How the model is built

Two limits worth knowing before you read the aspect scores. The first is about one aspect in particular: all four Industriousness questions are written so that agreeing means scoring lower, because the published item table has no question pointing the other way. Everywhere else in this battery, a habit of agreeing with statements partly cancels itself out. Here it does the opposite — it accumulates, and downward. Somebody who tends to agree with whatever is on screen will read low on Industriousness and slightly high on Orderliness, and those two are the halves of Conscientiousness, so the gap between them can be a fact about how you answer rather than about you. The averaged Conscientiousness score is the more trustworthy of the three numbers for that reason, which is the reverse of the usual advice here. Second: five factors, not six. The HEXACO model adds Honesty-Humility — sincerity, fairness, absence of entitlement and of a willingness to exploit — and it is not recoverable from any combination of the Big Five. Whatever it would have contributed is simply absent here rather than distributed among the other five.

PANAS resolves emotion into two factors, positive and negative, and the politically decisive distinction sits inside the negative one. Anger and anxiety both load on Negative Affect while pulling in opposite directions: anger predicts risk-seeking, mobilisation, punitive policy preference and searching for confirmation of what you already believe, where anxiety predicts attention to opposing arguments, information-seeking and conciliatory preferences. A single Negative Affect score averages those two together and can report the same number for people who behave nothing alike. That was the whole limitation here until September 2026, and it is now half closed. Fear is read from four of the ten negative adjectives — scared, nervous, jittery, afraid — and anger from two more, hostile and irritable, so the two are scored apart as well as together and the composite is unchanged. It cost nobody an extra question, because the answers were already being collected and only ever averaged. What remains is that two items is a thin reading of anger and four a thin reading of fear, and neither is a dispositional anger scale built for the purpose; they are a second look at a mood measure. Where the two subscales disagree, that disagreement is the more informative number, and where a reader has answered too few items for either to be stable the report says so rather than showing a difference that is noise.

The DS-R splits disgust into core, animal-reminder and contamination domains. The politically load-bearing split in the newer literature is a different one — Tybur, Lieberman & Griskevicius (2009, JPSP) separate pathogen, sexual and moral disgust, and it is sexual disgust that carries most of the association with social conservatism. The DS-R’s domains cut across that boundary rather than isolating it, so a disgust score here should not be read as the disgust-and-politics finding people have usually heard about.

The Cultural Capital number is close to a plain average of how warmly you rated the images you were shown, so it cannot separate a discriminating eye from an agreeable one — rate everything alike and it still returns a number. Only the contested register is reverse-keyed, and only three of its items are, so agreeing with everything cancels in one small corner of the task and accumulates everywhere else. The spread of your ratings is reported alongside as the quantity a habit of agreeing cannot move at all, and is deliberately left out of every score. Three of the four registers are loaded. Folk emptied when six objects tagged that way turned out to be their own societies’ court and ecclesiastical art, and was refilled with eleven genuinely vernacular pieces; contested — Bourdieu’s diagnostic subjects, which run in two directions and are scored with the sign he measured — followed. Pop is empty and likely to stay so: a thing is popular because it is in mass circulation now, and copyright is the legal shadow of exactly that, so a public-domain pop register would really be a history-of-pop register. Which register a piece belongs to remains somebody’s judgement about it rather than a property of the object.

Symbolic Capital has no source scale; those items are original, and asking somebody to rate their own standing is close to a category error for a construct Bourdieu defined by what other people concede to you rather than by what you claim. A reader who is modest and a reader who is unrecognised answer it the same way, and nothing here can tell them apart. Social Capital is on firmer ground — the items are behavioural rather than evaluative, which is why they were rewritten that way — but nine items is a thin reading of two constructs, and it is a self-report about a network rather than a measurement of one.

Reflective Taste is the one music reading that carries a number, and three things bound it. All four Reflective & Complex genres are asked so it is never computed from a two-item slice, and at least three must carry a real rating or it is not scored at all — somebody who listens to no reflective music is unmeasured here rather than low. It is read from a Music sitting at Surface, where the four genres are asked, and a retake there replaces it rather than widening it. And it is one factor of four: the other three, and every named artist and title, stay disclosed and unscored. Bandwidth has a limit of its own kind: it counts what you claimed against the list you were offered, and the list is this project’s own. Warde, Wright and Gayo-Cal showed that who comes out an omnivore depends on which genres were listed, so the number is honest about the bank it came from and says nothing about the genres that were not on it. Two limits sit on the news half as well. On the Iran track there is no over-claiming check, because that check needs outlet names that are provably invented and Iranian licensed press runs to hundreds of provincial papers built from exactly the words a plausible invented name would use — a name that turned out to be real would not weaken the check but reverse it, scoring a reader as over-claiming for correctly recognising their own local paper. And the breadth half assumes you know which register you just read, which is a larger assumption than it sounds: Pew followed people through a week of their real news reading and found they could name the outlet 56 per cent of the time, 52 per cent when the link came from social media. Three questions now ask about that directly, and where somebody cannot name their sources, a number about trusting across registers is measuring something they are not doing. The registers themselves rest on two different kinds of evidence — most on published content ratings, eight on Pew audience composition — and those can come apart: the New York Post is rated Right for what it publishes, and Pew grouped its audience as mixed.

