• No results found

Does people beauty discriminate because they pay more attention to pretty candidates?

N/A
N/A
Protected

Academic year: 2021

Share "Does people beauty discriminate because they pay more attention to pretty candidates?"

Copied!
40
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Does people beauty discriminate because they

pay more attention to pretty candidates?

Master Thesis

Author: Siqi Liu

Student number: 11615982

Thesis supervisor: Jeroen van de Ven

Date: July, 2018

Faculty of Economics and Business MSc Business Economics

Specialization: Managerial Economics and Strategy

(2)

Statement of originality

This document is written by student [siqi liu] who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The faculty of economics and business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

Beauty discrimination has been proved to exist in labor market by many research, but there is no clear answer what causes beauty premium. In this paper, I want to see if the difference in attention can explain part of the beauty premium. An online experiment was run to verify this assumption. The experiment uses 35 game contestants from a quiz show called “the weakest link” as material, testing experiment participants’ memory and attitude towards them. The result shows that a participant can recall more details about a contestant correctly when this contestant is prettier than other contestants, which means that they pay more attention to good-looking faces than plain faces. Participants distributed their attention unevenly before they show any sign of beauty discrimination against any game contestant. One more surprising result is that when an experiment participant cannot recall details about a contestant correctly, they tend to overestimate the performance of a bad-looking contestant and underestimate the performance of a good-looking contestant.

(4)

CONTENT

ABSTRACT ... 3

I. INTRODUCTION ... 5

II. LITERATURE REVIEW. ... 8

A.BEAUTY DISCRIMINATION ... 8

B.ATTENTION ... 11

III. THE EXPERIMENT ... 13

A.EXPERIMENTAL DESIGN ... 13

Why I chose “the weakest link” ... 14

B.EXPERIMENT PROCEDURES ... 15

The video clip, contestants, and questions ... 17

Incentivize ... 18

C.RATE ATTRACTIVENESS ... 19

D.HYPOTHESIS ... 20

IV. EXPERIMENT RESULT ... 21

RESULT 1 ... 21

RESULT 2. ... 28

RESULT 3. ... 32

V. CONCLUSION ... 36

(5)

I. Introduction

Since Hamermesh and Biddle (1994) first proved the existence of beauty premium, beauty discrimination has been studied a lot. Many studies tried to compare people’s wage with their appearance. Most studies confirmed the existence of beauty discrimination in the labor market and daily life. Lawyers earn more if they are pretty, Biddle & Hamermesh (1998) ; Unattractive players are more like to be sent away in TV game show even they perform no worse than attractive players, Michèle Belot et al (2012). However, there are also some different voices exist. Adolfo Sachsida et al (2003) verified the impact of physical appearance on the salespeople’s wage, their result showed that women with good appearance receive higher wages but this is related to productivity and not beauty discrimination.

From the perspective of discrimination theories, there are two main types of discrimination. The first one is taste-based discrimination which is first proposed by Becker (1957). Employers with taste-based discrimination would avoid having interaction with one ethnic group even if those applications’ ability is no worse than others’. and employers would like to pay financial prices to do so. The second theory is statistical discrimination. It says that when there is no clear information about a candidate’s ability or productivity, decision-makers will judge this candidate base on their group average performance, which leads to racial or gender inequality based on stereotypes.

Attractive people seems taking advantages from less attractive people simply from appearance. But the reason behind beauty discrimination is in fact much more complex

(6)

than race and gender discrimination. First of all, attractiveness is subjective. It differs with raters’ gender, age, education level and their personal taste. Many other factors, such as whether a person looks friendly or confidence, could determine whether a person is attractive or not. Secondly, workers with higher income have more spare money to improve their looks. Their high wage helps them to improve their confident level and appearance. It seems that both taste-based discrimination and statistical discrimination can be used in explaining beauty discrimination.

Currently, there is no clear answer for what causes beauty discrimination. Do people beauty discriminate because they judge others’ working ability, reliability or personality directly by appearance, or they prefer to interact with the attractive person even if this is costly? Mobius and Rosenblat (2006) “decompose the beauty premium in an experimental labor market. They identified three transmission channels: the visual and oral stereotype channels and the confidence channel”. They conclude that higher expectation by employers, higher confidence, and better communication skill are reasons why attractive workers have beauty premium.

In this paper, I would like to focus on the relationship between beauty premium and attention. Some studies in developmental psychology found that babies spent more time staring at good-looking faces than not good-looking faces. Judith H. Langlois et al (1987) found out that both the older and younger infants looked longer at attractive faces when the faces were presented in contrasting pairs of attractiveness. People not only spent more time on good-looking faces as babies, they as adults, also spend more time, attention or energy on people with pretty faces than bad looking face. People tend

(7)

to look longer at attractive than unattractive faces (Aharon et al., 2001). Those theories lead to a possible explanation for beauty premium: differences in attention levels leads to unbalanced familiarity and this cause beauty discrimination.

Attention is a valuable resource economically. A paper by Bartoš et al (2016) confirmed that differences in attention level could lead to systematic discrimination. This is achieved by two steps. Firstly, a candidate has some kind of trait that can attract more attention from HRs or decision maker, others don’t. HRs or decision makers do not have unlimited time to gather substantial information about each applicant. Attention, as a valuable resource, is distributed unevenly form this step. They gather more information form candidates who get more attention than others. Secondly, imperfect information encourages HRs or decision makers to statistical discriminate against candidates that they are not familiar with. Because the short of information about those candidates, the possibility that HRs or decision makers judge them by stereotype is higher than others.

