Instrument method: what the score supports and what it does not.

What the instrument is

The Evidence political position instrument records the position a person declares on 12 separate dimensions, each between two named poles. The current schema is evidence-pol-1. Dimensions are read one by one and are not added into a single label.

1. State structure Federal · Unitary
2. Representation Democracy · Autocracy
3. State power Security · Liberty
4. Immigration Assimilation · Multiculturalism
5. Diplomacy Militarist · Pacifist
6. Foreign intervention Non-interventionist · Nationalist
7. Ownership Public · Private
8. Economic coordination Planning · Free market
9. Foreign trade Protectionism · Globalism
10. Religion Irreligious · Religious
11. Social norms Progressive · Traditionalist
12. Technology Technology · Biology

Scale and direction

Each dimension runs from 0 to 100. 0 is pole A, 100 is pole B, and the direction is always from pole A to pole B. The number is kept as computed or imported, without rounding. A score of 50 is the midpoint between the poles, not a political center; a dimension with no answer stays UNKNOWN, not 50.

Where the score comes from

Questionnaire

The Evidence questionnaire has 48 original statements, 4 per dimension, with half pointing to each pole. For each dimension, the browser computes:

• the position of each answer, from 0 (pole A) to 1 (pole B), reversed when agreeing points to pole A;
• the score, the mean position times 100; with no scoring answer, the score is UNKNOWN;
• coverage, how many of the 4 statements scored;
• consistency, 1 minus 2 times the standard deviation of positions (with a single answer it cannot be measured and counts as 0.50);
• the confidence index, coverage times consistency: high from 0.75, medium from 0.50 and low below that.

Formula p = (resposta - 1) / 4; p_invertida = 1 - p; placar = 100 * media(p); cobertura = respondidas / 4; consistencia = 1 - 2 * desvio_padrao(p); confianca = cobertura * consistencia

Importing from another instrument

On the result page you can paste the result URL of another instrument (12Axes or Evidence's own portable format). The browser reads only the score parameters, without network access, and the vector is marked as imported. A value outside 0 to 100 rejects the whole import. Different instruments are not on the same ruler: the comparison shows differences, not errors.

Declared limitations of every imported vector

• Self-assessment: the vector records the answers a person gave, not observed behavior.
• Third-party method: the instrument, its questions and weights belong to the source and are not audited by Evidence.
• Scientific validation not established: the source describes itself as educational and not validated.
• The instrument may change without notice; vectors from different dates may not be comparable.

Evidence states

The six states are defined in the political glossary.

What the instrument is not

• It is not a voting recommendation and points to no candidate, party or side.
• It does not place candidates, parties or real people. There is no ideological ranking of anyone.
• It is not a personality diagnosis nor a measure of how extreme someone is.
• It does not infer position from behavior, likes, votes or third-party data.

Limitations

• Not a scientifically validated instrument in this version: there is no reliability test, validity study or population norm.
• Four statements per dimension is few: one answer moves the score a lot, which is why confidence is declared per dimension.
• Self-assessment on a date: the result records answers, not behavior, vote or anyone's political essence.
• The wording was written for the Brazilian context; the same text may read differently in another country.
• Scores from different instruments are not on the same ruler; the comparison shows differences, not errors.
• Social desirability: some people answer what seems acceptable, not what they think.
• Reading the statements: words such as liberty, State or tradition mean different things to different people.
• Acquiescence: some people tend to agree with any statement; that is why each dimension's statements point to opposite poles, which reduces the effect but does not remove it.
• Poles are conventions: the names of the ends may change in a future version of the schema without changing what the scale measures.
• Dimensions are not independent: answers on one dimension tend to go with answers on another; the instrument does not estimate those correlations and does not use them.
• Time: positions change. A result is dated and versioned; results from different versions may not be comparable.

Privacy

Political opinion is sensitive personal data under Brazil's LGPD (art. 5, II), with processing governed by art. 11. That is why the instrument is private by default:

• The score is computed in your browser. Answers stay in the open tab and in the fragment (#) of the result page address, a part the browser does not send to the server. Nothing leaves without an explicit action from you, such as copying the URL or taking the result to the workspace.
• The result page is not indexed by search engines (noindex), and no public page shows anyone's result.
• The portable URL carries the dimension scores and the date, with no name or identifier. Whoever gets the URL sees the scores: share it only if you want to.
• A public figure only enters with a documented public source (URL of a result published by the person or by an identified third party), never by inference.
• Evidence uses no tracking cookies or analytics; details in the privacy policy.

Readings on measuring position on several dimensions

• Chapel Hill Expert Surveyparty positions on separate economic and cultural axes.
• Manifesto Project (WZB)party positions by topic, from election programs.
• World Values Surveypopulation values by country, in survey waves.
• V-Demdemocracy indicators by country, from expert coding.

These projects measure parties, countries or populations; the Evidence instrument records one person's declaration. No question, weight or catalog of theirs was copied.

General sources declared by the instrument

• Feldman e Johnston (2014), Understanding the Determinants of Political Ideology: Implications of Structural Complexity, Political Psychology
• Treier e Hillygus (2009), The Nature of Political Ideology in the Contemporary Electorate, Public Opinion Quarterly
• Billiet e McClendon (2000), Modeling Acquiescence in Measurement Models for Two Balanced Sets of Items, Structural Equation Modeling
• Chapel Hill Expert Survey
• Manifesto Project (MARPOR), esquema de codificação