James Taranto of the Wall Street Journal seeks to understand why marriage polls never register the support for marriage we see in actual statewide votes, and can only find an anonymous source to help explain why -- read to the last line:

Every state where same-sex marriage has been on the ballot, it has lost--usually by considerably larger margins, but mostly in socially conservative states. Forty percent support and majority opposition seems in the right ball park.

A reader whose identity we'll conceal explains why the polls may be unreliable on this question:

With a marriage amendment on the ballot in Minnesota, we have been assaulted by the pro-gay marriage media and social-media coverage. I say assaulted because the message is not a positive argument for gay marriage, but rather a tarring as bigots of those who believe in the traditional definition of marriage. So of course polls would undercount support for the traditional view of marriage.

A person could tell a pollster that he believes in a position and risk the pollster thinking that he is bigoted, or he could toe the media line, give the pollster a fulsome answer of support for the measure and then vote his conscience privately.

I know what I do (and why not, anything more public than this email could risk my career).