Skip to main content

Examining Brand Strength of Political Candidates: a Performance Premium Approach


Despite the growth of research on political marketing, fundamental questions concerning brand value of political candidates and its relationship with marketing mix activities remain unanswered. This research extends premium-based brand valuation methods to the political context by presenting a performance premium approach to assessing the strength of politicians’ brands and exploring the relationship between politician brand strength and political advertising. We develop a joint hierarchical Bayesian model of brand performance and marketing activity and estimate our model using data on the election performance and advertising expenditures of political candidates from US House of Representatives elections. Our findings suggest that politicians who possess a strong brand—those that perform better than expected given the partisan leanings of their districts, advertising spending, and other model controls—face a consequence for their brand strength in which political advertising by their opponents has a stronger negative impact on their performance compared to the effect of political advertising against politicians with weaker brands. We discuss the implications of our findings for political marketing.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3


  1. See Gordon and Hartman [16] for more information on CMAG’s political advertising data.

  2. Four uncontested races and 19 races that did not feature a Republican vs. a Democratic were excluded (2.7% of the data). These races were excluded because vote share, fundraising, and advertising (by candidates, opponents, parties, and outside groups) are systematically different in such races. Thus, inclusion of these races may incorrectly impact our inferences about the relationship between brand strength and advertising effectiveness.

  3. Own advertising is defined in this way because the data provider codes candidate and party advertising spending together. This definition is fitting because the candidate has control over the message content for advertising funded by the candidate’s campaign and his/her national party.

  4. PVI captures district competitiveness, a key measure of expected candidate performance [19]. PVI is calculated from an average of a district’s past voting behavior, compared to national voting behavior, in the two most recent presidential elections; thus, the measure captures the Republican or Democratic leanings of districts. Positive PVI scores indicate a Republican-leaning district while negative PVI scores indicate a Democrat-leaning district.

  5. We construct this set of variables based on past research [16, 41]. For all districts, we were also able to collect a district-level measure of education (percentage of district population with bachelor’s degrees) and two additional measures of racial makeup (percentage of district population that is black or Hispanic). We found the education variable to be very highly correlated with median income. Given the importance of median income in predicting voting behavior [11], we include it in our analysis and exclude the education measure. The additional measures of district-level racial makeup are both strongly negatively correlated with percentage of district population that is white. We test an alternative model with all three measures of racial makeup. We find that this model has inferior model fit to the proposed model and that the two new measures are not significant in any equation. As such, we include only percentage of population that is white in our model.

  6. Future research may generalize our model framework to consider other outcomes, such as voter turnout.

  7. Modeling a log transformation of vote share as a function of the log of advertising spending is consistent with past work in marketing on political advertising [41]. Note that the advertising measures in Eq. (2) are the log of the summed total of own and outside advertising for a particular candidate (which differs from the advertising measures in Eqs. (6)–(9) which breakout own vs. outside advertising). As detailed below, we test several alternative specifications for the measures in Xi, including advertising without log transformations, share advertising variables, and breakouts of own vs. outside advertising.

  8. We assume that the coefficients β do not include time-varying unobservable factors. We leave the development of such a dynamic model to future research.

  9. The proposed model takes into account that campaign expenditures and political outcomes are intertwined. To test the robustness of the results, we also consider three more parsimonious model specifications in which we exclude the fundraising Eqs. (4)–(5), exclude the equations of advertising expenditures (6)–(9), or exclude the advertising expenditure and fundraising Eqs. (4)–(9). We find that the key result related to brand strength and advertising effectiveness holds in each of these alternative analyses.

  10. Priors, initial values, trace plots of each coefficients, and Gelman and Rubin statistics for each coefficient are available in the online supplementary materials. The supplementary materials also include model fit illustrations which plot the actual vs. predicted values for Eqs. (1)–(9).

  11. We also explore a model that excludes the demographic controls and find that it has inferior model fit compared to the proposed model (DIC = 32,599; MAE = 0.0365). This analysis further highlights the robustness of the findings as we see that the key result related to brand strength and advertising effectiveness holds in this alternative model.

  12. As of July 2018, 16 out of the 20 strong brand candidates in Table 4 were still in office. Of the four not in office, two retired and one ran for higher office. Of the weak brands, only four out of the 20 were in office as of July 2018.

