Chapter 25: Political Advertising: From TV Spots to Targeted Ads

Jake Rourke, Tom Whitfield's campaign manager, has a whiteboard in his office with two columns. The left column is labeled "TV." The right column is labeled "Digital." Under each column are numbers—not budget totals, but what he calls "effect estimates": his working model of how many persuadable voters each dollar spent in each channel is moving. Jake stares at the board most mornings during the final stretch of the Garza-Whitfield Senate race, recalculating after each new data point comes in.

Three offices away in a different building, Nadia Osei, Garza's analytics director, has a similar problem but a different budget constraint. Whitfield has significantly outraised Garza, and outside groups are spending heavily on his behalf. Nadia's question is not "how do we allocate the optimal budget across channels" but "how do we allocate a constrained budget across channels to maximize impact against a better-funded opponent." The answer to these strategic questions requires understanding how political advertising works—not just rhetorically, but causally, at the level of measured voter behavior change.

This chapter examines political advertising from both the strategic and research perspectives. We trace the medium's evolution, examine what experimental evidence shows about when and how advertising moves voters, and use the Garza-Whitfield race as a running example of how campaigns with different resources and strategic situations navigate the advertising landscape.

25.1 The Evolution of Political Advertising

From Print to Broadcast: The First Century

Political advertising in America predates television by over a century. Nineteenth-century campaigns communicated through print media—handbills, newspaper advertisements, and campaign posters—supplemented by mass-produced physical materials (buttons, banners, and the famous Lincoln log cabin imagery). Radio advertising, beginning in the 1920s, added audio—FDR's "fireside chats" were not paid advertisements, but they demonstrated the intimate persuasive power of the medium, which political advertisers quickly exploited.

The decisive transformation came on October 5, 1952, when Dwight Eisenhower's campaign aired the first political television advertisements ever broadcast in an American presidential campaign: a series of thirty-second spots featuring Eisenhower answering questions from "ordinary Americans" in a format pioneered by ad agency BBD&O. Eisenhower's spots were produced by Rosser Reeves, who had invented the "unique selling proposition" approach for commercial advertising and applied it to political communication with unsettling directness. The ads were derided by Adlai Stevenson (Eisenhower's Democratic opponent), who considered them demeaning to democratic discourse. Eisenhower won in a landslide.

The next landmark came twelve years later. Lyndon Johnson's 1964 "Daisy" advertisement—aired only once officially—depicted a small girl counting flower petals while a missile launch countdown played in the background, ending with a nuclear explosion and Johnson's voice warning of the stakes of the election. Barry Goldwater's name was never mentioned. The "Daisy" ad is widely cited as the first major demonstration of negative advertising's emotional power and the way emotional political advertising could set agendas through controversy that multiplied its reach far beyond its paid placement.

The Professionalization of Political Advertising

The period from the 1960s through the 1990s saw the creation of a specialized political consulting industry organized around campaign advertising. Firms like GMMB, Bully Pulpit Interactive, the Stevens and Schriefer Group, and dozens of others developed the organizational structure of modern political advertising: strategic advice, message development, polling, production, and media placement as integrated services.

The production economics of television advertising drove significant standardization: thirty-second spots are the dominant format because they are the standard television advertising unit, which determines both production conventions and media buying logistics. Sixty-second spots are used for more complex messages or introductory biographical ads; fifteen-second spots are used for rapid-response or cut-through contexts.

The media buying economics of broadcast television revolve around the gross rating point (GRP)—a unit that represents 1 percent of the target audience exposed to an advertisement once. Buying 500 GRPs means that, on average, your target audience was exposed to your advertisement five times each (100% coverage, 5 times = 500 GRPs; or 50% coverage, 10 times = 500 GRPs). GRP planning requires decisions about both reach (what share of your target audience will you reach at all?) and frequency (how many times will those you reach see the ad?).

📊 GRP Math in Practice

In the Garza-Whitfield race, Whitfield's campaign is targeting registered voters in the state between 35 and 65—his core persuasion demographic. The state's largest media market contains approximately 850,000 registered voters in that demographic. Jake Rourke estimates that 500 GRPs per week in that market will reach roughly 75 percent of those voters an average of 6.7 times. At a cost per rating point of approximately $1,200 for 30-second spots in that market's prime news programming, 500 GRPs cost approximately $600,000 per week.

This arithmetic explains why television advertising is the largest single line item in most competitive Senate and gubernatorial race budgets, and why it becomes prohibitive for campaigns in large states (California, Texas, New York), where media markets are so expensive that competitive television advertising requires budgets that dwarf what most campaigns raise.

25.2 Ad Spending Data: Following the Money

FEC Disclosure Requirements

Under the Federal Election Campaign Act and its amendments, campaigns are required to disclose all expenditures above $200 to the Federal Election Commission, with detailed vendor and purpose information. This disclosure infrastructure makes political advertising uniquely transparent among commercial activities—every television and radio ad buy by a federal campaign is, in principle, publicly traceable through FEC disclosure data.

The FEC database, available at FEC.gov, includes: committee name, payee (the media buyer or television station), amount, date, and purpose description. For advertising, purpose descriptions range from specific ("TV advertising placement—Albuquerque market") to vague ("media services"). The granularity varies by campaign; sophisticated campaigns often use intermediate vendors (media buying agencies) in ways that reduce the specificity of FEC-visible information.

The Wesleyan Media Project

The most important academic resource for tracking political advertising is the Wesleyan Media Project (WMP), housed at Wesleyan University. WMP uses a combination of FEC data and commercial ad monitoring services to track the content, volume, and targeting of political advertising across Senate, gubernatorial, House, and presidential races. Its publicly released datasets track ad airings by market, week, tone, and issue emphasis.

WMP data enables analysts to answer questions that FEC data alone cannot: not just how much a campaign spent on television, but what messages it was running, in which markets, in what quantities, and with what tonal valence. The WMP's annual reports on Senate race advertising are a standard reference in academic political communication research and increasingly used by campaign analysts.