Beauty is a trait which can cause difference in attention level. In this paper, I want to further confirm that do people recall information of attractive people more accurately. When there are limited opportunities, HRs or decision makers will consider people who they know that are of high ability firstly. If pretty candidates always attract more attention than other people, they get the chance to make HRs or decision makers be more familiar with them. Furthermore, people recall more details or recall details more correctly about pretty candidates than plat candidates. This means that People among those pretty candidates who also have a high ability are very likely to be chosen first,

(8)

and others are very likely to be ignored or statistical discriminated. Systematically, this leads to a bias toward pretty candidates.

The remainder of the paper is organized as follow. Section Ⅱ has a brief literature review of beauty discrimination and attention. Section Ⅲ introduce my experimental design and how I collect the data. Section Ⅳ gives and discusses three main results and Section Ⅴ provides the conclusion.

II. Literature review. A. Beauty Discrimination

The first paper that starts the study about beauty discrimination is by Hamermesh and Biddle (1994). They found out that plain people earn less than average-looking people, and good-looking people earn more than average-looking people in the labor market. In this part, I will have an overview of researches that proved the existence of beauty discrimination and studies that tried to decompose the beauty premium.

After Hamermesh and Biddle, there are some other studies confirmed that beauty discrimination exists in specific industries. Biddle and Hamermesh (1998) concluded that beauty cause differences in earning among lawyers. Sachsida et al (2003) verify the impact of physical appearance on the salespeople’s wage of Brasilia’s shopping malls. This wage discrimination among salespeople is from customers’ behavior. Gerard A. Pfann, (2000) found that in Dutch advertising firms, a company with a better-looking executive can also get higher revenue. Beauty also increases a candidate’s chance of being elected. Hamermesh (2006) used data from the American Economic

(9)

Association and found out that when a person improved his/her look, the chance of getting an electoral success also improved. It is clear that how people look is having an effect on their earning and their career path.

Meanwhile, many studies want to see what behind beauty premium. Does the difference in wage among plain people and beautiful people is actually due to productivity or due to beauty discrimination? Cipriani and Zago (2011) suggested that good looks could make people more productive. They concluded that good-looking people earn more not because the existence of discrimination, but because they are more productive. Hamermesh and Parker (2005) found a similar result. They used data from a group of universities. The result from their data shows that those instructors who are good-looking are rated higher by students about their performance and their classes. However, they are not sure that do student really learn more from good-looking instructors or do they beauty discriminate against bad-looking instructors when they are rating. Leigh and Borland (2007) tested the relationship between beauty premium, appearance and self-confidence. Their finding suggests that being more confidence does not explain the beauty premium. Customer-based discrimination, employer-based discrimination, or productivity are still reasons why beauty premium exists. However, based on what Mobius and Rosenblat (2006) confirmed, Higher expectation by employers, higher confidence and better communication skills are reasons why beauty premium exists.

Taste discrimination.

Becker (1957) defined discrimination as a difference in pay between two workers of equal productivity and he proposed the taste-based discrimination model. In his model,

(10)

if there are two group of applicants and they have same productivity, employers discriminate against one group of people because of their individual taste and they are willing to pay financial prices to do so. Becker also introduced three sources of discriminatory tastes: employers, co-workers and customers.

Statistical discrimination.

Inspired by Becker’s theory, Arrow and Phelps proposed models of discrimination based on rational optimizing behavior and limited information (Guryan and Charles, 2013). Based on their contribution, Statistical discrimination becomes another important discrimination theory. Statistical discrimination describes a situation that employers use average characteristics of a group to predict individual worker attribute. Stewart Schwab (1986) made his argument that Statistical discrimination maybe not efficiency. Even though statistical discrimination is based on free, accurate information, it does not necessarily allocate resources more efficiently than if firms ignored the information.

From all previous researches, there are no clear answers to what causes beauty discrimination. Good-looking people can be more productive, but this happens when good-looking helps productivity. For example, in sales, teaching, political, or law industry. Faces can play a role in these occupations. Instead of explaining the beauty discrimination happen in these occasion by productivity, we can also blame it on customer-discrimination. In this paper, I want to contribute decomposing the beauty premium by testing the relationship between beauty and how people pay their attention differently.

(11)

B. Attention

One part of the study about attention focus on infant attention and facial attractiveness. Studies about infant attention want to know how human infant distribute their attention because this should be connected to learning in the opening months of life (J. Kagan and M.Lewis (1965).

Studies about facial attentiveness compare human face to ordinary objects. These studies found that human faces attract more attention than ordinary objects. For example, Hershler & Hochstein,(2005) demonstrated that detection of human faces among a variety of objects is close to independent of the size of the search array. Theeuwes and Van der Stigchel (2006) used the occurrence of inhibition of return (IOR) and proved previous studies that attention may be preferentially directed and engaged longer by faces

Beautiful faces can also attract more attention than plat faces. Langlois et al (1987) did experiments on infants’ preference for attractiveness. Their results showed that both the older and younger infants looked longer at attractive faces when the faces were presented in contrasting pairs of attractiveness. This means that human infants distribute their attention to beautiful face and plate face differently before they are exposed to culture standard about beauty. (Rubenstein et al (1999) gave out a cognitive explanation that infants’ preference for attractive faces can be explained by general information-processing mechanisms. This can be explained socially and biologically. Annukka K. Lindell and Kaarina L. Lindell (2014) confirmed that beauty holds privileged attentional status because brain has evolved to activate neural networks

(12)

associated with reward in response to beautiful faces. People tend to look longer at attractive than unattractive faces. Even when people are doing other tasks, facial beauty automatically competes with an ongoing cognitive task for spatial attention. (Sui & Liu, (2009). Olson and Marshuetz (2005) also found that facial beauty can be appraised automatically and rapidly, which means that beauty is difficult to ignore, and this might explain why beauty premium exists and physically attractive people on average have higher wages and wider varieties of mate choice. People pay attention to a selected group. More attention further leads to more familiarity.

Attention economy.

