For most of the last century, the people in a given age group tended to move together politically. Knowing someone's generation told you something useful about how they saw the world. Among today's under-30s, that assumption is breaking down — and the fault line runs along sex.

Across a range of democracies, young women have moved steadily toward liberal positions over the past two decades, while young men have held roughly flat or drifted the other way. The result is a gap between the two that, in several countries, is now wider than at any point in the available record.

A pattern that repeats across borders

What makes the trend notable is not any single country, but how similar it looks in very different places:

  • South Korea shows one of the sharpest splits, with young women trending liberal and young men moving conservative over a short span.
  • The United States shows young women pulling away steadily since the mid-2010s, while young men stayed close to the center.
  • Germany and the United Kingdom show the same shape at different magnitudes — a widening fan between the two lines.

The specifics of measurement differ by country — some track self-described ideology, others support for liberal and conservative parties — but the direction of travel is strikingly consistent.

Why an average can hide the story

For anyone who works with survey data, the practical lesson is about aggregation. A single headline number for "young voters" now averages together two groups that are moving in opposite directions. Report only that average and you can miss a 30- to 40-point internal divide — and badly misread how a cohort will respond to a candidate, a message, or a policy.

The moment a subgroup splits internally, the group average stops describing anyone in it.

The fix is not exotic. It is disciplined cross-tabulation: read age and sex together, not one at a time, and check whether a within-cohort split is doing work that the headline conceals.

What is driving it

The causes are genuinely debated, and it is not our place to adjudicate them. Researchers point to a mix of factors — diverging experiences in education and the labour market, increasingly separate online media environments, and reaction-and-counter-reaction dynamics in which each group's shift reinforces the other's. What matters for measurement is less why the gap exists than the discipline to keep measuring it accurately.

What it means for research design

For our own work, the trend reinforces a few habits:

  • Weight and report by age × sex, not age alone, whenever a cohort may be internally divided.
  • Watch for mode effects. Groups that live in different media environments can be reached — and can respond — differently by phone, panel, and text-to-web.
  • Test messages separately. A frame that lands with one half of a cohort may fall flat, or backfire, with the other.
  • Resist cohort shorthand. "Gen Z thinks…" is now, more often than not, two different sentences.

None of this changes what a poll is: a careful measurement, taken with a known instrument, reported with its uncertainty. It simply raises the bar for reading the youngest part of the electorate — where the average has quietly stopped telling the whole story.