Political reporters love polls because they often feel like "proof" that a campaign is either gaining strength or sinking fast. But even statisticians will tell you that polling is only a best guess, and even then only a snapshot in time.
This morning, the combatants in the fall initiative campaigns are faced with a new poll that seems to reach a completely different conclusion about Governor Schwarzenegger's agenda. The new data comes from Survey USA, and was collected using a process that traditional pollsters often claim to be unreliable. But more on that in a moment.
The poll concludes that all three of the initiatives Schwarzenegger helped place on the ballot are ahead (Proposition 74- tenure- 55% yes; Proposition 76- budget- 58% yes; and Proposition 77- redistricting- 59% yes).
It also finds strong support for two other initiatives that he has endorsed: Proposition 73 (abortion- 59% yes) and Proposition 75 (union dues- 60% yes).
Survey USA gathers its data through automated phone calls. The phone rings, you hear a recorded voice, and you push the number that corresponds with your opinion. But suppose it's a 10-year-old who answers? The automated call doesn't have a filter for such errors... which is why many traditional pollsters say a live person asking the questions is more accurate.
Survey USA's data stands in contrast to the most recent statewide poll from the Public Policy Institute of California, which found much smaller support for Prop 74 (43%), Prop 76 (26%) and Prop 77 (33%).
How do you explain the difference? Perhaps opinions are shifting. Team Arnold's strategists certainly believe so, and are already pointing out to reporters that Survey USA conducted its poll two weeks after the PPIC, and after the governor's TV ads debuted.
But some of it could be methodology, and not just the issue of automated phone calls. First, there's the margin of error.
Let's examine the two polls in regards to Prop 74, which would extend a teacher's probation period. PPIC shows it as 43% yes, 47% no... with a margin of error of 3%. So if you're a Prop 74 fan, you can take heart that the correct results might actually be 46% yes, 44% no.
Survey USA shows Prop 74 at 55% yes, 44% no... with a margin of error of 4.3%. If you're a Prop 74 opponent, the margin of error this time could mean the correct results are 50.7% yes, 48.3% no.
Taking all of that into consideration, the two polls show a similar gap in opinions. The discrepancy between numbers for the other initiatives, however, is much larger.
The questions themselves should also be examined when looking for answers to why the two polls are so different.
Take Prop 76, the budget spending cap initiative. Survey USA's question was one simple sentence: "Proposition 76 limits growth in state spending so that it does not exceed recent growth in state revenues."
The PPIC question, on the other hand, was six sentences long, and included the controversial fact that the initiative could, under certain circumstances, lead to a reduction in funding for schools.
Is it possible that people who chose "yes" on Prop 76 in the Survey USA poll would have chosen "no" if they had heard the part about school funding?
Maybe, maybe not. But all of this is proof that while we reporters like them, no single poll is ever the final word about the mood of the electorate. The results of that poll come out on Election Night.