Davenport and Beck (2001) describe attention as the new currency of business. Many studies about attention economy see attention as a scarce resource, especially when it is combined with new media Industry. Marazzi (2008) defined human attention as a scarce but quantifiable commodity. The success of many new internet companies also proved that attention itself contain enormous economic values. For example, photo-sharing application Instagram was purchased by another internet giant Facebook by $1 billion in 2012. Instagram is selling users’ attention. “page views and clicks are synonymous with success and thus online status.” Alice E. Marwick (2015).

There is no doubt that attention can be seen as a valuable resource and the ability to attract other people’s attention can bring economic benefits to a person. Then, it is reasonable to relate the value of attention to beauty premium. If good-looking people do attract more attention from HRs and decision makers compared with other plain competitors, attention should explain part of the existence of beauty premium.

(13)

III. The Experiment A. Experimental design

In my experiment, my main purpose is to test if experiment participants can recall details more correctly about a pretty candidate than a bad-looking candidate. by doing this, I can see if people collect more accurate information form attractive candidates than plain people. Attention is a valuable resource. If pretty faces attract more attention than plain faces, it is reasonable to assume that people will have more information and be more familiar with candidates that are pretty because more attention is paid. The difference in familiarity leads to systematic beauty discrimination. Because imperfect information encourages people to ignore or statistical discriminate plain candidates. Based on this assumption, I designed an experiment to test whether attractiveness affects the familiarity level.

The experiment consists of two parts. In the first part, Participants watch a 5 minutes video fragment which is cut from a quiz show called “the weakest link”. In the game show clip, there are game contestants who join the game and answer questions in the show. In the second part, experiment participants are asked to answer sixteen questions, intending to test how familiar they are with each game contestant and do they beauty discriminate toward pretty contestants.

The material game show I used is a British show called “the weakest link”. In each episode of this show, there are nine contestants. I use those contestants as my materials, testing their attractiveness level and experiment participants’ memory of them. In my experiment, I use 35 contestants from 5 episodes show as the experiment material,

(14)

which means I used 7 contestants from one episode and cut it in one video clip. In this way, I used 5 video clips in total in my experiment. Each participant will only be shown one video clip. I made five versions of online experiments, each of them contains one video clip and questions refer to it.

In term of participants, I targeted at 30 experiment participants for each video clip, which means 150 experiment participants in total. In the end, I recruited more than 150. 164 participants to join this experiment. The experiment was designed online. Links were posted online and shared in social networks.

Why I chose “the weakest link”

“The weakest link” is a television quiz show broadcast on 14 August 2000 in British. In each episode, nine contestants answer general questions in turn. They answer several rounds of questions to accumulate prize money. After each round, all players will vote who they think is the weakest link, and that person is out. Any of the nine contestants in one episode could win up to 10,000 pounds. There are nine contestants in each episode, all nine people do not know each other before they begin the show. They need to cooperate and compete with each other. They cooperate to answer each question correctly and accumulate the prize money. But only one people could win money in the end. The game show began on 14 August 2000 and the last episode is on 31 March 2012. There are overall 1694 episodes from the British version.

I choose this game show as my experiment material because it has clear elements and stable time structure. In basically all British episodes, from 2000 to 2012, the game has just one host and nine contestants in each episode. By using this as material, participants’

(15)

attention is not likely to be disturbed in other directions. Because what they can see is controlled. When one contestant is introducing herself or himself, the background is black and there is no other element in the scene expect this contestant self and his/her name tag. Besides the clear element, another advantage of using this quiz show is that it has a stable time structure. In the five minutes video clip I used, every contestant shares the time equally and shows up using similar time. This character helps me control that each contestant is showed to experiment participant in a similar time. If each contestant is showed to experiment participants in same time length, but the participant recall detail from contestants differently. This could be explained by that they paid different attention to different contestants.

B. Experiment Procedures

The experiment contains two parts. In the first part experiment participants need to watch a video clip that cut from one full episode of “the weakest link”. After that, participants need to answer questions related to the video clip they just saw.

Part one. Each episode of this television show originally has around 45 minutes. For controlling the experiment time, I cut out unnecessary talking part and voting part and only kept self-introduction part and the second round of answering question part. The video clip ends up with about 5 minutes. In the video clip which the experiment will use, the video clip starts with all nine contestants introducing themselves. Then they will see the host telling all contestants the rules for the game. After this, there will be only one round showed, which is the second round. In this round, there is already a contestant voted out, which means only eight contests are competing and answering

(16)

questions. In the full episode, one contestant leaves after each round. but the video clip in our experiment will not show this. Experiment participant will just know each contestant, the rule of this quiz shows and what happened in the second round of this game.

I choose to cut the video in 5 minutes because, in the second part of the experiment, participants will be asked to recall their memories and answer questions about each contestant. If they watch a video that is too long, it is a challenge to their memorizing ability, and 5 minutes’ video already contain enough information and quiz questions to be used in part two.

Part two. In part two, after watching the video fragment, participants will need to answer 16 detail questions about what game contestants did in the video fragment. In each episode of “the weakest link”, there are nine contestants, only seven of them are used as experiment materials. The sixteen questions are about them, and questions can be sorted into three types based on my purpose of setting them.

One type of questions asks about color of contestants’ clothes and who answered which question. Answers of these questions are details that can easily recall. The purpose of setting this type of questions is testing if experiment participants remember more information about good-looking contestant than plain contestants.

Another type of questions is set as the following: How many questions did contestant “name of a contestant” answered correctly? I have two purposes of setting these questions. First is to test if participants can correctly recall each contestant’s information. Another purpose is to see if participants will overestimate or underestimate

(17)

game contestants’ performance when they cannot correctly recall the details. If they do, I could see if this estimation is based on attractiveness.

