Monthly Archives: January 2017

Reality Doesn’t Suck That Much for the Democrats

By Robert Dresden (Source:, January 19, 2017)

After reading my colleague’s column, “Reality Sucks for the Next DNC Chair,” I am compelled to reply:

In his January 17th article, he made a couple of points with which I strongly disagree. First, he claims the Democrats have become increasingly a ‘presidential party’ at the expense of down-ballot offices. This point has been made by many other pundits, most recently in Politico by Edward-Isaac Dovere. While the claim is defensible, the campaign finance evidence is mixed, if not contradictory.

Have the Democrats’ increasingly funneled money to the party’s presidential campaign when some of that money should have gone to Democrats in down-ballot races? The problem is with the data.  Its difficult to prove this claim one way or the other, in part, because there is no central resource for knowing exactly how much money is spent by politicians in state and local races.  Campaign finance compliance and reporting rules vary by state and aggregating all of this money into a total number is problematic.

Yet, we can look at the percentage of federal election campaign expenditures that go to the party’s presidential campaign versus the U.S. House and Senate races.  If the Democrats have become more of a ‘presidential party’ we should see a higher percentage of campaign expenditures spent by Democratic presidential candidates compared to the total spent by Democrats running for the U.S. House and Senate.

The chart below summarizes the campaign expenditures of Democratic presidential candidates relative to other federal races since 1976.  It may surprise some that Jimmy Carter’s 1976 campaign accounted for a higher proportion of total Democratic campaign spending than any campaign since.  Only the two Obama campaigns come close (56 and 54 percent, respectively).  Hillary Clinton’s 2016 campaign accounted for only 42 percent of all Democratic federal campaign expenditures — which is about average.

If you want to argue that the Democrats are losing U.S. congressional and state legislative elections because of an over-emphasis on presidential campaigns, it will require a more sophisticated understanding of how campaign dollars impact elections at all levels and how this spending interacts. For example, do local and state candidate gain more or less synergy from higher presidential-level spending than they do from spending by U.S. House and Senate campaign?  Or perhaps there is no relationship?

Its also important to note that the loss of local, state, and U.S. congressional seats also occurred for Democrats in midterm elections which are not affected by a presidential race. So, blaming the fundraising prowess of Democrats’ presidential candidates for the party’s declining success in other races seems unwarranted.

The second point from the Kroeger article that eats my grain is the contention that the Democrats are so far behind the Republicans at all levels of government that it will take a decade to recover.

When you recite the current numbers it does sound bad:  the Democrats are down 47 seats in the U.S. House, down three in the U.S. Senate, have just 16 Governors and control only 18 state legislatures.  But I will contend that the party’s troubles are exaggerated at all levels of government, but particularly at the national level.

I agree that the Clinton popular vote is a biased measure of the Democrats’ national strength, but it is equally foolish to think the Democrats aren’t close to having a winning presidential coalition.  They are close.  Very close.  And only minor shifts in turnout and support levels in a few key demographic segments (most notably working-class voters) would have changed the 2016 results in Clinton’s favor.

The task for the Democrats is greater at the congressional level, but far from irreversible. Regaining the Senate will have to wait until after the 2018 midterms where the Democrats are protecting 10 incumbent Senators in states Trump won.  They will be lucky to hold serve in 2018.  It’s 2020 where the Democrats will need to focus on getting back the Senate.

As for the U.S. House, it will take 24 seat flips for the Democrats to regain power. If we look to history as our guide, that is not a big number to turn in just one election cycle. Since Eisenhower, four out of nine midterm elections under a Republican president saw the Republicans lose at least 24 seats to the Democrats.

I do not dispute that the Democrats are in the political wilderness right now. Their most prominent leaders are old, they rely too much on a patchwork of unconnected issue positions, and don’t provide a coherent organizing principle voters can use as a heuristic device to navigate the political environment.

