10 Questions Need Answered (Please See Ten Questions Attached along with Supporting Documents)

Citation: 11 Just. Q. 119 1994

Content downloaded/printed from HeinOnline (http://heinonline.org) Wed Jul 29 10:40:15 2015

-- Your use of this HeinOnline PDF indicates your acceptance of HeinOnline's Terms and Conditions of the license agreement available at http://heinonline.org/HOL/License

-- The search text of this PDF is generated from uncorrected OCR text.

-- To obtain permission to use this article beyond the scope of your HeinOnline license, please use:

https://www.copyright.com/ccc/basicSearch.do? &operation=go&searchType=0 &lastSearch=simple&all=on&titleOrStdNo=0741-8825

GLOBAL AND SPECIFIC ATTITUDES TOWARD THE POLICE: DISENTANGLING THE

RELATIONSHIP*

STEVEN G. BRANDL University of Wisconsin-Milwaukee

JAMES FRANK University of Cincinnati

ROBERT E. WORDEN The University at Albany, State University of New York

TIMOTHY S. BYNUM Michigan State University

Since the 1960s, a substantial body of research has focused on citizens' attitudes toward the police. These studies tap a rather wide variety of out- looks: some ask about specific assessments of the police (e.g., satisfaction with the police in particular incidents), while others ask about more global assessments (e.g., satisfaction with the police in general, police in the com- munity, or police in the neighborhood). Using data obtained through a panel survey of 398 residents of a large midwestern city, we compare spe- cific assessments of police performance with more global attitudes toward the police. We also examine the effects of global and specific attitudes on one another. The results show that the two measures produce similar levels of support for the police. The results reveal further that global attitudes have substantial effects on specific assessments of police performance, and that the effects of specific assessments of police performance on global atti- tudes are modest by comparison.

Since the turbulent and riotous 1960s, many efforts have been made to measure and analyze citizens' attitudes toward the police. The role of the police in precipitating civil disorders and, more gen- erally, the tensions between the police and minorities initially di- rected attention toward the racial and ethnic correlates of citizens' attitudes. Over the past two decades, as evaluators and police prac- titioners have come to see citizens' support both as an important

* The authors are grateful for the comments and suggestions of the JQ refer- ees. The data on which this article is based were collected under Grant 89-DD-CX- 0049 from the National Institute of Justice, U.S. Department of Justice. The analy- ses and interpretations presented here are not intended to reflect the views of the U.S. Department of Justice.

JUSTICE QUARTERLY, Vol. 11 No. 1, March 1994 @ 1994 Academy of Criminal Justice Sciences

120 GLOBAL AND SPECIFIC ATITUDES

outcome in its own right and as an essential element in the "coproduction" of public safety, surveys of citizens have been incor- porated increasingly into evaluations of police strategies (e.g., com- munity policing). In the absence of market mechanisms whereby the public value of police services is established through consumers' willingness to pay, surveys are one medium through which the po- lice can obtain feedback-both positive and negative-on police practices.

The studies that have examined citizens' attitudes toward the police tap a rather wide variety of outlooks. Some surveys inquire about respondents' satisfaction with the police in their communities or neighborhoods; others elicit respondents' degree of satisfaction with their own contacts with the police. Unfortunately, we think, too little effort has been made to better understand these different outlooks as elements of citizens' systems of beliefs about the police. Existing research does not permit us to draw definitive inferences about how these attitudes differ in their valences (positive or nega- tive) or in their degree of intensity. Moreover, because the cross- sectional designs of previous research preclude the necessary analy- ses, previous research has not yet established the nature of the re- lationship between these attitudes: whether they are causal or spurious and, if they are causal, the direction of causality. Conse- quently analysts often claim that citizens' assessments of their con- tacts with the police influence their more general attitudes toward the police, but they rarely acknowledge that more general attitudes might influence assessments of individual contacts.

In this paper we report the results of analyses that extend our understanding of the structure of citizens' attitudes toward the po- lice. We compare specific attitudes about police performance in in- dividual contacts with more global or general attitudes toward the police in the neighborhood. We also examine the effects of global and specific attitudes on one another, using panel data that enable us to disentangle these effects. This research offers empirical evi- dence where only assumptions existed before. Insofar as the find- ings prove to be generalizable, the study may suggest that we reject some of those assumptions.

