Support for the Death Penalty and Question Wording

One great value of long term data archives for polling information is the ability to compare trends over long spans of time and across different question wording. The Roper Center at the University of Connecticut provides one of these archives. Its subscription only iPoll database includes over 500,000 questions that pollsters have asked since 1935. This is a uniquely valuable resource for opinion research.

Public support for the death penalty over time is a good example of the power of this collection. The plot above looks at death penalty support and opposition since 1936 in Gallup polls. Gallup has maintained a constant question wording over that span giving us the best long term look at this issue. Gallup has a summary of these data on a no-subscription required page here. It provides more data than I look at here.

My point today is two-fold. First, just to look at the trends over time. The standard Gallup question plotted above is

Are you in favor of the death penalty for a person convicted of murder?

Support for the death penalty declined to a low in 1960-75, with support around 50% and opposition around 40% (with some bouncing around a bit.) Following 1975 there was a long growth in support for the death penalty, with support rising to as high as 80% and opposition falling into the teens.

Since the decline in homicide rates in the 1990s, support has declined by about 10 percentage points since 2000, to around 70%, with opposition rising to the mid-20s.

The public overwhelmingly supports "the death penalty for a person convicted of murder."

But now let's consider what happens when we offer a specific alternative to the death penalty. In the classic Gallup question, there is no explicit alternative to the death penalty, so respondents must imagine for themselves what the alternative might be.

Since 1985, Gallup has also asked an alternative question which includes a specific alternative sentence:

If you could choose between the following two approaches, which do you think is the better penalty for murder--the death penalty or life imprisonment, with absolutely no possibility of parole?

This reflects a shift in the policy debate, as no-parole sentences were raised as an alternative to execution.

Presented with this alternative, opinion shifts substantially:
























With the "life without parole" option, support for execution falls to just over 50%, while support for life terms rises to the mid 40s. Still a majority in favor of the death penalty, but a substantially more closely divided public than with the classic question.

Two points from this. First, questions define the alternatives respondents are encouraged to consider. The classic question offers one option and leaves it to the respondent to imagine others. The death vs life introduces an explicit alternative, and finds much more support for a penalty short of death. However, we could imagine a question with a third option: life but with a possible parole. Or even specific sentence lengths (10 years? 20? 50?). In part the question reflects the policy debate. But survey questions must also necessarily limit the range of options under consideration. In this, there can be no escape from question wording effects, and no end to argument about whether questions "really" capture all the issues. The fact is every question includes some and excludes other options.

The second point is the political one. If one opposed the death penalty, one might be more likely to find support by arguing forcefully for hard time and no parole rather than arguing against the death penalty in principle.

Washington Scandals and Baby Names

Updated 5/26/07: Added Technical Appendix
























The appearance this week of Monica Goodling before the House Judiciary committee sparked a conversation in the Political Arithmetik household about a previous Monica related Washington scandal. It perhaps says something about our household that this provoked a search for empirical evidence concerning the effect of the Clinton-Lewinsky scandal on the popularity of Monica as a name. Was it urban legend that the scandal had an effect? Was the effect large or small? Was it immediate? Let's run the numbers.

Monica was a reasonably popular name in the early 1970s, ranking between 39th and 56th in the decade of the 1970s. As it happens both Monica Lewinsky and Monica Goodling were born in the summer of 1973, two weeks apart, when the name was ranked 40th, its second highest ranking. (Monica ranked between 59 and 141 in the decade of the 1960s.) [My thanks to my colleagues at the coffee shop for suggesting I check tennis player Monica Seles, who turns out to also be a 1973 baby. Granted, she isn't connected to a DC scandal despite being born in 1973, and being born in the former Yugoslavia makes the relevance to our current investigation a tad suspect.]

If we were going to pick a name to go with a DC scandal from babies born in 1973, better bets would have been Jennifer, Amy, Michelle, Kimberly, Lisa, Melissa, Angela, Heather, Stephanie or Rebecca, the top 10 girls names that year. But Monica at 40th wasn't rare by any means.

The 1970s were the peak years for Monicas. By the 1990s the name had slowly but steadily declined to rank between 76th and 88th during 1990-1997.

And then the events of 1998 intervened. The Clinton-Lewinsky scandal broke on January 21, 1998, reached its fevered peak by the end of 1998 with the impeachment of President Clinton and was resolved by the Senate's failure to convict on February 12, 1999. Of course that didn't prevent late night comics from continuing to milk the material for months, years, perhaps forever after.

The impact on parents was immediate, but not as drastic as I had expected. There were 11 months of 1998 in which the scandal's impact could be felt. And the ranking of Monica dropped from 79 in 1997 to 105 in 1998, a substantial but not precipitous drop. Of course events were unfolding during this year, so perhaps it is reasonable to focus on 1999, by which time surely every expectant parent in America would be aware of the Clinton scandal.

And in 1999 the ranking of Monica did fall dramatically, to 151, just a bit below where it stood in 1960.

So indeed, the impact of the scandal produced an immediate and substantial response, as one would surely expect. No urban legend this.

But what I find fascinating is the continued decline since 1999. I would expect the impact to be greatest in the immediate aftermath of the infamous episode and to level off or perhaps even abate thereafter. Instead, the data suggest a much slower response and a much longer diffusion of unpopularity through the population. Having dropped 72 places between 1997 and 1999, the popularity of Monica dropped ANOTHER 99 places from 1999 through 2006, the last year for which we have data, to now stand at the 250th name on the popularity list.

One interesting speculation is to consider the effect of the Clinton-Lewinsky scandal on the parents who are just now having baby girls. Many of them would have been in their teens or early 20s during the height of the scandal, compared to parents of 1999 or 2000 who would have been on average 7 or 8 years older. I wonder if the impact of the scandal was larger on teenage and college age parents to be. These are ages not noted for consumption of political news, but they are ages extremely well known for crude sexual humor, for which Kenneth Starr provided an abundant supply of raw material. So I wonder if this cohort that is now giving birth was somewhat more affected by the scandal than were even slightly older cohorts who were past the age of campus humor as well as early sexual development. That could explain the continued and steady decline in use of Monica as a girl's name. It would also predict a leveling off once cohorts start to dominate births who were too young to understand the Clinton-Lewinsky scandal at the time.

The alternative is a slow diffusion of unpopularity throughout the culture, which is having an increasing effect regardless of personal experience with the scandal. If so, there is little reason to expect a leveling off of ranking. But there is also a puzzle about why the cultural diffusion is as slow as it has been.

It seems unlikely that Monica Goodling's testimony will significantly reduce the already declining popularity of the name. But given the current standing of "Monica", it is much less likely that a DC scandal in 2035 or so will feature a Monica in the staring role.

Prospective parents may want to visit the source of these data, the Social Security Administration's Popular Baby Names site here.