Yes, the Democrats are in the forest, but its Virginia’s Shenandoah Forest not Alaska’s Tongass Forest. They are close to regaining power at the national level and are far from facing an existential crisis at the local and state levels. Add to this optimistic view that the long-term demographic changes in the U.S. will most likely work in the favor of Democrats and it is hard to feel to bad for them.  The greater challenge may be convincing the Republicans that can’t continue their success without winning the support of more women, minority, and younger voters. In my view, those are the battle lines that will define future elections and the Democrats are more than capable of winning on those fronts.

Democrats, here’s my short-term advice:  Don’t watch TV tomorrow or this weekend.  Go to the movies and see Rogue One for the second time, or just go to dinner and relax.


Reality Sucks for the Next DNC Chair

By Kent R. Kroeger (Source:, January 17, 2017)

Author’s email:

“The Democratic Party has always represented the people and now more than ever we must blow the walls off our tent so everyone feels welcome,” Sally Boynton Brown said as she announced her candidacy for the leadership of the Democratic National Committee. “If people haven’t noticed, our country is becoming more Republican.”

Sally Boynton Brown, Executive Director of the Idaho Democratic Party

Boynton Brown’s last comment shouldn’t startle most people. But within the Democratic Party and its scholar-bureaucrat class, her voice is almost alone in contending that this country has taken a definitive turn to the right.  In fact, the Democrat literati continue to push an astigmatic fiction that this country is becoming more liberal and the American people are more aligned with the Democrats than the Republicans on an issue-by-issue basis.

Issie Lapowsky’s Wired article, “Don’t Let Trump’s Win Fool You – America’s Getting More Liberal,” provides one of the better attempts at justifying this conclusion, but there are others as well.  Peter Beinart’s Atlantic article, “Why America is Moving Left,” comes to mind because it came out just as the 2016 primary season started and seemed destined to find its own proof in the election of a Democrat (or reasonable Republican) president in November.  Ooops.

But more people are probably exposed to the ‘liberal America’ case from the news entertainment class, where the argument burns white hot. Filmmaker Michael Moore made the claim in 2011, and did it again recently with Bill Maher in their 2016 post-election wake.  Moore even suggested firing the pollsters, which is a good idea if you want the policy preferences of Americans to be determined by media decree rather than by listening to the American people themselves.  Understand, I love Michael Moore.  He’s not just talented and funny, but unlike his ideological counterparts, he actually understands the lives of people outside the east and west coast elite circles.

If Hillary Clinton had heeded his advice in this past election, she would be the president-elect today, instead eating over-priced New York City hamburgers with billionaire designer Ralph Lauren.  And for those of you hard-working, 9-to-5 Democrats that thought Hillary Clinton understood your life and your issues, trust me, she’ll be spending the rest of her life avoiding people like you and your issues.  Only billionaires and Hollywood celebs need apply to be her next dinner companion.

Clinton’s tin ear to working-class people is a sharp contrast to Moore who is an empathic savant with respect to their daily lives.  Yet, when Moore says we live in a liberal nation, he is just plain wrong.

To refute his and others’ notion of  ‘liberal America,’ I turn to Lapowsky’s article as she offers the most compelling argument in its favor.  And, in her defense, I enthusiastically agree with her conclusion that America is increasingly a tolerant, inclusive society that welcomes immigrants, supports women’s privacy, is open to reasonable gun control legislation and expects the rights and dignity of all of our citizens to be protected, regardless of sexual orientation, religious preference, gender, race, and ethnicity.  That is the America we live in, even if it did just elect Donald Trump.

However, here’s the problem. Lapowsky (like others making the same claim) cherry-pick those issues where this country is more liberal and tolerant and subsequently avoid most of the bread-and-butter issues — attitudes towards taxes, regulations, national security, crime and the role of government — that play a much bigger factor in people’s vote decision.  When analysts and academics look at a broader palette of issues they see a country that is more conservative today than at anytime since 1980.  The chart below shows one such analysis conducted by the University of North Carolina’s James Stimson who has created a “policy mood’ index derived from thousands of responses to hundreds of specific policy questions over time.  A more detailed description of his methodology can be found on his website.

