What's likely for D-Day
March 4 is tomorrow. This is D-day according to Governor Bill Richardson. We don't have to read tea leaves to assert that it is the end of the road for one candidate. So, why are journalists hedging their bets and parroting the campaign spin that Clinton is still viable in a post-D-Day race? My judgement is that we're just biased towards believing in extremely rare outcomes. Take the example of how we get anxious about flying even though the chance of a catastrophic event is next to nil. Clinton threatening Obama's lead is about as likely as any of us perishing in a plane crash. Can Clinton win the big states on D-Day? Absolutely, but any likely outcome based on all the evidence still leads to an Obama nomination.
On to the data... Take a look at the first graph below which shows the trend based on polls conducted in Ohio from 1/31/08 through today, 3/3/08. Any individual poll may over- or under-estimate the true percentages of support due to random variability or bias in measuring the public's preferences. Think of taking your blood pressure at any given moment in order to ascertain your optimal blood pressure (variability) or of using an arm cuff that always reads too low or too high (bias). We have to take individual polls at a particular time point with a grain of salt. We'd have greater confidence in our conclusions if we looked at all the polls together. The first graph shows Ohio's trend. The solid black line is the trend line of support for Hillary Clinton. We can be pretty sure the true trend line that we're trying to measure lies somewhere between the two thin black lines. Barack Obama's trend is colored red. As of the day before the election, our best conclusion about Ohio's blood pressure is that Clinton has 50% support to Obama's 43%. This split could widen, but not enough for Clinton to overcome her delegate deficit. We could just as easily see a narrowing of the split tomorrow.
In Texas Obama is tied just as of today with a steady 2 point lead for a period. What would it take to break this stability by tomorrow? I can't think of anything.
We're not going to crash on this flight. We're going to have a normal, albeit choppy, landing, and any landing you can walk away from is a good landing. So, relax.
Actual Data
Nothing beats being able to study the actual data yourself. Here are the individual polls for Ohio. There were 7 pollsters that had more than one poll over this period, so we can see how their trends look. The remaining pollsters only provide information about a particular time point.
And here's the actual data for Texas.
On to the data... Take a look at the first graph below which shows the trend based on polls conducted in Ohio from 1/31/08 through today, 3/3/08. Any individual poll may over- or under-estimate the true percentages of support due to random variability or bias in measuring the public's preferences. Think of taking your blood pressure at any given moment in order to ascertain your optimal blood pressure (variability) or of using an arm cuff that always reads too low or too high (bias). We have to take individual polls at a particular time point with a grain of salt. We'd have greater confidence in our conclusions if we looked at all the polls together. The first graph shows Ohio's trend. The solid black line is the trend line of support for Hillary Clinton. We can be pretty sure the true trend line that we're trying to measure lies somewhere between the two thin black lines. Barack Obama's trend is colored red. As of the day before the election, our best conclusion about Ohio's blood pressure is that Clinton has 50% support to Obama's 43%. This split could widen, but not enough for Clinton to overcome her delegate deficit. We could just as easily see a narrowing of the split tomorrow.
In Texas Obama is tied just as of today with a steady 2 point lead for a period. What would it take to break this stability by tomorrow? I can't think of anything.
We're not going to crash on this flight. We're going to have a normal, albeit choppy, landing, and any landing you can walk away from is a good landing. So, relax.
************************
Actual Data
Nothing beats being able to study the actual data yourself. Here are the individual polls for Ohio. There were 7 pollsters that had more than one poll over this period, so we can see how their trends look. The remaining pollsters only provide information about a particular time point.
And here's the actual data for Texas.
Comments
Post a Comment