Estimating Risk: Stereotype Amplification and the Perceived Risk of Criminal Victimization Lincoln Quillian Northwestern University USA Devah Pager Princeton University USA
Group Perception as a Source of Discrimination • Discrimination remains important in labor and housing markets A common model of discrimination Perception or stereotype Discrimination • Models of discrimination conflict about the accuracy of social perception of other groups
Accuracy in Stereotype Perception • The theory of statistical discrimination and some psychologists (e.g. Arkes and Tetlock 2004) posit on-average accurate views of other groups • Emphasizes the cognitive utility of group characteristics in face of limited information
Accuracy in stereotype perception Scwab, “Is Statistical Discrimination Efficient”?, American Economic Review, 1996 “I sometimes refer to statistical discrimination as a situation where an employer acts on a ‘true stereotype.’ Examples of stereotypes include ‘blacks are less skilled than whites.’ ‘women quit more frequently than men,’ and ‘women live longer than men.’ The term stereotype reflects society’s moral distaste for statistical discrimination under the battle cry ‘judge me, not my group.’ Yet the word ‘true’ is a crucial modifier in the phrase true stereotype. The employer, I assume, responds only to correct group information (statements that are indeed true on average) . . .” (p. 228)
Bias in Social Perceptions • A common view in psychology and sociology; multiple theories of why this might be. • A classic view in psychology and sociology. LaPiere’s (1936) analysis of “type rationalizations.” Allport’s (1954) definition of prejudice as “antipathy based on a false generalization.” • Relatively few direct tests of accuracy.
Assessing Accuracy of Group Perceptions • Direct assessments face serious social desirability pressures. • Instead, examine role of race in crime-risk estimation. • Mental association of race and certain forms of crime are well established.
Problems in Comparing Perceptions and Reality of Neighborhood Crime • Difficult to compare realities of crime with perceptions using “vague quantifier” answer categories. • A handful of studies compare perceptions and reality in some form but problems with measures of crime.
A Different Approach • Ask respondents for quantitative assessments of risk (Juster 1966; Dominitz and Manski 1997). • Ask for actual realization of corresponding events on same, ongoing survey. • Allows a precise comparison of subjective risk and realizations. • Examine how accuracy varies with neighborhood composition.
Data: Survey of Economic Expectations • Designed by Charles Manski and Jeff Dominitz • Part of an ongoing 1994-2002 National Telephone Survey conducted by the U. of Wisconsin Survey Center • 50% response rate • For burglary models: 6745 whites, 541 black, 781 Hispanic/other • For robbery model: 5034 white, 413 black, 558 Hispanic/other • Matched to census tract zip code data (1990/2000 census, using nearest year)
Method Subjective Probability Estimates about Neighborhood “Now I will ask you some questions about future, uncertain outcomes. In each case, try to think about the whole range of possible outcomes and think about how likely they are to occur during the next 12 months. In some of the questions, I will ask you about the PERCENT CHANCE of something happening. The percent chance must be a number from zero to one hundred. Numbers like 2 or 5 percent may be “almost no chance,” 20 percent or so may mean “not much chance,” a 45 or 55 percent chance may be a “pretty even chance”, and a 95 or 98 percent chance may be “almost certain.” The percent chance can also be thought of as a number of chances out of 100.”
Survey Questions BUGLARY: What do you think is the PERCENT CHANCE (or CHANCES OUT OF 100) that someone will break into (or somehow illegally enter) your home and steal something, during the next 12 months ? ROBBERY: What do you think is the PERCENT CHANCE (what are the CHANCES OUT OF 100) that someone will take something directly from you by using force--such as a stickup, mugging, or threat--during the next 12 months ? HEALTH INSURANCE LOSS: Now please think about your health insurance coverage 12 months from now. What do you think is the PERCENT CHANCE (or CHANCES OUT OF 100) that you will have health insurance coverage 12 months from now?
Realizations Parallel assessments of victimization provide measures of actual risk: During the past 12 months, did anyone break into or somehow illegally get into your home and steal something? During the past 12 months, did anyone take something directly from you by using force--such as a stickup, mugging, or threat? Do you have any health insurance coverage?
Are Responses Meaningful? • Low rate of item non-response to estimated risk questions—less than 2%. • Some grouping at 5% probability intervals, but also intermediate responses, and many zero and low percentage responses. • Strong association between patterns of subjective risk and realized patterns over demographic groups.
Race and Victimization Risk Burglary Robbery No health insurance
Age and Victimization Risk Burglary Robbery Health Insurance
Contrasts of Estimated and Realized Risk Levels for Four Risk Events: Burglary, Robbery, Health Insurance Loss, Job Loss
Conclusion: Estimated and realized risk levels • Risks of crime are highly overestimated relative to realized rates and perceived/actual rates of health insurance loss and job loss (Dominitz and Manski 1997). • Many possible sources, but media reports likely to be important.
Predictors of Perceived vs. Actual Crime Zip population density; urban vs. rural compare predictors of perceived and realized
Models Both perceived and actual risk models use a logit specification: Logit of estimated probability (e): Ln(e/(1-e)) = a + b1x1+b2x2 + … +bkxk with 0 and 1 recoded to .001 and .999. Parallel logistic regression for realized victimization experiences.
Burglary Models Also included: controls on previous slide. * = p<.05; ** = p < .01; *** = p < .001
Figure 2: Estimated and Realized Burglary Risk and Zip Percentage Black
Robbery Models Also included: controls on previous slide. + = p < .1; * = p<.05; ** = p < .01
Figure 3: Estimated and Realized Robbery Risk and Zip Percentage Black Figure 3: Estimated and Realized Robbery Risk and Zip Percentage Black
Additional Conclusions • Similar results hold for percent Hispanic and burglary, but not robbery. • Higher level racial composition or economic status (MSA or county level) does not significantly predict estimated risk or realizations.
Basic Conclusions • Whites overestimate victimization risk by more than twice in much in predominately black zip codes as white ones. • Race of neighborhood drives perceptions, while socioeconomic status drives reality. • True even though ask about risk with no neighborhood reference. • Data from residents, not visitors. • Migration of respondents whose perceptions of burglary/robbery are most strongly affected by race away from non-white neighborhoods should suppress the association.
Understanding Group Stereotypes • White respondents’ perceptions are misattributed by race rather than class even though neighborhood affluence is observable. • Unclear if this mis-attribution is specific to crime: crime an especially race-linked issue and low personal experience. • Mass media may play a role.
Implications • Stigmatizing effect on black neighborhoods; affects mobility decisions of businesses and households. • Overestimation of victimization risk and link to race may contribute to policies that explain exponential increases in incarceration.