Swing State Update

I have expanded the list of states that might play a role in determining the outcome of the Presidential vote in the fall.  For each state in the list below, I have compiled all the available polls at Huffington Post Pollster and calculated the percent of polls in which Clinton held a lead.  For each state I then calculated a statistic called “chi-squared” to see whether her lead was sufficiently consistent to conclude she was truly ahead in the state.  Here are the results through today:


In Wisconsin, Hillary Clinton has led in every poll conducted in the state dating back to last fall.  She has nearly as impressive a lead in both Michigan and Pennsylvania, both states typically mentioned as targets for Donald Trump’s “rust-belt” strategy.  In those two states there is less than one chance in twenty that Clinton is truly behind given the number of polls in which she held the lead.  In the remaining states the results are still too mixed to draw any conclusions about which candidate is in the lead.  Clinton does especially poorly in the traditionally-Republican states of Arizona and Georgia, but there haven’t been enough polls taken to draw any conclusions there. The other states remain toss-ups.


Who Leads in the Swing States?

As in every Presidential election, the outcome will be determined by a very small number of states. As I did in 2012, I have compiled the polls in these “swing” states and counted up the number of times Hillary Clinton or Donald Trump was in the lead.  I have included every poll conducted so far that includes both candidates; the oldest poll was taken in late June of 2015.    I intend to update these results limiting them to only recent polls as the election nears.


Two states – Michigan and Pennsylvania – have supported Hillary Clinton consistently enough that there is just a small chance, less than one in twenty, the race is actually tied or she is behind Donald Trump in those states.  The other four states remain toss-ups.


Pennsylvania tempts Republicans to compete there every election cycle, and this one is no exception.  Still the state has trended Democrat in Presidential elections since the late 1960’s.


Race to the Bottom

As most everyone who follows politics knows by now, we enter the unprecedented 2016 Presidential election with the candidates of both major parties disliked by a majority of Americans.  In this posting I examine the trends in “favorability” for both Hillary Clinton and Donald Trump.

Using the data at Huffington Post Pollster I calculated the “net favorability” for each candidate, equal to the percent of respondents saying they view a candidate favorably versus the percent who say they view that candidate unfavorably. I begin with Hillary Clinton, for whom we have favorability data dating back to 2009.



It might be hard to imagine today, but during her tenure as Secretary of State in Barack Obama’s first term, Hillary Clinton was viewed quite positively by the American public. Between Fall, 2009. and Fall, 2012, about three out of five Americans surveyed reported that they viewed Secretary Clinton favorably.  Even as late as April, 2013, Clinton was favorably viewed by 64 percent of the adults surveyed by Gallup, compared to 31 percent who viewed her unfavorably.  That translates into a net score of +33 (=64-31) in the graph above. She would never attain that level of popularity again.

Opinions about Clinton did not fall right away after the attack on the U.S. Consulate in Benghazi, Libya, on September 11, 2012, but the downward trajectory began soon thereafter.  When she announced her candidacy for President on April 12, 2015, the proportion of Americans holding favorable and unfavorable views of Secretary Clinton were just about equal.  A few months later her favorability score was “underwater,” with the proportion of Americans holding unfavorable views outnumbering those with favorable ones by between ten and twenty percent.


Opinions about Donald Trump have also remained pretty constant, and consistently negative, since he announced his candidacy on June 12, 2015.   At no time since he began his campaign for President have more Americans reported feeling “favorable” toward Donald Trump than “unfavorable.”  His ratings improved somewhat after his announcement and through the summer of 2015, but when the primary campaign began in earnest starting in January of 2016, Trump saw his favorability score fall further south.  It has rebounded and levelled off since he became the presumptive nominee after winning the Indiana primary on March 26th.  Compared to Hillary Clinton’s ratings, though, Donald Trump’s net favorability score averages about -24 compared to her average net rating of -11.


If we now take the difference between these two net favorability scores, we can see whether both candidates are equally disliked, or whether one is disliked more than the other.  For most of the campaign so far, Hillary Clinton has been winning the contest over which of them is less disliked.  Her net favorability scores generally run around 11-12 percent less negative than Trump’s.  For instance, over the month of June, 2016, Clinton averaged 41 percent favorable versus 55 percent unfavorable, for a net favorability score of -14.  Trump’s scores were 35 percent favorable and 60 percent unfavorable, for a net score of -25, or eleven points worse than Clinton’s.


