An estimated $2.83 billion was spent on election-related political advertising on broadcast television and national cable in the 2015–2016 cycle, which represents about 4.29 million ad airings (Table 1). Just over 1 million ads aired in the presidential race – including the nomination races – at an estimated cost of $845 million. In Senate contests, over 926,000 ads aired on broadcast television at an estimated cost of $655 million, while 621,000 ads aired in races for the US House at an estimated cost of $335 million. Over 1.25 million ads aired in down-ballot races, in contests from state attorney general and state representative to mayor and county coroner.

To put the over 4 million airings in context, if every one of these ads aired back-to-back, they would be broadcast for nearly 1500 straight days without stop. And ad spending in federal races, at an estimated cost of over $1.8 billion, accounts for 27 percent of the reported $7 billion spent on federal campaigns in the 2015–2016 election cycle, according to estimates from the Center for Responsive Politics.²

The sponsorship of political ads in 2016 varied considerably by office, as Table 2 shows. Candidates sponsored about 71 percent of ads aired in the presidential race, with outside groups sponsoring almost all of the remaining spots. Interestingly, the political parties were largely absent from the presidential race (as we discuss in more detail below). Advertising was more varied in the races for US Senate, with candidates sponsoring the largest chunk of ads but groups sponsoring a considerable share: 38 percent. It is estimated, in fact, that groups, in spite of airing fewer ads, spent more on ads than candidates did in Senate rates. This cost discrepancy (the average cost per ad for Senate candidates was estimated to be $549 compared to over $1000 for groups) is likely because groups are not entitled to the lowest unit rate from television stations, meaning they pay whatever the market will bear for premium space in competitive races (Fowler, Franz, and Ridout, 2016, pp. 29–33). Candidates dominated advertising in House races, sponsoring 61 percent of ads. Parties were more relevant in House races, though, sponsoring 18 percent of House ads versus only 9 percent of Senate ads, and parties aired more ads than outside groups in House campaigns.³

Previous work on ad trends has noted the decline of party spending in Senate races. Fowler, Franz, and Ridout (2016, pp. 47–48) show that the party share of Senate ads declined from over 20 percent of ads aired in 2000, 2002, and 2008 to about 10 percent of ads aired since 2010. Party spending in Senate races in 2016 was at its lowest share in the time series of data from the Wisconsin and Wesleyan projects. This is particularly noteworthy given the intense focus this cycle on the possibility of a partisan flip in control of the chamber. As we will discuss below, the relative balance of power between political parties and outside groups has shifted to advantage groups, many of whom in the aftermath of the Supreme Court’s 2010 decision in Citizens United v. FEC are no longer beholden to limits or bans on the raising and spending of large donations for uncoordinated pro-candidate advocacy.

Also notable, advertising volumes in the presidential election of 2016 were down considerably from 2012. Figure 1 shows the volume of ads aired in the top 75 media markets in the presidential races between 2000 and 2016. We limit comparisons to the top 75 media markets because only these markets were available in the earlier years. Ad volumes in 2016 were down by over 300,000 airings from the previous cycle, though levels are comparable to 2004 and 2008. These declines are apparent despite both parties having competitive and long presidential primary campaigns, whereas 2012 featured only one party with a contested nomination race.⁴

Figure 1:

Volume of Ads in Presidential Race (Top 75 Media Markets, January 1–Election Day).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

In contrast, large declines in the volume of advertising in Senate and House races are not evident. Figure 2 shows the volume of broadcast advertising in US Senate races (left) and US House races (right) in each election cycle from 2008. Although the number of ads in 2016 was down slightly from 2014 to 2012, it remained above 2010 levels (the last cycle featuring races in the same states) and well above 2008 levels. In the House, broadcast television ads were up slightly from 2014 to just over 600,000 but down a bit from 2010 to 2012. These figures include all broadcast ads aired in the 210 media markets across the country, but even a comparison back to 2000 in the top 75 media markets shows only slight declines in congressional ads in 2016 in comparison to previous years.

