The Study

The Effects of Racial Diversity on Cognitive Complexity in College Students:

A True Classical Experiment

Anthony Lising Antonio and Kenji Hakuta, co-Principal Investigators

An experiment varying the racial (Black/White) and opinion composition in small-group discussions was conducted with college students (N=357) at three universities to test for effects on the perceived novelty of group members' contributions to discussion and on participants' integrative complexity. Results showed that the presence of racial and opinion minorities were both perceived as contributing to novelty. Generally positive effects on integrative complexity were found when the groups had racial and opinion minority members and when members reported having racially diverse friends and classmates. Findings are discussed in terms of social psychological theories of minority influence and social policy implications for affirmative action. The research supports claims about the educational significance of race in higher education, as well as the complexity of its interaction with contextual and individual factors.

Background

This study addresses the current legal debate on affirmative action in higher education by asking a specific question: does racial diversity in the educational environment produce positive intellectual gains for White majority students?  We raise this particular question because it is precisely on that question that the Supreme Court will deliberate this spring to determine the constitutionality of the use of race in college admissions.  In the famous Bakke decision, Justice Powell stated that race could be justified as a plus factor in admissions if it served a compelling educational goal of the institution.  The fifth circuit reversed Bakke in Hopwood v. University of Texas in 1996, declaring that "achieving a diverse student body is not a compelling interest under the Fourteenth Amendment."  Since Hopwood, the race is on to see whether there is any social science evidence to support or refute this assertion.

 The University of Michigan, in its defense, has drawn upon this evidence, including its own analysis of its graduates.  Other studies have tracked college students in representative samples of selective universities and looked at their patterns of interactions with racially diverse peers and their educational outcomes.  Still others have polled graduates of baccalaureate programs and law schools, asking whether racial diversity contributed to their education and training.  The studies as a whole support the assertion that White college students who are exposed to racially diverse educational environments receive benefits ranging from enhanced intellectual and social self-confidence to growth in academic skills to increased civic engagement.  The present study was conducted to test this assertion through the use of rigorous research methodology drawn from experimental social psychology.   In other words, we sought to put diversity to the test, to see if the effects of diversity reported by correlational studies could be detected using the most rigorous methods social science has to offer.  We had three primary goals for this study:

1. Conduct a study in the laboratory and experimentally control the presence of the variable in question -- racial diversity.

2. Infer a causal link between racial diversity and an educational outcome.

3. Measure the impact of diversity on cognitive ability using a behavioral outcome.

The first two goals dictated the design of a laboratory experiment.  Research using the experimental methodology has benefits as well as limitations.  The primary benefit of an experiment is the requirement of random assignment of individuals to experimental conditions.  Random assignment is the only methodology that enables a strong causal inference to be made.  On the other hand, a social science experiment is necessarily a simplification of real life – a petri dish rather than a real environment – and one has to be careful that the experimental conditions simulate real life as much as possible. 

 The third goal stems from our reading of the available social science evidence on the relationship between racial diversity and educational outcomes.  The pattern of the evidence clearly illustrates a positive relationship, and the bulk of it was derived from the analysis of survey data.  While compelling, the robustness of this pattern could be further informed by studies utilizing observable measures of additional outcomes in the cognitive domain. 

The Experiment

Our initial task in designing the experiment was to choose an appropriate outcome measure for the study.  After consulting with colleagues in psychology, we determined a best candidate, a measure of complex thinking processes known as Integrative Complexity (IC).  This measure determines the degree of complexity in the thinking processes of individuals as reflected in written materials such as essays and speeches.  Integrative complexity scores are assigned by expertly-trained coders who rate samples of writing with respect to two factors:  (1) differentiation of two or more perspectives, ideas, or dimensions; and (2) integration of those differentiated perspectives.  The score ranges from 1 (no differentiation present) to 7 (integration with explicit explanation of perspective interconnectedness and overarching principles) and is independent of specific attitudes, opinions, or decisions expressed in the writing.  Integrative complexity has been used in psychology for over 30 years and has also been linked with higher grades in college students.  Thus, IC is a well-established, performance-based measure of higher order cognitive functioning. 

 The key question addressed by the experiment was whether the racial composition of small groups discussions had a measurable impact on college students’ thinking, as measured by integrative complexity.  Because previous research had already shown opinion diversity to have a positive impact on integrative complexity, we designed our experiment to test whether there was an effect of race distinguishable from the effect of opinion diversity.  Thus, there were two main experimental variables that we manipulated: the opinion composition of groups, and the racial composition of groups. 