Conservative Policy Mood, 1952 – 2012 (Data Source: Dr. James Stimson)      (Graph Source: Dr. Larry Bartels)

In contrast to Lapowsky’s selection bias problem, Stimson’s comprehensive accounting for a much wider range of policy opinions reveals meaningful changes in America’s public mood that correlate with shifts in political and policy outcomes.  For example, we see the peak of conservatism in the late 1970s leading to Ronald Reagan’s election in 1980.  Subsequently, as Reagan’s policies were implemented, we see a public mood swing back in the liberal direction, a predictable “correction” process that often occurs when new presidents take office and inevitably over-reach in their pursuit of their preferred policy initiatives.  This correction process is also evident soon after Bill Clinton took office in 1993 and after Barack Obama entered the White House in 2009.

Quite markedly after Bill Clinton’s first presidential election, the change in public mood (in the conservative direction) has tracked  closely with the secular decline of Democrats in our nation’s state legislatures (see chart below).  Apart from a temporary recovery during Obama’s rise, the trend for Democrats has been a virtually uninterrupted, monotonic decline in our state legislatures.

Today, there are almost 1,000 fewer Democratic state legislators than Republican.  If past variation is an indicator, it will take a decade for the Democrats to close that gap.  Of course, an over-reaching Trump administration accompanied by disastrous results could speed up that recovery process; but at this point, if I were the Democrats, I make no assumptions in that regard.

The current obsession among Democrats with Russian electoral shenanigans and FBI Director James Comey’s letter postpones their need to assay their electoral decline.  Whatever the final conclusion of the likely congressional inquiries into Russian interference in this last election, it will not provide information about the health of the Democratic Party.  The impact of the Russians and the Comey letter are limited to 2016. They tell us nothing about why, at every level of government, the Democrats are worse off than any time since the Great Depression.  The Democrats’ problems are deep-rooted, not transitory phenomena.

Boynton Brown’s competition for the chairmanship include Minnesota U.S. Rep. Keith Ellison, Labor Secretary Tom Perez, South Carolina chairman Jaime Harrison, and New Hampshire party head Ray Buckley.  Among them, only Boynton Brown has taken the first step necessary to solve any problem.  She recognizes the problem.  The Democrats are in decline, in a serious way.

I’m not surprised someone from a red state like Idaho would be speak so realistically about her party.  In my own experience, having recently managed a Democrat challenger’s state legislative campaign in Iowa in 2016, I saw up close the disconnect many working-class Democrats feel with their national and state party.  The Democrats just don’t talk to them anymore.  Virginia Senator James Webb summarized this attitude best:  “They don’t think Democrats like them.”  Boynton Brown is correct when she says the Democrats need to do more than just expand their tent, they need to “blow the walls” off their tent.

So when Ellison talks about “championing the challenges of working families and giving voters a reason to show up at the polls in 2018 and beyond,” or Perez says,”I’m in this race because we’ve got a lot of fighting to do, we’ve got a lot of advocating to do,” it demonstrates their poor understanding of the Democrats’ intrinsic problem.  The Democrats have a strategic problem, not a tactical one.  By deceiving themselves into thinking they are still the majority in this country and if they just work harder they will return to power, the Democrats are avoiding the tough questions.  Boynton Brown nailed the issue right out of the gate:  This country IS becoming more Republican and if the party leadership doesn’t recognize that reality, the Democrats will be a minority party for a long, long time.

