Party was the primary vote determinant. The date of the poll shows up as a secondary variable. This implies some voters depended on new information generated by the campaigns. Their take on this new information was likely dependent on their political perspective.
The late movement was clearly among non Republicans. This makes sense if the measure was generally seen as regulation. This trend highlights the effectiveness of the “No” campaign.
QUESTION: Now, I would like to talk to you about some measures that will be on the November ballot. If the election were held today on Measure 92: Requires food manufacturers, retailers to label “genetically engineered” foods as such; state, citizens may enforce, would you vote yes or no?
IF YES/NO: Is that strongly or somewhat?
IF DON’T KNOW: Which way do you lean?
SURVEYS: 400 live telephone interviews were conducted of likely State of Oregon voters on October 13-16 and again on October 27-30 2014. Both landlines and cell phones were included. Likelihood was determined by a model overlaid on the voter file. The margin of error at the sample median is plus or minus five percent.
TREE ANALYSIS: Standard CHAID was applied to a pooled data set constructed from the above two surveys.The dependent variable in each case is the specific ballot choice. Independent variables were the same in all analyses. These include: Poll Date, Gender, Age, Party, Congressional District, Vote Frequency, Homeownership, Union Membership, Marital Status, Religious Importance, Religious Attendance, Education, and Income.