A superb academic study of the sociology of naming babies is A Matter of Taste: How Names, Fashion and Culture Change, by Stanley Lieberson.


Technical Appendix (added 5/26/07)

Warning: This is the really geeky part. Unless you think log2(x) is really cool, you might want to turn back now!

"Professor M" posted a comment on the cross post of this article at Pollster.com. Rather than "geek up" Pollster, I'm replying here. (This was supposed to be a "just for fun" post, after all.)

His/her comment is:

Hmmm. Try graphing the percent of babies given the name Monica in each year instead of the popularity rank. I think your discussion might change.

The good Professor M makes an excellent point. Let's think why. The actual rate of name use is quite small, even for the most popular names. For example, in 2006 the most popular name for girls was Emily. That name was used for 1.0267% of girls born. The number 2 name was Emma, 0.9159%. This difference in percentages is actually rather large. When we get down to ranks 101 and 102 we find Mya at 0.1602% and Amanda at 0.1599%. When we get down to Monica at 250, the rate is 0.0650% and for Carly at 251 the rate is 0.0649%.

So the rate of name use gets closer together for adjacent ranks as we go from more popular to less popular ranks. In my plots above, a change of one rank is the same vertical distance in the plot whether we are going from 1 to 2 or 100 to 101 or 250 to 251. But the percentage rates would not be changing by the same amount for each of those ranks. Instead, the difference in percentage rate would be getting smaller as we go from more popular to less popular rankings. In techie terms, the relationship between rank and percentage use is non-linear. And that can produce a different look to the plot, as Professor M suggests. So let's take a look.

I've converted the percentages into rate per 10,000 girls born, just to avoid the decimal points. That makes no difference for the look. So let's look at what Professor M suggests:























And behold! As Professor M suggested, the look is a bit different. What appears as a continued sharp drop after 1999 in my plot of rankings, now looks more like a continued decline but not so sharp, and much more of the decline came between 1997 and 1999. Also, the declining popularity of Monica between 1973 and 1997 appears more substantial, dropping from 41 per 10000 to 22 per 10000.

So Professor M's point is well taken. The change in rates are significantly different from the change in ranks. The popularity of Monica has continued to decline since 1999 but not nearly so dramatically as it appears in my ranking graph.

But...

Is the raw percentage (or per 10,000) rate the right measure either? As the rate approaches zero, it becomes impossible to decrease by a constant amount. From1973 to 1997, the rate of use of Monica fell from 41.0 to 22.1 per 10,000, a decline of 18.9. But in 1999 the rate was 10.96 per 10,000. It would be impossible for that to decline by another 18.9, lest we end up with a negative rate of name use! The point is, a constant change in the raw rate is impossible as we approach low incidence of the name. So perhaps linear change in the rate is also not a good way to model this.

An alternative is to think of the "half life" of the name use. This equates a fall of 1/2 from say 40 to 20 per 10000 with an equivalent proportionate change from 10 to 5 per 10000. This makes proportionate declines equal across the entire range of name rates. In effect, this says a fall of 1/2 in usage rate is the same wherever it occurs.

A simple way to measure this is to use the log base 2 of the rate per 10000. In base 2, each unit increase on the log2 scale is a doubling of the rate. So 1=log2(2), 2=log2(4), 3=log2(8), 4=log2(16), 5=log2(32) and 6=log2(64). Those values cover the range of Monica rates, and the critical point is that each 1 unit increase is a doubling and each 1 unit decrease is a halving of the rate of use.

Replotting the data on this log2(rate per 10000) scale produces the following:
























Now we see that from 1973 to 1997 the log2 rate fell from 5.4 to 4.5, or almost a full unit, representing a halving of the rate. From 1997 to 1999 it fell from 4.5 to 3.5, another halving. And from 1999 to 2006 from 3.5 to 2.7, a bit less than half again.

On this scale of proportionate change then, the drop from 1997 to 1999 is huge, a full halving of the rate (from 22.1 per 10,000 to 10.96) in just 2 years. The subsequent decline from 10.96 to 6.50 is a 41% decrease in rate over 7 years.

Now this plot is not identical to my ranking plot, but it is pretty close. The qualitative description in my original post applies pretty well to this one as it did to the ranking plot. So I stand by my original comments.

I had not looked at these issues before Professor M's comment, so I am very grateful to him/her for pointing this out. And indeed, as we saw above, the raw rates do look somewhat different. But on reflection, prompted by that comment, I think the log2 rate is probably the most reasonable way to look at this. The ranks alone can be misleading because the equal intervals between ranks distort the changes in rate. But the raw rates are also misleading because changes cannot remain constant when there is a lower limit of zero usage which we approach. Proportionate change seems more compelling in this case, and log2 is a convenient and easy to understand approach to this.

And one last technical point. The plot of rate against rank is strongly non-linear, as Professor M implies. The plot of log2(rate) against rank is much closer to linear, though with some continued bend. This is why my final log2 plot above more closely resembles the rank plot. Since log2 rate is close to linear with rank, the two plots must look quite similar.

Bush Approval: CBS at 30%, Trend at 32.2%

The new CBS News/New York Times poll, taken 5/18-23/07, finds approval of President Bush at 30%, disapproval at 63%. The April 20-24 CBS poll had approval at 30%.

With the addition of the new poll my approval trend falls to 32.2%. The blue trend line has taken a sharp dip over the past 6 or so polls and now stands right at the lower edge of the uncertainty around the trend of the past five months.

Since January we've seen a couple of periods in which the trend dipped below 33%, only to quickly revert back up to the 33-35% range. Inspecting the figures below casts a bit more light on the current polling.

First, the more sensitive "Ready Red" estimator pegs approval at 31.2%, still about a point below "Old Blue's" take. Still not much divergence between the two estimators.
























Of the last six polls, three usually fall a little above the trend estimate (Hotline, Fox and Gallup) while three usually fall below (CBS, ARG and AP). So the current reading isn't entirely due to a string of polls all of which usually have negative "house effects".
























Likewise, the residuals are pretty well behaved right now. While the Newsweek outlier sets the recent low, it is largely offset by the high (but not outlier) CNN result. The remaining 8 recent polls are rather close to the trend estimate.
























Finally, the range of uncertainty around the trend line, reflecting how much the trend can jump around as different polls come or go, is rather wide for the current estimate, suggesting a range of uncertainty that includes 30% but also 35%. That will shrink as new polls come in.

The bottom line: The model is not yet unambiguously insisting on a new downturn in approval. And it would be well to remember that we've seen this kind of a dip more than once this spring, only to quickly see a return to the recent equilibrium. So before declaring that decline is a certainty, we should remember that such a prediction has been wrong recently.