As you might expect, there is a strong correlation between this net favorability score and the proportion of respondents intending to vote for Clinton or Trump.  Net favorability alone explains about two-thirds of the variance in voting intention across the 113 polls where both questions were asked.  Given the relationship shown in the graph, a score of +11 in net favorability should yield about a five percent lead in voting intention.


One interesting finding from the regression results is that the constant term of 1.06 percent is significantly different from zero.  (It has a standard error of 0.38 with p<0.01.)  The constant predicts Clinton’s lead when net favorability is zero, or in a poll where the proportion of people favoring and disfavoring each candidate is identical.  When net favorability is zero, Clinton leads Trump on average by a bit over one percent.

A Simple Model of Senate Elections

Without a President seeking re-election at the top of the ticket, the Democrats face a substantial challenge if they wish to win back the Senate in 2016.

While the Presidential race gets all the media and pundit attention, the battle for control of the Senate also looms large in this election year.  Republicans enter the election holding 54 of the 100 Senate seats, so a net Republican loss of just five seats would put the Senate back in Democratic hands.  The Democrats have the advantage that many more Republican seats, twenty-four, are at risk in the 2016 election compared to only ten held by Democrats.  This lopsided margin reflects the result of the 2010 off-year election when Republicans picked up six seats from the Democrats.  In principle, some of those Republican senators may be more vulnerable in a Presidential year with higher turnouts and more visibility.  The Democrats certainly believe they can retake the Senate this November.  The Party aggressively recruited candidates for the Senate elections and had secured bids from all but one of its top-tier candidate selections by early October of 2015.

The Democrats also have the advantage that the party of the incumbent President usually wins a slim majority of the Senate vote in “on-year” elections when a Presidential election also takes place but loses in “off-year” elections.


Unfortunately for the Democrats the relationship between Senate electoral success and type of election is not so simple.  If I divide up on-year elections into ones when the President ran for re-election and ones when, like the upcoming election, he did not, a very different pattern emerges.  The President’s party fared substantially worse in the five open-seat elections since 1946 than it did in elections with the President at the top of the ticket.  While open-seat years gave the President’s party a one-percent boost compared to off-years, that difference is not statistically significant.  What matters is whether the President is running or not.


The Democrats’ optimism is also based on the much larger number of Republican seats at risk in 2016.  I find some support for the notion that a Senate “Class” with a comparatively lopsided division of the vote in one election becomes more competitive six years later.  Statisticians call this phenomenon “regression toward the mean,” where observations that were outliers at one time show more average scores when measured again. But this effect is weak, and the division of the Senate vote in 2010, 53-47 percent Republican, was not as lopsided as the margin in terms of seats, 65-35 percent Republican.  All told the estimated “rebound” effect given the Republican 2010 landslide is just 0.6%, raising the expected Democratic vote from 47.1% to 47.7%.

Where else might the Democrats gain some relief?  Perhaps the generally positive state of the economy might provide some help.  Political scientists and economists have tested many different measures of economic conditions in models of voting for President and Congress.  One simple measure that has consistently proven significant is the change in personal income, and that proves true for Senate elections as well.


This chart adds the effects of the year-on-year percent change in real disposable personal income to our simple model.  While the effect of rising incomes is positive and statistically significant, it alone cannot overcome the substantial deficit facing the Democrats in an open-seat election year.  In the six years of the Obama Administration, personal income rose by at most 2.2 percent in a single year, 2012.  With a likely figure for annual income growth in 2016 at around two percent, we should expect the Democrats to win only about 48 percent of the Senate vote in 2016.

Winning a majority of the Senate vote is not a requirement for winning a majority of the contested seats.  In 2004, and most dramatically in 1982, the Republicans managed to win a majority of the seats with a minority of the votes cast.


The high “swing ratio” of 2.4 means that a change of one percent in the percentage of votes won translates on average to a 2.4 percent increase in the percentage of seats.  So even fairly small changes in the division of the vote can have much larger effects on the composition of the United States Senate.

I tried some other possible influences like the approval rating of the President and the size of the President’s margin in on-year elections.  I found no “coattails” effect for the Presidential vote either in years when the President is running or years when he not.  Presidential approval does matter, but only in off-year elections, so I did not include it in this discussion about 2016.  That finding is consistent with a conventional view that off-year elections reflect public opinion about the President’s performance in office.


Iowa: So Many Polls. So Few Respondents.

