Full Paper at
Impact of a Racially Diverse Environment on Cross Race Effect
Hari Srinivasan
Seth Hammond
Sih-Ting Liao
Summer Hadla
Psychology 101, Research Methods in Psychology
UC Berkeley
Abstract
The well-investigated phenomena of Cross Race Effect (CRE) is the claim that all people have an inherent in-race bias in terms of facial recognition. Previous literature has suggested that environment can influence CRE. The current study was conducted by surveying participants from UC Berkeley. The intent was to test whether immersion in the culturally diverse environment of UC Berkeley would reduce CRE in its students. Participants were divided by grade level and tested on whether their grade level affected how easily they detect facial differences in other races. However, after running a two-way ANOVA test, results showed that there were no significant differences between each race or year level. A few outside variables that may have affected our data must be considered, such as the difficulty in gathering participants, and the limited time given to conduct this study. Given more time and a larger sample population, future studies might bear more statistically significant results. Nevertheless, these results give hope that a more diverse environment may help our society become adept to cultures and races from around the world, thereby building tolerance and reducing discrimination.
Keywords: CRE, Cross Race Effect, in-race bias, diversity, environmental impact
Impact of a Racially Diverse Environment on Cross Race Effect
The University of California, Berkeley, has long been known as the birthplace of the civil rights movements. The campus boasts of a body of students that is culturally diverse and a student environment that encourages tolerance and acceptance of all groups irrespective of race, gender and ability. Such environments are all the more critical in the broader socio-economic context of the US which has recently seen increasing incidents of violence and prejudice in the name of race. The phenomena of the Cross-Race Effect (CRE) is the claim that all people have an inherent in-race bias. That is to say, people seem to recognize minutiae differences in the facial features of their own race, which allows them to easily distinguish one person from another. At the same time, they are unable to easily distinguish between members of another race. This study attempts to test whether exposure to the culturally diverse environment of UC Berkeley decreases the CRE in its students.Literature Review
The idea of Cross-Race-Effect or “they all look the same to me,” has been well-investigated by psychologists. Studies have suggested that the cross-race phenomena becomes ingrained as early as three months, with infants showing a marked preference for the same race even when living in a different race environment, such as African-American babies in a Caucasian environment (Bar-Haim, Ziv, Lamy & Hodes, 2006). That is to say, it is hypothesized that we are genetically predisposed towards expertise on our own-race.
There has also been much debate on the nature versus nurture aspect of the CRE. On the nurture side, studies have attempted to investigate if the cross-race effect can be reversed or reduced. A 2005 study discovered that Korean adoptee kids were better able to identify Caucasian features suggesting that nurture does make a difference and that CRE is reversible and changeable (Sangrigoli, Pallier, Argenti, Ventureyra, & de Schonen, 2005). The current investigation is along these lines, except that instead of very young kids, the target population was the college-age students at UC Berkeley who would be exposed to different races. Past studies have also attempted to see if CRE can be reduced by changing the way we process information. For instance, when participants were exposed to videos of joy (positive emotion), they were better able to recognize other-race faces and reduce their own-race bias (Johnson & Fredrikson, 2005). Their study suggests this is due to, “improvements in holistic processing and promotion of a common in-group identity due to positive emotions.” It has also been suggested that what is identified as an “in-group” in a social context heavily influences even in-race bias. For example, a 2007 study found that middle-class whites were better able to recognize affluent whites than recognize poor whites as the latter was regarded as an out-group (Shriver, Young, Hugenberg, Bernstei, & Lanter, 2007). In effect, they conclude that, “social-cognitive mechanisms (modulate) differential perceptual expertise with same-race and cross-race faces.”
The implications for CRE is widespread in the multiracial society of modern-day America. One such example is in the legal system where an eyewitness testimony is given the utmost degree of importance by a jury. Hourihan, Benjamin, and Liuc (2012) assert that the meta-mnemonic judgments of eyewitnesses may well be influenced by CRE. Their study found that both Asian and Caucasian participants made judgments of learning when viewing photographs. As a result, they urge caution when a witness claims to identify a suspect in a lineup, especially if the suspect is not of the same race as that of the witness.
