Depending on sample size and mix of people questioned, public opinion polls can give us a reasonable insight into just where public opinion is trending. Repeated polls on a similar subject can reinforce the conclusion of past polls. This does not mean that a particular thing should be done, nor does it attest to the viability or falsifiability of a proposition however.
Yesterday, Quinnipiac polling institute released a study of 2,041 registered voters on the issue of gay marriage/civil unions. I found this conclusion of the study acutely interesting.
"By 55-38 percent, voters said they did not want their state to allow same-sex couples to marry. However, by 57-38 percent, they favored allowing such couples to form civil unions that would provide marriage-like rights and by 53-40 percent they supported allowing same-sex couples to adopt children."
The only difference is the choice of words here, "marriage" as opposed to "civil unions." The poll showed that people were slightly more in favor of civil unions as opposed to marriage. Legally speaking, civil unions are not qualitatively dissimilar from what we understand as marriage.
This leads me to conclude that the word marriage is a similarly loaded word like God. (Thus, I object to it placed on currency and in the pledge.) A magic word at that.
And here is room for some middle ground between both sides of this issue.
Advocates of gay marriage can protect the word marriage and refrain from using it without corruption of principles, and the civil right of civil unions between gay couples can be restored or realized in the remaining 47.