The two types of questions mentioned above contain 14 questions in experiment part two. All those 14 questions aim to test the familiar level participants have with each contestant. If participants watch the video clip with enough attention, they should recall enough detail about each contestant and answer most of the questions correctly. If they pay different attention to different contestants based on contestants’ appearance, they should recall more details of attractive players than plat players and have more accurate answers when questions are related to attractive players than unattractive players. Besides using these 14 questions to test the familiar level participants have with each contestant, I also ask participants two extra question to see if they are tending to beauty discriminate contestant themselves directly. The two questions ask them who they think is most likely to survive until the last and take all the prize money and who they think is most likely to be voted out before the next round. If participants predict pretty contestants are more likely to survive and bad looking contestants are more likely to be voted out. This could confirm that participants do beauty discriminate toward pretty people. If participants do not show any discrimination against bad-looking game contestants, we can see is there any other element that related to the decision that they made.

The video clip, contestants, and questions

I downloaded 5 episodes British “the weakest link” and cut each of them into a 5 minutes video clip respectively as the experiment material I was going to use. In each

(18)

episode, I picked 7 game contestants out of 9 contestants as my material. All together I picked 35 game contestants from 5 episodes. The five episodes are broadcasted on 10th November 2010, 18th March 2011, 2ed March 2012, 31st March 2012 and one episode’s broadcast time unknown. Among the 35 game contestants, 18 of them are male and 17 are female. Their average age is 37.91.

In term of the 16 questions in part two. Among all sixteen questions, 14 of them are about details of 7 contestants. Each experiment participant needs to answer 2 questions about one game contestant. In the video clip on participant watched, there are 7 contestants showed. This means that these seven contestants will be rated how attractive they are and each of them owns two questions aiming to test how familiar experiment participants are with them.

Overall, there are 35 contestants rated, 164 participants joined the experiment and 2296 questions about participants’ memory answered and collected (14 question*164 times). Each rated contestant owns 2 questions.

Incentivize

Each participant is payed 15RMB (2 euro) after finishing the experiment. The amount of money is not much nor too little, in consideration of the time length and the convince of doing it online. The experiment time is controlled within 10 minutes because I use a 5 minutes video clip in part one and ask 16 questions in part two. I am not going to pay participants based on how they perform in the experiment, for example, the number of questions they answer correctly. Because this is encouraging them to pay attention to every detail in the video clip no matter a contestant is attractive or not. It is better to let

(19)

participants spend as much energy and attention as they want as long as they watch the whole video clip.

C. Rate attractiveness

My method to rate the attractiveness of the 35 contestants that I used in my experiment is similar to the method which Michèle Belot, V. Bhaskar, and Jeroen van de Ven had used in their experiment (2012).

At the beginning of each episode in “the weakest link”, 9 contestants have a very short introduction about themselves. I use pictures cut from each contestant in this introduction part as material. I did an extra online experiment link using all these 35 pictures from contestants that need to be rated, there is a bar under each person’s picture. Raters can drag the bar form 0 point to 10 points so that they can rate the attractiveness of the picture upon the bar. The experiment link is shared on social network and participant join this voluntarily. There are altogether 21 Raters that I recruited online. 8 of them are males and 14 are females. Among those 21 results, two of those results are outliers because all 35 contestants’ attractiveness are rated under “3.0”. Those two results are deleted, so there are 19 effective answers. Raters are on average 29.87 years old.

Attractiveness raters firstly see a short introduction. It says that they need to rate the attractiveness of 35 people and the range is from 0 to 10. They are also told to use the benchmark average attractiveness in the population as 5.0. After they read this short introduction, they will see pictures of each contestant. What they do is to drag a bar

(20)

under the picture of one contestant and rate each contestant one by one. The results are shown in table1.

Table1

Attractiveness of contestants

Mean Standard deviation Minimum Maximum

All(N=35) 4.40 0.80 2.96 6.46

Men (N=18) 4.46 0.66 3.53 5.77

Women (N=17) 4.35 0.93 2.96 6.46

Age (N=35) 37.91 14.43 18 72

Table 1 shows that the 35 contestants that we used as material, their average attractiveness level is 4.40. Men among those 35 people are on average more attractive than women. The reason could be that we have 14 raters that are women, which takes 63% of raters. it also could be explained that because both the youngest and the oldest contestants are women, and people rate young people most attractive and old people most unattractive. from the result, it also shows that both the most attractive and unattractive person are women, rated 2.96 and 6.46 representatively.

Raters highly agree on the attractiveness they gave to each contestant. The Cronbach’s alpha coefficient is 0.916. This means the average attractiveness can is reliable.

D. Hypothesis

The main purpose of this thesis is to test if pretty people who are more attractive than people with plat face can gain more attention from others. Since attention is a valuable

(21)

resource, the difference in attention level might partly explain why beauty discrimination exist. Experiment are set to test 164 participants’ memory about all 35 game show contestants, to see if people remember more details about a contestant if he/she is more attractive than others.

Independent variable is facial attractiveness of 35 game show contestants, which are rated by 20 random raters. Dependent variable is the number of correct answers that experiment participants answered for each contestant in experiment part two. Controlled variables are contestants’ age, gender, experiment participants’ age, gender, and education level.

Hypothesis 1: Participants recall more details (have more correct answers) about attractive contestants than less attractive contestants.

Hypothesis 2: When a participant answers a question about a contestant wrongly, this participant will overestimate performance if contestant is pretty and underestimate if contestant is unattractive.

Hypothesis 3: Participants forecast attractive contestants to survival until the last. Hypothesis 4: Participants are more likely to vote out less attractive contestant if they are one of those players.