Figure 2:

Volume of Congressional Ads by Year (January 1–Election Day).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

Indeed, comparing volumes in House races across cycles needs to account for the number of competitive races. There were 43 House races in 2010 rated by Congressional Quarterly as “too-close-to-call.” This was the highest number in many decades (though 1994 was close with 40 such contests). The year 2012 featured 33 toss-ups, in comparison to 20 in 2014 and 19 in 2016. As competiveness is the primary driver of election spending, it is no surprise that 2010 featured the most political ads. As such, there is very little evidence in these totals of a decline in the emphasis placed by House and Senate candidates on traditional broadcast ad buys.⁵

There is much to consider in the presidential air war of 2016. In this section, we examine three highlights: (1) ad volumes from top sponsors, and in particular the efforts of the Clinton campaign in key states; (2) the increasing importance of local cable buys as a complement to traditional broadcast airings; and (3) the tone and focus of presidential ads in comparison to previous elections.

The top advertiser in the 2016 presidential race, including the primaries and general election, was the Clinton campaign, which aired over 400,000 ads on broadcast television at an estimated cost of $258 million (Table 3). In terms of ad count, the Sanders campaign – in spite of not airing any ads in the general election – came in second overall, airing more ads in the cycle than Donald Trump’s campaign. In fact, the Democratic presidential primary looked much similar to typical advertising campaigns, where rivals mutually rely on traditional local broadcast advertisements. As of May 8, for instance, the Sanders campaign had spent an estimated $73 million, and the Clinton campaign had spent approximately $62 million on ads on broadcast television. In comparison, only an estimated $2.3 million had been spent by outside groups in the Democratic contest.

Clinton’s huge ad advantage in the general election is evident in Table 4, which shows the number of broadcast ads and estimated spending on ads by each of the top ad sponsors during the period between September 5 and Election Day. The Clinton campaign aired just under 200,000 ads at an estimated cost of just under $120 million during the last 9 weeks of the campaign, dwarfing the Trump campaign’s 76,000 ads at a cost of $66 million. Clinton’s advantage becomes even more stark when one adds in the almost 50,000 spots aired by Priorities USA Action, a pro-Clinton super PAC. There was some outside group spending on behalf of Trump, including the NRA Institute for Legislative Action, which aired about 10,000 ads on broadcast during the general election period, and the Future 45 super PAC and its 501c affiliate, the 45 Committee.

Clinton’s advantage on broadcast television was considerable, though it did not extend to all media markets in the country. Figure 3 depicts in blue those media markets in which pro-Clinton ads outnumbered pro-Trump ads between June 8 and Election Day. In red are those media markets in which pro-Trump ads outnumbered pro-Clinton ads. Darker shades indicate a greater ad advantage. The only areas in which Trump held an ad advantage were in Utah, Hawaii, a few small markets in Ohio, northern Virginia and – most importantly – Wisconsin, a state that Trump ended up winning by fewer than 25,000 votes.

Figure 3:

Ad Advantages by Media Market (Presidential General).

However, these totals from June 8 to Election Day (both in total ads and across media markets) mask important variation over time. For example, although Michigan is colored blue in the map, Clinton’s advantage came only in the last week of the campaign. Figure 4 therefore shows the number of pro-Trump and pro-Clinton ads aired on broadcast television during each week, both overall (top left panel) and in three key states. Across all media markets, Clinton held an ad advantage in each week of the fall campaign. That ad advantage even grew in the final few weeks of the race when Trump’s advertising finally ramped up.

Figure 4:

Ads by Week (All Markets, Wisconsin, Michigan and Pennsylvania).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

In Wisconsin, however, Trump and his allies (primarily three groups – Reform America Fund, 45 Committee, and Rebuilding America Now PAC) started advertising in the state in mid-September and continued to advertise, albeit at fairly low but consistent volumes, until Election Day. Indeed, Trump was on the local broadcast airwaves unchallenged by Clinton forces until the last minute, when she barraged the state with over 4000 ads in the final week of the campaign.⁶

Michigan, a state that Trump won by a mere 12,000 votes, was not really on either campaign’s radar until November 1 when Clinton and Priorities USA Action went on the air aggressively. Still, Trump aired about 1500 ads in the first week of September and Right to Life of Michigan had a small ad buy in the final week of October, meaning pro-Trump ads were the only ads on local broadcast stations until the last week of the campaign, when Clinton overwhelmed the GOP camp.

The other state whose outcome surprised many prognosticators on Election Day was Pennsylvania, which Trump won by about 70,000 votes. Judging by the amount of advertising in the state in the fall, both campaigns must have deemed the state close. In each week, Clinton held an ad advantage, although the advantage tightened considerably in early October but mirroring the national trend, increased again during the final few weeks of the campaign.