📊 Wesleyan Data Applied: Reading the Garza-Whitfield Advertising Landscape

Pulling WMP-equivalent tracking data for the Garza-Whitfield race at the eight-week mark, Nadia Osei's team at the Garza campaign sees the following picture:

• Total ad airings to date: Whitfield campaigns (candidate committee + aligned outside groups): 12,847 airings. Garza campaigns (candidate committee + aligned outside groups): 7,204 airings.
• Tone breakdown: Whitfield ads: 43% positive (promoting Whitfield), 57% contrast/negative (attacking Garza). Garza ads: 61% positive (introducing Garza), 39% contrast/negative.
• Issue emphasis: Whitfield ads emphasize public safety (67% of airings), immigration (45%), and economic anxiety (38%). Garza ads emphasize healthcare (52%), education (41%), and Garza's prosecutorial record (37%). (Percentages exceed 100% because ads can address multiple issues.)
• Market targeting: Whitfield is heavily concentrated in the state's three largest markets (suburban and exurban). Garza is running advertising in 7 markets, including two smaller markets where Whitfield has no presence.

Nadia uses this data to construct a "competitive ad environment" score for each market—the ratio of Garza-favorable to Whitfield-favorable ad airings, weighted by audience size. The Garza campaign is losing the advertising air war in the large suburban markets and winning (by default, in the absence of Whitfield advertising) in the smaller markets. The question is whether the smaller markets have enough persuadable voters to justify the resource allocation.

25.3 Message Testing: How Campaigns Know What to Run

The Testing Infrastructure

Professional campaign advertising does not go on air untested. The major campaigns in competitive races run advertisements through multiple testing protocols before committing to media buys:

Focus groups are small (typically 8-12 participants), moderated discussions that provide qualitative insight into voter reactions to advertising concepts and finished spots. Focus groups are good at surfacing the language and concerns of specific voter segments, revealing unexpected interpretive frameworks, and identifying emotional responses that quantitative testing misses. They are not good at producing reliable estimates of population-level persuasion effects—sample sizes are far too small for statistical power, and the group dynamics of moderated discussions create conformity pressures that distort individual reactions.

Campaigns use focus groups strategically: for concept testing early in an advertising cycle (which messages resonate at the level of idea, before production?), for ad finishing (does this spot's final form communicate what we intended?), and for understanding voter segments that quantitative data cannot fully characterize.

Dial testing is a method in which focus group participants use hand-held dials to record their second-by-second reactions to video content, providing a continuous trace of emotional response throughout an advertisement. Dial testing reveals which moments in an advertisement are landing well and which are losing the audience—information that guides the editing of spots still in production. The dial trace also reveals where different demographic subgroups (Democrats, Republicans, independents; men, women; college-educated, non-college) diverge in their reactions, providing guidance for audience targeting.

💡 What Dial Testing Can and Cannot Tell You

Dial testing is excellent at identifying emotional response peaks and valleys within an advertisement. It is poor at predicting aggregate persuasion effects because: (1) focus group participants are not a representative sample; (2) laboratory viewing conditions (watching with strangers, holding a device, knowing you're being observed) differ from real-world viewing; (3) the dial captures immediate emotional reaction, not the persistent opinion change that constitutes actual persuasion. Campaigns that treat high dial scores as reliable predictors of electoral impact are making an inference the method does not support.

Online message testing has grown dramatically in sophistication and accessibility. Survey platforms like Qualtrics, Lucid, and specialized political testing firms like Swayable enable campaigns to deploy online survey experiments—randomizing which respondents see which advertisements and measuring opinion change between control and treatment groups—at relatively low cost and with large enough sample sizes to provide statistical power.

The standard online message test design: a representative (or targeted-representative) online sample is recruited; respondents are randomly assigned to see one of several ad versions (or a control condition with no ad); pre- and post-exposure measures of candidate favorability, vote intention, and issue salience are collected; differences between conditions estimate the ad's persuasive effect on each outcome.

Online message testing can be completed in 48-72 hours, making it usable for rapid response advertising. Nadia's team ran online tests of three potential Garza response ads to Whitfield's crime attack within 36 hours of the Whitfield ad appearing on television.

The Limits of Message Testing

⚠️ Testing Is Not the Same as Predicting

Message testing—whether by focus group, dial testing, or online survey—is conducted under conditions that systematically differ from the real-world media environment. In the real world: ads compete with thousands of other information inputs; advertising is repeated over days and weeks rather than shown once; voters encounter advertising in contexts (during favorite television programs, in targeted digital placements) that affect their reception; and the information environment at the time of viewing (what else has happened in the news?) shapes interpretation. Testing environments control for these factors in ways that produce clean experimental estimates—but those estimates may not translate accurately to real-world effects. The best message testing tells you something about relative ad effectiveness; it should not be taken as a reliable predictor of actual vote share change.

25.4 Negative Advertising: The Research and the Reality

The Case for Going Negative

Negative advertising—advertising that attacks an opponent rather than promoting the sponsoring candidate—has been the dominant form of political advertising in competitive American federal races since the 1980s. This is not a coincidence of campaign culture; it reflects substantive strategic logic supported by considerable evidence:

Negativity bias in human cognition. An extensive body of psychological research documents that negative information is cognitively weighted more heavily than positive information of equal objective magnitude. People remember negative information longer, process it more thoroughly, and integrate it more strongly into evaluations than comparable positive information. This "negativity bias" provides a theoretical foundation for expecting negative advertising to be more persuasively efficient per dollar spent than positive advertising.

Contrast advertising's dual advantage. Many of the most effective political advertisements are technically "contrast" ads—they simultaneously promote the sponsoring candidate and attack the opponent, often within the same spot. Contrast advertising gets two messages for the price of one and, in experimental research, is generally at least as effective as negative ads while producing somewhat lower backlash risk.

Negative information travels further in digital environments. Research on social media content spread finds that negative, emotionally arousing content is more likely to be shared than neutral or positive content. An effective negative ad has "earned media" potential: if it generates controversy, it may be covered by news media and shared on social media at multiples of its paid placement cost.

Issue-specific vulnerability exploitation. Negative advertising can be precisely targeted at a candidate's weakest position on an issue that is salient to persuadable voters. Positive advertising must build a multi-dimensional case for your own candidate; negative advertising can focus single-mindedly on the one dimension where the opponent is most vulnerable.