IV. Experiment Result

Result 1. Experiment participants recall details about good-looking people more

(22)

In the experiment, two types of detail questions are asked. One type of questions asks about a contestants’ dressing details or about a specific question a contestant answered. The other type of questions asks experiment participants to recall how many quiz questions each contestant answered correctly in the game show clip. Each participant’s answer to each question is counted as a dummy variable. It is one when a participant can answer a question about a contestant correct and zero if it is wrong. Attractiveness measures the average attractive points that each contestant owns, rated by 19 raters. Figure 1 and figure 2, show the number of attractiveness each contestant is rated and how many detail questions from a contestant are answered correctly by participants. There is no clear patent that we can see just from the figure.

Figure 1

Attractiveness is rated by19 raters recruited separately from experiment participants. Number in horizontal axis represent attractiveness. Number in vertical axis represent the number of correct answer participant recalled about each contestant.

0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7

Attractiveness of each contestants and how many detials about them are recalled by participants correctly

Correctly recall color and "who answered which" questions Correctly recall performance questions

(23)

Figure 2

Attractiveness is rated by 19 raters recruited separately from experiment participants. Number in horizontal axis represent attractiveness. Number in vertical axis represent the number of correct answer participant recalled about each contestant.

Table 2 presents results of 4 OLS regressions. It shows that when participants are asked to recall performance of a contestant, game contestants’ attractiveness is significantly related to how correct a participant can recall details. When a game contestant is prettier, a participant is more like to recall performance correctly. This means that when a game contestant is prettier, an experiment participant’s memory about how well this contestant performed in the game is more accurate.

0 5 10 15 20 25 30 35 2. 96 3. 04 3. 47 3. 53 3. 54 3. 76 3. 76 3. 79 3. 81 3. 89 3. 99 3. 99 3. 99 4. 02 4. 04 4. 23 4. 25 4. 28 4. 29 4. 31 4. 37 4. 47 4. 52 4. 73 4. 79 4. 82 4. 87 4. 96 5. 19 5. 42 5. 58 5. 69 5. 72 5. 77 6. 46

Attractiveness of each contestants and how many detials about them are recalled by participants correctly

Correctly recall color and "who answered which" questions Correctly recall performance questions

(24)

Table 2

Details about contestants’ performance that participants recalled correctly.

(1) (2) (3) (4) Attractiveness .0848116* (.0488057) .1294444*** (.044635) .0900337* (.0498538) .1297728*** (.0444698) Age Contestants .0052227*** (.0017753) .0051623*** (.0020037) Male Contestants .0849825 (.0574492) .0851043 (.0571056) Correct answer contestant gives .0226061 (.0226061) .0038298 (.0474873) _cons .1969719 -.2415564 .1375276 -.2469451 R² 0.0190 0.0408 0.0202 0.0408

“Answer contestant give” calculated how many questions in the game quiz show clip each game contestant actually answered correctly. Standard errors are clustered at the contestant level. Dependent variable is how many detail questions about a contestant’ performance each experiment participant recalled correctly. coefficients of OLS regression; * p<0.05, ** p<0.01, *** p<0.001

This result confirmed our Hypothesis 1 that Participants recall more details (have more correct answers) about attractive contestants than less attractive contestants. It can be explained by that when participants are watching the video clip, their attention on each game contestant is distributed unevenly. They pay more attention when a more attractive contestant is shown, and less attention when a less attractive contestant is playing. Attractiveness leads to an unbalance attention distribution when participant watching a 5 minutes video. It is like that during the five minutes, the video has a

(25)

participant’s full attention, but eight contestants and a host in this video in this five minutes are also competing for attention. We can say that pretty candidates win this competition because they get more attention during the five minutes compared to other plain candidates. Participants remember more information about them and also recall details about them more correctly than others.

This is very like a situation when HRs or decision makers is choosing from candidates. They would spend a specific time period to look over all candidate. During this time period, all candidates are competing with each other. Although when HRs or decision makers are making their decisions, they want to focus on the education background, working experience and personal skills, candidates’ faces play a role in the competition without their notice. Candidates’ look affect final decision because decision makers do not know each candidate equally. They know more about a pretty candidate; They recall information about a pretty candidate more correctly than others because they unintentionally spent more attention on pretty candidates than others. Based on this, when there are two candidates who have the same qualified background for a job position or a working opportunity, a pretty candidate is more like to be chosen. Not because this candidate is better, but because the decision maker cannot recall how good the other candidate is.

In our experiment, participants are asked to watch the video carefully, but they are not told to focus on which game contestant specifically. Also, all participants are paid equally. They get their payment as long as they finish the experiment. How they performed in the experiment, did they answered all questions in experiment part 2

(26)

wrong or they answered all questions right does not affect their final payoff. This means that they do not have any financial motivation to focus different game contestant differently. However, from the result in the row (1) in table 2, we know that without any incentive, they still recall information about pretty contestants more correctly than others. This is how I think that attention plays its role in beauty premium. It happens without any notification, but it does affect the familiarity level decision makers have with each candidate.

In table 2, row (2), (3) and (4), game contestant’s age (Age Contestant), game contestants’ gender (Male Contestant) and how well each game contestant performed in the video clip are add as control variable. “Answer contestant give” calculated how many questions in the game quiz show clip each game contestant actually answered correctly. Both age and gender could affect the attractiveness of a contestant. The results from row (2) confirmed this. We can see that after adding age and gender as control variable into the regression, coefficient of attractiveness increased from 0.848 to 0.130. Age and gender help us calculate the effect attractiveness have on how participant recall contestants’ performance more accurate than result in row (1). However, form row (3), it seems that the performance of game contestants is not related to their attractiveness. If good-looking contestants have higher ability and they performed better than plain contestants, the correct answers they gave should be related to attractiveness.