There has been some analysis that the immigration bill will turn Republicans against Bush, resulting in inevitable decline as his one remaining support group fails him. I think that is certainly possible, but must point out that it is exactly what did NOT happen last May. Then Bush had been in steady and strong decline since February. That trend hit a low estimate of 33.98% on May 12. On May 15 the president gave a televised address to the nation in support of immigration reform. His approval rating immediately turned up and was followed by a quite good summer (compared to his previous lows.) One can argue that the immigration issue has changed within the Republican party since that time and is now poison. Or one can argue that Bush has shown little recent leadership on immigration, and so will get no benefit from the issue. But I think immigration is a good deal more complicated than the confident assumption that the issue must be poison to the president's approval rating.

I'd also point out that opinion on Iraq remains strongly against the president even as he and his administration have stressed the negative implications of deadlines and benchmarks for Iraq funding. Often presidents have been effective in arguing that Congress cannot be commander in chief, or that any funding constraints must undermine the troops. Clearly the White House has expected those messages to ultimately move the public in their direction. So far there is little evidence of that.

So it may just be the fundamentals that are driving approval at this point. And the single most fundamental right now is Iraq. By most measures, the public has remained unsympathetic to Bush's recent arguments.

Pres08: National Primary Sensitivity Comparison

Updated: 8/7/07















































This post compares the estimates of presidential national primary support using the "stable and a bit stodgy" "Old Blue" estimator to the "sensitive and a bit gullible" "Ready Red" estimator.

For a full description of these see this post. Briefly, the Blue estimator is slow to react to just a few polls which suggest a change in trend but which may in fact just be random noise. The Red estimator is more sensitive to new polls and is therefore quicker to pick up change but also more likely to mistake random noise for new trend. When the two overlap, only the blue estimate is visible in the plots. In the long run, the two agree quite well. Put more trust in Blue if you prefer not to make mistakes. Red will alert you to changes quickly, but will also "cry wolf" more often.

This page is updated in place, so come back here for future updates. A link to this page appears in the right column's thumbnails.

Bush Approval: Hotline at 32%, Trend at 33.0%

A new Diageo/Hotline poll collected 5/16-20/07 of 800 registered voters finds approval at 32%, disapproval at 64%. The Hotline poll usually runs a little above the trend estimate, but in this case is a point under trend, which now stands at 33.0%.

Over the last several polls, approval has moved down a bit. It remains in the 33-35% range we've seen since January, but only just barely.

Since we've been talking about "Ready Red", the sensitive trend estimator in comparison to "Old Blue", the more conservative estimator I use here, I've also run a graph comparing the two. Red is willing to go a point lower in approval than Blue. But notice how well the two trends usually agree once all the data are in. If you can't see the red line below, that's because Blue is on top of it-- i.e. they agree. And usually they agree because Red eventually sees that Blue was right. HOWEVER, when approval or other trends DO change direction suddenly, it is Red that eventually convinces Blue to move along. At the moment, the difference is quite small and I'd say there is little additional evidence from Red at this point.

I've been impressed by the stability of approval for the past five months. I still am. I'll need to see four or five more polls all clearly below trend before I'm convinced that the current dip is real, and not just another false lead from a few polls.

The diagnostics below show no alarming indicators of anything, so nothing more to say.

Pres08: Short-term Giuliani-McCain Trends

I usually focus on the full set of polls available for an analysis. But sometimes it pays to zoom in on only some recent trends. In the case of the Republican presidential nomination battle, zooming in clarifies things a bit.

I've written about this, and compared my more sensitive estimator ("Ready Red") with my more conservative one ("Old Blue") in several places (here, here, and here.)

While I usually prefer the more conservative estimator because it is harder to fool it with a few polls, in this case it is clear that the Giuliani trend has taken a downturn since March. The evidence for a McCain upturn is much less compelling, but the red estimator thinks it sees a little bit of one.

In the plot above, it is clear that the sensitive red estimator follows the data rather well since early November when presidential nomination polls became frequent. The Giuliani trend clearly illustrates his substantial run-up but also his subsequent decline. The trend estimator now stands a point below where Giuliani began in November.

McCain, on the other hand, was in decline from December until about April 1. Since then he appears to have gained a couple of points.

This means that a substantial gap that Giuliani opened on McCain in the first quarter of the year has now largely disappeared. Giuliani has been ahead of McCain in the vast majority of polls, including 54 of the 56 polls taken since November 1 (see here as well). But that gap is narrowing.

Now what?

Bush Approval: ARG at 31%, Trend at 33.4%

A new American Research Group (ARG) poll taken 5/18-21/07 has approval of President Bush at 31% and disapproval at 64%. With this addition, the approval trend estimate is at 33.4%.

The recent downward wobble is still unconvincing as an indication of a shift in trend. We've seen exactly this kind of wobble for nearly five months now, with down wobbled converted to up wobbles and back down. So for the moment, pending a good deal more data, I continue to believe that approval remains in the 33-35% range we've seen since January.

This is now the longest period of stable approval we've seen in the entire Bush presidency. Evidently the Republican President and Democratic Congress are stuck in an equilibrium with each other-- unable to convince voters to shift their views of the president one way or the other. Events may eventually disrupt that equilibrium, but so far the Iraq debate and Department of Justice hearings have failed to do so, as has the White House critique of Democratic "surrender dates" and defense of Attorney General Gonzalez.

Curious. Very curious.

Pollster.com is a winner

From our sister site, Pollster.com. This is my partner Mark Blumenthal's post, hence he is the "I" in this. He speaks entirely for me in expressing our appreciation of this honor from AAPOR and our gratitude to our readers at Pollster (and I'd add, here at Political Arithmetik.)



Pollster.com Wins AAPOR's Mitofsky Innovator Award

I have the pleasure of sharing some very good news: On Saturday night, Charles Franklin and I had the very high honor of being named winners of 2007 Warren J. Mitofsky Innovator's Award by the American Association for Public Opinion Research (AAPOR) for our work on Pollster.com.

The award, first presented eight years ago, recognizes accomplishments in public opinion and survey research that occurred or had their impact during the last decade. This year, AAPOR renamed the award to honor late Warren J. Mitofsky, the great survey innovator who, as several speakers noted, would probably have won this award many times had it existed earlier in his career. Past winners include some of the most distinguished individuals in the field including Andrew Kohut (for the Pew Research Center) and Professor Robert Groves (for his leadership in establishing survey research as an academic discipline). The fact that AAPOR chose to grant this award to a "Weblog" says something very humbling about our efforts as well as the growing influence of the blogosphere. It is a huge honor.

This year, we were co-winners along with Arthur Lupia and Diana Mutz for their work on the project known as TESS (Time-Sharing Experiments for the Social Sciences).

Professor Franklin and I want to express our gratitude to Doug Rivers (a previous winner of this award) and his company Polimetrix for their continuing financial and technical support, to the many AAPOR members who provide advice and respond to our queries about their work and to our assistant Eric Dienstfrey who does much of the real work that makes Pollster.com possible. And finally, we thank you, our readers, for your continuing support and confidence.

