Why our barometer is not an opinion poll

This site’s barometer shows, from the answers of the people who take the test, how many have responded, how far they agree with each statement, nationwide and in each state, where they stand on average on the political compass and which issues divide their answers most. Once there are thousands of tests, it will be tempting to read it as an opinion poll. It is not one, and it is worth explaining exactly why.

What the barometer shows

The barometer has four parts, and the data are updated daily:

• Participation: the number of completed tests, how many in the last 30 days and how many states already have publishable data.
• Statement by statement: the share of answers in favor (agree or strongly agree), neutral and against (disagree or strongly disagree) for each of the test’s 45 statements, grouped by issue. A selector shows only the people who connected from a given state.
• Territory: the average position on the political compass of participants in each of the 32 federal entities (economic axis from −100 on the left to +100 on the right; social axis from −100 progressive to +100 conservative), based on the approximate location of their connection.
• Polarisation: for each issue, an index from 0 (everyone answers the same) to 100 (answers are split as much as possible between the two ends of the scale).

What it does not show matters just as much. The Mexican barometer publishes nothing by party: not which party participants say they would vote for, not which party comes out closest to their answers, nor any similar breakdown. It is a barometer of issues, not of parties.

Every chart carries the same warning: “This barometer is not a poll or an opinion survey on electoral preferences: it does not ask how people will vote and publishes nothing by party. It brings together the anonymous answers of the people who chose to take the test, so it is not representative and lacks the methodological guarantees of a poll.” The readings of the barometer we publish in our analyses follow the same rules.

The difference lies in how respondents are chosen

A serious poll starts with a sample design. The polling firm decides how many people to interview and with what breakdown by sex, age, education, state or size of locality, so that the result can be generalised to the whole population. It selects those people, often at random from electoral precincts or households, and if the final breakdown does not match the population, it weights the answers. That work is what allows it to say that a given share of Mexicans think something, with a margin of error.

The barometer has none of that. The people in these data came on their own: they searched for a political test, clicked a link on social media or got it from someone on WhatsApp. Nobody chose them. There are no quotas and no weighting, and there is no margin of error to calculate, because a margin of error assumes a random selection that never happened.

This group can be expected not to look like the electorate: it will probably be younger, more urban, more connected and more interested in politics than average. The test is also open to anyone, including people who do not vote: only citizens aged 18 or over who are on the electoral roll and hold a voter ID card can vote. That is selection bias, and the number of answers does not correct it. A million voluntary answers are still a million voluntary answers.

What Mexican law says

Electoral law regulates polls and opinion surveys throughout the electoral process, which for 2027 began in September 2026:

• The INE sets the rules. The INE’s General Council lays down the criteria that individuals and companies conducting polls or opinion surveys during electoral processes must follow (LGIPE, Article 213.1), and sample surveys must meet scientific criteria (Article 251.7). Those rules are in Chapter VII of Book Three of the Election Regulations.
• Whoever publishes delivers the study. Anyone who requests or orders the publication of “any poll or opinion survey on electoral matters” between the start of the process and the close of the polls must give a copy of the full study to the INE’s Executive Secretary (LGIPE, Article 251.5), no later than five days after publication (Election Regulations, Article 136). The Regulations also set out the methodological details that must accompany the publication, such as the dates, the population, the sample size, the wording of the questions and the margin of error.
• Silence at the end. During the three days before the election and until polls close, it is forbidden to publish or disseminate by any means the results of polls or surveys intended to show electoral preferences (LGIPE, Article 213.2; Election Regulations, Article 134). The ban lasts until the polls close in the country’s westernmost time zones (Article 251.6).
• And it is a crime. Publishing those results on those days is punishable by a fine and six months to three years in prison (General Law on Electoral Crimes, Article 7, Section XV).

The law does not require a study to be scientific for it to count as a poll or survey: method is a quality requirement, not the definition. And the INE also receives and reviews studies carried out online with participants who were not selected at random (for example, in its report on polls of 29 May 2024). So it is not enough to say the barometer is not representative: what sets it apart from a poll on electoral preferences is what it publishes.

Our rules

Hence the barometer’s rules, on our reading of the law:

• We publish no electoral preference. At the end, the test optionally asks which party you would vote for, only to show you which of that party’s positions you agree with and which you do not. That answer is never published, not even added up with other people’s. Nor do we publish which party comes out closest to participants’ answers, because it could be read as a preference.
• Issues only. The barometer shows how many people agree with each statement and how positions are spread across issues, not parties or candidates.
• We never call it a poll or survey, or “voting intention”, or present its figures as an estimate of what the country thinks or of what will happen in the election.
• We do not compare it with polls, and this site publishes no one’s poll figures.
• Every chart carries the warning quoted above, and the methodology note says where the data come from, between which dates and how many tests they cover.

There is one more reason for caution. This site is run from outside Mexico, and the Constitution states that foreigners may not interfere in the country’s political affairs (Article 33). Our job is to inform: we never ask anyone to vote, or not to vote, for any party, and we publish nothing that could be read as a forecast.

So what is it for?

To describe this group, which is no small thing. If most participants agree with a statement, or if an issue clearly divides their answers, that says something real about the people who take the test. What it does not allow is the leap to “Mexico thinks that”.

It is also useful for comparison, with care. Differences between states within the same self-selected group are informative, even if their absolute level cannot be generalised and the bias need not be the same everywhere. The same goes for comparing statements: that one has more support than another among participants is a fact about the participants.

How we protect participants

The test asks for no name, email or sign-up. Each result is stored with a random identifier. Before the result you may, if you wish, give your age bracket, your sex and the party you would vote for; none of these answers appears in the link to share your result, and none of them affects the calculation of your affinity. You are not asked for your state: it is estimated approximately from your connection.

Aggregated data have a minimum: no figure covering fewer than 100 tests is published. That is why, while there are few tests, the barometer will be almost empty, and a state that does not reach the minimum will not appear. It is not that data are missing: with few cases, a figure starts to say too much about specific people.

We never publish individual data or cross aggregates in a way that would allow anyone’s profile to be reconstructed. The details are in the privacy notice.

And the days before the election?

Election day is Sunday 6 June 2027. The ban on publishing poll results begins on Thursday 3 June and lasts until the last polling stations close on Sunday. Mexico has several time zones: Quintana Roo is the first to enter Thursday, and polling stations in Baja California and Sonora close at 18:00 local time, 19:00 in Mexico City.

The barometer is not a poll and publishes no electoral preferences, but its figures could be mistaken for one. That is why the whole barometer goes dark from Wednesday 2 June to Sunday 6 June 2027, full days in central Mexico time: a day earlier than the law requires and until hours after the last polling stations close. At the same time, the other places where we publish aggregated answers, such as the analyses that use them, are hidden, and on those days we do not publish or comment on their figures anywhere else, social media included. Nothing is deleted: the data return after election day.

The test stays open. Your individual result is neither a poll nor a forecast: it only shows how far you are from each party.

How to cite it

If you use these data, the correct way to present them is as aggregates from a self-selected sample of users of a political affinity test, stating the period, the number of tests and the warning that the sample is not representative. The barometer’s methodology note has all of this.

We ask you not to present them as a poll, a survey or a measure of how people will vote, because they are none of those things, and not to publish them at all from 2 to 6 June 2027. How each test result is calculated is explained in “How your affinity is calculated” and in the methodology; the election rules, in “How Mexico elects its 500 federal deputies in 2027”.