Indeed, Clinton clearly mobilized aggressively in the final weeks, both nationally and in these three states. Some political science research would suggest that this was a smart strategy, as political ad effects are assumed to be short-lived. In particular, research estimates that for most races, advertising effects decay quickly, generally lasting only days (Gerber et al. 2011), though in presidential races advertising effects may persist for up to 6 weeks (Hill et al. 2013). Perhaps the Clinton campaign and its allies assumed, given this, that late ad buys would score the maximal effect.

However, research on the question of ad decay is not settled, mostly because of a dearth of studies that test the robustness of the claim. Moreover, it seems reasonable to argue that such decay may be conditional on many factors [see especially the argument in Bartels (2014) on active processing], including pre-existing opinions of the candidates. Because both Clinton and Trump suffered from record-high levels of disapproval among voters, a more sustained advertising campaign may have been necessary to move pre-existing sentiment. Of course, it is also possible that because both candidates were so well known among the electorate no amount of advertising, even great imbalances of ads from one side, would have the ability to influence vote choice. If true, then the 2016 presidential contest would truly be sui generis. (And of course none of this counts out external influences such as the Comey letter.)⁷

A second factor in the Clinton-Trump contest is also important to consider. Much research on televised political ads uses data from the Wisconsin and Wesleyan archives of broadcast ad buys. But scholars and political operatives know that the media environment is changing, and candidates are increasingly turning to places like local cable to buy ads on shows with more homogeneous sets of viewers. Local cable refers to ads purchased in a media market but on channels other than the major local NBC, CBS, ABC, and Fox affiliates. This would include HGTV in Manchester, New Hampshire, and ESPN in Louisville, Kentucky. In addition to niche channels, local cable can also be purchased for zone areas within media markets, enabling even more targeted advertising than is possible through local broadcast buys, which reach the entire media market. Because the audiences of these cable networks are not as diverse as the broadcast stations and the purchasing can be designated for sub-regions of the media market, buying time on local cable allows candidates to reach a more desirable and niche audience.

One possibility is that the decline in broadcast ads noted above was replaced by a spike in local cable ads. We obtained market-level local cable buys from NCC Media, which tracks that venue. Drawing conclusions based on volume comparisons across broadcast and local cable can be difficult because the size of the audience watching ads on local cable shows can be very small, and the expanse of cable channels allows for more overall “real estate” on which to buy space. Moreover, the cost to purchase an ad on local cable is typically much cheaper than on broadcast. For example, despite the volume differences shown in Figure 5 described below, NCC Media estimates that local cable received roughly 21 percent of all candidate dollars devoted to television advertising. However, we can make some comparisons across cycles that are instructive.

Figure 5:

Volume of Candidate-Sponsored Presidential Ads on Broadcast (left) and Local Cable (right).

Source: Kantar Media/CMAG and NCC Media with analysis by the Wesleyan Media Project.

Although we do not have figures on all ad sponsors, we can show ad volume for Clinton and Trump in 2016 and for Obama and Romney in 2012. These are displayed in Figure 5. Local broadcast and cable totals in the figure cover the period from early June in each year through Election Day. These totals include only candidate-sponsored ads; party and coordinated buys are excluded. For the Democrats, note the differences across elections in each panel. Clinton purchased fewer ads than Obama on local broadcast (this is part of the volume declines noted above) but more advertising on local cable. Indeed, Clinton aired almost 150,000 more spots than Obama on local cable. On the Republican side, Trump aired far fewer ads on local broadcast than did Romney, and he aired just a handful of local cable spots, further widening the discrepancies on air. Both the broadcast and local cable totals tell a truly unparalleled story of imbalances in political ads.⁸

The literature is silent on the effects of local cable ads on measures of and changes in candidate favorability or reported vote choice, but the evidence on micro-targeting (of which this is a blunt example) suggests the strategy may be effective (Hillygus and Shields 2007). Of course, for targeting to be effective, key assumptions about the voter population must be correct, and in this atypical year, it is possible that the assumptions were faulty. It does seem clear that the Clinton campaign relied heavily on targeted strategies, especially in some key states, such as Wisconsin and Michigan, where her campaign aired local cable advertising rather than broader appeals on local broadcast. Here too the outcome of the election belies a clear inference that such efforts worked.