To see if how those three elements affect attractiveness, another OLS is run. Results are showed in table 3

(27)

Table 3 Attractiveness Age Contestants -.0268894*** (.0100561) Male Contestants -.020924 (.2493692)

Correct answer contestant gives -.0742769

(.1288896)

_cons 5.556697

R² 0.2622

Standard errors are clustered at the contestant level. Dependent variable is attractiveness. coefficients of OLS regression; * p<0.05, ** p<0.01, *** p<0.001

From table 3, we can now see that both age and performance of a contestant do not affect his/her rate of attractiveness. Good-looking contestants do not perform better, but their performance is remembered more accurate by experiment participants.

Table 4 again presents 4 OLS regression results. The results in table 4 does not show that there is a significant relationship between contestants’ attractiveness and participants’ memory. This result does not support our hypothesis 1. However, this does not mean that our hypothesis is wrong. Asking about the color of clothes and specific questions answered is more subjective than asking about the number a contestant answered correctly. One experiment participant might have a preference for one color or have a special interest in some kind of questions. This helps them to recall some questions more correctly than other questions, despite the attractiveness of a contestant.

(28)

Because of this, the insignificant result in table 4 could be explained by that our sitting of questions did not accurately capture the effect of attractiveness.

Table 4

Details about clothes color and “who answered which question” that participants recalled correctly. (1) (2) (3) (4) Attractiveness .0073731 (.0512031) -.0178451 (.0573322) .000128 (.0501758) -.0197499 (.0565775) Age Contestants -.0178451 (.0034329) -.0026249 (.0033804) Male Contestants -.0540655 (.082409) -.0547717 (.082371) Correct answer contestant gives -.0313637 (.0501758) -.0222118 (.0399685) _cons .6302736 .8820627 .7127468 .9133163 R² 0.0002 0.0085 0.0028 0.0097

“Answer contestant give” calculated how many questions in the game quiz show clip each game contestant actually answered correctly. Standard errors are clustered at the contestant level. Dependent variable is how many detail questions about clothes color and “who answered which question” each experiment participant recalled correctly. coefficients of OLS regression; * p<0.05, ** p<0.01, *** p<0.001

Result 2. When a participant answers a question about a contestant wrong, the

participant will underestimate performance if a contestant is pretty and overestimate if a contestant is unattractive. This is contrary to our hypothesis 2.

To see if the attractiveness of a people can affect how others predict his/her performance, we firstly need the data that a participant actually predicts a contestant’s performance.

(29)

Like what I have mentioned that in the experiment design, an experiment participant needs to watch 7 contestants’ performance and answer 2 questions about each contestant respectively. The second question is to recall how many quiz questions a contestant answered correctly in the video clip. When a participant answered this question wrong, this means that this participant does not know the actual performance of a contestant, and the answer this participant gave is a prediction.

In all questions that ask participants to recall performance of contestants, 475 of them are answered wrong and 631 are answered correct. It should be considered that the 631 correct answers could be given because participants recall details correctly or because participants guessed an answer right. To calculate the number of answered that participants guessed correctly, we could do a math. Each participant face four choices and only one is right, if he/she does not know the answer, 25% possible that this participant can still guess the right answer, Meanwhile, 75% possible that this participant will give a wrong answer. In the data, there are 475 wrong answers. This means there are actually 633(474/0.75) participants who do not know the actual performance of a contestant and gave a prediction.

I first delete all the data that experiment participants correctly answered type 2 question, because if they answered the type question correctly, they do know how much answers a contestant had given correctly. This means that they recalled the detail rather than predicted. 631 observations are dropped and 475 are left. Then, from the 631 dropped observations, I picked 158 observations randomly and mixed those data with the 475-left data. Those 158 observations represent those participants who cannot recall details

(30)

but guessed answers right. Now, I get a database which has 633 observations to test whether participants’ prediction about a contestant’s performance is related to attractiveness.

Table 5 shows that how many correct answer a contestant gave in the quiz show is not related to game contestant’s attractiveness. However, when a participant does not know the correct answer, he/her prediction is significantly related to contestants’ attractiveness. Contrary to my hypothesis, they tend to underestimate the performance of pretty contestants, and overestimate the performance of plain contestants.

Table 5 Correct answers contestants actually gave (1) Correct answers contestants actually gave (2) Correct answers contestants are predicted have gave

Attractiveness -.0857567 (.1523327) -.0516783 (.1600328) -.31379* (.1622584) Age Contestant .0157648* (.0081489) .0124869 (.0085808) -.0073296 (.0081161) Male Contestant -.0317944 (.2371781) .0014165 (.2543642) -.0145501 (.1897835) _cons 1.407068 1.367194 3.345815 R² 0.1188 0.0644 0.0512 observation 1106 633 633

Standard errors are clustered at the contestant level. coefficients of OLS regression; * p<0.05, ** p<0.01, *** p<0.001

(31)

In hypothesis 2, I assumed when a participant answers a question about a contestant wrongly, this participant will overestimate performance if a contestant is pretty and underestimate if a contestant is unattractive. To test this hypothesis, I first want to see if the actual performance of a contestant is related with the attractiveness of this contestant. In table 5, the first row shows that how many correct answers a contestant in the video clip is not correlated with attractiveness. This means that contestant who are more beautiful than others do not perform better than others. Face or attractiveness does not determine the performance of a contestant. Instead of appearance, age is the element that actually affects contestants’ performance. The first row in table 5 also shows that the correct answers contestants actually gave are significantly related to contestants’ ages. Elder contestants perform worse than other contestants. They give relatively less correct answers in quiz game video clip.

Row (2) and (3) used 633 observations. Row (3) in table 5 shows how attractiveness, contestants’ age and contestants’ gender could affect participants’ prediction. When participants are making predictions about the performance of a contestant, their prediction will be affected by the attractiveness of a contestant. They give higher performance prediction to less attractive contestants and lower prediction to prettier contestants. This result is surprising because it is contrary to what people normally would think. In order to make sure that attractive does not relate with a contestant’s actual performance, results in the row (2) confirms that pretty contestants do not give more or less correct answers compared with other contestants in the video clip.

