 253,717 eligible historical profiles into 126,859 higher and 126,858 lower scores. Marginal balance does not guarantee equal-sized combinations. The source spans 2008–2011; it is not a representative sample of current tastes or the general population.

Download aggregate rankings CSV · Aggregate results · Full scoring and matching methods