The full text of the award citation appears after the jump.

The American Association for Public Opinion Research

Presents the

2007 WARREN J. MITOFSKY INNOVATORS AWARD

To

MARK BLUMENTHAL

and

CHARLES FRANKLIN

For POLLSTER.COM

Modern day public opinion polling in the United States has come a long way from the early Gallup, Roper, and Crossley in-person surveys of the 1930s to the proliferation of polls conducted by telephone beginning in the 1970s, through today's movement toward increasing use of the Internet. The vast numbers of surveys today include not only those conducted by public pollsters in the above tradition but also by mass media organizations, academic survey centers, interest groups, political parties and consultants, and others. This expansion of polls and the reporting of them has resulted in an overload of often confusing information about public opinion during a time in which partisan conflict has grown and led to highly contentious and increasingly strident debates about what policies the public supports and which political candidates members of the electorate prefer as elections near-and, in the case of the recent controversies about exit polls-after they have voted.

Pollster.com is a Weblog that has provided an extraordinarily well-informed and critical forum for understanding contemporary public opinion research and poll results. Pollster.com's reach has extended beyond the survey research community to inform and educate in an accessible way those who visit and communicate with its website. Originally known as MysteryPollster.com, Pollster.com provides commentary on survey methodology and the interpretation and reporting of survey findings. It has extended existing, and developed its own, graphical methods for comparing poll results and for summarizing multiple polls and opinion trends. Its reports and commentaries provide transparency, as the data dictate, for multiple and conflicting interpretations.

With funding and technical support from Polimetrix, Pollster.com, provides a non-partisan source for the latest polling results, for state-of-the-art information about survey methodology, and for the latest debates and conflicts in the world of polling. AAPOR salutes Blumenthal and Franklin for this innovation in the fields of public opinion and survey research.

Presented at the 62nd Annual Conference
Anaheim, California
May 2007

Pres08: A Sensitive Update for National Primary

No time to write about this just now. The red "sensitive" trend estimator mostly agrees with conservative and stable "old blue". But the Clinton blue line now shows a very slight downturn, reflecting the earlier decline (and current stability) of the red line. Obama continues to look nearly flat.

On the Republican side, Giuliani's recent slide remains very apparent, while McCain's recent modest upturn also continues. Note how bimodal Giuliani's estimates have become in the second graph below, emphasizing the difference between short and long term estimates. This makes it more and more likely we'll see a downturn of the blue line for Giuliani.

For a full explanation of the methods and the charts, see the earlier post:

A closer look at primary trends

Gotta run.

Bush Approval: Fox at 34%, Trend at 34.2%

The Fox poll (taken 5/15-16/07) finds approval of President Bush at 34% and disapproval at 56%. That is down from 38% in the previous Fox poll, but in line with three Fox readings in February and March, so offers little evidence for any significant change in approval.

The trend estimate stands at 34.2%, also consistent with the stable approval level of 33%-35% we've seen for some time.

The usual diagnostics are below. No big news there.

Pres08: Giuliani sinks, McCain rises

I'm becoming convinced that my hasty, sensitive and mistake prone trend estimator, "Ready Red", has got the story right this time. For some time I've been watching the Giuliani and McCain trends. Since December, McCain has steadily trended down at a nearly constant rate. Giuliani has trended up over that time, but the trend estimator has been flattening out over the last couple of months, though still rising. That is based on "Old Blue", the conservative estimated trend line in the figure above. Blue is designed to be relatively slow to change direction but hard to fool with a handful of polls that represent more noise than new trend. But in this case, it is looking more and more like the trigger happy Red estimator has in fact picked up the current trend.

Charlie Cook at the Cook Political Report has graciously passed along the results of the latest Cook/RT Strategies poll, completed 5/11-13/07. This poll finds Giuliani at 25% and McCain at 24%. Two weeks earlier, 4/27-29/07, the Cook/RT Strategies poll found Giuliani at 28% and McCain at 21%. Given the sample sizes, which are typical of current national polling, this shift falls short of statistical significance. Nonetheless, this is in line with other recent polling trends and the sensitive Red estimator picks up this trend.

Giuliani spent a long time as the surprise front runner in the Republican nomination polls, leading McCain in the vast majority of polls. (I wrote about this in March: "Giuliani Leads, Press Finally Notices" and earlier in December: "The Republican Primary Race".) This was a mystery given Giuliani's liberal positions on abortion, gun control and gay rights. For people who follow politics closely, it was hard to believe that Republican constituents could set aside these positions which have been anathema to the "base" for nearly three decades. And yet he continued to lead and McCain fell steadily.

McCain's decline was somewhat less mysterious. His sponsorship of the McCain-Feingold campaign finance legislation, his opposition to the administration on torture, and his support for immigration reform have put him at odds with many on the right in the party.

But things have been changing recently. Giuliani appears to have benefited from a great many Republicans who simply did not know where he stood on these bedrock issues of conservative Republican principles. In the last two or three months that has begun to change as Giuliani has become the focus of news coverage that has emphasized his positions. Giuliani's recent clarification of his positions, and his performance in tonight's second Republican debate, is likely to further increase awareness among Republican voters.

While polling has found, somewhat surprisingly, that many Republicans say they are willing to overlook these issues I think there has been an inevitable drag on Giuliani, and that is now showing up in the Red estimator.

At the same time, McCain has tried to restart his campaign and it appears to have at least stopped his falling support and perhaps begun to produce some gains.

I am very slow to accept the responsive Red estimator because it is easy to be fooled by it. But the red trend for Giuliani has been quite consistent for some while. And the conservative Blue trend has been responding, if more slowly, by flattening out.

So I think the issue is now whether Giuliani can reverse his recent decline. His attempt at clarification and candor on abortion in particular is a huge risk if the Republican base cannot accept those positions. Given Giuliani's clarified position, McCain appears to have started looking a bit better to Republican voters. And to credit McCain, perhaps his restarted campaign is now playing to more of his strengths than it did in December through February.

Congressional vs Presidential Approval

The Associated Press analysis of their new poll (taken 5/7-9/07) used the lede:

People think the Democratic-led Congress is doing just as dreary a job as President Bush, following four months of bitter political standoffs and little progress on Iraq and a host of domestic issues.

And a paragraph later:

The survey found only 35 percent approve of how Congress is handling its job, down 5 percentage points in a month. That gives lawmakers the same bleak approval rating as Bush, who has been mired at about that level since last fall, including his dip to a record low for the AP-Ipsos poll of 32 percent last January.

The comparison of the two series is interesting. As the top figures shows, approval of Congress has generally been rising since the Democrats took control in January. The red line is the more sensitive trend estimate. It turns slightly down at the end, while the standard trend estimate continues to rise. The slight difference is not enough to convince me that there is any downturn at this point, despite the 5 point change in the AP poll over a month. As the graphic makes clear, there is lots of noise from poll to poll and the trend estimate is much more reliable than a poll to poll comparison. By contrast, the Presidential trend in the second figure has been quite flat since January, as both blue and red lines agree.

