Menu Adam R Brown

Notes navigation: Browse by titleBrowse by authorSubject index

Holbrook, Krosnick, Visser, and Gardner: Attitudes toward presidential candidates

Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I say who wrote them.

Holbrook, Krosnick, Visser, and Gardner. 2001. Attitudes toward presidential candidates: Initial optimism, inertial first impressions, and a focus. AJPS 45:930-50.

Research Question

How do people evaluate candidates and policies?



The literature (see Zaller 1992) assumes that we take the raw number of positive or negative considerations about a candidate and evaluate him accordingly. The authors call this the symmetric (equal magnitude) linear (additive) model, or SLM.

In reality, we tend to give greater weight to earlier information than to later information ("first impressions," making it a non-linear model). Our evolutionary background also makes us respond more strongly to potential threats than to potential rewards (cf. Brader 2005). Thus, we weight negative information more heavily than positive information (it's asymmetric). Thus, they present an asymmetric nonlinear model, or ANM.

Observable Implications of ANM vs SLM

  1. The first few bits of information acquired have stronger effects than later bits (in ANM), not equal effects (as in SLM).
  2. Negative information is weighted more heavily than positive information (in ANM) (though they fail to verify this empirically).
  3. In the absence of information, we have a slight "positive" bias (in ANM), but SLM predicts we'll be neutral without info.


Lots of data based on ANES surveys. Many different tests. Generally, they are looking at the number of positive and negative things that people say about each candidate, and finding that positive and negative considerations have unequal (asymmetric) coefficients, and that these coefficients are not linear.

Negative campaigning

As one part of the article, they discuss implications for their study on our knowledge of negative campaigning. They cite studies showing that campaigns tend to begin positive, then go negative. Their model would imply that, since negative information comes after a positive first impression has been formed, its effects will be small.


You're more motivated to turn out if you like one candidate and dislike the other than if you love one candidate and only like the other. It's not the absolute size of the difference; it's whether one has a positive evaluation and one has a negative evaluation. So the authors' variables (based on positive and negative considerations) have stronger effects than just feeling thermometers in predicting turnout. The point: Negatives matter more than positives; we don't weight them equally.

Main contributions (that they actually demonstrated empirically)

  1. Additional considerations have declining effects on your stated attitudes about something.
  2. Additional negative considerations have (slightly) more powerful effects than additional positive ones.

Research by the same authors

Research on similar subjects


Holbrook, Allyson (author)Krosnick, Jon (author)Visser, Penny (author)Gardner, Wendi (author)American PoliticsPublic OpinionMedia EffectsElectionsCampaign Advertisements

Wikisum home: Index of all summaries by title, by author, or by subject.