Men in red: A reexamination of the red-attractiveness effect
Elliot, Kayser, Greitemeyer, Lichtenfeld, Gramzow, Maier, and Liu (Journal of Experimental Psychology: General, 139(3), 399-417, 2010) showed that presenting men in front of a red background or with a red shirt enhances their attractiveness, sexual desirability, and status in the eyes of female observers. The purpose of the present research was to gain further insights concerning the robustness and the ecological validity of this red effect. In two experiments, we replicated the basic paradigm used by Elliot et al. Experiment 1 was a close replication of the first experiment in their original series. We presented the photo of a young man used by Elliot et al. on either a red or white background and asked participants (N = 89, female subsample n = 72) to rate it with regard to perceived attractiveness. Experiment 2 (N = 32) represents a somewhat more complex version of the first experiment; we increased the variance of the stimuli by showing photos of multiple men wearing different apparel styles (formal and casual, respectively). We did not find any significant impact of red in either of the studies. What we found, however, was a significant effect of apparel style with attractiveness ratings being higher for men wearing formal apparel than for men wearing casual apparel. Our results question the robustness and the ecological validity of Elliot et al.’s finding. On a more general level, they further point to limitations arising from (often necessary) restrictions in experimental designs.
KeywordsReplication Color Attractiveness Red effect Status
Overview of the seven experiments testing the red effect reported by Elliot et al. (2010)
Colors (CIE-Lch values)#
Test statistics (red effect: Exp. 1-4, 5a, 6a, 7; status effect: 5b, 6b
Red (49.6, 58.8, 30.4)
White (no color)
t(20) = 2.18, d = 0.95*
32 f. 25 m
Red (49.6, 58.8, 30.4)
White (no color)
Female observers: t(53) = 3.06, d = 1.11*,
Male observers: ns
Red (50.0, 59.6, 31.3)
Gray (50.0, — , 69.1)
t(32) = 2.44, d = 0.86*
t(32) = 2.43, d = 0.85*
Red (51.3, 51.7, 30.1)
Green (51.5, 51.6, 136.6)
t(53) = 2.03, d = 0.56*
t(53) = 2.25, d = 0.62*
Red (59.2, 104.9, 44.1)
Gray (59.3, — , 100.4)
Perceived present status
Perceived status potential
t(19) = 2.17, d = 1.11*
t(19) = 2.94, d = 1.35**
t(19) = 4.26, d = 1.96*
t(19) = 3.67, d = 1.68**
Red (52.3, 46.3, 29.0)
Blue (52.9, 46.1, 276.9)
t(35) = 2.23, d = 0.74*
t(36) = 2.82, d = 0.94**
t(36) = 2.10, d = 0.70*
Red (54.8, 43.2, 30.3)
Blue (55.1, 43.7, 283.0)
β = 0.40, F(1, 25) = 4.81*
β = 0.43, F(1, 25) = 6.43*
β = 0.42, F(1, 25) = 5.24*
The accumulated results of Elliot et al. (2010) seem strikingly convincing at first sight as the authors were repeatedly able to replicate their finding that a simple change in background or apparel color may have a strong and reliable impact on whether women classify a man as being attractive and sexually desirable or not. However, this flawless series of replications also turns out to be problematic as, in consideration of the estimated power of the experiments, it might indicate a publication bias in the reported, and thus accessible, results (Francis, 2013). In addition, the findings of Elliot et al. may indeed be generalizable to a certain extent (they tested participants from four countries, manipulated the color of two different features, and used four different contrasting colors); however, they lack ecological validity for a couple of reasons. First, in exclusively focusing on the impact of color, Elliot et al. neglect further important factors or features, such as apparel style (cf., for instance, Hill, Nocks, & Gardner, 1987; McDermott & Pettijohn II, 2011) that also may (or to a greater extent than color) affect a person’s perceived attractiveness, sexual desirability, and status in real life. Second, in each of the reported experiments only one photo of a moderately attractive man was presented to the participants, and over the complete series of experiments photos of only six different male targets were used. Being thus restricted, the stimulus material does not closely reflect the variability of physical appearance found in reality. So, it is unknown whether the enhancing effect of red is restricted to a medium base level of attractiveness, and it remains unclear whether the red effect will still be found if processes of comparison that typically occur in everyday interpersonal encounters take place (e.g., different kinds of context effects, Geiselman, Haight, & Kimata, 1984; Gerger, Leder, Faerber, & Carbon, 2011; Kenrick & Gutierres, 1980).