Many Democrats will now often respond, “Didn’t Hillary Clinton win the popular vote by almost 3 million votes and doesn’t that prove we out-number the other side?”  No, it doesn’t, and here’s why:

Yes, 66 million people voted for Clinton to Trump’s 63 million.  But, overall, 71 million people voted against her.  Furthermore, Clinton’s popular vote advantage reflects the significantly more money she had to spend on national TV advertising; such that, in states like California or New York, the presidential campaign they saw was measurably different from the election voters witnessed in Wisconsin, Michigan, Ohio and Pennsylvania.  Include the national media’s total coverage of Clinton being more positive than Trump‘s (even with the help he received from the Russians and Comey), and you can see why Clinton had a quantitative advantage in the non-swing states.  That is the source of Clinton’s popular vote victory.  It was not a function of the Democrats being more popular than Republicans; and by not realizing that, the Democrats postpone their own recovery going forward.

The Russians and Comey are tactical factors specific to one election.  To explain the 40-year decline in Democratic state legislators one must investigate strategic factors such as the Democrats’ core messages or the disproportionate share of fundraising dollars going to presidential campaigns (Clinton, Gore, Kerry, Obama) at the expense of local and state races.

The Clinton’s were master fundraisers.  By aligning their messages (especially their private ones) with the financial, high technology, pharmaceutical, and entertainment industries, the Clintons beat the Republicans at their own game.  It was brilliant and it kept them in control of the Democratic Party right up until they were eclipsed by Obama who had improved on their fundraising model.  This strategy was great for the Clintons and Obama. Unfortunately, it also sucked the oxygen out of the rest of the party.  For the Democrats to rebound in the legislatures and governors offices that fundraising model must change.

But the Democrat’s decline isn’t just caused by the flaws in their money model.  The core Democratic message is out-of-date.  Ellison and Perez, along with most other Democratic elites, seem to think message tweaks and redoubling organizational efforts to mobilize core constituencies (including working-class voters) will turn the tide.   The Daily Kos even posted recently a renewed call for Democrats to read Saul Alinsky’s seminal activist guidebook, “Rules for Radicals.”  “This classic work of political strategy not only explains much about our current dilemma but more importantly, provides a path forward to escape it,” writes the Daily Kos.  This is the “classic work” with such utter condescension for working-class people that it writes:  “(the working class) are a fearful people, who feel threatened from all sides (and whose) emotions can go either to the far right of totalitarianism or forward to Act II of the American Revolution.”

Put aside the book’s title that already alienates most voters, that attitude towards the working-class is indicative of a viewpoint that sees working-class people as lab rats to be manipulated, not as human beings.

Furthermore, the polling data and the empirical research continually show how un-radical most Americans are when it comes to politics.  And that’s a good thing because when the Trump administration over-reaches, it will be punished by the American voter.  You can count on it.

Sadly, that mainstream view Democrats can’t pull their messages from the same playbook and expect the electoral outcomes to be any different.  They must throw out the playbook and reinvent itself.  That will take new leadership from outside the beltway, be it someone like Boynton Brown, and a willingness to set aside past assumptions and to look at the American political landscape as if seeing it for the first time.

That means the stilted rhetoric that still dominates Democratic communications must be abandoned.  Through an open-minded inquiry, the Democrats must rebuild their brand by locating the strategic weaknesses in the Republicans’ brand — and there are many — and setting forth a vision that combines the party’s highest ideals, on the one hand, with the realities of governance, on the other.

In Part 2 of this essay, I will provide some empirical evidence to show that whatever policies are advocated by the Democrats and Republicans in the future, they will be severely constrained by demographic and economic realities.  As we stand today, the Republicans have a core message that works as well in today’s economic environment as it did when Ronald Reagan rolled it out in 1976 when he took on the American political establishment:  Less government, more freedom.

That message sells.  If you don’t think so, you need to stare at that second chart in this article showing the rise of the Republicans in the state legislatures.  The Republicans are growing their market share; the Democrats are not.  This didn’t just happen.  Its been going on since 1980.  And now the Democrats must finally address the problem.