  13. To provide an illustration of effect size of our brand strength measure (ηj), we use the posterior mean estimates of the coefficients, median values of the continuous measures, and most common value for dummy variables (e.g., most data is for 2010, so we use 2010 dummy). We vary the value of the brand strength measure (ηj) at its minimum, first quartile, median, third quartile, and maximum values. We estimate Eqs. (2)–(3) with these values and find the following Republican vote share (VS) estimates: ηj at minimum, VS = 23.03%; ηj at 1st quartile, VS = 44.40%; ηj at median, VS = 47.16%; ηj at 3rd quartile, VS = 49.21%; ηj at maximum, VS = 57.70%.


  1. Aaker DA (1996) Measuring brand equity across products and markets. Calif Manag Rev 38(3):102–120

    Article  Google Scholar 

  2. Agarwal MK, Rao VR (1996) An empirical comparison of consumer based measures of brand equity. Mark Lett 7(3):237–247

    Article  Google Scholar 

  3. Ailawadi KL, Lehmann DR, Neslin SA (2003) Revenue premium as an outcome measure of brand equity. J Mark 67(Oct):1–17

    Article  Google Scholar 

  4. Aragones E, Palfrey TR (2004) The effects of candidate quality on electoral equilibrium: an experimental study. Am Polit Sci Rev 98(1):77–90

    Article  Google Scholar 

  5. Campbell MC, Keller KL (2003) Brand familiarity and advertising repetition effects. J Consum Res 30(September):292–304

    Article  Google Scholar 

  6. Chattopadhyay A (1998) When does comparative advertising influence brand attitude? The role of delay and market position. Psychol Mark 15(5):461–475

    Article  Google Scholar 

  7. Citizens United v (2010) Federal Election Commission, 130 U.S. 876

  8. Danaher PJ, Bonfrer A, Dhar S (2008) The effect of competitive advertising interference on sales for packaged goods. J Mark Res 45(April):211–225

    Article  Google Scholar 

  9. Erikson RS, Palfrey TR (1998) Campaign spending and incumbency: an alternative simultaneous equations approach. J Polit 60(2):355–373

    Article  Google Scholar 

  10. Gelman A, Rubin DB (1992) Inference from iterative simulation using multiple sequences. Stat Sci 7(4):457–511

    Article  Google Scholar 

  11. Gelman A, Shor B, Bafumi J, Park D (2007) Rich state, poor state, red state, blue state: what’s the matter with Connecticut? Q J Polit Sci 2(4):345–367

    Article  Google Scholar 

  12. Gerber AS (1998) Estimating the effect of campaign spending on senate election outcomes using instrumental variables. Am Polit Sci Rev 92(2):401–411

    Article  Google Scholar 

  13. Gerber AS (2004) Does campaign spending work? Field experiments provide evidence and suggest new theory. Am Behav Sci 47(5):541–574

    Article  Google Scholar 

  14. Goldstein K, Freedman P (2000) New evidence for new arguments: money and advertising in the 1996 senate elections. J Polit 62(4):1087–1108

    Article  Google Scholar 

  15. Goldstein K, Schweidel DA, Wittenwyler MB (2012) Lessons learned: political advertising and political law. Minn Law Rev 96(5):1732–1754

  16. Gordon BR, Hartmann WR (2013) Advertising effects in presidential elections. Mark Sci 32(1):19–35

    Article  Google Scholar 

  17. Gorn GJ, Weinberg CB (1984) The impact of comparative advertising on perception and attitude: some positive findings. J Consum Res 11(2):719–727

    Article  Google Scholar 

  18. Green DP, Krasno JS (1988) Salvation for the spendthrift incumbent: reestimating the effects of campaign spending in house elections. Am J Polit Sci 32(4):884–907

    Article  Google Scholar 

  19. Hoegg J, Lewis M (2011) The impact of candidate appearance and marketing decisions on elections. J Mark Res 48(5):895–909

