There is something that doesn't make sense.

https://www.marketwatch.com/story/americans-overwhelmingly-support-roe-v-wade-but-still-favor-abortion-restrictions-2019-06-07 states that 77% of Americans want to keep Roe V Wade.

https://news.gallup.com/poll/1576/abortion.aspx states that only 46% of the population is pro choice.

Something is wrong with how people poll.

it's the way the question is phrased and the loaded terminology. For example, if they had asked something more like "do you think there are circumstances where a woman should have the right to end a pregnancy?", then it wold probably be higher. 

but "pro-life" and "pro-choice" are loaded terms. in the same way someone might tell you they are conservative, but then if you asked, they tell you they support medicare for all. they know they "should" be conservative so they identify that way, but on specific issues they are actually much more left than they would admit (even to themselves). 

lots of people will tell you they are "pro-life" even if they do support a woman's right to choose. But it is a terminology issue. 

polling has always been wrong

There was this other poll from Pew that said that 42% of people believe abortion should be illegal, but less than 30% want Roe V Wade to be overturned.  If you want abortion to be illegal, you easily should support Roe V Wade being overturned.

They might be conservative on other things.

but that's sort of my point. the term "conservative" can mean alot of different things to alot of people. So a KKK clan member's conservatism would be different from say, mitch mcConnell's (i hope). Similarly, saying you are "pro-life" in and of itself doesn't really mean anything. Does that mean you oppose all abortions? Does that mean you oppose late term abortions? Does that mean you don't really care but "pro-life" just sounds good? 

So circling back to my point. If you ask if someone is "pro-life" or "pro-choice", the answer doesn't actually tell you all that much. 

Oh.

Lots of problems with how polls are conducted. Even the same poll run by two different groups, will have skewed results.

Let's say NPR asks 10,000 people are they for or against the border wall, and Fox news does likewise... Even if they try their best to be random, you'll have massively different outcomes.

At 10,000 people, your not going to get a huge margin of error if done correctly.