GLOBAL ATTITUDES AND SPECIFIC ATTITUDES

Easton (1965) provides at least a general foundation for analy- ses of global and specific attitudes toward the police. He identified two forms of public support associated with political systems: dif- fuse and specific. Easton defined diffuse support as "that which continues independently of the specific rewards which the member may feel he obtains from belonging to the system" (p. 125). Specific

BRANDL, FRANK, WORDEN AND BYNUM 121

support refers to "input to a system that occurs as a return for the

specific benefits and advantages that members of a system experi- ence as part of their membership. It represents or reflects the satis- faction a member feels when he perceives his demands as having been met" (p. 125).

Dennis (1976) further defined and elaborated on Easton's framework. As he explained, specific support refers to an evalua- tion of a particular role incumbent within the institution, whereas "diffuse support involves a generalization of support for the institu- tion.., it is a reservoir of good will normally directed toward the institution rather than particular incumbents. It is a regime-level sentiment and support or rejection of institutional authority" (p. 50). Dennis hypothesized, first, that the levels of diffuse and spe- cific support normally differ substantially; the former is higher than the latter. Further, he suggested that specific support can af- fect the level of diffuse support for the institution.

At best, only weak evidence for the veracity of Dennis's first hypothesis can be gathered from the "attitude toward police" litera- ture. White and Menke (1982) compared measures that purport- edly capture general and specific support for police and found, in support of this hypothesis, that the general items produced more favorable attitudes than did the specific.1 Beyond White and Menke's study, we can only compare the results of studies that measured either specific or more global attitudes in order to draw inferences about the disparity between the two constructs. Several studies that measured global attitudes toward the police found very favorable attitudes (e.g., Benson 1981; Homant, Kennedy, and Fleming 1984; Koenig 1980); other studies which measured specific attitudes-that is, assessments of individual contacts with the po- lice-often found attitudes to be less favorable (e.g., Percy 1980; Poister and McDavid 1978). At the same time, however, some stud- ies report quite positive specific attitudes (e.g., Shapland 1983), while others reveal less favorable global attitudes (e.g., Carter 1985; Thomas and Hyman 1977; Zamble and Annesley 1987).

1 One could question, however, the extent to which the analyses presented by White and Menke actually parallel Dennis's (1976) hypothesis. Specifically, White and Menke (1982) measure general support in terms of responses to the statement "Most police are competent in their work," while they measure specific support in terms of responses to "A lot of criminals get off free because the police are not doing their jobs effectively." Although the second item may be more specific than the first in that it asks about a particular type of police "competence," the object of the judg- ment ("the police") is the same in both cases. A seemingly more valid measure of specific support would focus on a particular role incumbent, and thus on the police in a particular contact situation.

122 GLOBAL AND SPECIFIC AlTITUDES

These inconsistencies should not be surprising given other varia- tions across the studies (e.g., wording and format of questions, na- ture of the sample and sample characteristics, study site). Consequently we are left with little more than speculative conclu- sions about the disparity between specific and more global attitudes toward the police.

Equally weak are the inferences that we can draw about the causal relationship between specific and global attitudes. Several studies have concluded, on the basis of empirical analyses, that citi- zens' evaluations of police actions in particular incidents strongly influence citizens' global assessments of police performance (Dean 1980; Scaglion and Condon 1980).2 These (cross-sectional) studies, however, do not (indeed, cannot) control for global attitudes toward the police that existed before the measured contact. Instead they assume that specific attitudes affect global attitudes, and they do not test the opposite causal relationship: that preexisting global at- titudes (i.e., attitudes that exist before the measured contact) affect specific assessments. Koenig (1980) captures this dilemma when he states:

It is debatable whether perceived police rudeness and un- fairness [in a specific incident] lead to less favorable public evaluations of local police, or whether less favorable evalu- ations of local police lead to the perception of neutral be- haviors as incidents of rudeness or unfairness (p. 248).

The only study to date that controls for prior global attitudes (Sko- gan 1991) is able to estimate only the effects of police- and citizen- initiated contacts, but not the effects of citizens' assessments of those contacts, on global attitudes toward the police. The effects of such contacts presumably are mediated by citizens' reactions, which-as we show below-are not unanimous: not everyone who is stopped by the police is dissatisfied with the contact, and (of course) not everyone who calls the police is satisfied. Moreover, the effects of global attitudes on (the unmeasured) specific attitudes are not estimated.