Two limits, and the first is about this instrument rather than the theory. These twenty questions are not the Moral Foundations Questionnaire. They were written for this project to measure the five foundations Graham and Haidt describe, and nobody has ever checked whether they succeed — so the five numbers carry the theory’s authority and none of the instrument’s. Second, the five foundations do not cover the ground. Iyer, Koleva, Graham, Ditto & Haidt (2012, PLoS ONE) found that concerns about liberty are likely a moral intuition the theory had not measured, and libertarians scored highest on both kinds of liberty they asked about. Without it, someone whose morality is organised around non-interference reads on this scale as low on everything, which is a fact about the instrument and not about them. Quote Resonance carries a Liberty trait that reaches some of the same ground behaviourally, by measuring which rhetoric about freedom actually lands; that is a different kind of evidence from a moral-foundations score and not a substitute for the missing subscale.

Two limits, and they are different in kind. The first is that RWA is a right-coded instrument by construction: it measures deference to established authority, so aggression against authority in the name of justice is a different motion that it cannot register. That is a structural gap in the battery, not a finding about you, and it is not fixable by adding items to RWA. The second is about the words the questions use. Six of the ten name church, God’s laws, sinfulness, atheists or homosexuals, and for a reader who rejects the Islamic Republic those six can be rejected for what they sound like rather than for what they ask — which would read as low deference and would be wrong, in exactly the group this project is most interested in. Four forced choices between qualities in a child now run beside them, carrying no political or religious content at all, and neither reading replaces the other. If your two readings disagree, that is not an error to resolve: it is the clearest thing this instrument can tell you, and what it says is about the scale rather than about you. Quote Resonance now carries a Coercion trait aimed at the gap — resonance with the claim that force against opponents is warranted, sampled from both directions — but that is this project’s own unvalidated measure, not a substitute for a validated left-wing authoritarianism scale (see Costello et al. 2022 for what one looks like).

SDO is about hierarchy between groups, not dominance between people, and a person can be forceful in every personal dealing while scoring low here. The scale also collapses two subdimensions that Ho et al. (2015, JPSP) separate — support for active group-based dominance, and opposition to equality — which relate differently to policy preference. A single SDO number averages them.

Three separate numbers come off this, and the split between them is this project’s own. Kay and Jost’s scale produces one score and has never been shown to divide into these three, so a gap between your Fairness Perception and your Resistance to Change is not a finding about you — nothing has established that those are two different things to be. Read the three as one reading looked at from three angles rather than three readings to compare. Beyond that: system justification is measured as agreement with statements about how fair the existing order is, which does not separate believing the system is just from being satisfied with your own place in it. The two come apart most in exactly the people most worth distinguishing.

Left and right do not mean the same thing in every country, and self-placement on that axis is comparable only within a shared political context. For a respondent reading the Iran track this is not a technicality: the axis those words describe in Iranian politics is not the axis they describe in American politics, and a number on this scale carries whichever one the respondent had in mind. Nothing here can tell which. Two questions about religion sit here as well, and what they are for is worth saying plainly: several relationships this model draws elsewhere — between disgust, sanctity, deference to authority and belief that the existing order is fair — could turn out to be religion appearing four times under four names, and without asking, that cannot be told apart from four separate findings. They ask what you think about and what you believe, and deliberately not whether you attend anything. In a diaspora, attendance is suppressed for reasons that have nothing to do with belief, so an attendance question would have measured politics and called it religion — the same failure these two exist to detect. Two questions are a thin reading, and they are used to check other numbers rather than reported as a finding about you.

Stated confidence in institutions moves with events and with question wording, more than most attitudes do. A reading taken during a scandal, an election or a war is a reading of that moment as much as of the person, and this instrument has no way to date-stamp what it caught. There is also a question underneath the four registers rather than beside them: whether they are four readings or one. Five questions about whether important things are kept from the public are asked here for that reason — if they move together with all four registers, then low trust in parliament, courts, international bodies and the press is one disposition rather than four findings. Those five are all worded the same direction, so somebody who agrees with statements generally will read high on them; that pulls this reading and the four registers the same way at once, which weakens the comparison rather than inventing one.

Its three Reception scores are levels, not a contrast. Each is how warmly the quotes in one pool landed — anonymous, named, misattributed — and the pools hold different quotes in different numbers, so a gap between two of them can be a difference between the quotes, and it is never read as the effect of a name. The comparison that would be one, the same quote returning under a different name, belongs to Quote Resonance. A reaction also has no opposite pole, so there is no reverse-keyed item to write here and a habit of rating everything warmly goes uncorrected. Figure Sentiment depends entirely on recognising the figure: someone who does not recognise a name has told you nothing about the figure and something only about their own reading. And the Iran track assumes a familiarity most respondents will not have.

This bank is written for this project and has never been validated by anyone. Three things follow. Resonance is not endorsement — a line can land because it is well made, or familiar, or because you have argued against it for years. The bank is a selection, and someone chose it: nothing can move you that is not in it, so a flat result may say more about the bank than about you, which is why a sitting where nothing lands is reported as that rather than dressed as a profile. And the sixth trait, Coercion, exists to reach what a right-coded authoritarianism scale cannot register — that is a real gap to aim at, and aiming at it with an unvalidated instrument is still aiming with an unvalidated instrument. One of its numbers cannot exist after a single sitting: Source Sensitivity compares your reaction to the same quote under two conditions — anonymous, correctly named or falsely named — so that quote has to come back in a later sitting, and the screen says so rather than showing a number built from one.