The Case Against Going Negative—and the Evidence

The intuitive case against negative advertising is "backlash"—the idea that voters will react negatively to a candidate who attacks opponents, punishing the attacker rather than the target. This is the basis for the common folk wisdom that "going negative" is always risky. What does the evidence actually show?

The backlash evidence is considerably weaker than popular wisdom suggests. Most experimental and field research finds that negative advertising is, on average, at least as effective as positive advertising and often more effective, without producing meaningful backlash against the attacking candidate. When backlash does occur, it tends to be concentrated among:

• Voters who are already positively disposed toward the target of the attack (who process the negative information defensively)
• Voters with high "negativity aversion" toward negative campaigning in general (who are often less persuadable in any case)
• Contexts where the attack is perceived as unfair or inaccurate (where factual counterattacking is possible)

The most careful experimental work, including field experiments conducted during actual campaigns, finds that negative advertising effects are positive in direction (reducing opponent favorability and/or vote intention) but modest in magnitude—consistent with the broader finding that political advertising effects are real but rarely large.

🔵 The Demobilization Debate

A separate set of research on negative advertising focuses not on persuasion but on voter turnout. Ansolabehere and Iyengar's influential 1995 study "Going Negative" claimed that negative advertising demobilized voters—reducing turnout—particularly among independents. This was a politically significant claim: if negative advertising reduces the very turnout that campaigns are trying to generate, the rational strategic case for going negative is substantially weakened.

Subsequent research has substantially qualified this finding. Meta-analyses of negative advertising and turnout find effects close to zero on average, with some positive effects on turnout (mobilizing partisan bases who are energized by attacks on their opponents) and some negative effects (demobilizing moderates who are aversed to political conflict). The Ansolabehere-Iyengar finding does not replicate reliably across contexts. The demobilization hypothesis should not be dismissed entirely—there are specific contexts (sustained negativity in low-information races) where it may operate—but it should not be treated as a general rule about negative advertising.

25.5 Television Advertising: Media Markets and GRP Strategy

The Media Market Architecture

Television advertising strategy for statewide races begins with the media market (officially designated as Designated Market Areas or DMAs, as defined by Nielsen). The United States is divided into 210 DMAs, ranging from the New York DMA (over 7 million TV households) to small rural markets with fewer than 100,000 households. Media markets do not map neatly onto state boundaries—which creates strategic complexity in statewide races.

In a Senate race, a campaign must purchase advertising in every media market that has significant population of state voters. In a large state like California or Texas, this requires dozens of market buys with dramatically varying costs per thousand impressions. In a smaller state, a single or a few markets may cover the bulk of the population. The geographic structure of media markets creates a fundamental constraint on television advertising strategy: you cannot micro-target specific voters on broadcast television. You buy a market and reach everyone watching in that market, including voters in neighboring states who cannot vote in your race.

This "spillover" problem—paying to reach voters who can't vote for you—is economically most severe in markets near state borders. A campaign in northern Virginia must buy Washington, D.C. media market advertising, which reaches millions of Maryland and D.C. viewers alongside Virginia voters. The cost-efficiency of that market buy is significantly reduced by the spillover.

The Flight Strategy

Television advertising is purchased in "flights"—defined periods of advertising at specified intensity levels (measured in GRPs). A campaign's flight strategy specifies when to advertise, at what intensity, in which markets, and with which creative.

The basic strategic considerations in flight planning include:

Early vs. late: Advertising early in a race builds name recognition and establishes initial frames before opposition advertising complicates them. Advertising late in a race targets the final persuadable voters when attention is highest. Most competitive Senate campaigns do both, with a trajectory from biographical introduction ads early to contrast/attack ads in the final weeks.

Base vs. persuasion: Television advertising can target either base mobilization (energizing strong partisans to turn out) or persuasion (shifting less committed voters). These are not distinct audiences in practice—the same ad runs in the same markets—but ad content can be calibrated toward one goal or the other. Base mobilization ads emphasize partisan identity and core issue positions; persuasion ads emphasize crossover appeal and candidate character.

Concentration vs. distribution: A campaign with a limited budget can either concentrate its GRPs in one or two markets (achieving high frequency in those markets) or distribute them across all markets (achieving lower frequency everywhere). Research on advertising wear-in suggests that a threshold of exposure (typically 3-5 exposures) is needed for an ad to register; below this threshold, additional reach without sufficient frequency may produce less effect than concentrated frequency in fewer markets.

📊 Jake Rourke's TV Strategy

With 12 weeks remaining in the race, Jake and the Whitfield campaign's media consultant are planning their final television flight strategy. The state has four significant media markets; Whitfield's polling shows very different competitive situations in each:

• Market A (largest, suburban): Whitfield leads by 4 points. Crime messaging is moving voters; economic anxiety frame is testing well. Strategy: maintain 400 GRPs/week of contrast advertising.
• Market B (mid-sized, mid-state): Race is tied within margin of error. This is the decisive market for the race's outcome. Strategy: surge to 650 GRPs/week; test two different ad versions to find which performs better.
• Market C (smaller, rural): Whitfield leads by 11 points. Base is solid. Strategy: minimal buy (150 GRPs/week) for base maintenance; redeploy resources to Market B.
• Market D (smaller, college-town): Whitfield trails by 8 points. University-area voters heavily skew Garza; marginal TV is unlikely to move many voters at acceptable cost. Strategy: pull TV entirely; maintain digital-only presence targeting specific segments.

This market-by-market resource allocation—concentrating resources where they can move the race rather than where the candidate is already winning or definitively losing—is the core logic of television advertising strategy in competitive races.

25.6 Digital Advertising: The Revolution in Targeting

What Changed

The digital advertising revolution in political campaigns is not fundamentally about the internet as a distribution channel; it is about targeting. Television advertising reaches everyone in a market regardless of their political characteristics. Digital advertising, at its fullest development, can deliver different messages to different individual voters based on their voter file records, behavioral data, and inferred characteristics.