A theoretical basis exists for the proposition that global atti- tudes toward the police influence specific assessments. If the cogni- tive processes associated with stereotyping characterize the formation of citizens' attitudes toward the police, we would expect that information received during a contact with a police officer would be processed in accordance with the citizen's existing beliefs or attitudes (Hamilton 1981). Existing stereotypes may even bias the processing of information so that that information which is processed appears to reinforce the existing belief system (Hamilton

2 See Hero and Durand (1985) for an example outside the police context.

BRANDL, FRAN, WORDEN AND BYNUM 123

and Rose 1980). In addition, evidence that confirms a stereotype may be more cognitively available, and thus may be recalled more easily (Synder, Tanke, and Berscheid 1977). The data collected here allow us to examine the extent to which global attitudes affect specific assessments of police performance, as well as the reverse.

METHOD

Sample

Data for this study were collected through a three-wave panel survey that was conducted as part of a larger study of narcotics en- forcement. To minimize the effects of panel mortality, we use only the second and third waves for the analyses described here. Re- spondents were drawn from four areas (each consisting of several census tracts) in a large midwestern city. A cluster sampling proce- dure was used to ensure that all or most blocks within the areas were represented; households were selected randomly from each block in each area and were contacted by telephone. For the Wave 1 interviews, the interviewer asked to speak with someone age 18 or older who had lived in the home for at least six months.3 Efforts were made to contact the same individual for Waves 2 and 3. If the respondent from the previous wave could not be recontacted, a re- placement household was selected. The Wave 2 interviews, which for our purposes are called Time 1(T 1), were conducted in October 1990 and had a response rate of 75 percent (N=571). Wave 3, which we call Time 2(T 2), was conducted in April 1991 and had a response rate of 74 percent (N=560). The panel contains 398 respondents.

Table 1 illustrates the demographic composition of the Wave 3 sample, the panel, and on the basis of 1990 census data, the four areas from which the sample was drawn; these are similar to the city in race, gender, age, and other demographic characteristics. The panel is similar to the Wave 3 sample; this fact suggests that the findings are not distorted by panel mortality. The panel differs from the larger population, however, in that the panel contains larger proportions of whites, females, and individuals who 6wn their own homes. These areas were not selected because they are representative of any larger population; thus even a perfectly repre- sentative sample of their residents would form the basis for only limited generalization. The more telling points to be made with ref- erence to Table 1 are that these areas are predominant black and of low to moderate income, as are the panel respondents. Therefore

3 If a person possessing these characteristics was not at home, a return call was scheduled; this process was repeated until a refusal was received or until sev- eral contacts had been attempted without success.

124 GLOBAL AND SPECIFIC ATTITUDES

the generalizability of the findings we report below may be limited to such neighborhoods. We will return to this issue in our discus- sion.

Table 1. Demographic Characteristics of Wave 3, Panel, and Sample Areasa

Characteristic Wave 3 Panel Sample Areas

Totals (N) 560 398 65,084

Race White 187 (33.8) 146 (36.9) 13,530 (20.8) Black 338 (61.0) 228 (57.6) 48,316 (74.2) Native American 8 (1.4) 5 (1.3) 247 ( .3) Asian 3 (.5) 3 (.8) 391 (.6) Hispanic 4 (.7) 2 (.5) 1,663 (3.0) Other 14 (2.5) 12 C3.1) 939 (1.4)

Gender Male 164 (29.3) 115 (28.9) 30,348 (46.6) Female 396 (70.7) 283 (71.1) 34,737 (53.4)

Age 18-24 65 (11.9) 40 (10.2) 8,026 ( 1 9 . 1 )b 25-44 218 (39.8) 147 (37.6) 19,726 (47.0) 45-64 149 (27.2) 112 (28.6) 8,871 (21.2) 65+ 116 (21.2) 92 (23.5) 5,309 (12.7)

Household Type Own 394 (70.6) 291 (73.5) - (48.1) Rent 164 (29.3) 105 (26.5) - (51.9)

Median Income $15,001 - $25,000- $15,318

Education Elementary 49 ( 8.8) 35 ( 8.8) 4,276 (1 2 .9 )d Some high school 109 (19.5) 82 (20.6) 10,382 (31.4) High school graduate 189 (33.8) 133 (33.4) 9,026 (27.3) Some college 153 (27.4) 105 (26.4) 6,109 (18.5) College graduate 59 (10.6) 43 (10.8) 3,299 (10.0)

a Percentages are in parentheses. b N = 41,932. 1 Income was collected as a categorical variable. The $15,001-$25,000 category was the median category for Wave 3 and for the panel. d N = 33,092 because the census includes only individuals who are age 25 and over for the education measure.

