Cognitive Representations of Social Groups and Support for Public Policies

Download data and study materials from OSF


This experiment was fielded as part of a TESS telephone survey. Data and materials for all the studies included on this survey is available here.


Principal investigators:

Elizabeth Levy Paluck

Princeton University

Email: epaluck@princeton.edu

Homepage: http://www.betsylevypaluck.com/

Donald Green

Columbia University

Email: dpg2110@columbia.edu

Homepage: http://polisci.columbia.edu/people/profile/82


Sample size: 707

Field period: 10/2003-11/2003

Abstract

This experiment tests the hypothesis that a differentiated cognitive representation of a social group fosters greater support for programs and policies for marginalized people within that group. People who hold a differentiated cognitive representation of a given group recognize differences among individuals and subgroups within the group. In the context of this experiment, we focus on differences that are linked to political needs and identities, such as race, class, and sexual orientation. Recent research on diversity and multiculturalism (e.g. Markus, Steele, & Steele, 2000; Wolsko, Park, Judd, & Wittenbrink, 2000) suggests that emphasizing social differences rather than similarities can lead to less social stereotyping and to greater social tolerance. The present investigation extends this research by investigating whether recognizing differences among American women as a group can lead to increased support for social and political benefits for marginalized American women.

Hypotheses

1. Participants who hear the experimental message emphasizing differences among American women will be more likely to think of examples of American women who represent "difference" in significant social and political ways: e.g., women with a disability, with different sexual preferences, who belong to racial, ethnic, or religious minority groups (these responses will be coded as demonstrative of a differentiated cognitive representation of women.) On the other hand, participants who hear the experimental message describing "American women" as a general group will be more likely to think of American women who do not differ from the socially and politically dominant female prototypes in American society: women who are mothers, who are heterosexual, who are European-American, etc (these responses will be coded as demonstrative of an undifferentiated cognitive representation of women.)

2. Participants who have positive feelings toward American women as a group will be more likely to support social programs and political policies aimed at women more than participants who do not have positive feelings toward American women as a group.

3. Participants who have positive feelings toward American women as a group and who have a differentiated cognitive representation of women will rate and rank the programs and policies aimed at marginalized subgroups of women as important, to a greater degree than participants who do not have positive feelings toward American women, or who have an undifferentiated cognitive representation of women.

4. Participants who have positive feelings toward American women as a group and who have a differentiated cognitive representation of women will be more likely to donate money and/or larger sums of money to the programs and policies aimed at marginalized subgroups of women than participants who do not have positive feelings toward American women, or who have an undifferentiated cognitive representation of women.
5. Female participants will be more likely to support programs and policies aimed at women than male participants.
Experimental Manipulations

Participants' cognitive representations of women were manipulated using a message from a fictitious nonpartisan women's policy research group. In the experimental manipulation phase, participants were randomly assigned to hear one of two mission statements from the women's policy group. The message emphasized either 1) the importance of differences among women in policy research or 2) the importance of women in general in policy research. The first message was expected to evoke ideas about women as a diverse group, and images of individual women who illustrate the many differences contained in the group, including differences from the participants themselves. In short, the first message was intended to evoke a differentiated cognitive representation of American women. The second message was expected to evoke ideas of American women as a more homogenous group, and images of women who do not differ from the socially and politically dominant female prototypes in American society and/or who do not differ from the participants themselves. Thus, the second message was intended to evoke an undifferentiated cognitive representation of women. To determine whether the messages had this intended effect, participants were asked to name two women who came to mind when they heard the mission statement (the statement was repeated after this question.) These examples of women were submitted to a coding scheme developed according to our definition of differentiated and undifferentiated cognitive representations.

Outcomes

Independent variables:
Prior to the manipulation, participants will be asked to rate their feelings toward women, their political orientation, and their endorsement of multicultural ideology (so as to assess openness to difference and diversity in general.) Responses to the open-ended question about individual women who came to mind when they heard the mission statement were coded as to whether they represented a differentiated or undifferentiated cognitive representation of the two groups (coding schema available from authors.)

Dependent variables:
After the experimental manipulation, participants rated their support for and ranked the importance of eight social programs and political policies that the fictional women's policy group researches. We classify these programs and policies as "dominant women" or "marginalized women" programs and policies. The four dominant programs and policies are designed to meet the needs of American women who are politically and culturally dominant:
1. Programs to increase the number of women working in high-powered positions in business, such as the number of female presidents of top corporations.
2. Programs that connect female law students with the same kinds of high-powered internships and job opportunities as their male classmates.
3. Policies that encourage corporations to provide on site childcare for women employees.
4. Tutoring programs for female college students in advanced math and science programs.

The four programs and policies designed to target and uplift a marginalized subgroup of American women are:
1. A welfare policy that counts the care a mother gives her children as a job.
2. Programs to help immigrant women who work as housekeepers to organize for higher wages and better working conditions.
3. Laws that allow lesbian women to be married.
4. Policies that raise the level of health care for women in prison, including mental health care for women who have experienced substance abuse or physical abuse.
Participants rated and ranked the importance of these programs and policies, and then they were asked if they were willing to donate money, stuff envelopes, or sign a petition for their two highest-ranked programs or policies. Self-reported willingness to help was one measure of support; all participants were then informed they could donate any amount of money out of their $20 honorarium from the survey company toward research for the program or policy of their choice. Participants were then asked to specify how much money they would like to be taken out of this honorarium, and the program or policy of their choice.

Summary of Results

There were no differences in support for subgroup policies and for general policies between the two experimental groups. However, differentiated vs. undifferentiated cognitive representations are still being coded (using the open-ended responses to the experimental messages.) Results based on these codes are forthcoming. Results will become available following the TESS evaluation conference, October 15 2004.

Additional Information

A mini experiment was built into the general experimental design to test the effects of donation suggestions on amount of money donated. When participants were told that they could donate some of their $20 stipend for participation to their favorite program or policy, they were randomly assigned to one of three suggestions:
"Fifty cents, one dollar, or even two dollars"
"Fifty cents, one dollar, or even five dollars"
"Fifty cents, one dollar, or even ten dollars"