The level of approval of Congress was very low in late 2006 and remains well short of its high points in early 1998 and following 9/11 in 2002. (See the entire Congressional approval series since 1990 here.) But Congress is rarely loved. In the 17 years since 1990, Congressional approval has risen above 50% only twice-- rising just above 50% in early 1998 before dropping sharply after the impeachment of President Clinton late in 1998, and for a few months following 9/11. For the vast majority of the time since 1990, approval of Congress has been below 45% and below 40% for much of the time.

In 1995, following the Republicans capture of control, approval stood at just over 30% and did not break 40% until mid-1997. By contrast, Presidents routinely enjoy approval over 50% and are seen as in some trouble politically when their approval falls below 50%. President Bush's lengthy record below 40% is unusually low and long in comparison to previous presidents (though a number dip briefly to the 30s.)

In this light, while approvals of 35% apiece may be numerically equal, the political implications in light of historical polling are not the same. The most obvious difference is that in 2007 Congressional approval has been rising while that of the President has been stagnant. Democrats in Congress are not enjoying very high levels of approval, but they are doing considerably better than Republican members. (See the graph below.) While Democrats lead Republicans by 8 points on approval, Republican disapproval is a whopping 21 points higher than disapproval of the Democrats.





























There is a famous question in political science: "why do voters hate Congress but love their Congressman?" The simple answer is that the institution is a convenient whipping boy for the President but also for its own members. Many members (who often enjoy personal approval levels well above 50%) run for reelection by running against Congress as an institution. The result is approval ratings of the Congress that are poor in comparison to those of the President or of its individual members. Thus it becomes a dangerous thing to make direct comparisons of Presidential and Congressional approval. The two are quite different in their norms and dynamics. It would be better to look at Congressional approval in light of its own history, and the Republican victory in the 1994 elections and in government after 1995 provides an excellent point of comparison.

Bush Approval: AP 35%, Gallup 33%, Trend 34.4%

Two new polls continue to reinforce the estimate of stable presidential approval. The AP/Ipsos poll, taken 5/7-9/07, finds approval at 35%, disapproval at 61%. A new Gallup poll taken 5/10-13/07 has approval at 33%, disapproval at 62%. With these polls, my trend estimate is 34.4% approval.

Approval has remained between 33% and 35% since January, with scattered polls above and below but little evidence of any enduring change from that range.

The current polls do not raise any reevaluation of recent polling, so I'll just present the usual diagnostics below without further comment. See my previous posts in May on approval for a more extended discussion.

Approval of Parties in Congress

Updated: 11/6/07

























Approval of Republicans and Democrats in Congress trend.
(Click image once or twice for full resolution)

This post is updated in place. Come to this page for the latest tracking of the approval of congressional parties series. The index in the right column provides a direct link from the thumbnail to this page.

Typical question wordings:

ABC/WP: "Do you approve or disapprove of the way the Democrats/Republicans in Congress are doing their job?"

Pew: "Do you approve or disapprove of the job the Democratic/Republican leaders in Congress are doing?"

Harris: "How would you rate the job Democrats/Republicans in Congress are doing: excellent, pretty good, only fair, or poor?"

Bush Approval: AP at 35%, Trend at 34.8%

A new AP/Ipsos poll taken 5/7-9/07 is out today with approval at 35% and disapproval at 61%. With this addition my approval trend estimate stands at 34.8%.

There is not much to discuss here. The approval trend estimate has remained in the 33-35% range since January, and is still there. Once or twice we've bumped into those limits, only to promptly return to the range. At the moment we see a slight upward trend, but earlier this week saw a slight DOWNward trend. So I'd say the evidence so far gives no reason to think approval is doing anything other than remaining flat, with poll to poll variation bumping the trend up or down briefly.

Everything looks pretty normal in the diagnostics below, so I'll not add any more commentary. See the earlier posts on approval this week for more analysis, all of which holds true in light of today's AP poll.

Bush Approval: NPR at 37%, Trend at 34.7%

The polls are really rolling in yesterday and today and they provide a striking contrast and hence a good lesson.

The latest to arrive is actually a bit old but only just released. The NPR poll conducted 4/26-29/07 is actually the sixth oldest of the recent polls. It found approval at 37% and disapproval at 59%. The NPR poll is the product of a bi-partisan team of Greenberg Quinlan Rosner Research (D) and Public Opinion Strategies (R). They sample likely voters, rather than adults or registered voters. This sample population mixes with house effects to generally puts the NPR poll about 1.9 percentage points above the trend estimate. In fact, the trend estimate is now 34.7 and NPR is 37%, a 2.3 point discrepancy which is reasonably close to what the house effect would predict.

The interesting comparison and lesson is that the Newsweek story over the weekend stressed a new all time low in approval of President Bush while other polling has failed to find any strong downward trend (except for a Harris poll taken 4/20-23 that also had approval at 28%.) In my trend estimator, when Newsweek was the endpoint, the trend estimator was pulled down to 33.2%. I warned then that Newsweek was an outlier and that an estimate of 34 or so was more likely. The flurry of new polls since Newsweek help reinforce that conclusion.

Here are the last twelve approval polls in order:

38 33 35 35 28 32 37 35 35 28 34 38

The mean is 34% and the median is 35%. The two polls at 28% stand out pretty strongly from the rest.

The lesson for interpreting polls is that a single poll does not a trend make. The strong interpretation Newsweek gave its poll, and the extra hype it got on Sunday talk shows, was not supported by other evidence. I stress that I'm complaining here about interpretation of poll results, and not about the poll itself. The Newsweek poll was an outlier, but outliers happen and that should not lead to a conclusion of anything other than a bad random outcome about the poll. The news story that was based on the poll, on the other hand, took the findings and made a very strong claim about approval at a new low when other evidence is clear that approval has not fallen into the 20s. One simply cannot ignore all other polls when making claims about a trend.

The ultimate irony of this is that Newsweek is right: President Bush IS at an all time low for his presidency, and has been since January! His stable approval rate of 33%-35% in my trend estimator is indeed the lowest of his presidency, and he has shown no signs of recovery (though no sign of further decline either.) But that all time low for the trend is 33-35%, not 28%. The desire to pick each poll number that happens to be well below the trend estimate and emphasize how low it is misleads us from attending to the central trend estimate which has far more to say about the actual level of presidential support. Each poll is a random draw around the trend. Some will be lower, some higher. But that poll to poll variation is largely random. It is the trend that captures the systematic path of approval.

And so I conclude as I did originally about the Newsweek poll. The best estimate of approval remains in the 33-35 percent range. At the moment the trend is at 34.7, but it could easily drop a bit with new polls, or possibly rise a little. My money is on an average of 34, for the moment.