Data

The same questions were asked at both T1 and T 2. For analyses

concerning the disparity between specific and more global attitudes

(Table 2), we use only the T 1 data. For the causal analyses (Tables 3 and 4), we use the panel data to provide the proper temporal order of variables.

We measured citizens' global attitudes toward the police ac- cording to their responses to this question: "In general, how satis- fied are you with the police? Are you very satisfied, somewhat

BRANDL, FRANK, WORDEN AND BYNUM 125

satisfied, somewhat dissatisfied, or very dissatisfied?" This ques- tion was prefaced with the statement "Now let's talk about the po- lice in your neighborhood."

We measured four specific attitudes, each involving a different type of contact with the police: when requesting information ("How satisfied were you with how the police handled the problem?"), when requesting assistance ("How satisfied were you with the way the police handled the problem?"), when stopped and questioned ("How satisfied were you with the way you were treated?"), and when victimized ("How satisfied were you with the police in their handling of the incident?"). Respondents who had each type of con- tact within the past year were asked about their most recent con- tact of that type.4 For all of the specific attitude items, the response options were the same as in the global measure: "Were you very satisfied, somewhat satisfied, somewhat dissatisfied, or very dissatisfied?"

Items that potentially could confound the relationship between specific attitudes and more global attitudes toward the police are included as control variables in the multivariate analyses. These variables are the citizen's age, education, race, income, and gender.5

RESULTS

To assess the disparity between specific and global attitudes toward the police, and to explore the association between the two, we compare the two measures across types of contact: first, by com- puting differences in mean item scores with t-tests of statistical sig- nificance; second, by examining the joint distribution of scores; and third, by estimating correlations.

6

As shown in Table 2, the mean level of satisfaction expressed by individuals with each specific contact is equal to or greater than

4 The Wave 2 and Wave 3 interviews were six months apart, but in the Wave 3 interview, respondents were asked about contacts that occurred within the past year. To ensure proper temporal order of contacts and (global) satisfaction, those respondents who stated during the Wave 3 interview that their most recent contact with the police occurred before the Wave 2 interview are either coded as not having a contact (Table 3) or excluded from the analyses (Table 4).

5 We used the following coding schemes: education (1 = less than high school, 2 = some high school, 3 = completed high school, 4 = some college, 5 = completed college/some advanced college); race (1 = nonwhite, 2 = white); income (1 = 0 - $8,000, 2 = $8,001 - $15,000, 3 = $15,001 - $25,000, 4 = more than $25,000); gender (1 = male, 2 = female). Age was coded in its raw form.

6 All of the respondents included in these analyses (N=262) had had either an "information," a "help," a "victimization," or a "stopped and questioned" contact with the police. Only two (of 262; .8%) had had all four types of contact, 22 (8.4%) had three types of contact, and 66 (25.2%) had two types of contact. The remaining 172 (65.6%) had just one type of contact. Therefore, because well over half of the sample only had one type of (measured) contact with the police, the possibility of con- founding experiences is of little concern.

126 GLOBAL AND SPECIFIC ATITUDES

their mean level of global satisfaction, although none of the differ- ences achieves a conventional (.05) level of statistical significance. Thus we find no support in these data for the hypothesis that gen- eral support for the police is greater than specific support.

