Method and limits

Kosmatrich · chapter 6

What the test does, how it scores and how far it can go.

Personality and condition

Four principles of trait psychometrics

The construction follows standard practices of classical test theory: Likert-type scales, as in the Big Five inventories.

Unipolar scales per pole, not bipolar per dimension

A single bipolar scale cannot tell equilibrium (both poles present) from absence (neither). The eight poles are scored separately; the ontology is recovered afterwards: both high, equilibrium; one high and one low, bias; both low, absent.

Balanced key

Each pole has two direct items and two reverse ones, to detect acquiescence, the tendency to agree with everything. The reverse items measure the absence of the pole, not its opposite, so that equilibrium stays measurable.

Few items

Four per pole, 32 in all, about six minutes, with five answer options.

Plain language

The items describe everyday behavior and experience; none uses the technical vocabulary of pole or dimension. That vocabulary stays on the side of the analysis, never on the form: only the scoring key knows the mapping.

How it is scored

Score per pole

Weighted mean of its four items (the reverse ones inverted), rescaled from 0 to 1. The weights are provisional, set by the author's judgment.

State of each dimension

With thresholds of 0.6 (high) and 0.4 (low): equilibrium, bias towards a pole, absent or intermediate.

Prototype of each landa

A landa is read nested: the first pole, the outer wrapper, weighs 0.4; the second, 0.3; the third, 0.2; and the fourth, the core, 0.1. The expected intensity of a pole is the sum of the weights of the positions it occupies, so a repeated pole weighs more.

Resolution

Pearson correlation between your eight-pole profile and the prototype of each of the 512 landas. The top three and the gap to the second are reported. The choice is always among real landas of the catalog: a sequence is never built by hand, which avoids impossible paths in the graph.

Validity indicators

Extreme raw mean (acquiescence), near-zero deviation (uniform answers) and inconsistency between the direct and reverse items of the same pole.

Synthetic validation and limits

What is missing for it to be a test properly speaking

Until then the test is a tool for exploration, not a validated instrument.

Language

The instrument and the texts of each personality exist in Spanish and in English. The English version of the 512 personalities is composed from the same pieces as the Spanish one, translated: it is the same reading, not an adaptation.