The current research study attempts to test the CRE phenomena amongst the students of UC Berkeley. UC Berkeley presents a racially diverse student body in addition to promoting equal opportunity for its students. The research question of this study is whether exposure to this diverse environment of Berkeley significantly reduces the CRE. The expectation based on results from past studies is that there will be a significant difference in the CRE across the four years, the difference being most significant between freshman and seniors.
Hypothesis
The alternative hypothesis of this study is that seniors are less influenced by CRE than first-year students, as seniors are more exposed to the diverse environment at Berkeley; that CRE will reduce when the years of a student spent at UC Berkeley increase. It means that when a student spends more time in Berkeley, the better a student is at recognizing faces of people belonging to a race other than their own. On the contrary, the null hypothesis of the study is that there are no significant differences between year and racial groups in identifying facial variations in people belonging to another race.
Method
Participants
Participants were sampled across all four years of study at UC Berkeley and across three different races: Caucasian, African American, and Asian. In order to conduct a study that was representative of the population of students at UC Berkeley, participants were randomly selected from students passing by Sproul Plaza. The break up of participants is given below.
The sampling method was chosen because it avoids creating a possibly biased sample that might have been the case if participants were chosen from the same course, scholastic department, or social circle. Unfortunately, this sampling method limited the accessibility of the whole population, resulting in an irregularity between the number of participants within each group. For example, Asian freshman participants were not able to be included in the study as they were not as available for testing as participants in the other groups. Also, more Asian seniors were included in the study as opposed to relatively small numbers of participants in the African American groups. Given more time and resources, the participants within the groups of the study could have been more evenly distributed.
Materials
Thirty face cards (photos) were taken for each of the three races (Caucasian, African-American, Asian). Each set of race cards were numbered on the back from 1 through 30 to help with identification. Thumbnails of the face cards used can be found in Appendix A.
Experiment Design
The experiment conducted in this study was designed to test CRE across race and years in the racially diverse culture of UC Berkeley. The goal of the study was to use a participant’s ability to recognize the faces of people within a racially homogenous group in an effort to assign a scale variable to the influence of CRE on them. To accomplish this, a participant was shown a group of face cards and asked to remember and recognize the face cards amongst a group of similar looking people of the same race. This was the dependent scale variable at the center of the study.
In order to use this data to find differences among race and year at UC Berkeley, two nominal independent variables were introduced into the study. At the beginning of each experimental test, the answers to the participant’s identified race and year in school were recorded. In order to ensure that the study would be concise, there were three levels possible for race for the participant to assign themselves: Caucasian, African-American, and Asian. These three levels were chosen because they represent the largest sub-populations of students attending UC Berkeley. Participants could also align with four levels of study at UC Berkeley: Freshman, Sophomore, Junior and Senior. These were the two nominal independent variable that categorized the groups for the statistical tests of this study.
Procedure
Each participant was shown face cards from a race other than their own. Demographics of the race in which they identified as well as their year in school were collected at the beginning of the procedure. There were two parts to administering the test as outlined below in Part A and Part B.
Part A. Cards # 1-20 face cards were shuffled and laid out on the table in front of the participant. The participant was then asked to look at them for one minute.
Part B. Unbeknownst to the participant, cards #1-10 were removed and substituted with cards #21-30. The second set of cards was shuffled and laid out in front of the participant. (Note: this set includes cards #11-20 from the previous set). The participant was then asked to point to ones he/she recognized from the previous set.
Data Collection.
The number of cards the participant correctly recognized from Part A (cards #11-20) were tabulated as “hits.” The number of additional cards the participant chose from cards #21-30 that were not in Part A of the test were tabulated as “misses.”
Statistical Tests Performed.
In order to test if there was any statistically significant difference in the data, a two way ANOVA test was performed. This test was chosen because it would compare the data within the cells, between columns (racial differences), as well as between rows (year at UC Berkeley). Given the aim of the research, a two way ANOVA was the most efficient test to run because it takes into account all aspects of the study within one test. A two way ANOVA not only allows us to compare the levels of two variables but also allows us to see their joint effects.
The effect size was also calculated as the effect size statistic neutralizes the influence of sample size. Hypothesis tests are affected by sample size. The measure of effect size on the other hand do not use sample size in its calculation and uses standard deviation instead of standard error in its calculation. It is therefore useful to have the information on effect size in addition to the results of a hypothesis test. Effect size essentially tells us how practical the finding of a hypothesis tests is and whether it is worthwhile continuing or replicating that research.