Table 2. Relationships between Global Satisfaction and Specific Satisfaction Measures

Specific Global Satisfaction Satisfaction Measures r N 1=vd 2=sd 3=ss 4=vs

Satisfaction/Info .50 160 39 42 60 19 1=very dissatisfied 66 33 15 16 2 2=somewhat dissatisfied 15 1 6 8 0 3=somewhat satisfied 25 2 8 13 2 4=very satisfied 54 3 13 23 15

Global mean = 2.37 Specific mean = 2.42 t-value = -. 53; p=.60

Satisfaction/Assistance .58 103 23 21 41 18 1=very dissatisfied 26 12 7 7 0 2=somewhat dissatisfied 13 4 4 5 0 3=somewhat satisfied 28 7 5 15 1 4=very satisfied 36 0 5 14 17

Global mean = 2.52 Specific mean = 2.72 t-value = -1.91; p=.06

Satisfaction/Stopped .24 56 10 12 25 9 l=very dissatisfied 15 6 1 7 1 2=somewhat dissatisfied 10 1 4 3 2 3=somewhat satisfied 12 1 3 6 2 4=very satisfied 19 2 4 9 4

Global mean = 2.59 Specific mean = 2.63 t-value = -. 20; p=.85

SatisfactionNictimization .43 49 15 18 11 5 1=very dissatisfied 16 8 4 4 0 2=somewhat dissatisfied 9 5 4 0 0 3=somewhat satisfied 11 0 6 3 2 4=very satisfied 13 2 4 4 3

Global mean = 2.12 Specific mean = 2.43 t-value = -1.82; p=.08

Notes: Missing data are excluded from the analyses. Raw frequencies are presented in cross-tabulations.

Although the means of these attitude measures do not differ widely, an examination of the joint distribution shows some diver- gence. In each cross-tabulation in Table 2, more than half of the cases (53.4% to 64.3%) fall off the main diagonal. In addition, a sizable percentage of respondents (28.2% to 41.1%) in each cross- tabulation are "very" or "somewhat" satisfied on one attitudinal measure but "very" or "somewhat" dissatisfied on the other. More- over, the correlations range only from .24 to .58. Thus, although

BRANDL, FRANK, WORDEN AND BYNUM 127

individuals who are satisfied ("very" and "somewhat" combined) with the police in a particular contact are also more likely to be satisfied than dissatisfied with the police overall, and although in- dividuals who are dissatisfied ("very" and "somewhat") with the po- lice in a particular contact are more likely to be dissatisfied than satisfied with the police overall, the association is only moderately strong. It would be an error to conclude that the global and the specific attitudinal measures are statistically indistinguishable.

With reference to Table 2, we would add that satisfaction with specific types of contacts varies markedly among citizens who had the same types of contacts. More than half of those who had been stopped by the police were satisfied with the way they were treated, whereas half of those who requested information and more than one-third of those who had requested assistance were dissatisfied with the police response. Analyses of global attitudes toward the police should estimate, if possible, the effects of specific attitudes, not of contacts as such, because citizens are so diverse in their as- sessments of seemingly similar types of experiences.

To test the hypotheses that assert a causal relationship be- tween the global and the specific attitudes, we perform ordinary least squares (OLS) regression analyses. 7 Table 3 displays findings concerning the hypothesis that specific attitudes influence global attitudes. Here the dependent variable is T2 global satisfaction. The independent variables are T1 global satisfaction, the specific satisfaction measures, and (as controls) the demographic items. By including prior (T1) global satisfaction in the analyses, and thus controlling for its impact on T2 global satisfaction, we can more ac- curately identify the effects of specific attitudes on general atti- tudes. Table 3 shows, in parentheses, the estimated effects of specific satisfaction and demographic characteristics when T1 global satisfaction is omitted from the model; these are the results of a cross-sectional analysis that makes the customary assumption about causal direction.

In both sets of analyses presented in Table 3 (with and without T1 global satisfaction), each of the four specific attitudes appears as two variables that are coded to represent degrees of positive (satis- fied) or negative (dissatisfied) evaluations of each experience (2=very satisfied or very dissatisfied; 1=somewhat satisfied or some- what dissatisfied; O=did not have a contact of that type). For exam- ple, a respondent who was "somewhat satisfied" with the officer in a

7 The purpose of the regression analyses is to test the hypotheses concerning the causal effects of global and specific satisfaction on one another. The analyses are not intended to explain satisfaction levels. Therefore the proportion of variance ex- plained in the models is of secondary importance.