This targeting capability creates a qualitatively different advertising architecture. Rather than buying a media market and reaching all its voters with the same message, digital advertising enables:

• Delivering a healthcare-focused message exclusively to suburban women who have searched for healthcare information
• Delivering a crime-focused message exclusively to exurban men who have consumed crime news content
• Delivering a "get out the vote" message exclusively to sporadic voters who voted in 2020 but not 2022
• Excluding strong opposition partisans from seeing any advertising (because they're unlikely to be persuaded and their seeing your ad provides no benefit)

This is a fundamental change in the relationship between campaign communication and democratic representation—a point we return to in the ethical analysis section.

Programmatic Advertising: How It Works

Programmatic advertising refers to automated, algorithmic buying of digital advertising inventory through real-time auctions. When a registered voter loads a webpage or opens an app, that page load triggers an auction (conducted in milliseconds) among advertisers who want to reach that specific user. The campaign's demand-side platform (DSP) submits a bid based on the user's voter file match, behavioral characteristics, and the campaign's targeting criteria. If the campaign wins the auction, the ad appears; if not, a different advertiser's ad appears. The user experiences a seamless page load; the entire auction and delivery process occurred in the background.

For political campaigns, programmatic advertising is primarily used through three vectors:

Display advertising (banner ads, sidebar placements) across the open web—news sites, weather apps, local information sites. Display advertising has relatively low click-through rates but high impression volume at low cost, making it useful for building name recognition.

Video advertising on YouTube and streaming video platforms, including connected television (CTV) advertising on Roku, Hulu, Peacock, and similar platforms. CTV advertising is one of the fastest-growing categories in political advertising because it combines the visual storytelling of traditional television with the targeting precision of digital. CTV can deliver video advertising specifically to targeted voter profiles—something cable TV is beginning to enable but broadcast television cannot.

Social media advertising on Facebook, Instagram, Twitter/X, and TikTok. Social media advertising offers native placement within users' feeds rather than display-style placement, generally achieving higher engagement. The major social platforms all offer political advertising targeting using voter file data (through advertiser data uploads matched against platform user accounts) or platform-native targeting characteristics.

Search Advertising: The Underappreciated Channel

Search advertising—paid placement in response to specific search queries—is often overlooked in discussions of political advertising but is strategically significant. Search advertising reaches voters at the moment they are actively seeking information about candidates or issues, rather than interrupting them with messages in other contexts.

A voter who searches "Garza immigration policy" or "Whitfield crime record" is signaling both interest and uncertainty—exactly the profile of a persuadable voter. Both campaigns in the Garza-Whitfield race run search advertising against each other's names and against key issue terms, ensuring that searchers encounter their preferred framing before clicking through to news coverage or the opponent's site.

Search advertising is also used defensively: campaigns buy their own candidate's name to ensure the first search result is the campaign's framing rather than opposition attacks or negative news coverage.

25.7 Micro-Targeting: Voter File Data Meets Digital Advertising

The Voter File Foundation

The foundation of political micro-targeting is the voter file—the official registration record maintained by state election authorities that documents each registered voter's name, address, party registration, and voting history. The voter file is the starting point; data vendors (Catalist for Democrats, i360 for Republicans, and nonpartisan vendors like TargetSmart) enhance the voter file with:

• Demographic appends: Age, gender, household composition, estimated income, homeownership—derived from commercial data and census records.
• Consumer data overlays: Product purchase history, magazine subscriptions, vehicle registrations, and other commercial behavioral data from data brokers.
• Modeled issue positions: Propensity scores estimating individual voters' positions on specific issues, derived from survey data calibrated to voter file characteristics.
• Turnout and persuasion scores: Predicted probabilities of voting and of being persuadable on key issues, derived from statistical models trained on prior election data.

The result is an individual-level database of every registered voter in the state, with hundreds of appended variables, that campaigns use to segment voters for different advertising messages.

How Micro-Targeting Works in Practice

Nadia Osei's targeting team at the Garza campaign has segmented the state's electorate into twelve distinct voter segments based on combinations of issue priorities, partisanship, and turnout propensity. For each segment, the campaign has developed distinct advertising messages designed to address the specific concerns of that segment.

For example:

Segment: Soft Republican women, suburban, healthcare priority: Women registered Republican, voting history suggests split-ticket voting, consumer data suggests household with health insurance concerns (medical bill subscriptions, insurance comparison search history), propensity model scores as potential Garza persuadable. Message: Garza's record on protecting pre-existing condition coverage; does not emphasize party affiliation.

Segment: Low-propensity Democratic men, younger, economic anxiety: Men registered Democrat or no party, low prior turnout history, consumer data suggests working-class economic profile, mobilization target rather than persuasion target. Message: Economic dignity framing; emphasizes what Garza will fight for rather than contrasting with Whitfield; strong turnout activation call to action.

Segment: Independent women, college-educated, suburban: Women registered independent, high turnout propensity, issue survey models show moderate positions on most issues, high concern about democratic norms. Message: Garza's character and professional record; "the candidate you can trust" framing; minimal partisan cue.

Each segment receives different creative content delivered through the digital programmatic stack, with the voter file as the targeting spine. The campaign is delivering twelve different campaigns simultaneously—more accurately, twelve different conversations with different voter groups about why Maria Garza deserves their vote.

⚖️ Micro-Targeting and Democratic Representation

Micro-targeting raises profound questions about democratic representation and political accountability. When campaigns deliver different messages to different voter segments—messages that are invisible to each other and to the public—several democratic functions are compromised:

Accountability: Candidates are accountable to voters in part through their public statements. If a candidate communicates different things to different groups about what they will do in office, the public accountability function of campaigns is undermined.

Deliberation: Democratic theory assumes that voters share a common informational space and can deliberate about political choices. Micro-targeted advertising creates parallel, non-overlapping information environments for different voter groups, making shared deliberation about campaign claims impossible.

Transparency: In traditional broadcast advertising, anyone who watches television can see any candidate's advertising. In micro-targeted digital advertising, only the specific targeted audience sees a given message. Journalists, academics, and opposition researchers cannot monitor what campaigns are saying to specific voter segments unless they are members of those segments.

Discrimination: Voter file micro-targeting can be used to exclude specific demographic groups from receiving political information—a form of digital voter suppression when applied to target groups defined by protected characteristics. The civil rights implications of racially targeted political advertising exclusion are not merely theoretical.