The governments in today’s advanced economies are about as big and intrusive as they can get relative to their total size.  What voters will expect, therefore, is a party that can address the seen and unforeseen needs endemic to all societies (security, growth, opportunity, health), hold the government and its elected officials accountable when it fails to address those needs, all while ensuring the long-term fiscal health of the state well into the future.

Not an easy task.  And it will require ending government intrusions where they fail to help and moving some government functions to the lowest level of jurisdiction possible so that people can once again feel like they have some control over their lives.

Author’s email:


The Trump Rallies Mattered…A Lot

Questions and press inquiries regarding this article and the data analysis contained within can be directed to: or call Kent Kroeger at 515 512 2776

By Analytic Team (Source:, January 5, 2017)

One Los Angeles Times writer called it the “scorched-earth” campaign.  Even his critics will agree, Donald Trump’s 2016 presidential campaign was like no other in American political history.

Yes, we’ve had negative campaigns before this one.  In the 1828 campaign, John Quincy Adams called Andrew Jackson’s mother a prostitute and Jackson’s wife an adulteress.  More recently, George W. Bush’s 2000 primary campaign was accused of indirectly promulgating a rumor among South Carolina Republicans that John McCain had a “black baby” out of wedlock.

But most examples, like those, were ancillary to broader campaign motifs that generally leaned more positive.  In 1988, George H.W. Bush’s campaign infamously used the “Willie Horton” ad to accuse his opponent, Michael Dukakis, of being soft on crime. But, even in that election, Bush’s message was much more about “no new taxes” than it was about attacking his opponent.

But now comes along Donald Trump who seemed to change the rules of American political campaigns that, heretofore, were assumed to be inviolable.  Trump’s most notable deviation was in his ground game, or lack thereof, according to many political campaign experts.  Where the political consultants expected door-knockers and phone banks, Trump gave us campaign rallies and more campaign rallies.  And not just rallies, but human waves, often more than 10,000 attendees, that strode across cornfields and stadium parking lots, often waiting hours just to see their new political liberator.  Trump rallies were events, similar to the atmosphere one experiences while tailgating prior to a Big 10 or SEC football game.

In contrast, Hillary Clinton ran a more traditional campaign that put a much higher premium on advertising and get-out-the-vote efforts than did the Trump campaign.  This difference in campaign strategy was, in part, a function of Clinton having more money and a more established campaign organization; but, it was also a function of expectations.  Modern presidential campaigns are supposed to look like Hillary Clinton’s campaign.  Who needs creativity when you have a proven model for winning presidential elections.  Barack Obama didn’t re-create how modern campaigns are structured.  He simply did it better than anyone else in 2008 and 2012.

Trump has changed the rules.  So what if he was outspent by $600 million.  He’d make up for it through free media (euphemistically called “earned media” nowadays by the media) and campaign rallies.  He also employed a level of negativity unseen in modern presidential elections.

“(Trump) sketched out conspiracies involving global bankers, casually threatened to jail his political opponent, and warned in increasingly specific terms that a loss by him would spell the end of civilization,” wrote Los Angeles Times writer, Noah Bierman.

Here is an example of a Trump rally in Grand Rapids, Michigan:

Yes, this campaign was different.  But did it really make a difference in the final election outcome?

Using state-level event data from the website and aggregated polling data from, we analyzed the relationship between the candidates’ travel schedules and changes in their state-level of support.

Since over 95 percent of 2016 candidate events and most state-level polling occurred in just 15 “swing states,” our analysis is restricted to those states (see list of states in table below).  The table shows the opinion polling data averages at the state-level at the beginning of October (RCP Poll Avg. as of Oct. 1st,) and the candidate’s actual vote percentage on Election Day (Actual Vote %).  The column on the far-right summarizes this information by showing the extent to which Trump’s support changed between Oct. 1st and Election Day.  As you can see, in the final month of the campaign, Trump gained support in all by two of the 15 swing states (Colorado and Nevada).