In the present research, we addressed some of the problems mentioned above and reexamined the red effect as reported by Elliot et al. (2010). We tested the robustness of the red effect and aimed to attain further insights concerning its ecological validity. We conducted two experiments replicating the basic paradigm employed by Elliot et al. to test the effect of red on perceived attractiveness and perceived status: Experiment 1 is a close replication of the first experiment in Elliot et al.’s original series with the stimulus material and the operationalization of the dependent measure (perceived attractiveness) being the same as used by Elliot et al. Experiment 2 represents an extended version in which we used stimuli with a wider range of variation and included apparel style as an additional factor.
Eighty-nine students (17 males; age M = 21.9 years, SD = 5.0, range: 18-51 years) voluntarily participated in the experiment that was run in the context of a psychology lecture. Participants were grouped according to the location (left-hand side, right-hand side) of their seats in the auditorium. The groups were then assigned to the color conditions red (n1 = 44, 9 male; age M = 21.2 years, SD = 5.2, range: 18-51 years) and white (n2 = 45, 8 male; age M = 22.6 years, SD = 4.8, range: 18-39 years), respectively. Participants were naïve to the purpose of the experiment. It was ensured that participants in the condition red did not see the test material used in the condition white and vice versa.
For each participant, we generated a two-page sheet printed on orthochromatic natural paper (100 g/m2, type ColorCopy by Mondi Ltd.) with the stimulus being displayed on the first page and three questions for assessing perceived attractiveness on the second page. The questions were the same as those used by Elliot et al. but translated into German: 1) “Wie attraktiv ist diese Person Ihrer Meinung nach?” [“How attractive do you think this person is?”]; 2) “Wie angenehm ist es, diese Person anzusehen?” [“How pleasant is this person to look at?”]; and 3) “Wenn ich die gezeigte Person real träfe, würde ich denken, dass sie attraktiv ist.” [“If I were to meet the person in this picture face to face, I would think he is attractive.”]. The items were to be evaluated using 9-point rating scales ranging from 1 = gar nicht [not at all] to 9 = sehr [very much].
Depending on the condition to which they had been assigned, participants received a sheet that included the photo of the target person on a red or white background. They looked at the photo for 5 seconds, then turned it over and evaluated the displayed man’s attractiveness by means of the three questions printed on the second page of the sheet.
Results and discussion
Thirty-two female students (age M = 20.2 years, SD = 2.1, range: 19-26 years) voluntarily participated in the experiment for extra course credit. Half of the sample was randomly assigned to the color condition red, the other half of the sample to the color condition white. All participants had normal or corrected to normal vision (verified by a Snellen chart test) and normal color vision (verified by a short version of the Ishihara color vision test). Participants were naïve to the purpose of the experiment.
Participants were tested individually. Each participant was assigned one of the eight sets of stimuli; assignment was counterbalanced across the sample so that four participants were tested per set. Participant and experimenter were sitting next to each other, with the experimenter’s view of the participant being obscured by a screen. The experimenter sequentially presented all stimuli of the picked set to the participant. After looking at a stimulus for 5 seconds, the participant evaluated the depicted man’s attractiveness by means of one item (“Wie attraktiv ist die gezeigte Person?”/ “How attractive is the presented person?”), using a 9-point scale ranging from 1 = gar nicht attraktiv [not at all attractive] to 9 = sehr attraktiv [very attractive]. After a short break, the experimenter presented the set to the participant for a second time. Stimuli were again presented sequentially for 5 seconds each and in the same order as before. This time, the participant evaluated the status of each depicted man by means of one item (“Als wie hoch schätzen Sie den Status der Person ein?”/ “How high in status do you think this person is?”), using a 9-point rating scale ranging from 1 = gar nicht hoch [not high at all] to 9 = sehr hoch [very high].
Results and discussion
In the present research, we reexamined the red effect as reported by Elliot et al. (2010). According to their findings, women perceive men presented in front of a red background or wearing a red shirt as being more attractive, more sexually desirable, and higher in status. We conducted two experiments replicating the basic paradigm used by Elliot et al.
Experiment 1 was a close replication of the first experiment in Elliot et al.’s (2010) original series—“close replication” in terms of: 1) having used a reproduction of the original stimulus material (same photo as used by Elliot et al., same image dimensions, also printed on paper), 2) having operationalized the dependent variable perceived attractiveness by means of the same three questions that had been asked in the original experiment, and 3) having used the same 9-point answering scale. As Brandt et al. (2014) pointed out, a replication in psychological research will never be absolutely exact or direct (see also, Stroebe & Strack, 2014), which is, of course, also the case in the present research. Major differences compared to the original study are: 1) In contrast to Elliot et al. who tested each participant individually, we tested the whole sample at once in a group session; 2) we tested German participants instead of a sample from the United States (note, however, that Experiments 5a and 5b of Elliot et al.’s original series were conducted in Germany as well), which also required using German translations of Elliot et al.’s original questions for the assessment of perceived attractiveness; and 3) we utilized a larger sample size (total N = 89, female subsample n = 72), which yielded a high test power (total sample 0.993, female subsample 0.978).