25.8 Ad Transparency: Disclosure, Databases, and Gaps

What Gets Disclosed

Political advertising transparency is a rapidly evolving landscape. The baseline disclosure requirements are:

FEC disclosure covers expenditures by candidate committees and some political committees, but not all outside spending and not the content of what is being advertised—only that advertising spending occurred.

Platform transparency reports from Facebook, Google, and (previously) Twitter provide searchable databases of political advertising placed on those platforms. Facebook's Ad Library (now Meta Ad Library) provides the most accessible transparency: any user can search for political advertising by any page, including seeing active and recently inactive ads, approximate spend ranges, and demographic targeting breakdowns. Google's Transparency Report similarly documents political advertising spending by advertiser.

The Wesleyan Media Project provides academic-quality tracking of television and digital advertising content, spending, and targeting in competitive races, derived from commercial ad monitoring plus FEC data analysis.

AdImpact and Kantar are commercial advertising tracking firms that provide real-time data on political advertising volume and spending; their data is available to campaigns, outside groups, and journalists on a paid basis.

The Gaps in Transparency

Despite these disclosure mechanisms, significant transparency gaps remain:

Connected television: CTV advertising is not consistently covered by existing transparency infrastructure. As campaigns shift spending from traditional television (tracked by Kantar and others) to streaming platforms (where ad monitoring is more limited), a growing share of political advertising spending is occurring in a less transparent environment.

Platform dark patterns: Facebook's transparency database shows ads but not the micro-targeting criteria used to deliver them. A campaign can run an ad targeting only one specific demographic segment, and that targeting choice is not visible in the public-facing transparency interface. Users see that an ad ran; they cannot see who it was targeted to.

Outside spending complexity: Super PACs and other outside groups are subject to different disclosure requirements than candidate committees. "Dark money" nonprofits that don't disclose their donors can make independent expenditures (including advertising) subject to disclosure of the spending itself but not its source. The full picture of who is funding political advertising in any given race often requires piecing together data from multiple disclosure sources.

🔴 The Transparency Asymmetry Problem

There is a troubling asymmetry in political advertising transparency: well-resourced political actors (campaigns with large analytics teams, outside groups with sophisticated legal counsel) have far greater knowledge of the advertising landscape than ordinary voters or even journalists covering the race. Campaigns can track their own advertising, their opponents' advertising, and outside group advertising through commercial data subscriptions. Voters see the advertising targeted to them and have no visibility into what is being told to other voter segments. This information asymmetry is not politically neutral; it systematically advantages well-resourced campaigns and outside groups relative to ordinary democratic actors.

25.9 Effectiveness Research: What the Field Experiments Show

The Problem with Observational Evidence

Before the field experimental turn in political science, most evidence about advertising effectiveness came from observational analysis: correlating advertising spending with vote outcomes across races, controlling for other factors. This evidence consistently suggested that campaign spending has significant effects on vote share—but causal interpretation was compromised by severe selection problems (candidates in competitive races spend more, and competitive races are competitive for underlying reasons unrelated to advertising) and measurement limitations (spending proxies for advertising but doesn't directly measure exposure).

The development of randomized field experiments in political campaigns, beginning in earnest in the early 2000s, produced a more credible evidence base—and, initially, more modest effect estimates than observational analyses had suggested.

What Field Experiments Find

The most important field experimental evidence on political advertising effectiveness comes from several landmark studies:

Gerber, Gimpel, Green, and Shaw (2011) conducted a large-scale field experiment in the 2006 Texas gubernatorial race, randomly varying the timing and intensity of television advertising across media markets. They found significant but short-lived advertising effects: television advertising substantially increased candidate favorability, but the effects decayed rapidly after the advertising flight ended—half-life of approximately one week. This decay finding is among the most important in the advertising literature because it has major implications for flight timing strategy.

Sides and Vavreck's "The Gamble" (2013) examined the 2012 presidential election and found that campaign advertising moved preferences in the expected direction but that effects were quickly offset by subsequent communication from the opposing campaign. The net effect of advertising in competitive environments—where both candidates are advertising—is close to zero because campaigns advertise to roughly equal levels and cancel each other out.

Green and colleagues' GOTV field experiments found that targeted voter contact (including through digital advertising) could increase turnout by 1-3 percentage points—a smaller effect than many campaigns assume but meaningful in close races. Digital advertising specifically produced smaller turnout effects than direct mail or canvassing in most experimental comparisons, though effect sizes vary considerably across contexts.

Kalla and Broockman (2018) conducted a comprehensive meta-analysis of experiments estimating the persuasive effects of political campaign contact, including advertising. They found average persuasion effects of essentially zero in general election contests, with some positive effects in primary elections and for down-ballot campaigns where voters have less prior information. This sobering finding has been highly influential and controversial: it suggests that most general election advertising is approximately canceling the other side's advertising rather than generating net persuasion.

💡 The Kalla-Broockman Finding and Its Limits

The near-zero average persuasion effect in Kalla and Broockman's meta-analysis is an average—it encompasses variation across candidates, contexts, voter segments, and advertising content. Several important qualifications apply:

1. "Approximately zero net effect" in a competitive race doesn't mean advertising is ineffective; it means the two campaigns' advertising is approximately canceling each other. Unilateral disarmament by one campaign would produce substantial disadvantage.

2. The meta-analysis emphasizes persuasion effects; base mobilization effects (energizing already-supportive voters to turn out) may be more durable and important.

3. Timing effects may be substantial: the Gerber et al. finding of short-lived but real effects implies that advertising timed precisely before an election (when the decay period extends to or past election day) may be more effective than the overall average suggests.

4. Some voter segments may be substantially more persuadable than others, and targeted advertising reaching only those segments may produce larger-than-average effects.

25.10 The Garza-Whitfield Advertising Strategies: A Contrast Study

Whitfield's Strategy: The Resource Advantage

Jake Rourke's approach to the Whitfield campaign's advertising strategy is built around two structural advantages: a fundraising lead that enables more total advertising and outside group support concentrated in television. The strategic logic is to win the television air war in ways the Garza campaign cannot match, while using the campaign's digital program primarily for base mobilization and small-dollar fundraising rather than persuasion.