(32)

Why the result in table 5 is contrary to what people would normally it should be? There are two things that worth mention. Firstly, people make predictions because they cannot recall the correct answers. This means that when preparing the data, more data about pretty contestants are deleted compared to data related to bad-looking contestants. In result one we confirmed that pretty contestants get more attention than other contestants, and participants are more likely to recall details about them correctly. This means that the data that participants give predictions about a contestant’s performance are more likely to be related to less attractive contestants. In fact, when we using 1106 observations the average attractiveness is 4.40, and when using 633 observation it is 4.36. Results in table 5 do not reflect the whole picture. Meanwhile, it is possible that participants are tending to reverse discriminate against pretty contestants. If they think that people who have a beautiful face always get more chances and opportunities than others, it is reasonable that they predict pretty candidates to have lower ability compared with other plat candidates who are doing a same task.

Result 3. Experiment participants do not tend to beauty discriminate against

unattractive game contestants after watching the five minutes video clip showed in experiment part 1.

Although we confirmed in result 1 that people do distribute their attention differently based on how attractive a contestant is. There is still one concern. Is attention a cause of beauty discrimination or a result of beauty discrimination? Maybe people recall details about pretty game contestants more correctly than other plat contestants because

(33)

the beauty discrimination happened before attention distribution happened. However, from the results in table 6, we can see that experiment participants’ predictions about which game contestant will win till the end and who will be voted out next are significantly related only to the actual performance of each contestant. Attractiveness does not relate to their prediction. This could mean that beauty discrimination does not happen after they watched the video clip, and their distribution of attention does not cause by pre-happened beauty discrimination.

Figure 3 and figure 4 shows the result of participants’ future performance prediction for each contestant. In figure 3, a contestant gets high ballot of prediction to win always comes with a high correct answer he/she gave. At the same time, in figure 4, we can see if a contestant gave less correct answers compared with others, he/she will get a lot vote expecting this contestant to be voted out in the next round.

Figure 3

“predicted win” means the answer experiment participants gave in the experiment, about which contestant they think will win till the last. Attractiveness is rated by 22 raters recruited separately from experiment participants.

-5 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40

Performance, attractivveness and participants' prediction (win)

(34)

Figure 4

“predicted voted out” means the answer experiment participants gave in the experiment, about which contestant they think will be voted out in the next round. Attractiveness is rated by 22 raters recruited separately from experiment participants.

From the result in table 6, we can see that one point improvement in attractiveness can help a contestant get on average 2.2 more participants (form total 30 participants) expecting he/she to win till the last. Although the effect is not very significant, the t-test result to this coefficient is still pretty big (1.20). Not like the prediction about the winner, attractiveness has less effect on predictions about who will be voted out. The coefficient is 0.59 and not significant at all. When predicting which game contestant will win at the end and which contestant will be voted out next, experiment participants seem to be quite rational. Only the number of correct answers that a game contestant gives in the video clip has a significant effect on their prediction. One more correct answer helps a contestant gain on average 2.9 support votes form 30 participants, and

-5 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 40

Performance, attractivveness and participants' prediction (out)

(35)

one wrong answer will encourage on average 3.9 participants decide to vote out a game contestant.

Table 6

Times a contestant is predicted to win at last

Times a contestant is predicted to be voted out

Attractiveness -2.200163 (1.826494) .5985707 (1.341117) Correct answers contestant gave 2.879346* (1.508816) -3.893018*** (1.295822) Age Contestant -.0234503 (.0950282) -.0416632 (.0578538) Male Contestant -.3352088 (1.934158) 2.62687 (1.704998) _cons 10.69863 8.361064 R² 0.2248 0.3773

“correct answers contestant gave” means how many correct answers a contestant gave

in the game show video clip. Standard errors are clustered at the contestant level. coefficients of OLS regression; * p<0.05, ** p<0.01, *** p<0.001

This result, combined with result one, shows that people pay more attention and recall more details about good-looking contestants than plain contestants, but they do this not because they beauty discriminate again plain contestants. They distributed the attention resource unevenly without conscious. This also means that the uneven distribution of attention happened before beauty discrimination can possibly happen. It is clear that difference in attention is not a result of beauty discrimination. On the contrary, it proves the main assumption of this paper, which is that different attention level is paid to

(36)

people with different attractive level. People who get more attention from others also make other people more familiarly with them. This, in turn, explain part of the beauty premium, because pretty candidates can always be known first and be chosen before others with the same productivity are chosen.

V. Conclusion

By doing an online experiment, this paper tests whether the distribution of attention is one of many reasons that beauty premium exists. I use a TV show called “the weakest link” as my experiment material and recruited 164 experiment participants. The experiment results show that people do distribute their attention differently. They recalled details about good-looking game contestants more correctly than details about plain contestants. This means that people should be more familiar with pretty candidates in labor market and whenever there is an opportunity, a pretty candidate with high ability always get the chance first. The experiment also shows that when people do not know a contestant, they tend to under estimate the ability of pretty candidate and overestimate the performance of candidates with plain faces. This paper also found that people do not pay their attention differently because they beauty discriminate against bad-looking contestants. The action of distributing attention happened before they begin beauty discriminate anybody.

(37)

Reference

Aharon, I., Etcoff, N., Ariely, D., Chabris, C. F., O’Connor, E., & Breiter, H. C. (2001). Beautiful Faces Have Variable Reward Value: fMRI and Behavioral

Evidence. Neuron, 32(3), 537–551. https://doi.org/10.1016/S0896-6273(01)00491-3 Bartoš, V., Bauer, M., Chytilová, J., & Matějka, F. (2016). Attention

Discrimination: Theory and Field Experiments with Monitoring Information Acquisition. American Economic Review, 106(6), 1437–1475.

https://doi.org/10.1257/aer.20140571

Becker, G. S. (1957). The economics of discrimination: an economic view of

racial discrimination. University of Chicago.

