It does matter, you’re assuming that they surveyed equal number of men and women, but I don’t see anywhere on that site that indicates that. Example in which op could be correct along with the statistics from (or close enough because I am on my phone and can’t get too complicated). Let’s create our dataset that could fit both results, let’s say they surveyed 100 people, 74 men and 26 women. From that 36 people are against abortion, 28 men and 8 women. So 36% of people surveyed are against abortion( This part matches the overall %s from the survey). Those 28men make up 78% of the people against abortion (this part matches OP). If we calculate them within each gender, 38% of the men are against abortion and 29% of women are against abortion (numbers got a little different here, but it’s close to the survey results). I am assuming that they didn’t have equal number of men and women, that’s why they stayed away from that calculation.