Discounting extreme positions: party normalization and support for the far right
Discounting extreme positions: party normalization and support for the far right was written by Laia Balcells, Sergi Martínez, and Ethan vanderWilden in 2024. It was published in Political Science Research and Methods.
The authors argue that the electorate distrusts extremist policy statements taken by a party that is not stigmatized as extremist. A party label that is associated with/expected to be extremist leads to the same statements being received as authentic. The crux of this discounting theory is that norms and stigmatization around a party label have meaningful impact on perceived extremism.
The authors test this theory using a survey about Vox. Respondents are experimentally treated with framing to suggest that Vox is either a stigmatized party (stigmatized cohort) or not (normalized cohort). Furthermore, respondents are randomized into three cohorts:
Primed with a statement from a Vox deputy: "hoy es el día de gritar: el pasto es verde, los niños tienen pene y las niñas vagina, el sexo es binario, la biología es real, la ideología de género es un cáncer" ("Today is the day to shout: the grass is green, boys have penises and girls vaginas, sex is binary, biology is real, gender ideology is a cancer")
Primed with statements by Santiago Abascal, leader of Vox, about Franco's legacy (e.g. denying that he executed a coup)
- null/no priming
Survey sample is a web panel (Respondi&Bilendi) recruited exclusively from political center and right. (Specifically, used an 11-point ideology scale and screened out those scored between 0 and 4.) Survey fielded for 3 weeks between municipal elections (May 28) and national elections (July 23). Also implemented an attention check and removed speeders.
Regarding representativeness of the same:
- Survey used quotas on age, sex, and region.
Authors compared demographics to those collected in the CIS post-election barometer; greatest marginal differences were ~3%.
Authors used ANOVA to demonstrate that random assignments are not significantly correlated with any demographic covariates.
Measures are:
- sympathy to Vox: 0-100 scale
- "feeling thermometer"
- this is the primary outcome
- Vox's authenticity in extremist positions: Likert scales
- "Vox is hostile to LGBTQ+ people"
- "Vox would re-install dictatorship"
- these are discounting measures
The outcome measure is regressed on indicators for exposure to stigmatization, exposure to either #1 or #2, and an interaction term. The discounting measures are also regressed on the same indicators, bearing in mind that e.g. "Vox is hostile to LGBTQ+ people" is only measured for those assigned statement #1, and so on.
Authors find that exposure to statement #1 leads to decreased sympathy for the party in the normalized cohort, but not in the stigmatized cohort.
Authors find no movement following exposure to statement #2.
Implication is that the electorate can be persuaded to support extreme parties on some issues (esp. LGBTQ+ issues) but not others.
Authors used mediation analysis (with the discounting measures as the mediator) to establish that there is a significant mediation effect. Notably it is very significant among moderates, although the direct effect is also significant among moderates, suggesting there are more causal factors.
Reading Notes
I have read similar research that I think was more rigorous in the experiment design, especially in using a non-ideological statement as the null priming.