Whitfield's advertising message arc:

Weeks 1-4 (Introduction): Biographical spots establishing Whitfield as a successful business owner, family man, and community member. The frames: authentic outsider (never been a politician), economic success story (built a real business), values-grounded (church community, family). These spots ran heavily in the two mid-sized markets where Whitfield needed to build name recognition outside his political base.

Weeks 5-9 (Definition): Attack advertising against Garza, primarily focused on her record as Attorney General. The central message is that Garza was "soft on crime" (citing specific case decisions) and "out of touch" (citing fundraising sources and public statements). Research on these frames showed they were most effective with soft Republicans and independents in suburban areas. The campaign ran A/B versions in different markets to test which attack frame produced greater favorable-unfavorable shifts.

Weeks 10-final (Closing): Combination of contrast (Whitfield vs. Garza on public safety) and positive closing spots emphasizing economic vision. The closing positive was deliberately designed for news cycles—a format optimized for earned media coverage of the campaign's final message.

Throughout, Whitfield's outside group advertising supplemented the campaign's buys with heavier attack content that the campaign itself could disclaim. The outside groups' advertising was tonally more aggressive than the campaign's own creative, operating in the documented strategic space where campaigns benefit from attack advertising by nominal independents while maintaining the campaign's own relative positive tone profile.

Garza's Strategy: The Resource Constraint Problem

Nadia Osei's challenge throughout the race has been managing scarcity. The Garza campaign is outspent on television roughly 1.7 to 1 when outside group advertising is included. The strategic response involves three elements:

Selective market concentration: Rather than competing across all markets, Garza concentrates resources in the markets where persuadable voter density is highest relative to cost. This means accepting a disadvantaged advertising environment in some markets in order to be competitive in others.

Digital as asymmetric advantage: Garza's campaign team has significantly stronger digital advertising capabilities than Whitfield's—a function of the Democratic campaign infrastructure's more developed digital ecosystem and the campaign's deliberate investment in digital targeting expertise. The campaign uses its digital budget to run highly targeted advertising to specific voter segments in markets where television competition is too expensive: sporadic Democrats, soft-Republican suburban women, and high-propensity independents identified through the voter file.

Earned media leverage: Where paid advertising is limited, earned media (press coverage, debate performance, viral moments) becomes relatively more important. Garza's campaign has been deliberate about creating moments—the viral TikTok clip from week 6 (described in Chapter 23), the debate performance that shifted media framing—that generate coverage that extends beyond what the campaign paid for.

Message differentiation: Garza's advertising emphasizes issue dimensions where she is stronger (healthcare, constitutional governance, professional competence) rather than fighting Whitfield on his strongest ground (public safety). The strategic logic is to make the election about the dimensions where your candidate has advantages rather than conceding that the opponent's preferred terms of debate define the race.

📊 The ROI Calculation

Nadia runs a weekly return-on-investment calculation for the campaign's advertising across channels. The metric is simple: estimated vote margin improvement per $10,000 spent, estimated from panel surveys and voter file modeling. At the current stage of the race:

• Television (mid-sized markets): approximately 0.3 percentage points per $10,000 (within margin of error; confidence in this estimate is low)
• CTV targeting: approximately 0.5 percentage points per $10,000 among targeted segments (higher confidence, smaller absolute numbers)
• Digital social targeting (persuasion): approximately 0.2 percentage points per $10,000 (confidence: low)
• Digital social targeting (mobilization, low-propensity Democrats): approximately 0.8 percentage points per $10,000 (confidence: moderate, drawing on experimental literature)

These estimates have very wide confidence intervals and should not be taken as precise predictions. But the ordering—mobilization digital being more efficient than persuasion digital or television—is consistent with the field experimental literature and is driving Nadia's recommendation to shift resources toward the mobilization program in the campaign's final weeks.

25.11 Who Gets Targeted, Who Gets Left Out

The Ethics of Exclusion

Digital micro-targeting's targeting capability implies an equally important exclusion capability: campaigns can choose not to show advertising to specific voter groups. This exclusion capability has several ethically troubling applications:

Demobilization targeting: A campaign could in principle run discouraging or demobilizing content specifically to segments of the opponent's base—content designed to reduce turnout among groups likely to vote for the other candidate. This is a form of voter suppression through information manipulation, and while evidence of deliberate demobilization targeting by major campaigns is limited, the capability is technically straightforward.

Differential information by race: Voter file data enables targeting by race-correlated characteristics. Delivering different political information to different racial groups—or excluding specific racial groups from political advertising altogether—is both a legal civil rights concern and a democratic accountability concern.

Suppression of opposition information: A candidate who faces a serious policy critique can, through micro-targeting, ensure that voters who are most likely to be moved by that critique never encounter the candidate's advertising—while reaching all other voter groups. This is a subtle form of message management that traditional broadcast advertising could not accomplish.

⚖️ The "Who Gets Counted" Question in Advertising

The textbook's recurring question—who gets counted, who gets heard?—applies to political advertising with particular force. Micro-targeting systems optimize for persuadability and efficiency: campaigns deliver messages to voters most likely to be moved at acceptable cost. The voters who are excluded from these calculations are those deemed either too committed (already strongly supporting the campaign, not worth persuading) or too unlikely to vote or be persuaded (not worth the resource expenditure). Both categories disproportionately include specific populations: reliable partisans in "safe" demographic groups who are taken for granted, and low-propensity voters in communities that data models identify as hard to move. The communities that receive the least targeted political advertising are often those whose political engagement is lowest—and receiving less political information further reduces the conditions for engagement. The efficiency logic of micro-targeting can systematically deprioritize already-marginalized political communities.

25.12 The Future of Political Advertising

Connected Television and the Closing Gap

The most significant near-term development in political advertising is the rapid growth of connected television advertising. CTV combines television-quality video delivery with digital precision targeting—offering the storytelling power of traditional television without the blunt geographic precision of broadcast media markets. As streaming viewership continues to grow at the expense of traditional broadcast and cable television, CTV advertising's share of political budgets is growing correspondingly.