Given this, it is important to consider the content of each candidate’s ads. We know from prior work that ad imbalances are strongly correlated with changes in vote share, but we also know that the content of ads can condition those effects (Ridout and Franz 2007). Consider first simple measures of ad tone. For all of the vitriol in the 2016 presidential election (in rallies, the debates, on cable news programs), the tone of political advertising was actually less negative than it was in 2012. The Wesleyan Media Project (consistent with the Wisconsin Advertising Project before it) classifies negative ads as those that solely mention an opponent, positive ads as those solely mentioning the sponsor, and contrast ads as those mentioning both. As shown in Figure 6, which compares the 2016 campaign to the previous four cycles, 2016 was less negative than 2012 as measured both by the proportion of pure attack ads and the proportion of positive advertising on the air. The 2016 election did, however, earn the distinction of the second most negative in the last decade and a half.

Figure 6:

Tone of 2016 General Election Presidential Advertising (June 8–Election Day).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

The candidates took different strategies when it came to the tone of their advertising, though, as Figure 7 shows. Nearly half of all Clinton campaign airings were negative whereas over half of Trump campaign airings were contrast spots, which discussed Clinton negatively but also provided information about Trump. Similar to prior cycles, the party and outside groups were attack dogs; their airings on behalf of both candidates were much more negative than the candidate-sponsored advertising.

Figure 7:

Tone of 2016 General Election Presidential Advertising by Sponsor (June 8–Election Day).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

The advertising tone breakdown alone, however, hides an important difference in strategy that made 2016 advertising very unusual. Namely, the majority of the Clinton campaign’s negative advertising attacked Trump’s characteristics and personality. In other words, the attack ads were personal-focused as opposed to policy-focused (Figure 8). Fewer than 10 percent of ads attacking Trump focused on his policies whereas about 90 percent was focused on Trump as an individual. Clinton’s contrast ads were similarly devoid of policy discussion. By and large, it was only in ads promoting Clinton that the campaign actually discussed policy, and those ads comprised only 30 percent of her overall mix on air. Clearly, the Clinton campaign’s strategy was to disqualify Trump based on his temperament, not on his policy positions, in ads like “Role Models” and “Mirrors” featuring Trump’s voice and children and/or young girls listening. By contrast, about 70 percent of ads from Trump and his allies that attacked Clinton contained at least some discussion of policy, and when there were contrasts drawn between the two candidates, those contrasts were almost all policy-based such as the “Two Americas” ads, which explicitly compared how Hillary Clinton’s America would differ from Donald Trump’s America.

Figure 8:

Tone and Substance of 2016 Presidential Candidate Advertising (June 8–Election Day).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

In total, over 60 percent of ads supporting Clinton were solely about candidate characteristics, while only about 25 percent were focused on policy. This is a huge difference from Trump’s advertising, over 70 percent of which was focused on policy, and it is a huge difference from every other presidential campaign for which we have comparable data (Figure 9). In a typical campaign, ads that focus on candidate character have comprised less than 20 percent of total ad airings, and in some years like 2000, there were hardly any ads that focused on the candidates’ character.

Figure 9:

Substance of Presidential Advertising over Time (June 8–Election Day).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

For all of the talk of the unusual advertising campaign that Trump ran in 2016, his message strategy was more traditionally policy-focused. Ironically, it was the Clinton campaign that deviated sharply from the conventional playbook when it came to messaging despite following conventional norms in terms of volume, placement and targeting of ads. Indeed, only one in four Clinton campaign ads focused on policy, which is by far the lowest percentage we have seen since data from Kantar Media/CMAG have been available. How exactly this message strategy may have played into voting decisions will be discussed at more length in the conclusion.

Outside groups (non-candidate, non-party organization) continued to be an important sponsor of political advertising in 2016. In the wake of the Supreme Court’s shift to a more deregulatory perspective on campaign finance, begun in earnest with Citizens United v. FEC, outside group spending has jumped considerably. In 2016, outside groups accounted for about 28 percent of all ads aired in federal races, as Figure 10 shows. This percentage is very similar to 2012 and 2014 but is a large increase over the previous six election cycles, none of which saw more than 15 percent of advertising come from groups.

Figure 10:

Percentage of All Ads in Federal Races Sponsored by Groups.