Below are the usual poll diagnostics. I commented on these more extensively in the last two approval posts, so I'll not say more about them here.

Bush Approval: 3 New Polls, Trend at 34.3%

Three new polls put President Bush's approval at 34% or above, and bring my trend estimate to 34.3%. The Hotline poll, taken 4/26-30/07 but only released yesterday, has approval at 35%, disapproval at 62%. USAToday/Gallup, 5/4-6/07 has approval at 34%, disapproval at 63%. CNN/Opinion Research Corp., 5/4-6/07 has approval at 38% and disapproval at 61%.

These results contrast with the weekend release of a Newsweek poll at 28%. Newsweek's story on the poll stressed the new low rating for President Bush. A Harris poll some 10 days earlier had also pegged approval at 28%. However, other recent polls at 32-35% failed to offer evidence for a sharp decline in Bush's approval. With the addition of these new polls at 35%, 34% and 38%, the trend estimate has returned to "flat", continuing a long period of stable approval in the 33-35% range which has persisted since December.

The recent polls are plotted below. The CNN poll at 38% is a good bit above the trend estimate, while Gallup and Hotline are quite close. The CNN result is also a bit high compared to previous CNN polling which generally tracks the trend quite closely.
























The addition of these new polls brings the trend estimate up, making the previous Harris and Newsweek surveys look like larger outliers than before. The upward shift also keeps the rather high CNN estimate within the 95% confidence interval, meaning that while it is high it is not a statistical outlier.
























The current trend remains within the uncertainty range of the trend estimate. The current uncertainty, with polls ranging from 28-38 percent, is rather large but not suspiciously so.
























Finally, the various estimates of trend for the last 20 polls shows a strong clustering in the 34-35 percent range, so the evidence remains good for approval in this narrow range.

Partisan variation in 2007 polling

The Newsweek poll released this weekend was quickly criticized for under-representing Republican partisans. For the adult sample, Newsweek had 22% Republicans and 35% Democrats, while for registered voters their numbers were 24% and 36% for Reps and Dems. Independents were 36% of adults and 37% of registered, with 7% of adults and 3% of registered refusing or unable to answer the question.

How out of line are these results? The graph above shows the distribution of party id for all polls taken since January 1, 2007. The groups are for three populations (Adults, Registered Voters and Likely Voters) and for whether the party ID question is "leaned", that is does it include those who say they "lean" to one party or the other, after first saying they are independent or non-partisan. In the legend for the graph A.N is adults, not leaned, RV.L is Registered Voters, Leaned, and so on.

It is clear that the distribution of Republicans, Democrats and independents depends quite a bit on the population and the question wording. For the adult, not leaned, sample that Newsweek used for presidential approval, the 22% Republican they found is at the low end of samples in 2007, where the median is 25%. Among registered voters, not leaned, the median in 2007 is 29%, compared to Newsweek's 24%. For Democrats, the Newsweek sample at 35% for adults compared to a median of 35%, while among registered voters their 36% compares to a median of 35%.

So the Newsweek results are a bit low (-3%) for adult Republicans, and a bit lower (-5%) among registered voters, but quite close to the median for Democrats. While the adult Republican number is the lowest of the year in comparable samples and questions, it is only 3 points from the median. That is indicative of the relatively low spread of estimates of Republicans in the adult population this year, where half of all polls fall between 24% and 28% and only 3 of 24 polls fall above 30%. Among registered voters with unleaned questions, half of all polls fall between 27% and 32%, with four polls above 32% and six polls below 27%. Newsweek's registered voter sample is the second lowest reading for Republicans of this group.

The relationship between percent Republican and percent Democrat is not terribly strong within categories of population and question wording, as shown below.























If anything, there is a positive correlation of the two partisan groups, ranging from .44 to .81 for adults and registered voters. (The four cases of likely voters have little variation and produce negative correlations, and I discount this.) The positive correlation is a good indication that what is at work with partisanship is that some questions or survey organizations produce more or fewer independents, driving down or up both Reps and Dems simultaneously.

The variation we see above is a reminder that party identification is a central concept, but subject to quite a lot of variation in measurement depending on population and question wording (and probably "house effects" due to survey organization, a topic ignored in this post.)

Given the sample size of each partisan group, the range of variation we see across polls is not surprising, and that raises an important point. One of the first things we ask about a poll is what its partisan split is. That is an important diagnostic to have available, and Newsweek deserves to be praised for openly disclosing this bit of data. All pollsters SHOULD release this information but a number do not. The current criticism of Newsweek is exactly the disincentive these organizations have for such release. If your party id is even a bit off, it becomes an immediate target for criticism and partisan cries of "fraud". Such is an uninformed view. Normal sampling variation will produce a range of party id estimates, and the data above show that in general this variation is well behaved, doing about what we'd expect given the sample sizes involved. But it does provide a tempting target for anyone unhappy with a particular poll.

This is not to say we should not scrutinize pollsters based on their data. That is entirely what we do here, of course! And I think I was the first to point out that the latest Newsweek poll appears to be an outlier on presidential approval. But that conclusion was based on a great deal of evidence from over 1400 polls and a consistent methodology I apply to every new poll regardless of source and regardless of results.

Finally, it is easy to overstate the impact of the distribution of party id on other survey estimates, such as presidential approval or horserace results. While party id is a powerful individual level predictor of those things, a discrepancy of 3-5 percent in partisanship, as in this case, does not translate directly into a similar discrepancy in every question with partisan overtones. Partisans are not completely homogeneous in their views, so the impact of variation in party id is mitigated by the variation within partisan category in approval or candidate preference. If we want to analyze a poll's results, it is better to focus directly on the question of interest, rather than on another variable like party id which may influence but not entirely determine the result we care about.

For further analysis of variation in party id across pollsters see here, and for variation over time see here.

Iraq Opinion: War was right thing or a mistake

Updated 11/6/07

























Iraq war was right thing to do or a mistake trend. (Click image once or twice for full resolution)

This post is updated in place. Come to this page for the latest tracking of the series. The index in the right column provides a direct link from the thumbnail to this page.

The question wording varies for this item. Here are some examples:

ABC/Washington Post: "Considering everything, do you think the United States did the right thing in going to war with Iraq, or do you think it was a mistake?"

Gallup: "In view of the developments since we first sent our troops to Iraq, do you think the United States made a mistake in sending troops to Iraq, or not?"

CBS/New York Times: "Looking back, do you think the United States did the right thing in taking military action against Iraq, or should the U.S. have stayed out?"

Pew: "Do you think the U.S. made the right decision or the wrong decision in using military force against Iraq?"

Fox: "Do you think going to war with Iraq was the right thing for the United States to do or the wrong thing?"

