"Hillary Hatred" and other questions !!
by SevenStrings, Tue Mar 25, 2008 at 01:49:44 PM EDT
In this diary, I will examine two questions:
(1) What effect did the YouTube Sermons and Sen. Obama's speech have on the dynamics of the race ?
(2) What drives Hillary hatred ?
As an added bonus, you can also examine the effect of the "Tuzla sniper" incident.
I will do this by examining trends in the favorability- unfavorability numbers for Sen. Obama and Sen. Clinton, as provided by the tracking polls of Rasmussen. I have diaried on this before, but now I am able to provide additional information.
(1) Effect of the YouTube Sermons, and Sen. Obama's speech
First, a discussion is in order about polling methods, use of Rasmussen vs Gallup etc., and how this analysis differs from the negligible changes in the Obama/Clinton numbers available from Rasmussen and Gallup.
In examining the effect of the YouTube Sermons, and that of Sen. Obama's speech, we should look at the trends in the raw Favorability/Unfavorability numbers, and not the Obama/Clinton comparison. Sen. Obama could be badly damaged due to the YouTube sermons, and still appear to have gained ground with respect to Sen. Clinton because Sen. Clinton could also have been damaged by other unrelated events. In fact, as I will show that both Sen. Obama and Sen. Clinton have suffered as a result of the YouTube sermons, but the short-term loss was greater for Sen. Clinton (go figure ~ this is a topic for the "What drives Hillary hatred" question).
Second, a lot of people question my use of Rasmussen. "Didn't they just say that HRC would lose to McCain by 7% in AR. That must mean that they are a crappy outfit, and one should not use their numbers". This is a somewhat fair point ~ one should not rely on their raw numbers: their methodology may mean that there exists a large and systematic difference between their raw numbers and the right value. However, we can still rely on "trends" within their numbers. They can be systematically off, but as long as their methodology is consistent from day to day, they will be systematically off by about the same amount. Thus, we can still examine trends within their numbers, provided those trends are outside the "margin of error"
And that brings us to the final point: what is their margin of error ? They do not report this for the favorability/ unfavorability numbers, so we are left on our own for this. We could infer the margin of error from the charts, but this assumes that we have an accurate enough model for the underlying trendline (in nerd-speak, the margin of error originates from the standard deviation of repeat measurements of the same entity; for accurate standard deviation estimates, you need a very large sample size). In our case, the trendline is unknown (because that is what we are trying to estimate), and two data points refer to different entities (public opinion on different days) so we do not know the margin of error. And so, when looking at my conclusions, keep that in mind and draw your own conclusions (as noted on the charts, the solid lines are meant to guide the eye, and are not fitted models).
And finally, to the important question:
Qu 1.a: Did the YouTube Sermons damage Sen. Obama ? Did his speech work ?
The graphs above shows the (favorability - unfavorability) and the strongly favorable-strongly unfavorable) numbers for Sen Obama, with the 2 events marked. You can clearly see the "trend reversal" brought upon by the "YouTube sermons", and the lack of any trend reversal brought upon by Sen. Obama's "A more perfect union" speech. This clearly proves that the YouTube sermon did "damage" Sen. Obama (I use the word "damage" loosely here ~ the damage appears to be ongoing, and has not been fully realized yet. Bleeding would be a better word!). Finally, we can also see that Sen. Obama has been unable to stop the bleeding with his speech.
Qu 1.b: If the YouTube Sermons damaged Sen. Obama, why does it not show up on the Obama/Clinton matchups ?
I was puzzled by the answer to this, which is why I refrained from commenting on this in earlier diaries. However, the evidence is now overwhelming: The YouTube sermons, or something else "damaged" Sen. Clinton even more than the damage sustained by Sen. Obama. You can see this clearly in the charts for Sen. Clinton presented in the next section. Unlike the "damage" sustained by Sen. Obama (wherein, the damage is ongoing, and is more akin to bleeding than to instantaneous damage), the damage sustained by Sen. Clinton was immediate. Perhaps people thought she was behind it ? Perhaps people blamed her for not defending Sen. Obama ? Also, Sen. Clinton appears to have weathered that damage, if today's data point is any indication. As they say, the next few days will be interesting (the next few days are always interesting).
(2) What drives "Hillary hatred" ?
Hillary hatred is unique in that it is very difficult to understand the underlying reasons for this. Consider her (like her-dislike her) and her (love her-hate her) charts depicted below.
It is clear from this chart that Hillary hatred subsided substantially when Sen. Obama appeared to have locked down the Democratic nomination (i.e., in the Feb time scale), and flared up again when it became apparent that she was fightin on to the TX/OH primaries. Prior to the February, Hillary was hated by 33% of the overall population. The chart does not show this, but the underlying reason for the apparent decrease in Hillary hatred is that about half of all Hillary haters (about 17% of the total population) went from hating Hillary to merely disliking her when it appeared that she was out of the race.
Thus, it is safe to conclude that about half of the "Hillary-hatred" is generated by an intense desire to see her lose, and not to any logical reason (examples of logical reasons: she is a liar, she cares for herself etc. etc.)
Now, consider the second example: the dates corresponding to the release of the "YouTube sermons", and Sen. Obama's speech are also marked in Sen. Clinton's charts. Notice the obvious: both events result in a marked increase in "Hillary hatred". In fact, the short-term loss for Sen. Clinton was more than the short-term loss for Sen. Obama, both as a result of the YouTube sermons, and also as a result of Sen. Obama's speech. I consider this as another example of irrational Hillary hatred.
And now, consider the last example. These past few days, it has become clear that Sen. Clinton "misspoke" about sniper fire in Tuzla. This has been caricatured as proof that she is a liar ~ you would normally expect Hillary hatred to increase. Now notice the significant increase in her (love her-hate her) numbers for today (I try to avoid commenting on data points from a single day, but in this case, the increase is fairly significant). Of course, it is only natural that Hillary hatred subsides when apparent bad news about Sen. Clinton makes the headlines ~ it makes as much sense as other facets of Hillary hatred I discussed above.
It is my opinion that Sen. Clintons (love her - hate her) chart is oscillating between monster (levels < -10%) and "just another politician" levels for no logical reason. If anyone has any insight into what is driving these numbers, I would sure love to hear them.
PS: I used Red for Sen. Clinton so as to conform the claim by Sen. Obama's supporters that she is tryign to destroy the democratic party. I am, in fact, supporting Sen. Clinton in this particular race.
As always, comments are welcome.