Experiment 2 was an extended, thus more complex version of the original paradigm, in which we assessed the impact of red on perceived attractiveness and status using stimuli with a wider range of variation. While in Experiment 1 we used only one photo of a young man (just as had been done by Elliot et al.), we used photos of multiple men dressed in different apparel styles (formal or casual) in Experiment 2. The stimuli thus reflected at least part of the variation to be found between different persons in real life, which allowed for comparative processes common in interpersonal encounters to occur within the experimental context as well.
Neither of our replications yielded a statistically significant effect of the color red on the dependent variables (Experiment 1: perceived attractiveness, Experiment 2: perceived attractiveness and perceived status). This indicates that the red effect probably is not as robust as the multiple successful replications reported by Elliot et al. (2010) might initially suggest. Francis (2013) put forth that Elliot et al.’s results might be contaminated with publication bias. In reaction to Francis’ comment, Elliot and Maier (2013) replicated the third experiment of the original series (comparing the impact of red and gray) with a larger sample (N = 144). This time, they did not find a significant red effect either.
Besides challenging the robustness of the red effect, our results also question its general ecological validity. We have shown that other factors may be more powerful in affecting the perception of (male) attractiveness and status than the color red: In Experiment 2, again no significant impact of red on perceived attractiveness and status occurred; both variables, however, were significantly affected by the factor apparel style: Men wearing formal suits were perceived as being more attractive than men wearing casual sportswear. Concerning perceived status, this difference was even more pronounced.
The resulting conclusion that apparel style can have an impact on how we perceive other persons is in line with findings from earlier studies (Nielsen & Kernaleguen, 1976). Hill, Nocks, and Gardner (1987), for instance, showed that apparel displaying higher status can enhance different attractiveness measures (physical, dating, sexual, and marital attractiveness) under certain circumstances. This was especially the case for female raters evaluating male models. In Experiment 2 of the present research, the formal suits obviously conveyed a higher status than the casual sportswear as is indicated by the significant difference in the average ratings of perceived status that were given for men wearing these different apparel styles. According to Elliot et al. (2010), status potentially functions as a variable mediating higher perceived attractiveness. This also might be the case with regard to the impact of apparel style and perceived attractiveness that was found (status and attractiveness were significantly correlated, r = 0.576, p = 0.0005). Detailed and profound conclusions about such a mediation chain, however, require further experiments, and the inclusion of further potentially associated variables and constructs.
The revealed effects of apparel style with regard to perceived attractiveness and status may not have come unexpectedly and they were reasonably large (ηp2 = 0.387 and ηp2 = 0.888, which equals Cohen’s d = 2.82 and d = 5.63, respectively). Yet so far this is a singular finding obtained by only one experiment in the lab, and it may well be the case that in a more naturalistic setting or in real life there are further factors having greater impact than apparel style or factors levelling its effect. Even if the effect occurs in real life, we cannot conclude that it is based on the same processes as the effect found here in the lab. Wearing formal apparel, for instance, might change the wearer’s feeling and manner, which will further influence the way observers perceive him (“wearer effect”; with regard to color-associated impacts a wearer effect was shown by Roberts, Owen, & Havlicek, 2010). Using static stimuli with an a posteriori color manipulation, as we did in the present research, does not allow for the reproduction of such processes; the effect of apparel style that we found in the lab must ipso facto be based on something else. In this regard, our experiment and its results are an exemplary case of a problem that is not too uncommon in experimental research: the potential difference between the processes taking place in an experiment with sharply specified, controlled conditions and those taking place in a markedly more complex real-life setting.
In sum, our results challenge the red effect as it was reported by Elliot et al. (2010)—which does not, however, automatically imply that red (or any other color) is principally without effect. Our conclusions are limited to the specific paradigm that was used by Elliot et al. and replicated in the present research. On a more general level, our results paradigmatically point to the consequences that arise from (necessary) restrictions in experimental designs. This does not call into question experimental psychological research per se, but it should remind us to keep in mind the scope and limitations of our findings obtained by means of the experimental approach we follow and the methods we use.
- McDermott, L. A., & Pettijohn, T. F., II. (2011). The influence of clothing fashion and race on the perceived socioeconomic status and person perception of college students. Psychology & Society, 4(2), 64–75.Google Scholar