    Article  Google Scholar 

  20. Jacobson GC (1978) The effects of campaign spending in congressional elections. Am Polit Sci Rev 72(2):469–491

    Article  Google Scholar 

  21. Keller KL, Lehmann DR (2006) Brands and branding: research findings and future priorities. Mark Sci 25(6):740–759

    Article  Google Scholar 

  22. Kennedy E (2012) Citizens United turns two: democracy is not a game. Huffington Post (Jan. 19) Accessed 2 August 2018

  23. Kenny C, McBurnett M (1994) An individual-level multiequation model of expenditure effects in contested house elections. Am Polit Sci Rev 88(3):699–707

    Article  Google Scholar 

  24. Kent RJ, Allen CT (1994) Competitive interference effects in consumer memory for advertising: the role of brand familiarity. J Mark 58(July):97–105

    Article  Google Scholar 

  25. Kumar A, Krishnan S (2004) Memory interference in advertising: a replication and extension. J Consum Res 30(March):602–611

    Article  Google Scholar 

  26. Li S, Sun B, Wilcox RT (2005) Cross-selling sequentially ordered products: an application to consumer banking services. J Mark Res 42(2):233–239

    Article  Google Scholar 

  27. Manchanda P, Rossi PE, Chintagunta PK (2004) Response modeling with nonrandom marketing-mix variables. J Mark Res 41(Nov):467–478

    Article  Google Scholar 

  28. Matsa KE (2014) Interest in midterms may be low, but local TV awash in political ad spending. Pew Research Center (Oct 31) Accessed 2 August 2018

  29. Moe WW, Schweidel DA (2012) Online product opinions: incidence, evaluation, and evolution. Mark Sci 31(3):372–386

    Article  Google Scholar 

  30. Mondak JJ (1995) Competence, integrity, and the electoral success of congressional incumbents. J Polit 57(4):1043–1069

    Article  Google Scholar 

  31. O’Connor P (2015) TV remains king in political ad spending. Wall Street Journal (Aug. 30) Accessed 2 August 2018

  32. Park YH, Bradlow ET (2005) An integrated model for bidding behavior in internet auctions: whether, who, when, and how much. J Mark Res 42(4):470–482

    Article  Google Scholar 

  33. Park CS, Srinivasan V (1994) A survey-based method for measuring and understanding brand equity and its extendibility. J Mark Res 31(May):271–288

    Article  Google Scholar 

  34. Park DK, Gelman A, Bafumi J (2004) Bayesian multilevel estimation with poststratification: state-level estimates from national polls. Polit Anal 12(04):375–385

    Article  Google Scholar 

  35. Parti T (2013) $7 billion spent on 2012 campaign, FEC Says. Politico (Jan 31) Accessed 2 August 2018

  36. Pew Research Center (2011) The generation gap and the 2012 election (Nov. 3) Accessed 2 August 2018

  37. Schweidel DA, Knox G (2013) Incorporating direct marketing activity into latent attrition models. Mark Sci 32(3):471–487

    Article  Google Scholar 

  38. Stone WJ, Maisel LS (2003) The not-so-simple calculus of winning: potential U.S. house candidates’ nomination and general election prospects. J Polit 65(4):951–977

    Article  Google Scholar 

  39. Thomson M (2006) Human brands: investigating antecedents to consumers’ strong attachments to celebrities. J Mark 70(July):104–119

    Article  Google Scholar 

  40. Van Diepen M, Donkers B, Franses PH (2009) Dynamic and competitive effects of direct mailings: a charitable giving application. J Mark Res 46(Feb):120–133

    Article  Google Scholar 

  41. Wang Y, Lewis M, Schweidel DA (2018) A border strategy analysis of ad source and message tone in senatorial campaigns. Mark Sci 37(3):333–355

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations


Corresponding author

Correspondence to Beth L. Fossen.

Ethics declarations

Conflict of Interest

The authors declare that they have no conflict of interest.

Electronic Supplementary Material


(DOCX 410 kb)


(ODC 6401 kb)


(ODC 14 kb)


(TXT 4 kb)


(ODC 281710 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Fossen, B.L., Schweidel, D.A. & Lewis, M. Examining Brand Strength of Political Candidates: a Performance Premium Approach. Cust. Need. and Solut. 6, 63–75 (2019).

Download citation

  • Published:

  • Issue Date:

  • DOI:


  • Political marketing
  • Brand strength
  • Advertising
  • Human brands