Pollsters have conducted over 44,000 interviews among Iowa’s 160,000 Republicans, but they probably interviewed just 15,000 unique people.  A majority of those polled took part in at least three interviews over the course of the campaign.

It seems like new Presidential polling figures are released every day.  We generally talk about each new poll as a unique snapshot of the campaign with some fresh sample of a few hundred caucus-goers.  That concept of polls might apply to national samples, but when polling in states as small as Iowa and New Hampshire, the number of eligible respondents is not that large compared to the number of interviews being conducted.

Making the problem worse is the falling “response rate” in polling, the proportion of eligible respondents who complete an interview.  Mobile phones, caller-ID, answering machines, all have enabled potential respondents to avoid the pollster’s call.  Pew reports that response rates have fallen from 21 percent to 9 percent just between 2006 and 2012.  If we assume a response rate of ten percent, only some 16,000 of Iowa’s 160,000 eligible Republican caucus-goers might have agreed to take part in a poll.

Huffington Post Pollster lists a total of 94 polls of Republican caucus-goers through today, January 31, 2016, constituting a total of 44,433 interviews.  I will use this figure to see how the composition of the sample changes with different response rates.1

How large is the electorate being sampled?

Around 120,000 people participated in the Republican caucuses in 2008 and 2012.  While some observers think turnout in 2016 could be higher because of the influx of voters for Donald Trump, I have stuck with the historical trend and estimate Republican turnout in 2016 at just under 124,000 citizens.

To that baseline we have to add in people who agree to complete an interview but do not actually turn out for the caucuses.  In my 2012 analysis I added 20 percent to the estimated universe to account for these people, but recent findings from Pew suggest 30 percent inflation might be more accurate.  With rounding, I will thus use 160,000 as my estimate for the number of Iowa Republicans who might have been eligible to be polled about the 2016 Republican caucuses.

How low is the response rate?

Most of those 160,000 people will never take part in a poll.  Pew estimated 2012 “response rates,” the proportion of eligible respondents who actually complete an interview, in the neighborhood of 9 percent.  To see what this means for Iowa, here is a table that presents the average number of interviews a cooperating respondent would have conducted during the 2016 campaign at different response rates.  At a ten percent response rate like Pew reports, the 16,000 cooperating prospects would each need to complete an average of 2.78 interviews to reach the total of 44,433.


How many people gave how many interviews?

Finally, I’ll apply the Poisson distribution once again to estimate the number of people being interviewed once, twice, three times, etc., to see the shape of the samples at each response rate.


Even if everyone cooperates, random chance alone would result in about 13 percent of respondents being interviewed at least twice.  When the response rate falls to 10 percent, most respondents are being interviewed three or four times, with fifteen percent of them being interviewed five times or more.  Even with a 20 percent response rate, about double what Pew reports, a majority will have been interviewed at least twice.

Certainly someone willing to be interviewed three, four, five times or more about a campaign must have a higher level of interest in politics than the typical Iowa caucus-goer who never cooperates with pollsters.  That difference could distort the figures for candidate preferences if some candidates’ supporters are more willing to take part in polls.

Basing polls on a relatively small number of cooperative respondents might also create false stability in the readings over time.  Polling numbers reflect more the opinions of the “insiders” with a strong interest in the campaign and may be less sensitive to any winds of change.  We might also imagine that, as the campaign winds down and most everyone eligible has been solicited by a pollster, samples become more and more limited to the most interested.

Overarching all these findings remains the sobering fact that only about one-in-ten citizens is willing to take part in polling.  Pollsters can adjust after the fact for any misalignments of the sample’s demographics, but they cannot adjust for the fact that people who participate in polling may simply not represent the opinions of most Americans.  We’ll see how well the opinions of those small numbers of respondents in Iowa and New Hampshire match the opinions of those states’ actual electorates on Primary Day.


1For comparison, Pollster archives 65 polls for the 2012 Iowa Republican caucuses totalling 36,300 interviews.  The expanded demand for polls has increased their number by 45 percent and increased the number of interviews conducted by 22 percent in just one Presidential cycle. (To afford polling at greater frequencies, the average sample size has fallen from 558 in 2012 to 473 in 2016.)


A Tale of Two Candidacies

Trump fares better in polls conducted by robots; Sanders polls better when humans conduct the interview.  Sanders also shows greater strength in polls of “likely” voters.

Commentators often treat Donald Trump and Bernie Sanders as representing two “insurgencies” within the Republican and Democratic Parties.  While there are certainly some surface similarities between the two candidacies, national polling data for the two candidates show substantial differences as well.  I begin with two charts comparing the trends in their national polling support using data from Huffington Post Pollster since each candidate announced he was running for President of the United States.