AP: "All in all, thinking about how things have gone in Iraq since the United States went to war there in March 2003, do you think the United States made the right decision in going to war in Iraq or made a mistake in going to war in Iraq?"

Iraq Opinion: War worth cost

Updated 11/6/07

























Has the Iraq war been worth the cost trend. (Click image once or twice for full resolution)

This post is updated in place. Come to this page for the latest tracking of the series. The index in the right column provides a direct link from the thumbnail to this page.

There is a variety of question wordings for this item. Some examples are:

ABC/Washington Post: "All in all, considering the costs to the United States versus the benefits to the United States, do you think the war with Iraq was worth fighting, or not?"

Gallup: "All in all, do you think it was worth going to war in Iraq, or not?"

CBS/New York Times: "Do you think the result of the war with Iraq was worth the loss of American life and other costs of attacking Iraq, or not worth it?"

NBC/Wall Street Journal: "When it comes to the war in Iraq, do you think that removing Saddam Hussein from power was or was not worth the number of U.S. military casualties and the financial cost of the war?"

Bush Approval: Newsweek at 28%, Trend at 33.2%

A Newsweek poll taken 5/2-3/07 finds President Bush's approval rating fallen to 28%, with disapproval at 64%. This is a new all time low for Newsweek polling on Bush and ties the lowest readings of Harris polls (taken 4/20-23/07) and an older CBS poll taken 1/18-21/07. No poll has found a lower approval reading for the President.

With the addition of this Newsweek poll, my trend estimate stands at 33.2%, also an all time low.

Before concluding that Bush's approval has crashed, however, the Newsweek poll should be compared to the recent trend and to other recent polls. The last six polls have registered approval at 35, 28, 35, 32, 35 and 28. So the last six readings are three at 35, one at 32 and two at 28, a seven point range.

In the figure below, the Newsweek poll is clearly well below the current trend estimate, and more importantly, is below the typical Newsweek poll, which is normally about 2-3 points below the trend. The current poll is 5.2 points below trend.
























This means that Newsweek is a bit atypically low, compared to both other polls and to its own normal trend.

A check of the outlier analysis shows that Newsweek, like Harris 10 days earlier, is in fact an outlier when compared to the variation around the trend estimate for all polls.
























The Newsweek and Harris polls are exerting a significant pull on the trend estimate, even when balanced against the three recent polls at 35%. The current small downward slope in the trend estimate is due to the addition of Newsweek. It was flat prior to adding the new poll.

The uncertainty of the trend estimate can be seen in the gray area in the plot below. While the trend is moving a bit down since early 2007, it has remained well within the gray region, suggesting we have no reliable evidence that approval is moving either way. Since January the approval trend has sometimes moved up slightly and sometimes down slightly, but has never sustained either trend long enough to move out of the uncertainty band. The current downward slope may again be reversed with new polling (or not-- let the data speak.)
























The sensitivity analysis below shows how the last 20 trend estimates have moved around as each new poll has been added. the substantial effect of the new Newsweek poll is apparent in the gap between the current estimate and the previous cluster of estimates between 34% and 35%.

























The bottom line is that the evidence for a new decline in approval remains quite mixed, with two polls claiming a rather dramatic drop in approval, while several others suggest continued stability around 32%-35%. Given the outlier analysis and the fact that the Newsweek poll is clearly below that poll's own trend, we must still wait before concluding that approval has taken a sharp turn down. The evidence remains in favor of rough stability around 34%, though the trend estimate now stands a little below that, in deference to Newsweek's evidence.

Right Direction or Wrong Track

Updated 11/6/07
























Right direction or wrong track trend. (Click image once or twice for full resolution)

This post is updated in place. Come to this page for the latest tracking of the right direction/wrong track series. The index in the right column provides a direct link from the thumbnail to this page.

Congressional Approval

Updated 11/6/2007













Congressional approval since 1990. Click once or twice on the image for a full size view.

This post is updated in place so you can always come here for the latest data on Congressional approval. The index in the right column provides a direct link from the thumbnail to this page.

Pres08: A closer look at primary trends

Who is up and who is down in recent presidential nomination polling? What are the current trends? The question is not as trivial to answer as it might seem. If we look at different polls, we can find some bouncing up while others bounce down. Commentators often reach different conclusions because the are comparing different polls. None of the recent polls have the order of finish significantly different--- all have Clinton and Giuliani in first place and Obama and McCain in second, with Edwards and Gore together and Romney, Fred Thompson and Gingrich mixed together. But the gaps between the candidates, and who has moved up or down since the last poll varies quite a bit across polls.

The goal of my kind of analysis is to avoid the trap of focusing on only one or a couple of polls. My approach is quite skeptical of the evidence provided by any single poll, but quite confident in the information from all the polls taken together. The problem is how to combine the polls to get a good estimate of what is "really" happening, and not to be deceived by the random variation from poll to poll. So let's see how the presidential nomination races are shaping up when we take all the polls seriously.

Regular readers know that my "standard" trend estimate is the blue line in the charts. This is a line that is calculated to go through the "middle" of the data, with an average error of zero, meaning the points below the line balance the points above the line. "Old Blue", as I affectionately call this line, is deliberately conservative in the sense that it takes quite a bit of new polling data to convince it to change trend direction. The reason for this is that we know there is quite a bit of noise in the polls (just look at the spread of points around the line!) so when a new poll comes in high it might mean an upturn in support, but it is just as likely that it simply reflects random noise and the next poll is as likely to come in low. If we allow the trend estimate to chase each new data point too much, we'll just plot random noise rather than the best estimate of the trend in support. Experience with these and other data (such as presidential approval) has shown that Old Blue is seldom misled about new trends, though it does take a while (about a dozen polls) to notice changing trends.

While it is good to avoid responding too much to a single poll, it is also true that Old Blue may stick to a trend longer than it should. A more sensitive estimator would notice a change in direction quicker, and would jump on the new trend while it is still news-- and before others notice it. "Ready Red" is the answer to this. The red line in the charts is twice as sensitive to change as is Old Blue. As a result it will pick up changes in momentum more quickly, letting us spot new trends early. Unfortunately, it will also sometimes be misled and will think it sees a new trend when in fact none exists-- just a few polls that happen to be "down" or "up" but which really don't represent any significant shift.

Of course you can adjust the sensitivity of the trend estimator to anything between Ready Red and Old Blue (or outside them too, for that matter) to see how much difference the sensitivity makes. There is no perfect way to choose a "best" estimator. I've settled on the more conservative Blue estimator as my standard because I find the hasty red estimator has often jumped the gun on presidential approval trends, which more data has subsequently shown were not really changing. But that doesn't mean we shouldn't examine the more sensitive trend estimate-- it tells us a lot, even if we have to be a bit cautious. (I'd rather be right but slow. Others prefer to be quick, and adjust to mistakes as necessary. Both approaches have their virtues.)