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

Of course, the extent to which groups were involved varied considerably across races. Figure 11 shows the percentage of ads aired after September 4 that was sponsored by outside groups in the presidential race and in the most-advertised House and Senate races. During that time period, groups (including super PACs, 527s and 501cs) sponsored about a quarter of all ads in the presidential race. But group sponsorship ranged from under 10 percent in some House races to just about 50 percent in California’s 21st congressional district. Groups also paid for just about half of all ads in the Senate races in Nevada and Pennsylvania – and over 40 percent in New Hampshire, Missouri and Florida.⁹

Figure 11:

Percentage of Group-sponsored Ads by Race in 2016 (September 5–Election Day).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project. Only races with at least 10,000 ads are shown.

The increased involvement of groups in electoral politics, especially since 2010, has coincided with a decline in ad sponsorship by parties. We noted this earlier with respect to Senate elections, and we show the trend for presidential elections in Figure 12. In 2000, over half of the ads that aired in the presidential race were sponsored by political parties. These were primarily “soft money” ads where parties used unregulated accounts for spots they claimed were “issue advocacy” or “party building,” despite the ads’ containing mostly promotional or attack appeals for or against the party nominees. The Bipartisan Campaign Finance Reform Act of 2002 forced parties to raise and spend only regulated “hard money.” Concurrently, outside groups benefited from the deregulatory perspective of the Supreme Court, noted earlier. Moreover, Barack Obama in 2008, followed by all of the major party nominees in 2012 and 2016, opted out of general election public funding, allowing them to raise hundreds of millions of dollars outside of the party. As such, the percentage of party ads has steadily fallen each election cycle, bottoming out in 2016.

Figure 12:

Proportion of Group and Party Airings in Presidential General Elections.

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

While advertising in the presidential race in 2016 did not follow traditional patterns, advertising in Senate and House races generally did. For one, advertising was concentrated in the states with the most competitive races. Table 5 shows that the Senate races with the most ads were in Pennsylvania, Indiana, Ohio, North Carolina and Missouri – all races that were ranked as highly competitive at some point in 2016. There is no evidence of any major candidate disengaging from political ads or being outspent in the same manner as Donald Trump.

Second, the political parties continued to be comparatively more involved in these races than in the presidential, though with considerably fewer ads than outside groups. All told, the parties sponsored 10.6 percent of general election ads in US Senate races, and another 4.3 percent of ads were paid for with party coordinated spending (Table 6). Groups sponsored almost 40 percent of ads aired in the general election (and 30 percent in primary races), while candidates paid for almost 46 percent of the spots. The candidate share of ad airings was much higher in primary races, where candidates paid for two in three ads.

The tone of Senate advertising was also quite similar to recent years, with just over 50 percent of ads being negative ads (Figure 13). Although this percentage is higher than for any year for which we have data (now nine election cycles), it is very similar to the level of negativity in every election cycle since 2008, a year in which negativity spiked somewhat.

Figure 13:

Tone of Senate Advertising Over Time (September 1 to Election Day).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

For contests in the US House of Representatives, shown in in Table 7, we see that the top race was that for Maine’s 2nd congressional district, a race that most pundits labeled a pure tossup. The race drew almost 27,000 general election ads at an estimated cost of $8.2 million. Montana’s at-large congressional race and Minnesota’s 8th congressional district also drew substantial general election advertising.

In contrast to the presidential race, both the Democratic and Republican parties poured considerable resources into these House contests. Including both general election and primary ads, parties were responsible for 24.3 percent of ads, if one includes both party and coordinated expenditures together (Table 8). This is significantly greater than the 14.9 percent of ads sponsored by outside groups. Clearly, although parties have almost abandoned the presidential race – and have taken a backseat to groups in Senate races – they remain an important player in races for the House.

The tone of House races was also relatively similar to recent election cycles, with just slightly more negative ads and slightly fewer positive ads than 2014 (Figure 14). But levels of negativity in 2016 were down from 2010 to 2012. As with volume comparisons across cycles in House races, though, levels of negativity depend on the number of competitive elections, and 2010 and 2012 featured more competitive races.

Figure 14:

Tone of House Advertising over Time (September 1 to Election Day).

Source: Kantar Media/CMAG with analysis by the Wesleyan Media Project.

There was a huge imbalance in advertising at the presidential level in 2016, with Clinton and her allies pummeling Trump and his allies in almost every media market in the country. Pro-Clinton advertising dominated on local broadcast, national cable and even local cable television. Decades of research (e.g. Zaller 1992) suggests that in the face of such imbalanced message flows, the advantaged candidate should benefit at the polls. And yet in spite of those advantages, Hillary Clinton lost many of the states that saw the most political advertising. Is the message from 2016 that political ads just do not work anymore?