128 GLOBAL AND SPECIFIC ATITUDES

Table 3. OLS Regression Analyses of T2 Global Satisfaction

Standard Variable Coefficient Error t-ratio Beta

T, Global Satisfaction

Education

Sex

Race

Income

Age

Positive Info

Negative Info

Positive Assistance

Negative Assistance

Positive Stop

Negative Stop

Positive Victimization

Negative Victimization

Constant

R 2

Adjusted R 2

N

.50

.08 (.10) -. 03 ( .00) -. 14 (-.28)

-. 01 (-.04)

.01

.01)

.01

(-.02) -. 21 (-.36)

.26

.28) -. 21 (-.28)

.01 (-.01)

.03 (-.14) -. 17 (-.13)

-. 38 (-.68) 1.23

(2.59) .42

(.20)

.39

.17) 314

(327)

10.92**

1.81* (2.17)*

- .32 (-.03)

-1.48 (-2.69)**

- .24 (-.82)

1.93* ( 2.89)**

.12 (-.35) -3.16** (-4.90)**

3.20** ( 3.06)**

-1.96 (-2.35)

.05 (-.07)

.20 (-.98) -1.12 (-.74)

-1.51 (-2.29)**

3.67** ( 7.18)**

.50

.09 (.12)

-. 01 (.00)

-. 07 (-.14)

-. 01 (-.05)

.09 .16)

.01 (-.02) -. 14

(-.25)

.14

.15) -. 09

(-.12)

.00

.00)

.01 (-.05) -. 05 (-.04)

-. 06 (-.11)

* p < .05; ** p < .01 (one-tailed test) Notes: Missing data are excluded from analyses. Results in parentheses do not include T1 global satisfaction.

request for information contact is coded 1 on the Positive Info varia- ble, while another respondent who was "very dissatisfied" in the

same type of contact is coded 2 on the Negative Info variable. This treatment is necessary because previous research has shown that

BRANDL, FRANK, WORDEN AND BYNUM 129

negative evaluations have a stronger impact than positive evalua- tions on overall attitudes (see Dean 1980) and because a single vari- able would not allow us to examine the differential impacts of negative and positive assessments.

The results of these analyses show, first, that T1 global satisfac- tion has a substantial effect on T 2 global satisfaction; this coeffi- cient, even without a correction for attenuation due to measurement error, suggests that the global attitude is fairly stable over at least a six-month period. Second, the results show that T 2 global satisfaction is affected by three specific attitudes, namely positive and negative assessments of assistance contacts and nega- tive assessments of information contacts, even when prior (T1) global satisfaction is controlled. Third, because a cross-sectional analysis-the results of which appear in parentheses- cannot con- trol for T, global satisfaction, it somewhat overestimates the impact of specific satisfaction on T 2 global satisfaction. The cross-sectional analysis yields an estimate of the impact of citizens' negative as- sessments of victimization contacts which is substantively and sta- tistically significant, and which falls to 0 when T, global satisfaction is controlled. This analysis also produces inflated esti- mates of the effects of other specific attitudes: positive and nega- tive assessments of information contacts, positive and negative assessments of assistance contacts, and negative assessments of stopped and questioned contacts.

To examine the hypothesis that global attitudes influence spe- cific attitudes, we estimate OLS regression equations once again. The dependent variable in each equation is the specific satisfaction measure (1=very dissatisfied; 4=very satisfied); T 2 global satisfac- tion and the demographic characteristics are explanatory variables. As shown by the results in Table 4, global satisfaction has a sub- stantively and statistically significant effect on specific satisfaction with each of three types of contact (the 'ictimization" contact is the exception). Moreover, in these three equations, the strongest im- pact on specific satisfaction is that of T, global satisfaction. The effect of global satisfaction in victimization-related contacts is sub- stantially large but does not achieve statistical significance, per- haps because of the small size of the subsamples on which the estimate is based. One then might infer that global attitudes to- ward the police influence evaluations of police in particular contact situations.

DISCUSSION

This study represents an attempt to understand more clearly the relationship between specific and more global attitudes toward

130 GLOBAL AND SPECIFIC ATTITUDES

t- CD -4 U) 00 -- c + ~ ci ca-

I I-' ci

co ,, qcoU)00t

t- t 0 01 ~ -4C*

0

0D1

o o) oI i

ci)

to-

cow

m 0 e) - r'i 1 C -I C

00 Uo Pdq

E- z *Q,-4qc-

BRANDL, FRANK, WORDEN AND BYNUM 131

the police. The comparison of specific with global attitudes shows that most of the differences-in means and in proportions-are neither substantively large nor statistically significant; insofar as global and specific attitudes differ, respondents who have had a contact with the police tend to be more satisfied with their individ- ual contact than with the police more generally. These findings are contrary to the hypothesis offered by Dennis (1976) and to White and Menke's (1982) conclusion.