While both men’s support has continued to grow over the course of the campaign, the trajectories of their support are radically different.  Sanders’ polling figures have increased consistently over the course of the campaign, but the rate at which his support has risen has slowed as the campaign progressed.  Trump’s support seems to have gone through three phases — rapid growth at the outset, a plateau over the fall, and a second surge beginning around Thanksgiving that slowed at the turn of the year.  Extrapolating out to February 1st when the Iowa Caucuses take place, Trump would be approaching just under forty-five percent in national polls with Sanders  reaching thirty-five percent.


I have examined two types of polling effects: the method of interviewing and the type of sample drawn. Trump does over four percent worse when interviews are conducted by a live human being.  Sanders does worse by an essentially identical margin in polls that use automated telephone methods.  The result for Trump may reflect an unwillingness on the part of his supporters to admit to preferring the mogul when talking with an actual human interviewer.

Sanders’ poorer showing in polls that rely on automated polling may have to do with their exclusion of cell phones which cannot by law be called by robots. Usually this problem is adjusted for after the poll has been conducted by weighting the data to conform to expected demographic breakdowns.  However Sanders large lead among younger voters who are much less likely to have a landline phone may be suppressing his support in automated polling.  In a recent FoxNews poll, for instance, Sanders holds a 61-31 lead over Hillary Clinton among respondents under 45 years of age; older voters strongly prefer Clinton 71-21.  That same demographic explanation does not work for Trump, however, since he drew an identical 35 percent among voters in that same poll from both age groups.  The “social desirability” explanation probably has greater force when accounting for his poorer showing in polls conducted by human interviewers.

Turning to the differences in sampling frames, we find that polls that screen for “likely” voters show surprisingly greater levels of support for Bernie Sanders than do polls that include all registered voters or all adults.  Trump’s support shows no relationship with the sample of voters drawn.  Both candidates, as “insurgents,” are thought to suffer from the problem of recruiting new, inexperienced voters who might not actually show up at the polls for primaries and caucuses.  That seems not to be an issue for either man, and in Sanders’ case it appears that the enthusiasm we have seen among his supporters may well gain him a couple of percentage points when actual voting takes place.

The Kentucky Gubernatorial Election

Political observers were shocked when the Republican candidate for governor in Kentucky, Matt Bevin, won a resounding victory over Democrat Jack Conway. Conway, the incumbent Attorney General, had held a small lead in the polls throughout the campaign, about 2.1 percent according to the Huffington Post’s Pollster charts.  That difference translated into an 88 percent chance that Conway was actually ahead in the electorate as a whole.

The actual results were not even close.  Bevin won with 52.5 percent of the ballots cast compared to 43.8 percent for Conway and 3.7 percent for third-party candidate Drew Curtis. Bevin received nearly 218,000 more votes than the Republican candidate in 2011, David Williams, while Conway lost ground winning some 37,000 votes fewer than the total won by the outgoing Democratic governor, Steve Beshear, when he ran in 2011.  In the two largest counties, Jefferson and Fayette, where the Louisville and Lexington metropolitan areas are located, Conway actually won about 10,000 fewer votes in 2015 than he had received when running for Attorney General in 2011.

In 53 of the 120 counties in Kentucky, Bevin’s share of the vote increased by more than twenty percent compared to that won by Williams four years ago.

Following my earlier report on the 2013 Virginia gubernatorial race, here is a simple model based on county-by-county data predicting the vote for Bevin as a function of the vote for Williams in 2011, the proportion of black and Hispanic adults in 2011, and the change in turnout measured as a proportion of the total population aged 19 and older.  (As usual, I have transformed these proportions into their “logits.” )  The voting data comes from the official returns on the site run by the Kentucky Secretary of State.  The demographic data were compiled from Census estimates reported here.  Because of the way the Census groups people by age, my definition of “adults” excludes 18 year-olds.

