Spoken for: the gender gap in Pakistan’s vote

In February 2024, 42.7 per cent of registered women voted and 51.8 per cent of registered men did, across the National Assembly constituencies with available figures. The difference was nine percentage points, close to the 9.1 recorded in 2018. Turnout fell by about four points for both sexes over the six years.

The Elections Act of 2017 sets a different test: where women cast fewer than 10 per cent of all ballots, the Election Commission may presume that an agreement restrained them from voting and may declare the poll void. That threshold concerns women’s share of votes cast, not the difference between women’s and men’s turnout rates. Neither the threshold nor the national average tells us where turnout gaps are widest.

This analysis links the available Election Commission returns for National Assembly seats in 2018 and 2024 to census estimates of mother tongue and urban population. In 2024 the gap was 8.3 points in Punjabi-majority seats, 7.1 in Saraiki-majority seats and 5.8 in Sindhi-majority seats. In Pashto-majority seats it was nearly eighteen, and in the selected urban seats of most of the nine cities examined, eleven or more. The national gap changed little, alongside narrowing gaps in the Pashto-majority group and several city samples and a widening gap in rural Punjab.

The data

The figures come from Form-47, the provisional consolidated result for each constituency. It reports constituency totals of registered voters and votes polled by sex. Section 91 of the Elections Act requires separate statements of men’s and women’s participation at polling stations; 2018 was the first general election under that requirement. The sex split used here is available for 270 of 272 seats in 2018 and 257 of 266 in 2024. Adaad did not identify a sufficiently reliable constituency-level series for an earlier election.

The census does not count by constituency, so language and urban share were carried from tehsils to seats, through the census block codes in the 2018 polling scheme and the tehsils and patwar circles named in the 2024 delimitation list. A seat that takes part of a tehsil inherits that tehsil’s language mix, so the estimates cannot see differences between neighbourhoods. Mother tongue comes from Table 11 of the 2017 census for 2018 and of the 2023 census for 2024.

Two measures run through the piece. The turnout gap is defined as the share of registered men who voted minus the share of registered women who voted, in percentage points. If half of registered men and four in ten registered women vote, the gap is 10 points. The roll gap is defined as men’s share of registered voters minus women’s, also in points. A woman who never registered is missing from both sides of the turnout rate and so invisible to the first measure. Group figures pool every voter in the group rather than averaging seats, and a seat is called Pashto-majority, or Punjabi-majority, when that language is the mother tongue of over 70 per cent of its residents.

Where the gap is twice the national figure

Figure 1 groups the seats where one mother tongue is spoken by over 70 per cent of residents and shows the turnout gap for each group in both elections.

In the 37 Pashto-majority seats, 28.3 per cent of registered women voted in 2024 against 46.0 per cent of men, a gap of 17.6 points and nearly twice the national figure. In 2018 it had been wider still, at 21.2 points. The 87 Punjabi-majority seats had a gap of 8.3 points in 2024, the Saraiki seats 7.1 and the Sindhi seats 5.8. In the three Balochi-majority seats about four in ten registered voters of each sex voted, leaving a gap of 1.2 points.

Grouping hides the spread, so Figure 2 plots every seat on its own: the share of its residents who speak Pashto against its turnout gap, in either election, with a fitted line and its correlation.

Across all seats, the share of Pashto speakers correlates with the turnout gap at +0.64 in 2018 and +0.56 in 2024, and no other language is beyond 0.3 either way. Moving the language threshold from 70 per cent to 50 or 90 changes the 2024 Pashto gap by less than half a point. The association is also visible within Khyber Pakhtunkhwa. Within that province, the five seats where Hindko has the most speakers had a gap of 6.4 points against 17.8 where Pashto has the most, with the ex-FATA merged districts included throughout. The Pashto seats are not alike. In 2024 the gap ran from 5.3 points in Pishin, 5.8 in Kurram and 7.6 in Karak to 29.5 in Shangla, 30.6 in Battagram and 31.8 in Swat-II.

These are seat-level associations. The returns do not record the language of anyone who voted or stayed home, so they cannot show that a Pashto-speaking woman was less likely to vote than her Punjabi-speaking neighbour. After adjusting for the sex composition of the roll and the provision of polling stations and booths for women, a larger Pashto-speaking population still went with a wider gap in 2018. In 2024 the adjusted result is less certain once station provision is measured separately for women and men. No part of the analysis measures how far women must travel to their assigned station, because this analysis does not have sufficiently complete, verified station locations. Whether the gap is distance, custom, mobilisation or something else is the open question the returns cannot settle.

Wider gaps in urban seats outside the Pashto belt

The second cut is by how urban a seat is. Outside the Pashto belt, the most urban seats have a wider turnout gap than the most rural ones. Figure 3 groups seats with fewer than 50 per cent Pashto speakers by their estimated urban population share. Its lower panel shows selected urban constituencies in nine large cities, including seats within the Pashto belt.