Results
The raw data collected from the 49 respondents is tabulated in Appendix B. The respondents were from all four years of undergraduate study at UC Berkeley and from three races: Caucasian, African American, and Asian. Unfortunately, freshman Asian students proved difficult to recruit and were not able to be included in the calculations.
Due to the nature of the experiment, it quickly became evident that participants were not only selecting correct response cards (hits) but were also selecting cards which were not included in Part A of the experiment (misses). In order to distinguish between a participant that correctly selected all 10 of the original cards, and only those 10 cards, and a participant that selected all 10 correct face cards as well as 4 other incorrect cards, the d’ sensitivity index was calculated for each of the participants based on their number of “hits” and “misses,” a hit being a correctly chosen face card and a miss being an incorrectly chosen card. The d’ calculation is frequently used in Signal Detection Theory to account for noise; noise is essentially anything that interferes with detection. In the context of our study, it presented a more accurate record of how well the participant performed in recognizing the differences between the face cards. This, in turn, displayed a better picture of the CRE in action. To calculate the d’ result of each participant, an online calculator was used (Neath, n.d.), taking into account the hit and miss rates of each participant. The d’ variable from each participant was then used in the two way ANOVA test. The d’ results are tabulated in Appendix C.
After the data from our 49 participants was collected and analyzed as outlined in Appendix D. The resulting d’ index score grand mean (GM=3.08) was relatively high. Surprisingly, on average, the highest scoring group was the Caucasian freshman (3.966). The highest mean score among the different races was the Asian group (3.551). The highest scoring year at UC Berkeley was the freshmen (3.603). The single group with the lowest mean d’ score was the Caucasian seniors (1.188). Overall, based purely on observations of the resulting means of d’ scores, the chances of the alternate hypothesis being true, that each year in a diverse environment at UC Berkeley would lower the CRE, did not look strong. To be sure, the sums of squares was calculated to continue the pursuit of finding a significant difference in the data.
The sums of squares was calculated to find the dispersion of the data points within the groups, between the rows, and between the columns. In conjunction with the calculated degrees of freedom, we were able to find the F statistic for each group. Based on the data collected, the calculated column F statistic (F=.905) was not higher than the critical F value (F(2,37)=3.252), based on the set alpha (alpha=0.05). Also, the observed interaction F value (F=.516) was not higher than the F critical (F(6,37)=2.356). Additionally, the obtained F statistic for the row (F=.22) was lower than the critical F value (F(3,37)=2.859), meaning there was no statistically significant difference between years at UC Berkeley. The calculations of the data, as shown in Appendix D, support the decision to retain the null hypothesis that there is no difference in CRE between race or year in school at UC Berkeley.
Discussion
As mentioned in the Results section, the retained data was not statistically significant, preventing any known difference to be found in the Cross Race Effect between race or year of participants. It was concluded that the environment of UC Berkeley does not have an impact on whether individuals can distinguish between people of other races as well as they can distinguish between their own. Therefore, seniors are no more likely to identify other races’ faces better than their own. Meanwhile, a participant who considers themselves to be Asian is no better at identifying other races’ than those who considers themselves to be Caucasian or African American. There were a relatively equal number of strengths as well as shortcomings in this research project. One of the strengths was the method in which the research was conducted. The pictures that were chosen to represent each race were varied and unknown to the participants, and were organized in a neat and efficient way as to not confuse the participant or the experimenter. Another strength was the way in which the participants were divided, by year and race; experimenters made sure to place participants in correct categories and attempted to get an equal number of participants in each cell. However, there were also a few shortcomings. The time in which the experimenters had to find the participants and conduct the research was a bit tight. Additionally, the sample may have not been completely representative of the population, which could partially accounted for our small to medium effect sizes. A future study which allows more time to find a more representative sample must be considered. Nevertheless, the findings were not disappointing, since many of the participants did very well in distinguishing between people of other races. This gives hope to believe that environments such as UC Berkeley house individuals who have almost mastered the cross race effect, thus limiting their in-race bias, and therefore encouraging a more diverse and accepting environment. An environment like UC Berkeley, though, should not be special; as a society, we should strive to teach everyone to be open and accepting towards all races. This then brings up the question of whether or not an environment such as UC Berkeley is representative of the nation, and if people in today’s society have overcome the cross race effect. In order to test this theory, experimenters must change their sample from students at UC Berkeley to citizens of the nation. While this may seem daunting, an experiment such as this one may change the way people view other races forever.