Ordinary Least Squares; 120 Kentucky Counties
Dependent variable: Republican Vote for Governor, 2015
All variables measured as "logits."

                 coefficient   std. error   t-ratio   p-value 
  const            0.480953     0.116231      4.138    6.70e-05 ***
  Rep Gov 2011     0.646872     0.0354867    18.23     1.04e-35 ***
  Black 2014      −0.0545183    0.0206335    −2.642    0.0094   ***
  Hispanic 2014    0.0132545    0.0328316     0.4037   0.6872  
  Turnout Change   0.322859     0.136171      2.371    0.0194   **

Mean dependent var   0.341255   S.D. dependent var   0.388213
Sum squared resid    3.832535   S.E. of regression   0.182555
R-squared            0.786303   Adjusted R-squared   0.778870

Bevin did less well in counties with higher proportions of blacks, though not Hispanics, even after controlling for the 2011 Republican vote.  He also apparently fared better in counties where turnout increased.

However the turnout effect largely depends on three “outliers,” Cumberland, Elliott, and Menifee Counties, all with adult populations under 7,000. In the first two of these, turnout fell by more than fifteen percent, while in Menifee it rose by about the same amount.  If we exclude these three counties, the effects of turnout change are much more modest:

Ordinary Least Squares; 117 Kentucky Counties
Dependent variable: Republican Vote for Governor, 2015
All variables measured as "logits."

                 coefficient   std. error   t-ratio   p-value 
  const            0.455381     0.118921      3.829    0.0002   ***
  Rep Gov 2011     0.682421     0.0379599    17.98     8.51e-35 ***
  Black 2014      −0.0486127    0.0205614    −2.364    0.0198   **
  Hispanic 2014   −0.00571780   0.0342821    −0.1668   0.8678  
  Turnout Change   0.147086     0.184871      0.7956   0.4279  

Mean dependent var   0.344539   S.D. dependent var   0.383348
Sum squared resid    3.617365   S.E. of regression   0.179716
R-squared            0.787798   Adjusted R-squared   0.780220

The effects for the prior Republican vote and the proportion black and Hispanic remain about the same after these three counties are excluded, but the effect for changes in turnout is about half its prior value and fails to achieve statistical significance.

The Kynect Effect

One of the major issues in the campaign was the “Kynect” program, Kentucky’s implementation of the exchanges provided for under the Affordable Care Act.  Bevin opposed Kynect and first threatened to abolish the program entirely if elected.  He has since relented somewhat agreeing to grandfather all current enrollees but not accept any new applications.  We might thus expect that counties with higher Kynect enrollment rates might show lower levels of support for the Republican.  Using 2014 enrollment data from the Kentucky governor’s site, I find no effect for Kynect enrollment when measured as a proportion of a county’s total population.  When added to the model above, the coefficient is trivially small (-0.014) and statistically insignificant.

It turns out, though, that if we look at the factors influencing Kynect enrollments, we get what might be considered a counter-intuitive result:

Ordinary Least Squares: 120 Kentucky Counties
Dependent variable: Proportion of Total Population Enrolled in Kynect
All variables measured as "logits."

             coefficient   std. error   t-ratio   p-value 
const         −2.88070      0.205054     −14.05    8.61e-27 *** 
Rep Gov 2011   0.147432     0.0639254      2.306   0.0229   ** 
Black 2014    −0.0979391    0.0366472     −2.672   0.0086   *** 
Hispanic 2014 −0.154206     0.0587630     −2.624   0.0099   *** 

Mean dependent var  −1.955458   S.D. dependent var   0.388445 
Sum squared resid    12.62110   S.E. of regression   0.329852 
R-squared            0.297104   Adjusted R-squared   0.278926

Kynect enrollments are higher in counties that voted Republican in 2011 and lower in counties with larger proportions of black or Hispanic citizens.

One possible theory might be that because the ACA was designed to provide insurance to less well-off Americans not already covered by programs like Medicaid, Kynect rates should be higher in counties where Medicaid rates are relatively lower.  This is certainly false.  The bivariate correlation between 2014 Kynect enrollment rates and 2011 Medicaid enrollment rates is 0.89.  Kynect enrollments are highest in counties where Medicaid enrollments are also higher.  The real determinant of Kynect (and Medicaid) coverage rates is whether a county is urban or rural.  If we use total county population as a rough measure of urbanity, then we have this relationship:

Kynect enrollments are higher in the smaller counties.  Not surprisingly, those more rural counties gave a larger share of their votes to Bevin.


However Bevin fared worst in two largest counties where Lexington and Louisville are located.

Many commentators suggested that Bevin’s success came more from his appeal to social and religious conservatives than anything having to do with economics or programs like Kynect.  Kentucky ranks eighth among the states based on weekly church attendance rates, and Bevin appealed directly to religious conservatives with his strong endorsement of Kim Davis, the local official who refused to issue marriage certificates to homosexual couples after the Supreme Court’s decision in June.  It seems much more plausible that Bevin’s victory was powered more by these religious appeals than by anything having to do with his policy stands.