For campaigns, CTV creates both opportunity (precise targeting of high-quality video to specific voter segments at costs lower than broadcast television in major markets) and challenge (the fragmented streaming landscape is technically complex to buy across multiple platforms, and ad frequency management across platforms is difficult, creating the risk of severe over-exposure to specific viewers).

AI-Generated Content and the Authenticity Question

Artificial intelligence tools for advertising content generation—including AI-generated video, synthetic voice, and dynamically generated text—are entering political advertising at an accelerating pace. The implications are significant in multiple directions:

Scale: AI tools enable campaigns to generate far more advertising creative at far lower cost, enabling more extreme message personalization than was previously feasible.

Authenticity attacks: AI-generated synthetic media (deepfakes) can create false video or audio of candidates saying things they did not say—a new category of attack advertising that is difficult for voters to identify as fabricated.

Detection arms race: Campaigns, platforms, and regulators are developing detection tools for AI-generated political content, but the detection tools are in a continuous arms race with generation tools.

These issues are addressed more fully in Chapter 40 (AI and the future of political analytics); the point here is that the advertising landscape in which campaigns like Garza-Whitfield are operating is already changing around them, and the strategic environment of political advertising a decade hence will be substantially different from what this chapter describes.

25.13 The Message Architecture: Positive, Contrast, and Attack in Strategic Sequence

Why Campaigns Don't Just Run Their Best Ad

One of the most consistent observations from campaign practitioners is that advertising is not just about what you say—it is about the sequence in which you say it. A campaign that begins with heavy attack advertising before introducing itself to voters forecloses the biographical narrative that justifies the attack ("she's a better choice than him because X"). A campaign that runs only positive advertising late in a race misses the window when attack advertising's priming effects can shape election-day evaluations. The architecture of an advertising campaign—which messages run when—is as strategically significant as the content of any individual ad.

The standard three-phase advertising architecture in competitive Senate races follows a logic derived from both the research literature and practitioner experience:

Phase 1: Introduction. Early in the race, the priority is building name recognition and establishing a positive first impression. Biographical advertising—establishing who the candidate is, what they value, and why they are seeking office—creates the positive evaluative foundation that subsequent advertising builds on. The research on persuasion suggests that candidates with high favorability can take more risks with contrast advertising; low-favorability candidates are less credibly positioned to attack.

Phase 2: Definition. Once the candidate is known, the midgame shifts to defining the race on favorable terms—establishing the frames and priming the considerations that the candidate wants voters to use in their final evaluation. This is when issue advertising becomes most important: not just "I support healthcare" but "this race is about healthcare" (agenda-setting) and "judge your Senator on healthcare" (priming).

Phase 3: Contrast and Close. In the final weeks, the advertising intensity typically increases and the tone shifts more negative. This is when the advertising decay research (Gerber et al.) is most strategically relevant: advertising in this phase lands close enough to election day that its persuasive effects don't decay before voting. Attack and contrast advertising in the closing phase serves multiple functions: it drives unfavorable evaluations of the opponent, it reinforces base motivation by activating partisan threat perceptions, and it establishes the final evaluative frame through which voters approach election day.

The Whitfield campaign followed this architecture closely: biographical introduction in weeks 1-4, crime-focused definition in weeks 5-9, and closing contrast advertising in weeks 10 through election day. The Garza campaign, operating under resource constraints, compressed its introduction phase and entered definition and contrast earlier—accepting a shorter introduction window in exchange for more time on the contested evaluative terrain.

The Role of Outside Groups in the Architecture

One important complexity in modern campaign advertising architecture is the role of independent expenditure groups, which operate on their own timelines and message strategies. Outside groups cannot coordinate with campaigns legally, which means the overall advertising architecture the voter experiences is a composite of independently planned advertising programs that may or may not be strategically coherent.

In practice, campaigns and aligned outside groups often maintain informal coordination through publicly available information—each can see the other's FEC filings, their TV buys through commercial monitoring, and their digital advertising through platform transparency tools. This creates a de facto coordination structure through transparent public observation that doesn't technically violate coordination prohibitions. The result is that outside groups often fill specific roles—doing the heaviest attack advertising, reaching markets where the campaign isn't advertising—in ways that appear coordinated without being legally coordinate.

25.14 International Perspectives on Political Advertising Regulation

The American Exception in Advertising Regulation

The United States is unusual among developed democracies in permitting essentially unlimited political advertising by candidates and outside groups, with disclosure rather than restriction as the primary regulatory tool. Most other democracies take a substantially more restrictive approach:

United Kingdom: Political advertising is prohibited on television and radio. Political parties receive free broadcast time allocated by formula. Paid political television advertising has been banned since the 1950s. Digital political advertising is allowed with spending limits.

Canada: Campaign spending limits restrict total advertising expenditures by parties and candidates. A separate limit applies to third-party (outside group) advertising. Advertising is limited to a defined writ period (the formal election campaign period).

Germany: Political parties receive public funding and free broadcast time. Paid political advertising exists but is subject to strict spending limits. Outside group advertising in the American Super PAC sense essentially does not exist.

Australia: Caps on campaign spending apply to both parties and candidates. A "truth in political advertising" law in South Australia (and pending federal legislation) creates accountability mechanisms for factually false advertising claims.

The American advertising regulatory environment—essentially unlimited spending with disclosure—produces a political advertising ecosystem that is unique in the world for its scale, intensity, and the dominance of outside group advertising. The $14.4 billion spent on the 2020 federal elections is an order of magnitude larger than comparable democratic systems.

🌍 What Cross-National Comparison Reveals

The fact that most other democracies heavily restrict political advertising without obvious harm to democratic quality challenges the premise that unlimited political advertising is a necessary feature of democratic competition. Critics of American exceptionalism argue that the advertising-dominated campaign model serves incumbent financial advantages, raises the cost of electoral competition in ways that benefit wealthy donors, and produces information environments in which resource disparities translate directly into information environment disparities. Defenders argue that advertising restrictions amount to speech restrictions with their own anti-democratic implications, and that disclosure-based transparency is more consistent with free speech norms than content or spending limits.

This debate is unlikely to be resolved here; the point for political analysts is to understand the regulatory context within which American political advertising operates and to recognize that this context is not the only conceivable one.