Biddle, J. E., & Hamermesh, D. S. (n.d.). Beauty, Productivity, and Discrimination: Lawyers’ Looks and Lucre, 30.

Cipriani, G. P., & Zago, A. (2011). Productivity or Discrimination? Beauty and the Exams*: Productivity or discrimination? Oxford Bulletin of Economics and

Statistics, 73(3), 428–447. https://doi.org/10.1111/j.1468-0084.2010.00619.x

Davenport, T. H., & Beck, J. C. (2001). The attention economy: Understanding

the new currency of business. Harvard Business Press.

Guryan, J., & Charles, K. K. (2013). Taste-based or Statistical Discrimination: The Economics of Discrimination Returns to its Roots. The Economic Journal,

(38)

Hamermesh, D. S. (2006). Changing looks and changing “discrimination”: The beauty of economists. Economics Letters, 93(3), 405–412.

https://doi.org/10.1016/j.econlet.2006.06.004

Hamermesh, D. S., & Biddle, J. E. (n.d.). BEAUTY AND THE LABOUR MARKET, 46.

Hamermesh, D. S., & Parker, A. (2005). Beauty in the classroom: instructors’ pulchritude and putative pedagogical productivity. Economics of Education Review,

24(4), 369–376. https://doi.org/10.1016/j.econedurev.2004.07.013

Hershler, O., & Hochstein, S. (2005). At first sight: A high-level pop out effect for faces. Vision Research, 45(13), 1707–1724.

https://doi.org/10.1016/j.visres.2004.12.021

Jerome Kagan, & Michael Lewis. (n.d.). STUDIES OF ATTENTION IN THE HUMAN INFANT. Merrill-Palmer Quarterly of Behavior and Development, 11(2). Retrieved from http://www.jstor.org/stable/23082701

Langlois, J. H., Roggman, L. A., Casey, R. J., Ritter, J. M., Rieser-Danner, L. A., & Jenkins, V. Y. (n.d.). Infant Preferences for Attractive Faces: Rudiments of a Stereotype?, 7.

Leigh, A., & Borland, J. (2007). Unpacking the beauty premium: Is it looks or ego. Unpublished Manuscript, Australian National University.[ADM].

Lindell, A. K., & Lindell, K. L. (2014). Beauty captures the attention of the beholder. Journal of Cognitive Psychology, 26(7), 768–780.

(39)

Marazzi, C. (2008). Capital and Language: From the New Economy to the War Economy, trans. Gregory Conti, Introd. Michael Hardt (Los Angeles: Semiotext [e],

2008).

Marwick, A. E. (2015). Instafame: Luxury Selfies in the Attention Economy.

Public Culture, 27(1 75), 137–160. https://doi.org/10.1215/08992363-2798379

Michèle Belot, V. Bhaskar, & Jeroen van de Ven. (2012). Beauty and the Sources of Discrimination. The Journal of Human Resources, 47(3), 851–872.

MOBIUS, M. M., & ROSENBLAT, T. S. (2006). Why Beauty Matters. THE

AMERICAN ECONOMIC REVIEW, 96(1), 22.

Olson, I. R., & Marshuetz, C. (2005). Facial Attractiveness Is Appraised in a Glance. Emotion, 5(4), 498–502. https://doi.org/10.1037/1528-3542.5.4.498

Pfann, G. A., Biddle, J. E., Hamermesh, D. S., & Bosman, C. M. (2000). Business success and businesses’ beauty capital. Economics Letters, 7.

Rubenstein, A. J., Kalakanis, L., & Langlois, J. H. (1999). Infant preferences for attractive faces: a cognitive explanation. Developmental Psychology, 35(3), 848.

Sachsida, A., Dornelles, A. C., & Mesquita, C. W. (2003). Beauty and the Labor Market - Study one Specific Occupation. SSRN Electronic Journal.

https://doi.org/10.2139/ssrn.505742

Stewart Schwab. (1986). Is Statistical Discrimination Efficient? The American

(40)

Sui, J., & Liu, C. H. (2009). Can beauty be ignored? Effects of facial

attractiveness on covert attention. Psychonomic Bulletin & Review, 16(2), 276–281. https://doi.org/10.3758/PBR.16.2.276

Theeuwes, J., & Van der Stigchel, S. (2006). Faces capture attention: Evidence from inhibition of return. Visual Cognition, 13(6), 657–665.

Referenties

GERELATEERDE DOCUMENTEN

To identify whether individual differences in exposure can explain inter-individual variability in response to telmisartan, linagliptin, and empagliflozin, we successfully

Bovendien is het van belang om te onderzoeken hoe persoonlijkheidsfactoren van de jongere een mogelijk risico vormen voor het slachtoffer worden van online grooming aangezien

Since the ultimate goals of performance appraisal is increased positive organizational outcomes and since organizations increasingly strive for a committed workforce,

Hoewel er in deze hoofdvraag nog altijd een nadrukkelijke rol is weggelegd voor de koloniale verbinding tussen Nederland en Suriname, mag er ook niet worden ontkend dat deze

Healthy relations with others: Participants expressed their opinion in words: ‘I have a healthy relationship with other people and that’s why, am I a better person

• “The execution of national language policies requires the participation of all levels of society. With regard to this, there must be formed a National Language Cultivation

De v raag die als eerste beantw oord moet w orden is: hebben uw verzekerden in beginsel aanspraak op kostenvergoeding van een niertransplantatie die in het buitenland

Aangezien LCDD geen enkelvoudig ziektebeeld is maar een gevolg van een reeks niet altijd goed gedefinieerde aandoening van plasmacellen, kan de vraag of de toepassing van