Many Republicans and Independents See the Benghazi Committee as Politically Motivated and Approve

In today’s New York Times Charles Blow cites a finding from a recent CNN/ORC poll where 47 percent of Republican respondents agreed that the House Select Committee on Benghazi was “using the investigation to gain political advantage.”  At face value this is a rather surprising result.

Most questions that ask people to approve or disapprove of the actions of politicians generate partisan results.  Democrats are more likely to approve of the performance of President Obama while Republicans generally disapprove.  So, at first glance, for half of all Republicans to agree that the Republican-controlled Committee acted for political gain could seem unusually critical of the Committee’s actions. As it turns out there is a much larger group of Republicans who see the proceedings as politically motivated and are cheering the Committee on.

It turns out that the question Blow cites was asked of only half the sample.  Another half were asked whether “Republicans have gone too far” in the way they have handled the hearings, or whether they have handled them “appropriately.” The left-hand table reports that 71 percent of Democrats said “Republicans” had “gone too far” while 20 percent believed “Republicans” had handled the hearings “appropriately.”  For Republican respondents the reverse held true; only 16 percent of them say “Republicans have gone too far” while 74 percent say Republicans acted “appropriately.”

These figures do not sum to one hundred percent because of “don’t know” responses.  Nine percent of Democrats (=100-(71+20) = 9) have no answer on the “gone too far” question as do ten percent of Republicans (= 100 – (16+74) = 10).

The question on the left constitutes a referendum on “Republicans” while the one on the right asks about the “House Select Committee” with no partisanship attached.  When asked to judge the Republicans’ behavior, we see the usual pattern of partisan response. However when asked to judge whether the Committee conducted an “objective investigation” or one to “gain political advantage,” the difference between Republicans and Democrats is considerably smaller. Republicans split about equally between the two alternatives, with 47 percent choosing the “objective” response and 49 percent the “political” one. Democrats almost uniformly see a political motive behind the Committee’s actions. 85 percent of them choose the political answer while just 10 percent see the Committee as “objective.”

Since these sub-samples were randomly chosen from the overall pool of respondents, both represent equally valid samplings of public opinion.  One thing we do not have are the answers of citizens when asked both questions because they are in separate half-samples. We can, however, run some experiments to see what proportion of Republicans think the Committee is conducting a political investigation and approve of it.

We start with the basics — half of Republicans believe the Committee’s actions are politically motivated, and three-quarters of them approve of the its conduct of the investigation.  We can combine these two measures to estimate how many Republicans endorse the Committee’s following a political agenda.


This table uses the responses for Republicans from the first table.  Republicans’ answers to whether the Committee was objective or political appear on the columns and how they judged the Committee’s actions along the rows.  We know the percentage of Republicans who gave each of these answers, but we do not have data for the cells of the table because no one was asked both questions.  We can generate a “baseline” estimate for these cells by assuming that there is no relationship between answers to one question and answers to the other.  Under that assumption we get a most-likely estimate of the proportion of Republicans endorsing a politically-motivated Committee of about 36 percent.  That figure is calculated by taking 49 percent, the proportion seeing the Committee as politically motivated, and applying it to the 74 percent of Republicans who thought the Committee’s actions were “appropriate.”  Multiplying those figures together yields the estimated proportion of Republicans holding both opinions,  49% x 74% = 36.3%.

That figure represents our best guess since it makes no assumptions about how opinions on the two questions might be related. However we can also set upper and lower bounds for this value because it is constrained by the “marginals,” the row and column totals that each must sum to one hundred percent.  The minimum, or “benign” estimate assumes every Republican who thinks the Committee has “gone too far” also believes the Committee is acting politically. That produces a table like this:


In this extreme case the 7.5 percent in the original “objective/gone too far” cell is added to the corresponding “political” cell on its right.  Since the “political” column must still sum to 49 percent, the proportion of Republicans who think the Committee’s action appropriate must fall to compensate and reaches its minimum of just under 29 percent.

Likewise we can add the 7.8 percent in the original “gone too far/political” cell to the “objective” cell on its left.  That more “aggressive” model assumes all Republicans who see the Committee acting politically also endorse its actions, and none think it has gone too far.  That increases the estimate to its maximum of 44 percent.