Furthermore, this study suggests that global and specific atti- tudes toward the police are causally related, and that the causal effects are asymmetrically reciprocal. That is, citizens' global atti- tudes toward the police affect their assessments of specific contacts with the police, and citizens' assessments of specific contacts affect their global attitudes, but the former effect is stronger than the lat- ter. These findings are consistent with the proposition that citi- zens' evaluations of their personal experiences with the police are affected by stereotyping and selective perception; those who hold generally favorable views of the police are more likely to evaluate their contacts with the police favorably, and those who hold gener- ally unfavorable views are more likely to evaluate their contacts unfavorably. The findings also support propositions about the ef- fects of specific attitudes on global attitudes. They indicate, how- ever, that these effects might have been overestimated in previous research, which did not control for the confounding effects of prior global attitudes.

These results have important implications for community polic- ing reforms, which are directed toward building police-citizen part- nerships to address community problems. One objective of community policing, especially in neighborhoods whose residents traditionally have regarded the police with skepticism or hostility, is to improve citizens' views of the police in order to establish a firmer foundation for cooperation. Inasmuch as citizens' assess- ments of their contacts with police are influenced by their existing global attitudes more strongly than the reverse, the police may face a particularly difficult-though perhaps not insuperable-barrier to forming closer police-community relationships. The failure to overcome citizens' preconceptions of police may contribute to the differential outcomes that Skogan (1990:104-107) found in Houston, where some forms of community policing benefited some social groups- minorities, renters, low-income households-less (if at all) than they benefitted higher-status groups.

Whether the findings of our research are generalizable is a question that invites replication of this research. The survey re- spondents whose attitudes we analyzed live in neighborhoods that

132 GLOBAL AND SPECIFIC ATTITUDES

are predominantly black and of low to moderate income- neighbor- hoods in which global attitudes toward the police may be least favorable, and in which cultural forces may have pronounced effects on residents' (global and specific) attitudes toward the police (Erez 1984; Jacob 1971). The relationship between global and specific at- titudes among different populations in different social contexts may differ from those found here. Future research should collect panel data in various neighborhoods in various cities, and from larger samples.8

Future research also should be directed toward understanding more clearly how global attitudes toward the police are formed, and how (if at all) global or specific attitudes can be influenced by the quality of police contacts. One potentially informative direction for such research would be analyses of socialization, focusing especially on the forces-including familial and peer attitudes and exper- iences-that influence youths' attitudes toward the police. Another potentially useful approach would involve intensive interviews with small samples of selected social groups. Most research on citizens' attitudes toward the police has been based on responses to closed- ended items on highly structured survey instruments; these items provide data on outlooks that the researchers conceived. More loosely structured interviews that provide for more open-ended re- sponses (see, e.g., Hochschild 1981; Lane 1962) could permit deeper insights into the form and nature of citizens' views of the police: their definitions of the police role, their expectations of their local police, their perceptions of police practices, the criteria by which they evaluate the police, and so on. Such research might form the basis for survey instruments which are designed to capture the ele- ments of citizens' belief systems more completely and more accu- rately, and on which more informative quantitative analyses could be based.

In another logical extension of this study, researchers would examine the consequences of citizens' attitudes toward the police. It is a common but untested assumption that citizens who are satis- fied with the police are more likely to engage in behaviors which support the police (e.g., reporting crimes, providing information to the police, engaging in crime prevention activities). Citizens with more favorable attitudes presumably are more likely to become

8 It is worthwhile to highlight the methodological difficulties associated with providing larger sample sizes in a panel survey with the purposes intended here. To do so, one must obtain a (representative) sample of people who will have a certain form of contact with the police within some period in the future. On the assumption that 30 percent of the general population has such contacts with the police in a six- month period, the addition of 100 respondents in the Wave 1 survey would yield only 30 additional respondents for the subsample analyses.

BRANDL, FRANK, WORDEN AND BYNUM 133

"coproducers" (Goldstein 1987) of crime prevention. An empirical examination of this assumption could prompt additional inquiries into the complexities of citizens' attitudes toward the police.

