Date Thesis Awarded


Document Type

Honors Thesis

Degree Name

Bachelors of Arts (BA)




Jaime Settle

Committee Member

Christine Nemacheck

Committee Member

Cheryl Dickter


Individuals do not always express their private political opinions in front of others who disagree with them. Neither the political science literature nor the psychology literature has been able to firmly establish why this behavior occurs. Previous research has explored how social network composition can influence political attitudes and how political attitudes can be resistant to persuasion. However, the concept of conformity does not involve attitude change or persuasion; it more accurately involves self-censoring to match a socially desirable norm. In an effort to improve our understanding of this behavior, I conduct a lab experiment in which participants discuss political issues with actors who deliberately disagree with them. I measured the differences between the responses participants gave on a private survey compared to their publicly stated attitudes in the discussion group. Results indicate that regardless of the order in which participants gave their responses, individuals do indeed conform to the group’s opinion or censor their views. Conformity and censorship were most frequent among introverted, emotionally stable, and racial minority participants. Significant differences were also found between the types of issues; specifically, non-social and ideologically ambiguous issues yielded higher levels of conformity. Political conformity and censorship could lead to a distorted view of public opinion and may challenge the execution of freedom of expression.

The following errata was provided on March 5, 2016:

The main results of this analysis reported on page 17 were calculated incorrectly due to an error in the pretest survey. One question on the pretest survey, “On a scale of 1 to 10, with 10 being ‘very likely’ and 1 being ‘very unlikely,’ how likely would you be to vote for a candidate who supports the Patient Protection and Affordable Care Act, also known as Obamacare?” However, the Omnibus Survey coordinator accidentally mislabeled the scale such that 1 represented “very likely” and 10 represented “very unlikely.” As such, this question should have been reverse coded so that it matched the direction of this question in the lab and on the post-test. After correcting this error and reverse-coding the responses to this question, the following main results hold: Participants in the treatment condition conformed significantly more frequently than participants in the control condition for standard (p<.01) and strict (p=.105) conformity. 88.9% of participants conformed on at least one question by standard conformity measures (94.1% in the treatment group and 82.8% in the control group), and 58.7% of participants conformed at least once by pure conformity measures (65.2% in the treatment group and 52.2% in the control group). There are no statistically significant differences in censorship between the treatment groups. More details about the results are available upon request ( An updated version of this paper is published as “Political Chameleons: An Exploration of Conformity in Political Discussion” in Political Behavior DOI: 10.1007/s11109-016-9335-y.

img009.pdf (771 kB)