Everyone Agrees, Until It Is Specific
Almost everyone in this survey says they believe in gender equality. Ask who is responsible when someone is harassed on the street, and that consensus falls apart. This Beporsim report follows that gap across five questions, and across what men and women report living through.
On paper, gender equality looks settled. In the answers people actually give, it is not.
Between 2017 and 2019, Beporsim asked thousands of people across Iran the same short set of questions about equality, discrimination, and everyday experience. The pattern that emerges is not a simple yes or no. It is a ladder. At the top, almost everyone agrees with the principle. A few rungs down, when the question names a job, a courtroom, or a street, that agreement thins out. At the bottom, on who is to blame when harassment happens, the sample splits along gender lines that the earlier questions barely hinted at.
This page is the narrative cut of that work: what the answers say, where men and women part company, and what the survey can and cannot claim. The full PDF holds the tables, the province scores, and the statistical detail. If you are citing a number, download that file.
A survey that measures the gap between principle and practice
Two indexes sit behind the story. The Gender Equity Index gathers five questions that move from the abstract idea of equality to more concrete judgments. The Perceived Discrimination Index asks whether people see women treated badly in the country, in school, and at work, and whether they say they have lived that treatment themselves. Each index turns those yes-or-no answers into a score from 0 to 100.
The cleanest comparison the data supports is between male and female respondents. They answered the same items. The gaps are large enough to survive ordinary sampling noise. What the data does not support is treating the combined totals as a portrait of Iran as a whole. This was a self-selected mobile panel: heavily male, and skewed young. Weighting brings men and women to an even share in the combined figures. It does not fix age or education. Read the methodology before you quote a national-sounding number.
Everyone agrees, until it is specific
Ask whether someone believes in gender equality, and the egalitarian answer is overwhelming. Ask whether the person who is harassed is responsible for what happened, and the egalitarian answer becomes a minority view. Between those two poles sit questions about leadership, the bench, and equal worth. The chart below walks that ladder from the easiest affirmation to the hardest one.
Weighted share giving the egalitarian answer on each Equity Index item.
2017–2019 · %
That drop is the report’s central finding in one picture. People are not rejecting equality as a slogan. They are disagreeing about what equality requires once a concrete role or a concrete duty is named.
Where men and women part ways
On the most general statements, men and women are closest. The distance opens once the question leaves the abstract. The widest single crack is harassment. A majority of men in this sample name the targeted person as responsible. A majority of women name the aggressor. Both groups had already said, a few questions earlier, that they believe in gender equality.
A majority of men locate responsibility with the person it happened to. A majority of women locate it with the aggressor.
Harassment responsibility, by respondent gender
Experience sits underneath belief. When people could name things that had happened to them because of their gender, lower pay showed up often for both groups. Harassment did not. Women reported it far more often than men. About one man in three said none of the listed experiences applied. Among women, that figure was closer to one in thirteen. The gap in what people believe about discrimination sits on top of a gap in what they say they have encountered. This survey cannot separate the two.
What people blame, and what they think might help
Asked what stands in the way, respondents pointed to culture more often than to law. Asked what might help, the most popular answer was also the vaguest: raise awareness. More concrete options, such as teaching equality in schools or speaking up when something happens, were chosen less often, even though they line up with the cause people had just named.
The stereotypes people say they hear about women and about men tell a related story. The loudest claim about women is a duty: that they should cook and keep house. Nothing about men dominates in the same way. One list tells women what they should do. The other mostly describes what men supposedly are.
How far these numbers travel
Education raises both indexes, but modestly, and it moves perception more than principle. Age does not behave the way a simple generational story would predict. Over the long field period, neither index moves beyond its confidence interval inside this sample. Province rankings exist in the workbook, including for smaller provinces with wide intervals. Those details belong in the PDF and the source tables, not as a throwaway map on this page.
Other published surveys do not ask this exact battery, so exact matches are impossible. Broader World Values Survey work on Iran points in a similar direction on leadership attitudes and on who tends to hold more pro-equality views. That is context, not validation of every figure here.
Read the full report
This article is the invitation. The downloadable PDF is the source: full question wording, weighted and unweighted tables, province scores, confidence intervals, and the methods behind the rebalancing. Use the download panel on this page after you subscribe, then cite from the file rather than from a paraphrase.
Analysis for this edition is credited to Yasaman Ranjbaran. The underlying records come from the Beporsim platform’s 2017 to 2019 field period.
How it was done
Methodology
Results are based on 4,621 self-administered mobile questionnaires collected between 2017-07-02 and 2019-11-03, comprising 18 questions.
Population and sampling. The study population consists of 140,000 users of the Beporsim platform (9 applications and the website, beporsim.com) who can access the platform from across Iran. Users participated after being informed about the purpose and consenting to data collection. Sampling used a random sampling method and Cochran's formula, with a 95% confidence level and a 3% margin of error. Recruitment was through mobile network operators. Participation was voluntary and self-selected.
Weighting. Every figure appears twice: unweighted, and weighted to an even split between men and women. The weight corrects one thing only: this sample is 77.4% male. Men and women are taken as equal halves of the adult population as a matter of arithmetic rather than of any official statistic. No further post-stratification is applied: weighting a self-selected panel toward additional margins corrects composition without correcting self-selection.
Completeness and indexes. All 4,621 respondents answered every single-choice question, so index denominators are the full sample throughout. Province and age group are recorded for every record; education, age and province are broken out only where a band holds at least 50 respondents. Each index is the respondent-level mean of five binary items multiplied by 100. "Don't know" is treated as a non-affirmative valid response and remains in the denominator.
Inference. Differences between groups are tested with chi-square. Family-wise tests use a Benjamini-Hochberg correction for the false discovery rate. Cramér's V is reported alongside because with a sample this size a difference can be reliable and still be small. Confidence intervals are 1.96 times the standard error of the respondent-level index mean. Weighted intervals use the effective sample size implied by the weights. Across all 31 provinces the half-width runs from ±1.9 to ±12.1 index points. The intervals assume simple random sampling within each group; that assumption is approximate here, and true uncertainty may be wider. They are reported as a floor on error, not an estimate of it.
Field period. Collection spanned 29 months across 2017 to 2019. Attitudes may have moved within a window that long, and no adjustment for interview date has been made, so the figures describe the period as a whole rather than any point inside it.
Stay in the conversation
If you see a chance to collaborate, have a question about the method, or want to reuse this work, write to us.