How did the polls measure up to what the voters actually said on election day on ballot issues?
The bottom line is that the 9News/Survey USA polls were far less accurate than those conducted by the Rocky Mountain News.
This may have to do with a Survey USA methodology of asking voter’s to state only if they were completely certain that they would vote for a measure, instead of their usual survey question format, which is “If you were in the ballot booth right now, how would you vote on . . . . ?”
Here is some the the final pre-election polling on ballot issues and the actual results. The actual results are in bold:
Amendment 38 (Petitions):
RMN I 39-32; 31-69
This measure was unpopular even in early polling, and unanimous newspaper editorial opposition to the measure was probably a factor in the continuing decline in support for it over time.
Amendment 39 (Education Spending):
RMN I 58-25; 38-62
The dramatic shift from the support Amendment 39 received in early polling shows that the unanimous newspaper editorial opposition and news coverage of Amendment 39 clearly had an impact.
Once voters understood Amendment 39, they didn’t like it.
Amendment 40 (Judicial Term Limits):
RMN II 38-54; 43-57
The Rocky‘s predicted margin of defeat for this measure was within its margin of error. Undecided voters favored the measure more than they opposed it.
Amendment 41 (Gifts To Goverment Officials):
RMN II 60-30; 62-38
The Rocky was well within the margin of error, and as is usually the case, the undecided vote went strongly against the measure.
Amendment 42 (Minimum Wage):
RMN II 53-42; Survey USA 44-36; 53-47
Survey USA got the margin of victory right, but the absolute numbers far oof. The Rocky was on the money in prediction for vote in favor, and as one would usually expect, the undecided vote went strongly against the measure.
Amendment 43 (Marriage Definition):
RMN II 53-43; Survey USA 40-41; 56-44
Survey USA blew this one. The Rocky came within ints margin of error. Unusually, undecided voters swang in favor of, rather than against, this ballot measure.
Amendment 44 (Marijuana Possession):
RMN II 34-61; Survey USA 36-51; 41-59
Amendment 44 (Marijuana possession) over performed polls, suggesting that proponents belief that young cell phone users were undersampled, or alternately that people didn’t answer honestly, but not enough to make a difference in passage or failure.
Referendum I (Domestic Partnerships):
RMN II 47-49; 42-58
The Rocky overstated support for this measure. Whether this is a product of that fact that many Denver votes (not enough to make a difference in the outcome) have not yet been counted, or some other reason, is still unclear.
This is a measure where unanimous newspaper endorsements across the state didn’t seem to have much of an effect.
Referendum J (Education Spending):
RMN I 66-21; 42-58
Support for both Referendum J and Amendment 39 declined dramatically over time. The six percentage point edge J had over 39, winnowed to 4 points in the final result, proportionate to the decline in overeall support for the measures.
Only about 10% of Referendum J supporters didn’t also support Amendment 39.
Referendum K (Immigration Lawsuit v. Feds):
Survey USA 26-26; 56-44
Survey USA blew it here, again. Who says that Coloradans are opposed to frivilous lawsuits?