REFERENCES

Benson, P.R. (1981) "Political Alienation and Public Satisfaction with Police Serv- ices." Pacific Sociological Review 24:45-64.

Carter, D.L. (1985) "Hispanic Perception of Police Performance: An Empirical As- sessment." Journal of Criminal Justice 13:487-500.

Dean, D. (1980) "Citizen Ratings of the Police: The Difference Contact Makes." Law and Policy Quarterly 2:445-71.

Dennis, J. (1976) "Who Supports the Presidency?" Society 13:48-53. Easton, D. (1965) A Framework for Political Analysis. Englewood Cliffs, NJ: Pren-

tice-Hall. Erez, E. (1984) "Self-Defined 'Desert' and Citizens' Assessment of the Police." Jour-

nal of Criminal Law and Criminology 75:1276-99. Goldstein, H. (1987) "Toward Community-Oriented Policing: Potential, Basic Re-

quirements, and Threshold Questions." Crime and Delinquency 33:6-30. Hamilton, D.L. (1981) Cognitive Processes in Stereotyping and Intergroup Behavior.

Hillsdale, NJ: Erlbaum. Hamilton, D.L. and T.L. Rose (1980) "Illusory Correlation and the Maintenance of

Stereotypic Beliefs." Journal of Personality and Social Psychology 39:832-45. Hero, R.E. and R. Durand (1985) "Explaining Citizen Evaluations of Urban Services:

A Comparison of Some Alternative Models." Urban Affairs Quarterly 20:344-54. Hochschild, J.L. (1981) What's Fair? American Beliefs about Distributive Justice.

Cambridge, MA: Harvard University Press. Homant, R.J., D.B. Kennedy, and R.M. Fleming (1984) "The Effect of Victimization

and the Police Response on Citizens' Attitudes toward the Police." Journal of Police Science and Administration 12:323-32.

Jacob, H. (1971) "Black and White Perceptions of Justice in the City." Law and Soci- ety Review 6:69-89.

Koenig, D.J. (1980) "The Effect of Criminal Victimization and Judicial or Police Con- tacts on Public Attitudes toward Local Police." Journal of Criminal Justice 8:243-49.

Mastrofski, S. (1981) "Surveying Clients to Assess Police Performance." Evaluation Review 5:397-408.

Lane, R.E. (1962) Political Ideology: Why the American Common Man Believes What He Does. New York: Free Press.

Parks, R.B. (1984) "Linking Objective and Subjective Performance." Public Admin- istration Review 44:118-27.

(1986) "In Defense of Citizen Evaluations as Performance Measures." Urban Affairs Quarterly 22:66-83.

Percy, S.L. (1980) "Response Time and Citizen Evaluation of Police." Journal of Po- lice Science ancl Administration 8:75-86.

Poister, T.H. and J.C. McDavid (1978) "Victims' Evaluation of Police Performance." Journal of Criminal Justice 6:133-49.

Scaglion, R. and R.G. Condon (1980) "Determinants of Attitudes toward City Police." Criminology 17:485-94.

Shapland, J. (1983) "Victim-Witness Services and the Needs of the Victim." Vic- timology 8:233-37.

Skogan, W.G. (1990) Disorder and Decline: Crime and the Spiral of Decay in Ameri- can Neighborhoods. Berkeley: University of California Press.

(1991) 'The Impact of Routine Encounters with the Police." Paper presented at the annual meeting of the American Society of Criminology, San Francisco.

Snyder, M., E.D. Tanke, and E. Berscheid (1977) "Social Perception and Interper- sonal Behavior: On the Self-Fulfilling Nature of Social Stereotypes." Journal of Personality and Social Psychology 9:656-66.

134 GLOBAL AND SPECIFIC ATTITUDES

Thomas, C.W. and J.M. Hyman (1977) "Perceptions of Crime, Fear of Victimization, and Public Perceptions of Police Performance." Journal of Police Science and Administration 5:305-17.

White, M.F. and B.A. Menke (1982) "On Assessing the Mood of the Public toward the Police: Some Conceptual Issues." Journal of Criminal Justice 10:211-30.

Zamble, E. and P. Annesley (1987) "Some Determinants of Public Attitudes towards the Police." Journal of Police Science and Administration 15:285-90.