All told then, between 29 and 44 percent of Republicans see the House Select Committee on Benghazi as acting politically and approve.

Charles Blow views the 49 percent of Republicans who believe the Committee is politically motivated as showing widespread “skepticism” about the Committee’s motives that extends even to Republicans.  With three-quarters of Republican endorsing the Committee’s investigation, I see more cheering than skepticism in Republican ranks.  The true skeptics, those who think the Committee has “gone too far,” make up just sixteen percent of Republicans.  Twice as many Republicans or more endorse the Committee’s actions precisely because it has pursued a political agenda.

Surprisingly,  independents prove even more likely to see the approve of a Committee with partisan motivations..  Three-quarters of independents think the Committee see a political motive, but a majority of them, 57 percent, also think the Committee has acted appropriately. Applying the baseline assumption of no-correlation as before, and multiplying those two figures together, indicates that nearly 43 percent of independents endorse a politically-motivated investigation, higher even than the Republican figure of 36 percent.

Elsewhere in the CNN/ORC poll we see that independents are vastly more unhappy with Hillary Clinton’s handling of the Benghazi affair than are Democrats.  Twice as many independents report being “dissatisfied” than “satisfied,” 65 percent to 31 percent.   Democrats hold the reverse set of opinions with 63 percent satisfied and 30 percent dissatisfied.  The Republicans are the most extreme, of course, with 85 percent dissatisfied and only eleven percent satisfied. Those dissatisfied independents could play an important role in next fall’s general election.

Honey, It’s the Pollster Calling Again!

Back in 1988 I had the pleasure of conducting polls during the New Hampshire primaries on behalf of the Boston Globe.  The Globe had a parochial interest in that year’s Democratic primary because the sitting Massachusetts governor, Michael Dukakis, had become a leading contender for the Presidential nomination.  The Republican side pitted Vice-President George H. W. Bush against Kansas Senator Bob Dole, the upset winner of the Iowa caucuses a week before the primaries. Also in the race were well-known anti-tax crusader Jack Kemp and famous televangelist Pat Robertson.  Bush had actually placed third in Iowa behind both Dole and Robertson.

We had been polling both sides of the New Hampshire primary as early as mid-December of 1987, but after the Iowa caucuses, the pace picked up enormously. Suddenly we were joined by large national polling firms like Gallup and media organizations like the Wall Street Journal and ABC News.  As each day brought a new round of numbers from one or another pollster, we began to ask ourselves whether we were all just reinterviewing the same small group of people.

Pollsters conducting national surveys with samples of all adults or all registered voters never face this problem.  Even with the volume of national polling conducted every day, most people report never being called by a pollster.  In a population of over 240 million adults, the odds of being called to participate in a survey, even ones with a relatively large sample like 2,000 people, are miniscule.  That is still true even if we account for the precipitous decline in the”response rate,” the proportion of households that yield a completed interview.  A wide array of technological and cultural factors have driven survey response rates to historic lows over the past few years as this table from Pew shows clearly:

In 2012, fewer than ten percent of households were represented in a typical poll.  Still, even at such a low response rate, the huge size of the United States population means that any individual has only a tiny chance of being selected from a sampling universe numbers of 24 million homes.  Even for a large survey of 2,000 people, the chance of any individual household being selected is a mere 0.000008.

Those odds change drastically when we narrow the universe of eligible people to “likely” voters in an upcoming New Hampshire Republican primary.  Even including people who claim they will vote but later do not, the total universe of eligible respondents in 2012 was probably just 300,000 people.   To reach that figure I started with the total of 248,485 ballots cast in the Republican primary.  To those voters we need to add the other people who reported that they would take part in the primary but did not actually turn out on Primary Day.  For our purposes, I have used an inflation factor of 20% which brings the estimated the total number of self-reported likely Republican primary voters to 298,182 people.  I rounded that figure up to 300,000 in the tables below.

Over a dozen polling organizations conducted at least one survey in New Hampshire according to the Pollster archive for the 2012 Republican primary.  In all there are 55 separate polls in the archive representing a total of  36,839 interviews, or about 12% of the universe of likely voters.  If all 300,000 likely Republican primary voters had been willing to cooperate with pollsters in 2012, about one in every eight of them would have been interviewed.  If we choose a much more realistic response rate like ten percent, there are actually fewer cooperating likely voters than the total number of surveys collected, so some respondents must be contributing multiple interviews.  Can we estimate how many there are?