References
Bar-Haim, Y., Ziv, T., Lamy, D., & Hodes, R. M., (2006). Nature and nurture in own-race face processing. Psychological Science. 17(2). 159-163 https://doi.org/10.1111/j.1467-9280.2006.01679.x
Hourihan, K.L., Benjamin, A.S., Liuc, X., (2012). “A cross-race effect in metamemory: Predictions of face recognition are more accurate for members of our own race.” J Appl Res Mem Cogn. 1(3). 158–162. https://doi.org/10.1016/j.jarmac.2012.06.004
Johnson, K. J., Fredrikson, B. L.., (2005). “We all look the same to me: Positive emotions eliminate the own-race bias in face recognition. Psychological Science. 16(11). 875-881. https://doi.org/10.1111/j.1467-9280.2005.01631.x
Neath, I. (n.d.). $d'$ Calculator. Retrieved December 15, 2017, from https://memory.psych.mun.ca/models/dprime/
Sangrigoli, S., Pallier C., Argenti A.-M., Ventureyra V.A.G., de Schonen S., (2005). Reversibility of the other-race effect in face recognition during childhood. Psychological Science, 16, 440–444
Shriver, E. R., Young, S. G., Hugenberg, K., Bernstein, M.J., Lanter, J.R., (2007). “Class, race and the face: Social context modulates the cross-race effect in face recognition” Personality and Social Psychology Bulletin. 34(2). 260-274. https://doi.org/10.1177/0146167207310455
Appendix B: Raw Data
Raw Data*
| |||
Caucasian
|
African American
|
Asian
| |
Freshmen
|
8/0
|
5/0
| |
5/0
|
6/2
| ||
6/0
| |||
5/1
| |||
6/0
| |||
Sophomores
|
6/1
|
8/0
|
8/3
|
4/2
|
5/3
|
4/0
| |
9/0
|
7/2
|
8/1
| |
5/0
|
7/0
| ||
Juniors
|
10/3
|
6/1
|
8/2
|
5/1
|
5/0
|
8/0
| |
8/2
|
8/0
|
7/1
| |
7/2
|
6/2
|
9/1
| |
6/0
|
5/2
| ||
8/3
|
10/1
| ||
9/1
| |||
10/0
| |||
Seniors
|
6/2
|
10/2
|
5/1
|
5/1
|
8/1
|
1/0
| |
5/2
|
4/0
| ||
6/1
|
8/1
| ||
10/2
| |||
10/2
| |||
10/1
| |||
*Results were recorded as follows: “(hits)/(misses)”
Appendix C: d’ Calculation
d’ Scores
| |||
Caucasian
|
African American
|
Asian
| |
Freshmen
|
5.142
|
4.3
| |
4.3
|
1.095
| ||
4.553
| |||
1.281
| |||
4.553
| |||
Sophomores
|
1.535
|
5.142
|
1.366
|
0.588
|
0.524
|
4.047
| |
4.3
|
1.366
|
2.123
| |
5.582
|
4.824
| ||
Juniors
|
4.824
|
1.535
|
1.683
|
1.281
|
4.3
|
5.142
| |
1.683
|
5.142
|
1.806
| |
1.366
|
1.095
|
2.563
| |
4.553
|
0.841
| ||
1.366
|
5.582
| ||
2.563
| |||
8.6
| |||
Seniors
|
1.095
|
5.142
|
1.281
|
1.281
|
2.123
|
3.018
| |
0.841
|
4.047
| ||
1.535
|
2.123
| ||
5.142
| |||
5.142
| |||
5.582
| |||
Source Table
SS
|
df
|
MS (s2)
|
F
| |
Row
Column
R X C
Within
|
2.718
7.457
12.372
152.522
|
3
2
6
37
|
.906
3.729
2.129
4.122
|
.220
.905
.516
|
Totals
|
175.469
|
48
|
3.656
|
No comments:
Post a Comment