25.15 The Analytics Infrastructure Behind Modern Advertising: From Voter File to Attribution

The Data Stack

Modern political advertising is embedded in a data infrastructure that would have been unimaginable to the consultants who developed the 30-second spot format in the 1950s. Understanding this infrastructure is essential for analysts who want to understand not just what campaigns are doing but how they are doing it and what decisions the data enables and constrains.

The core data stack for political advertising optimization consists of:

Voter file + enhancements: The foundation, as described in Section 25.7. Individual-level registration records enhanced with commercial data, modeled scores, and turnout/persuasion propensity estimates.

Audience matching: The process of matching voter file records to platform user accounts on Meta, Google, and other platforms. Match rates typically range from 50-70 percent for robust first-party data matches. The 30-50 percent of registered voters who cannot be matched are reached only through proxy targeting—behavioral interest signals or demographic characteristics that correlate with but do not directly identify the targeted voter.

Attribution modeling: The most analytically challenging component of the advertising data stack. Attribution asks: which advertising exposures caused which voter behaviors (turnout, vote choice, donation)? The challenge is that voters are exposed to multiple advertising messages across multiple channels over weeks or months, and disentangling which specific exposures caused which specific behaviors requires either experimental designs (randomly assigning advertising exposure) or statistical models with known limitations.

Campaign attribution modeling typically uses a combination of: voter file matched control/treatment comparisons (comparing voters in targeted segments to similar voters in untargeted control groups), digital platform attribution reports (which have self-reporting biases toward platform value), and ad effect decay models calibrated to the Gerber et al. and related research. None of these approaches produces clean causal estimates; together they provide directionally useful guidance for resource allocation decisions.

A/B testing infrastructure: Well-resourced campaigns run continuous A/B tests of advertising creative, targeting parameters, and flight strategies using the voter file as the randomization frame. A voter file can be randomly split into test and control cells; different cells are exposed to different advertising; outcomes are measured through voter file match-backs (did these voters donate? return surveys? show up to vote?). This testing infrastructure is the practical application of the field experimental methods described in Section 25.9.

The Analytics Division of Labor in the Garza Campaign

Nadia Osei's analytics operation is organized around three distinct functions that feed into advertising decisions:

Targeting: Voter file segmentation, propensity score modeling, and audience match maintenance. This function answers "who should we reach?" and maintains the data infrastructure that makes targeted advertising possible.

Measurement: Attribution modeling, testing design and analysis, and media environment monitoring (the competitive ad environment analysis described throughout this chapter). This function answers "what is our advertising doing?" and provides the evidence base for resource allocation decisions.

Intelligence: Competitive analysis, ad monitoring (both Whitfield's program and outside group activity), and opposition research on advertising claims. This function answers "what is the opponent doing?" and enables rapid response when competitor advertising launches new attacks.

The three functions are operationally distinct but analytically integrated: targeting decisions depend on measurement results (which segments are responding to which messages?), measurement requires intelligence context (what competitor advertising were targeted voters also seeing during the test period?), and intelligence findings drive targeting adjustments (if the opponent is concentrating resources in a specific market, does that change our targeting priorities?).

This integrated analytics operation represents the state of the art in well-resourced campaign advertising infrastructure. It is also substantially beyond the capacity of most state legislative and local races, where the advertising analytics infrastructure is typically much thinner—a gap that has political consequences for the quality of advertising strategy in down-ballot races.

Summary

Political advertising has evolved from a blunt instrument—broadcast television reaching everyone in a geographic market—to a precision system capable of delivering tailored messages to individual voters identified through voter file data and commercial behavioral data. This evolution has transformed both the strategic opportunities available to campaigns and the democratic accountability questions raised by political advertising.

The evidence on advertising effectiveness is more nuanced than either optimistic campaigns or skeptical academics sometimes suggest. Television advertising has real but short-lived persuasive effects; these effects decay rapidly, requiring sustained flight strategies rather than one-time investments. Digital advertising's targeting precision creates efficiency advantages for mobilization programs but does not show reliably strong persuasion effects in experimental research. Negative and contrast advertising is generally at least as effective as positive advertising, without the backlash effects popular wisdom predicts, though contextual factors matter.

The Garza-Whitfield race illustrates both the strategic logic and the practical constraints of political advertising. Whitfield's resource advantage in television translates to a structural advantage in the traditional advertising competition; Garza's digital targeting capabilities and earned media strategy represent an asymmetric response to resource scarcity. The outcome of that strategic contest depends on which channels reach the genuinely persuadable voters in the proportions and at the times that matter.

The ethical dimensions of advertising strategy—particularly the exclusion of specific voter communities from targeted communication and the accountability gaps created by micro-targeted message differentiation—deserve ongoing attention from analysts who want to understand not just how political advertising works technically, but what it does to the democratic communities it operates within.

Key Terms

Ad buy: A planned purchase of advertising time or digital inventory by a campaign committee, specifying market, timing, quantity, and creative content.

Connected television (CTV): Digital video advertising delivered through internet-connected televisions and streaming devices, enabling the targeting precision of digital advertising with the storytelling format of television.

Dial testing: A focus group method in which participants record second-by-second reactions to political advertising using hand-held dial devices, providing a continuous response trace.

Gross rating points (GRP): A measure of advertising weight equal to the percentage of a target audience reached multiplied by the average number of times they are reached; 100 GRPs = 100% reach x 1 exposure, or 50% reach x 2 exposures.

Media market (DMA): A Nielsen-defined geographic area representing a distinct television market, used as the primary unit of broadcast advertising strategy.

Micro-targeting: The practice of delivering tailored advertising messages to specific individual voters identified through voter file data, behavioral data, and predictive modeling.

Negative advertising: Political advertising that attacks an opposing candidate rather than promoting the sponsoring candidate.

Programmatic advertising: Automated, algorithmic buying of digital advertising inventory through real-time auctions, enabling targeting of specific users across the web.

Voter file: Official registration records maintained by state election authorities, documenting each registered voter's name, address, party, and voting history; enhanced with commercial data to form the foundation of campaign targeting.

Wesleyan Media Project: An academic project tracking the content, volume, and spending of political advertising in major federal and state races, providing the primary academic resource for political advertising research.