This experiment is designed to research further into the Halo and Devil effects and replace the independent variable from Nesbitt & Wilson’s study (the warmness of the interviewer) with the attractiveness of an individual and see whether the Halo and Devil effects are generalisable to a situation when all an individual knows of a person is their attractiveness.
Independent variable: The attractiveness of a picture shown to an individual.
Dependent variable: The ratings between 1 (lowest) and 5 (highest) participants give on popularity, intelligence and success.
Hypothesis (directional): As participants see a more attractive person, their ratings for that persons’ popularity, intelligence and success will go up, despite these being unrelated characteristics.
Hypothesis (null): As participants see a more attractive person there will be no difference for their ratings for that persons’ popularity, intelligence and success.
Method
Design
The experiment will follow a repeated measures design; participants will be given both conditions one after another, this will show what affect the Halo and Devil effects have on that particular individual. This also allows the researcher to test the Halo and Devil effects on a large sample of participants, therefore gaining a more representative sample then an independent measures design would allow.
Participants
Participants will be chosen on an opportunity sampling basis; participants will be selected from the internet forum and asked to complete an online questionnaire. No more then 38 participants will be able to take place, the questionnaire links will be removed once this has happened. This enables the researchers to gain a large sample with relative ease and enables the selection of participants from around the United Kingdom whilst restricting it to students to form a generalisable random sample of students.
Materials
Pictures will be collected from the website to ensure participants do not identify with other characteristics of the person in the picture; if a celebrity is used then participants may be affected by the Halo and Devil effects for different attributes, for example, if a picture of Dylan Moran is used then participants may not find him attractive but may rate this high due to the fact that they find him funny (the Halo effect). This would add a bias to any data collected and make it difficult to generalise from this. The questionnaire will contain four Likehert scales numbered 1 (lowest) to 5 (highest) and asked to rate the person they see in terms of intelligence, success, attractiveness and popularity.
Procedure
Thirty-eight participants will be e-mailed the link to a questionnaire and asked to complete it. The questionnaire will contain four pages, each page will contain the image of one person with four Likhert scales below the image, participants will be asked to rate these people between 1 and 5 in terms of their intelligence, success, attractiveness and popularity (in that order). The four images will be one attractive male, one unattractive male, one attractive female and one unattractive female. This will enable the measurement of the Halo and Devil effects along with any gender differences between the two.
Results
Figures one, two, three and four show a significant gender divide; when the largest difference between the Halo and Devil effects for a male is in the rating of a males success (1.21 change in the mean) whilst in a female the largest change is in the rating for popularity (1.24 change in the mean). In fact, the Halo and Devil effect have only been shown as correct due to these factors; figure three demonstrates that despite a significant change in the perceived attractiveness for a male the only major difference is in the success rating. The 95% error bars for each other rating (intelligence and popularity) overlap significantly so the results cannot be deemed statistically significant. The interesting rating is within the rating for intelligence; the less attractive male is rated as more intelligent then the more attractive male (by 0.6). This adds question to the Halo and Devil effects as it goes against the prediction of the theory.
The female ratings show general support for the Halo and Devil effects, however, Figure four demonstrates that due to the 95% error bars overlapping significantly, the only rating that is significant for females is popularity; as attractiveness decreases that rating given for popularity also significantly decreases.
Fig1. Table showing the figures of central tendency for males
Fig2. Table showing the figures of central tendency for females
Fig3. Bar Chart showing the rating of each question for males
Fig4. Bar Chart showing the rating of each question for females
Discussion
The results of this experiment show very little support for the Halo and Devil effect; the majority of the variables have error bars that overlap significantly, however, there is a significant difference in the success ratings for males, and the popularity ratings for females, this can lead to an acceptance of the directional hypothesis (“As participants see a more attractive person, their ratings for that persons’ popularity, intelligence and success will go up, despite these being unrelated characteristics.”). This research can be applied in everyday life in terms of advertising, if a well known celebrity sponsors a particular brand then people may become attracted towards that brand merely because of the positive traits of the sponsor, rather then the actual quality of the brand.
However, more research is needed to clarify that the Halo and Devil effect exist; this research demonstrated very weak support that could lead to a false conclusion; despite the acceptance of the hypothesis there was still no significant difference between two of the three traits, furthermore, the intelligence rating for the less attractive male went up, this implies severe questions to whether the conclusions of this study can be supported. Furthermore, an unreliable sampling method was used, opportunity sampling, whilst easy for a researcher, will not always yield a representative sample making the sample lack generalisibility.
References
Cook, G. I., Marsh, R. L., & Hicks, J. L. (2003). Halo and devil effects demonstrate valenced-based influences on source-monitoring decisions. Consciousness & Cognition, 12, 257-278
Nisbett, R. E., Wilson, T. D.. 1977. The halo effect: Evidence for unconscious alteration of judgements. Journal of Personality & Social Psychology, 35, 250-256
Zebrowitz, L. A., Hall, J. A., Murphy, N. A. (2002) Looking smart and looking good: Facial cues to intelligence and their origins. Personality & Social Psychology Bulletin, 28, 238-249