Among these seats the gap was 6.1 points where fewer than one in ten residents is urban and 11.1 where more than nine in ten are. Men’s turnout falls from 54.9 per cent in the most rural band to 46.7 in the most urban, and women’s falls further, from 48.8 to 35.6. The roll gap barely moves across the bands, staying at seven or eight points, so the wider urban gap appears in voting rates among registered voters, rather than in a wider imbalance between men’s and women’s shares of the roll.

City by city, Peshawar’s two urban seats had the widest gap in 2024 at 21.2 points, then Quetta at 13.7, Faisalabad 13.3, Gujranwala 13.1, Hyderabad 12.4, Multan 11.9 and Lahore 11.6. Rawalpindi’s five urban seats, at 7.8 points, were the exception. Karachi, at 10.1, sits near the bottom of the list only because its men vote so little too: 42.3 per cent of its registered men voted and 32.2 per cent of its women. Quetta’s were lower still, at 34.3 and 20.7 per cent.

Every seat where Urdu is the largest language lies in Karachi or Hyderabad. In a model pooling both elections, those seats had a gap 2.7 points wider than seats where Punjabi comes first. Adding the urban measure reduces that difference to 0.1 points, while the Pashto–Punjabi difference is 11.5 points in the adjusted model. The Urdu difference therefore largely disappears after adjustment, while the Pashto difference remains in this specification. This is an association, not evidence that urbanisation causes one gap or that language causes the other. The urban measures also differ: 2018 uses registered voters assigned to urban census blocks, while 2024 uses census population allocated to constituencies.

Different movements behind a similar national gap

Between 2018 and 2024 the gap moved in opposite directions across the groups examined. Figure 4 shows the roll gap and the turnout gap for each language group in both elections. These are comparisons of changing groups: boundaries were redrawn and language-group membership changed. City and rural samples also use different urban measures in the two years, so their changes are indicative comparisons, not measurements of change within an identical set of places.

In the Pashto-majority group the turnout gap narrowed from 21.2 points to 17.6. Several city samples narrowed too: Hyderabad from 17.3 to 12.4, Multan from 15.8 to 11.9 and Peshawar from 24.4 to 21.2. The most urban band outside the Pashto belt went from 12.2 to 11.1; Quetta widened. Over the same six years the Punjabi-majority group went from 6.8 points to 8.3 and the Saraiki group from 5.2 to 7.1. In Punjab’s rural sample, defined as seats below 30 per cent on the relevant urban measure, the gap was 5.0 points in 2018 and 7.2 across 73 seats in 2024. Two of those 73 have unusually large gaps, confirmed as transcribed from their Form-47 scans; omitting them leaves 6.3 points. These opposing movements coexist with a similar national gap. This analysis does not decompose their contributions to the national change, which also depends on each group’s share of registered women and men.

The roll gap narrowed nationally and in every language group shown. Women were 44.1 per cent of registered voters in 2018 and 46.1 per cent in 2024, so the roll gap narrowed from 11.8 points to 7.7, though the 2024 roll in the seats with usable returns still held 66.7mn men and 57.1mn women. In the Pashto-majority seats both gaps narrowed, the roll gap from 15.8 points to 10.0. The Balochi-majority seats combined a 9.9 point roll gap with a turnout gap of 1.2, and the two Brahui-majority seats a roll gap of 10.2 with a turnout gap of 5.0.

Where women are missing from the roll, the turnout gap understates how many women are missing from the vote, because the women who never registered do not count against it. Women cast 24.4mn votes in 2024 against 21.7mn in 2018, and their share of all votes rose from 39.8 per cent to 41.4 even as their turnout rate fell, because the roll became more balanced. Had registered women voted at men’s rate in each constituency in 2024, they would have cast about 5.2mn more ballots.

What the law can see

Section 9 of the Elections Act 2017 allows the Election Commission to presume that an agreement restrained women from voting if women cast fewer than 10 per cent of a constituency’s votes, and to declare polling at one or more stations, or the whole election, void. This is a specific presumption, not the limit of the Commission’s powers: section 9 also provides for action where grave illegalities or violations materially affect the result, including agreements restraining women from voting.

In the 2018 sample, two seats fell just below the 10 per cent threshold, Shangla and a seat in the ex-FATA merged districts, both at 9.9 per cent. None of the 257 observed constituencies in 2024 fell below it; the lowest women’s share was Battagram’s 18.8 per cent. Yet nineteen had a turnout gap of at least 20 points, including sixteen Pashto-majority seats. Passing the threshold therefore does not establish equal participation, and does not rule out restraint or other grounds for intervention.

The returns cannot say why Swat-II’s gap is six times Pishin’s, or why the selected urban seats of Faisalabad have a wider gap than those of Rawalpindi. Testing possible explanations would require better data on polling-station locations and the voters assigned to them, alongside surveys that ask women themselves. The returns and census estimates identify where further investigation is needed: a gap of about nine points across the observed constituencies, nearly eighteen in Pashto-majority seats, and eleven or more in the selected urban seats of most cities examined. A similar national average conceals substantial geographic differences; these data describe those differences without settling their causes.