Based on this evidence alone, you might conclude that Trump had the superior campaign.  However, it is fair to ask, was this domination simply the result of two exogeneous factors:  The Comey letter and the Podesta emails?

If those two factors alone drove Trump’s dominance, we’d expect the impact to be similar across all 15 swing states.  After all, voters’ exposure to these two factors come primarily from national media outlets, not local or state-level sources.

Yet, we see from the table above that Trump’s support gains were not evenly distributed across the “swing” states.  In fact, there is a lot of state-level variation that warrants explanation.

The next table shows the number of post-convention candidate visits for each of the swing states and the percentage of the eligible voter population that was Hispanic/Latino as of January, 2016.  Without formally testing the relationships, it is evident that in states where the Hispanic voting population was large (>10%), Trump did not make significant inroads into Clinton’s early October lead (Arizona +1.1%, Colorado -3.4%, and Nevada -2.6%).

Less evident is the relationship between the number of candidate events and the relative change in Trump’s support.  In Ohio, where Trump’s relative position improved by 10.6% between Oct. 1st and Election Day, the Trump campaign had 12 more events than Clinton’s campaign.  Yet, in Colorado where Trump’s campaign had 13 more visits, his relative vote position deteriorated by 3.4%.   The bivariate correlation (p = 0.45) between the number of Trump campaign events and the size of the Hispanic/Latino population (p = -0.60) with changes in vote support were statistically significant; however, the number of Clinton events and the relative difference in the number of events were not statistically significant in the bivariate correlations.  Without a more formal test, it is difficult to know the relative contribution of each factor to changes candidate support.


We estimated a linear model to explain the vote change in the 15 “swing” states (the full regression model summary and dataset can be accessed here).  We included four variables in the initial regression model:  (1) Number of Clinton campaign events in each, (2) Number of Trump campaign events, (3) Percentage of Hispanic/Latinos in the voter eligible population, and (4) a indicator variable for Colorado.  Given the small sample size (n=15) it was not feasible to include more than four variables (see the Methodological Note at the end of this article for other issues related to our regression model).


The regression model revealed two significant findings.  The number of Trump campaign events was significantly related to changes in support for Trump.  All else equal, one Trump event translated into an increase of 0.2 percent in Trump’s vote support relative to Clinton’s.  Conversely, given the small sample size, the impact of Clinton’s events was too small to be distinguished from zero, though it was negative (-0.17) as expected.  The regression also revealed a strong negative relationship between the size of the Hispanic/Latino voter eligible population and changes in voter support for Trump.  In Colorado, Trump’s campaign events were particularly ineffectual.

Our model’s implications are summarized in the table below.  By setting the Hispanic/Latino population at two levels – 17.8% (for large Hispanic/Latino population states) and 3.2% (for low Hispanic/Latino population states – we analyzed the impact on Trump’s relative vote support given different numbers of Trump campaign visits.  In low Hispanic/Latino population states, five Trump events translated into a 3.1 percentage-point increase in his support relative to Clinton’s.  Twenty events gave Trump a 5.7 percentage-point bump up in those states.

In contrast, in high Hispanic/Latino population states, the impact of Trump events were dampened.  It would take more than 10 Trump campaign events to overcome his disadvantage in those states.

A final note, while we did not include the impact of advertising and news coverage in our regression model, we do believe those factors might help explain the intrinsic and pervasive disadvantage Trump had with Hispanic/Latino voters on a national level.  Whatever the strategic and tactical flaws in the Clinton campaign overall, it does appear her campaign was very successful at checking Trump’s support gains in states with large numbers of Hispanic/Latino eligible voters.