Although the impact of advertising in 2016 on the outcome of the general election race was likely minimal, we urge caution in concluding that television advertising is no longer effective. For one, we never expect advertising to have a large effect in a presidential race. Research has demonstrated that advertising effects are smaller in presidential contests than in other down-ballot races (Ridout and Franz 2007), and this may be especially true in a presidential race that features two well-known candidates, a former First Lady and a celebrity who has been on Americans’ television screens for decades. Because voters have a large store of pre-existing information about the candidates, they are less open to influence by advertising. Nevertheless, advertising is an important way in which candidates can talk directly to voters, and research does suggest that advertising advantages do correlate with movement in the polls (Ridout and Franz 2007; Sides and Vavreck 2014). Despite the fact that pro-Clinton advertising dominated the airwaves overall, in several of the pivotal states – especially Michigan and Wisconsin – Donald Trump actually had ad advantages on local broadcast up until the very last week of the campaign, which may well have mattered.

Second, in a presidential race, advertising must compete with almost limitless media coverage of the race, which tends to neutralize the impact of the advertising. Perhaps the intense media coverage in 2016 was driven less by the messages of political advertising as it has been in the past and more by Donald Trump’s Twitter account. Although the Clinton campaign may have viewed the news media’s focus on Donald Trump as beneficial to them, they also claimed to have trouble getting coverage of policy statements. Yet the Clinton campaign itself did not focus on policy in its own messaging either, a tactic that very well may have hurt their candidate, which leads to our next point.¹⁰

Third, all ads are not equally effective. Even though the ad advantage may have been in Clinton’s favor, it does no good if the message does not resonate with voters. Of course, the jury is still out on how effective Clinton’s message was, but evidence suggests that negativity in advertising can have a backlash effect on the sponsor (Pinkleton 1997) and that personally-focused, trait-based negative messages (especially those that are uncivil) tend to be seen as less fair, less informative and less important than more substantive, policy-based messaging (Fridkin and Geer 1994; Brooks and Geer 2007). In stark contrast to any prior presidential cycle for which we have Kantar Media/CMAG data, the Clinton campaign overwhelmingly chose to focus on Trump’s personality and fitness for office (in a sense, doubling down on the news media’s focus), leaving very little room for discussion in advertising of the reasons why Clinton herself was the better choice. Trump, on the other hand, provided explicit policy-based contrasts, highlighting his strengths and Clinton’s weaknesses, a strategy that research suggests voters find helpful in decision-making (Mattes and Redlawsk 2014). These strategic differences may have meant that Clinton was more prone to voter backlash and did nothing to overcome the media’s lack of focus on Clinton’s policy knowledge, especially for residents of Michigan and Wisconsin, in particular, who were receiving policy-based (and specifically economically-focused) messaging from Trump. As such, it may very well be that Clinton misallocated advertising funds (both hyper-targeting on local cable and advertising in non-traditional battlegrounds like Arizona rather than in the Midwest, for example) and a lack of policy messaging in advertising may have hurt Clinton enough to have made a difference.

While advertising patterns in the 2016 presidential race were certainly a departure from the past, it seemed to be business as usual in the House and Senate. The tone of these races was very similar to the past few election cycles, as was the volume of advertising. And the sponsorship of advertising, with outside groups taking on a heavier role in the Senate than House, also reflected the past few elections. Indeed, it is almost surprising that so much has stayed the same with television advertising given the recent rise of digital media and social media.

But will down-ballot candidates in the future read the outcome of the 2016 presidential race as a signal that spending millions on television advertising just is not necessary? While many might consider such an approach in 2018, we think it would be unwise for candidates to use Trump as a model. For sure, future campaigns will invest heavily in new approaches to reach voters, from local cable to digital and web-based advertising. But at least in 2016, it seems that these additional tactics are viewed as just an “add-on” way to reach voters (a complement if you will) rather than as a replacement for more traditional ad buys on local broadcast stations. In sum, the race between Donald Trump and Hillary Clinton violated so many of our rules of political science that generalizing from it strikes us as risky – at least for now.

We thank Laura Baum, Dolly Haddad, Matt Motta and our student coding team for their help. We also thank Tim Kay of NCC Media for access to local cable data.