It turns out the chances a person will be interviewed, once, twice, etc., or never at all can be modelled using the “Poisson distribution.”  Usually a statistical distribution relies on two quantities, its average and its “variance,” but the Poisson distribution has the attractive feature that the mean and variance are identical.  Thus we need only know the average number of interviews per prospect to estimate how many people completed none, one, two, or more interviews.  Here are estimates of the number of interviews conducted per potential respondent at different overall cooperation rates.  At a 20 percent cooperation rate, only 60,000 of the 300,000 likely voters are willing to complete an interview.  Dividing the number of interviews, 36,839, by the estimated number of prospects gives us an average figure of 0.614 interviews per prospect.


Now we plug those values into the Poisson formula to see how many people are interviewed multiple times during the campaign.


In an ideal world where every one of the 300,000 likely primary voters is willing to be interviewed, 88.4% of them would never be interviewed, 10.9% would complete one interview, and 0.7% would be interviewed twice.  If response rates fall to  8-10%, only 20-30% of likely voters are never interviewed.

Though only a few prospects would be interviewed more than once in the ideal, fully-cooperative world, at more realistic response rates closer to what Pew reports, many people were interviewed multiple times in the run up to the 2012 primary.  If only eight percent of likely voters were willing to complete an interview, about a quarter of the prospects were interviewed twice, and one in five of them were interviewed at least three times.

We can use those estimates to see how the size and composition of the actual survey samples change as a function of response rate.


At 100% cooperation, obtaining nearly 37,000 interviews from 300,000 people means a small number, about 2,000 people, would be interviewed twice merely by random chance.  So those 37,000 interviews represented the opinions of  32,000 people who were interviewed once, and another 2,000 people interviewed twice.  As response rates fall, the total number of unique respondents, the height of each bar, declines, with a larger share of interviews necessarily coming from people interviewed multiple times.  At a 10% response rate the proportion of people interviewed multiple times just about equals the proportion of people interviewed only once.  Below that rate the proportion of people interviewed only once declines quickly.

A Sea Change in Congressional Attitudes on Surveillance

About two years ago I posted an analysis of the vote in the House of Representatives on the so-called “Amash Amendment,” an attempt to restrict the bulk collection of telephone records by the National Security Agency. Observers at the time were startled when, despite widespread opposition from the Obama Adminstration and House leadership, the amendment failed on a close vote of just 217 to 205 in July, 2013.

This past week the House had another chance to vote on the question of NSA surveillance, and this time the House voted to put new restrictions on the mass collection of telephone data. In particular, the telephone and Internet providers will retain the data and provide it only in response to government requests. This so-called USA FREEDOM Act of 2015 passed the House by a vote of 338 to 88. Perhaps the May 7th decision by the Second Circuit Court of Appeals ruling the entire NSA program illegal encouraged the lop-sided outcome.

Opponents of the surveillance programs were quick to argue that this Act does not go far enough in terms of reining in the activities of the FBI and NSA. Justin Amash himself, the author of the 2013 amendment, voted against the FREEDOM Act, arguing that:

H.R. 2048 threatens to undo much of the progress resulting from the Second Circuit’s opinion. The bill’s sponsors, and unfortunately some outside advocacy groups, wrongly claim that H.R. 2048 ends “bulk” collection. It’s true that the bill ends the phone dragnet as we currently know it—by having the phone companies themselves hold, search, and analyze certain data at the request of the government, which is worse in many ways given the broader set of data the companies hold—but H.R. 2048 actually expands the statutory basis for the large-scale collection of most data.

Even House leaders who opposed his 2013 amendment supported the FREEDOM Act.  This included both House Majority Leader Kevin McCarthy and Minority Leader Nancy Pelosi.  (The Speaker typically does not cast roll-call votes and did not vote on the FREEDOM Act.  He did choose to vote against the Amash Amendment, perhaps to indicate his strong opposition to that provision.)

I have compared the votes on the FREEDOM Act with those on the Amash amendment two years ago.


The top half of the table presents the actual votes; the lower half presents the percentages based on the Members’ votes on the Amash Amendment.  Only seven Members voted against both provisions.  Those who opposed the Amendment in 2013 voted overwhelmingly, 94 percent, in favor of the FREEDOM Act.  Most of the opposition to the Act came from the supporters of the Amendment.  Nearly two out of five of those Members opposed the Act as did Amash himself.  New Members of the House also voted heavily in favor of the Act; only eleven of the 67 new Members in the 114th Congress voted no, or just 16 percent.