As for the impact of the Comey Letter or the Podesta emails, that will be left for a more thorough analysis.  Again, we do not believe inclusion of these factors would change our state-level findings.  Those are national events filtered through national media sources.  In our four-variable model, the impact of those factors may be buried in the ‘regression intercept’ which represents the starting point for changes in Trump’s support, setting all other factors to zero.  Here, the intercept was estimated to be 3.2.  That is, Trump would have gained about a three percentage-points versus Clinton in October even if all other factors — candidate visits and the size of the Hispanic/Latino vote eligible population — were non-factors.  If there is a Comey and/or Podesta effect, it is in there somewhere.  In a close election, like the last one, those could have been decisive factors, even if they weren’t the biggest determinants of the election outcome.


It is fair to ask, how does a candidate visit change voter opinions (or, more likely, the likelihood that the voter will vote) given that so few people attend a campaign rally?  Even Trump rallies attracted only about 5,000 people per event – and most of those people were already Trump supporters.  How does a really change an election outcome?

We see it as a multi-stage process powered by word-of-mouth and television coverage.  We don’t think many voters change their vote choice because they attended a rally.  Instead, based on telephone interviews conducted with Trump supporters in late October in Polk County, Iowa, we heard Trump supporters repeatedly talk about the excitement surrounding the Trump campaign.  Had they attended a Trump rally?  In all but a few cases, no.  Instead, their sense of the excitement came from television coverage and world-of-mouth from family, friends, and work colleagues, some of whom had attended a Trump event.  That is a one-two-punch that, anecdotally at least, leads us to believe a single campaign event can indirectly move voters into the voter booth in favor of one candidate over another.  It’s a dynamic that we don’t think worked as consistently or as often in the Clinton campaign’s favor.


Admittedly, our analysis doesn’t address the impact of news coverage and advertising (TV, radio, direct mail).  We know those campaign factors matter – a lot.  Nonetheless, we can explain over 63 percent of the variation in the state-level vote change during the last month of the campaign by referencing just three variables:  The number of Trump visits, the relative size of the Hispanic population in each state and a deflation factor unique to Colorado.

The Clinton campaign spent much more money on advertising, according to Kagan Media; and, we have yet to see an analysis suggesting Clinton’s TV and radio were of a lower quality than Trump’s.  In our judgement, Clinton’s TV and radios were qualitatively superior to Trump’s – and we would say, it is not even close.  Clinton’s TV spots were brilliantly produced and powerful.  As the father of an autistic son, the Clinton spot with the mother of an autistic child lamenting Trump’s mocking of a handicapped New York Times reporter, was as moving as any political ad I’ve ever seen.

Furthermore, contrary to the current media narrative, Trump’s news media coverage in the general election period was significantly more negative than Clinton’s (see the charts below provided by Harvard’s Shorenstein Center).  The ‘horse-race” coverage was particularly tilted in Clinton’s favor.

Thus, we are skeptical that the inclusion of advertising or the tone of media coverage would change our main finding that campaign events were a significant factor in Trump’s electoral college triumph.

Combine the impact of Trump’s rallies with the cold-hearted efficiency of the Trump campaign’s allocation of campaign resources into the key “swing” states, and you get the most improbable electoral victory in US. Presidential history.


Questions and press inquiries regarding this article and the data analysis contained within can be directed to: or call Kent Kroeger at 515 512 2776


METHODOLOGICAL NOTES:  There are a number of caveats to our analysis.  First, we did not include state-level media buys by the two campaigns.  Our initial source for media buy data (The University of Maryland’s Center for Political Communication and Civic Leadership) lacked the fidelity we required at the state-level; therefore, we did not formally test for the impact of media buys.  However, in the media data we did obtain, the variation between non-swing states and swing states far exceeded variation between each of the 15 swing states.  We are confident that the inclusion of the media buy data would not change the results reported here.  Second, the candidate event data includes both the presidential nominee visits (Trump, Clinton) and vice-presidential nominee visits (Pence, Kaine).  Third, we only have 15 data points (one for each state) making it difficult to discern weaker relationships within the data.  For example, we find Clinton campaign visits had no significant impact on voter support.  We believe an analysis at the state and weekly levels might better reveal the nature of that relationship.