To operationalize this design, we used the classic experimental social psychology technique of using research collaborators who pretend to be a research subject, but are actually operating under the direction of the researchers.  We convened discussion groups in the laboratory composed of three white students and one collaborator.  Depending on the predetermined experimental condition, the collaborator is carefully instructed to either agree or disagree with the majority sentiment of the group in which s/he participates.  In addition, the collaborator is either White or Black.  To ensure minimal collaborator intervention other than their race, the collaborators were given scripted statements for discussion and kept blind to the purposes of the study during data collection.  In order to control the opinion composition of our groups, we screened potential participants with a survey of their opinions regarding social issues and recruited White students holding similar views regarding the small group discussion topics, either child labor or capital punishment.  All discussion groups were also all men or all women.  A total of 357 undergraduates participated in experiments conducted at Stanford, UCLA, and the University of Maryland.

 Measurements were taken as follows.  First, a facilitator led the three participants and the collaborator into the laboratory and asked them to be seated around a small table. They were then given an issue prompt describing the same target social issue on which they had indicated their opinion on the screening survey.  After reading the prompt, and seated with the group but before any discussion took place, participants were asked to indicate their agreement or disagreement with the issue as described in the prompt and to write a short essay describing their support for or opposition to the issue (the pre-discussion essay).  They were given 15 minutes.

After completion of the first essay, participants were asked to discuss their opinions on the issue.  The facilitator asked each member of the group to begin by orally stating his or her opinion.  This was followed by an unstructured 15-minute discussion during which the collaborator followed his/her script designed to agree or disagree with the majority view held by the participants.

Participants were then asked to write a second essay on the same topic (the post-discussion essay), for which they were given 15 minutes.  After completion of the second essay, participants were given a second prompt asking them to indicate their agreement or disagreement with a different social issue (either child labor practices or the death penalty, the alternate of the first issue) and to write a short essay describing their support or opposition to this second issue. We call this the transfer essay because it tests whether any stimulation of complex thinking due to the group discussion on the first topic transfers to thinking on a second topic. Participants were given 15 minutes to complete their essay.  Finally, participants were asked to complete a questionnaire in which they rated each member of their group, including the collaborator, on the extent to which he or she made others think about the issue in different ways, introduced a novel perspective to the discussion, and was influential in the group. We averaged these three ratings of the collaborator (Cronbach's alpha of .90) to form an index of Perceived Novelty (scale range from 1 to 7).  The two essays were scored for Integrative Complexity. 

Results

 Overall, we found generally positive effects due to the racial composition of the discussion group in each of our analyses.  Specifically:

• Prior to discussion, integrative complexity was higher among students in groups with the Black collaborator. 

• Participants who reported more racially diverse social contacts in their everyday lives exhibited higher complexity in their post-discussion essays compared to those reporting more racially homogenous contacts.

• Students who had a Black collaborator in the group had higher IC scores on the transfer essay than those with a White collaborator, particularly for those writing on the topic of child labor.

• Students who reported more racially diverse social contacts exhibited greater integrative complexity in their transfer essays than students reporting less diverse contacts.

• Participants in the discussion groups judged the collaborator’s contribution to the discussion as more novel when the collaborator was Black, independent of the opinion held by collaborator.

These results are highly consistent with earlier research.  The main contributions of this study do not reside necessarily in the effects measured as much as the methodology used to detect them.  Questions concerning the causal attributions to racial diversity reported in previous non-experimental research as well as the validity of self-reported outcome data are in part addressed in this study with the use of random assignment, experimental procedures, and a performance-generated outcome measure.  The findings reported here warrant further experimental and naturalistic exploration of these effects to inform policy and practice in higher education.

Before exploring the resources, please read the full research report (PDF), which has been published in the August 2004 issue of Psychological Science, and the FAQ, which answers many common questions regarding the design and conduct of the experiment.
 

Contact Information

Postal address

Anthony Antonio

Assistant Professor of Education

Stanford University

CERAS Bldg.

Stanford, CA. 94305

650-723-4053

 

Kenji Hakuta

Vida Jacks Professor of Education

Stanford University

CERAS Bldg.

Stanford, CA. 94305

650-725-7454 / fax 650-723-7578

 

Electronic mail

Co-Principal Investigators: Anthony Antonio & Kenji Hakuta
Webmaster: William Perez