The chart above shows the data and the trend estimates for the top Democrats. Republicans are in the chart below. In addition to Old Blue and Ready Red, there are a number of gray trend lines (81 in each figure). These show the estimated trends for levels of sensitivity from quite a bit MORE sensitive than Ready Red to MORE conservative than Old Blue. If you can see the gray lines, this means that at least for some levels of sensitivity the estimated trend differs from either Blue or Red. If you cannot see the gray lines, or only barely, this means that the estimated trend hardly depends on the amount of sensitivity and the many gray lines all lie under Red or Blue in the plot, and so are covered up. This typically happens when there is a smooth, steady trend with no bends in it.

So enough statistics, let's look at the politics.

In the Democratic race, the Old Blue estimator says that Clinton has been flat since January, after a bit of decline in 2006 and a slight rebound late in 2006. Not much action.

But if we look at Ready Red, the Clinton campaign appears to be falling off in recent polling, declining by about 3 or 4 points since her peak in January. That isn't a large drop, but it does suggest that the stable picture of Old Blue may be masking some short term decay.

The Obama campaign is similarly interesting in comparison of the two trend estimates. Old Blue sees a sharp rise in early 2007 with a slower but still upward trend recently. Red sees more of a plateau in recent polls, with some indecisive and quite small up and down bounces. If I believe Red, I say Obama has stalled. If I believe Blue, I say he has slowed but is still moving up a little. (If I'm really crazy, I say the last little uptick in Ready Red suggests Obama is about to move up again, but that would be giving an awful lot of weight to the very last polls on a sensitive estimator. I'm not that crazy.)

The two trends are in pretty close agreement for Gore, but with Red suggesting a slight downturn at the very end, while Blue says the trend remains up a bit. Again, the difference is driven by only the polls at the very end and I'm not willing to bet much on them.

The Edwards campaign could take heart in Red's somewhat higher rate of climb in support compared to Blue. Both agree Edwards has been moving up, but Red sees the upturn as sharper and ending at a higher level. The best that can be said here for Red is that this trend has been supported by more polls than is the last little change of Obama or for Gore.

As for the numbers, the estimates are not far apart regardless of which estimator we pick.

Clinton: 35.7 (Blue)/34.1 (Red)
Obama: 24.0/25.1
Edwards: 15.1/16.6
Gore: 15.5/14.0

On the Republican side, Giuliani has enjoyed a long and sustained rise based on Old Blue, but suffers a recent downturn if Ready Red is to be believed. If the sensitive estimate is right, there has been over a five point decline in Giuliani's recent standing since early February. If Blue is right, then don't be hasty and Giuliani has continued to gain, though at a slower rate than in late 2006.
























Red and Blue agree that 2007 has been a bad time for the McCain campaign. After a flat 2006, McCain has dropped over five points in both estimators. Sensitive Red thinks there may be a chance of a recent reversal of that slide, but Old Blue remains entirely unconvinced that McCain's fortunes are improving.

Blue and Red also agree that Gingrich has suffered a bit of a recent decline (more or less coinciding with talk of a possible Fred Thompson candidacy.) This is a nice example of both estimators reaching the same conclusion, even with late trends. Red sees Gingrich slightly worse off than does Blue, but the difference is slight.

Likewise, both estimators are largely in agreement that Romney has sustained his tortoise-like slow but steady increase. Despite some campaign gaffes, both trends remain up, with Red being a little more bullish than Blue.

Fred Thompson lacks enough data to provide a fair assessment of the estimators, but who can ignore him at this point. The only rational approach would be to be conservative in the face of very limited data for a trend estimate. By that account, Old Blue says the sudden possibility of a Thompson campaign has generated 10 points of support, but with no evidence of a trend either way since polling on Thompson began. The Red estimator jumps around-- there just isn't enough data for a sensitive trend.

The current estimates for each trend are:

Giuliani: 34.8 (Blue)/30.4 (Red)
McCain: 19.4/21.3
Romney: 9.4/10.5
FThompson: 9.9/10.1
Gingrich: 8.3/7.3


We can check the sensitivity of these estimates to the amount of smoothing used to estimate the trend. Here I use 81 separate estimates of the current standing of each candidate, with the smoothing ranging from MORE sensitive than Red to MORE conservative than Blue. This is a wider range than I think anyone would reasonably want. The most sensitive end produces trends that jump around way more than anyone could believe, and the most conservative fit is basically just a straight line with hardly any change at all. But somewhere between these limits of silliness are a range of reasonable estimates. If the bottom line estimate for a candidate is pretty compact, then the amount of smoothing doesn't matter. If the estimates are spread out, then we at least know that sensitivity matters and we should be cautious. The summary of the data are presented below.



















The top half of the plot shows that the estimates for most candidates are in fact within a fairly small range regardless of how sensitive the estimates happen to be. Gore and Edwards are quite close, with some overlapping estimates of support. But Obama and Clinton are clearly distinct from each other and from Edwards and Gore. Similarly, Gingrich, Thompson and Romney show similar estimates and considerable overlap, while McCain is clearly above them and Giuliani clearly ahead of McCain.

Giuliani stands out among all the candidates in demonstrating more dependence on the sensitivity of the estimator. His box is more spread out than those of other candidates in the top half of the plot. In the lower half, which plots the distribution of all the estimates, Giuliani shows a bi-modal distribution. If we pick a more sensitive estimator, Giuliani support falls in the lower "hump" of the distribution, while less sensitive estimators suggest a stronger standing, producing the right hump. This difference is not trivial-- the more sensitive estimator says Giuliani is at about 30% support, while the more conservative one says 35%. No other candidate shows as large a discrepancy. This is due to the rather substantial downturn that Ready Red sees in Giuliani's polling over the last three months, but which Old Blue is still reluctant to accept. Which is right? Well, that's the whole point here: If you are a bit more daring, believe what Red has to say. If you like to buy municipal bonds, go with Blue.

One final way to look at sensitivity is to plot the estimated support for each candidate against the degree of smoothing used for each of the 81 estimates. Low values are less smoothing and more sensitive trends, while high degree of smoothing are more conservative and less sensitive.
























The good news from my point of view is that most of the lines do not demonstrate a strong relationship between amount of smoothing and the estimated support. While there is a little movement, it isn't sharp for anyone. The Giuliani line is the one showing the greatest variation across degree of smoothing, as I already noted. The upshot of this is that while I constantly worry about how much my estimates are affected by my preference for Old Blue, the data show that mostly it doesn't matter a lot, and certainly not within a reasonable range of smoothing. (In the plot above, Old Blue is a degree of smoothing of .7, while Ready Red is at .35.)

It is good to compare Old Blue and Ready Red-- both offer helpful insights into the nomination race. Your acceptance of the risk of being too slow to recognize change versus the risk of chasing phantom blips should help you decide which to give more credence.