THE REASONS FOR COMPLIANCE WITH LAW Margaret Levi, University of Washington Tom Tyler, New York University Audrey Sacks, University of Washington DRAFT Paper for the Workshop on the Rule of Law. Yale University. March 28-29, 2008.
1 THE REASONS FOR COMPLIANCE WITH LAW People s willingness to comply with the law is a litmus test of the effectiveness and viability of the state. Without compliance, there is no rule of law, no matter how well the institutions and regulations are designed. Governments unable to motivate their citizenry to pay taxes, fight on behalf of the state, and generally refrain from lawbreaking (Levi 1988, 1997; Tyler 1990) are unlikely to survive in the long run. In emerging and transitional societies, a central question is how to establish compliance with law. Even in established societies such as the United States compliance cannot be taken for granted; the government must continually work to secure and sustain deference to at least some of its policies. Our concern is with one aspect of this broad question: the role of the rule of law in generating and sustaining compliance. We argue that rule of law arguments have two core propositions. The first is that people are influenced by their views about the legitimacy of the law. Here we propose that individuals behavior is responsive to more than the ability of the state to promise rewards or threaten punishments. People are also responsive to their judgments about the degree to which the state is an appropriate political authority, entitled to be obeyed. Our second proposition is that legitimacy itself rests, at least in part, on assessments of the state as operating according to the rule of law. These assessments have both positive and normative bases. Individuals have a normative idea about what government should be doing, and they then evaluate what government objectively does according to those normative criteria. Our argument is that people form legitimating
2 beliefs about their government in response to four aspects of governance: performance, competence, procedural justice and trustworthiness (Levi, Sacks, and Tyler Forthcoming). This analysis We test our propositions using two datasets. These include cross-sectional data from the United States and cross-national survey data from Africa. In the United States, where legitimacy is well established and there is strong community-based social capital, we expect to find to find a robust positive relationship between legitimacy and compliance. Where there is deep-seated legitimacy, we expect to observe high rates of compliance. We also expect to observe high levels of legitimacy, where there is a strong basis for the rule of law. In Africa, where there is very little evidence that legitimacy is widespread within communities, we expect to observe a positive relationship between legitimacy, compliance and the rule of law but we do not expect these relationships to be as robust. Theoretical Framework Legitimacy and Compliance Legitimacy derives from the beliefs citizens hold about the normative appropriateness of government structures, officials, and processes. Of central importance is the belief that rules and regulations are entitled to be obeyed by virtue of who made the decision or how it was made (Tyler 2006, 377). When members of the polity believe that government is legitimate, they are more likely to defer to political authorities and uphold laws. Legitimacy denotes popular acceptance of government officials right to govern. As such it is distinct from the ability of the government to compel compliance through
3 the threat or use of sanctions and/or its ability to motivate compliance by the provision of rewards or services. A major effect of legitimacy is an increased likelihood of compliance with governmental rules and regulations (Levi and Sacks 2007). A government perceived as legitimate can expect widespread public cooperation for such voluntary acts as voting, volunteer military service and participation in community problem solving as well as quasi-voluntary compliance with taxes and enlistment (Levi 1997; Levi and Sacks 2007). Legitimacy also leads people to be more willing to defer to the law and to the decisions of legal authorities such as the police and the courts (Tyler 1990). Without legitimacy, people may be less willing to support government programs that redistribute economic resources (Hetherington 2005). Legitimacy shapes citizens reactions to government policies (Weatherford 1992) and provides government with grounds for eliciting citizen support other than appeals to immediate self-interest. Citizens are more likely to support and volunteer to fight in wars (Leff 1991; Levi 1997) and comply with health regulations during an epidemic (Lieberman 2007) when they perceive government as legitimate. They are also more willing to work with government authorities to solve problems in their communities (Sunshine and Tyler, 2003; Tyler and Fagan, in press). Although it is possible to rule using only coercive power, legitimate power makes governing easier and more effective. Without legitimacy, governments have to expend more resources on monitoring and enforcement to induce sacrifice and compliance. The existence of legitimacy reduces the transaction costs of governing by reducing reliance on coercion and monitoring. Hence, while scholars of politics disagree about whether legitimacy is a necessary component of an effective government, everyone recognizes
4 that governments benefit when they have legitimacy. This is especially true of emerging governments, which find motivating their publics to be a key element in viability, and during periods of crisis or change, when governments are least able to either reward their citizens or effectively deploy system of surveillance and sanctioning. Governments are most dependent upon the cooperation of their citizens under those circumstances in which they are least able to obtaining it via the mechanisms or reward and punishment. Our goal in this paper is to present a model of legitimacy and provide some empirical evidence supporting that model. We first model legitimacy as a sense of obligation or willingness to obey authorities (value-based legitimacy) that then translates into actual compliance with governmental regulations and laws (behavioral legitimacy) (see figure 1). The stronger the bases of value-based legitimacy in a society, the more compliance, or behavioral legitimacy, we expect to observe. We focus upon responsibility and obligation because it is the core feature of theoretical discussions of legitimacy (Tyler, 2006b). When members of a polity view the political authorities and institutions of that polity as legitimate, they defer to the policies they enact out of a normative sense of obligation to do so because it is right and proper for those authorities and institutions to make rules and, hence, the responsibility of citizen to defer to those rules, separately from judgments of whether they or their groups benefit or are harmed by them. It is this sense of doing what is normatively appropriate that is the essence of the idea of legitimacy. And, from our perspective here, the key issue is whether the members of a polity judge their leaders to be legitimate that is the concern. There is a large and robust literature on the objective criterion that should be used to decide whether or not authorities or institutions ought to be obeyed. While we draw upon
5 this literature for guidance concerning issues that might be of concern to members of the public, our analysis is psychological in nature and concerns the judgments made by mass publics within the various societies we consider. From the psychological perspective there are several benefits to basing political authority on legitimacy. The principle advantage is that the government is freed from the continual need to provide citizens with rewards or credibly threaten them with punishments to obtain desired behavior. Even when effective systems are in place to provide such resources, the government must do continue to do so, or as least must make a credible case that if it cannot do so in the short term, its prospects for doing so in the long terms are good. Loyalty that is purchased is fleeting. As Machiavelli said long ago friendships that are obtained by payments and not by greatness or nobility of mind, may indeed be earned, but they are not secured and in time cannot be relied upon. A second advantage is that those citizens whose own values lead them to support the state are more likely to do so proactively, finding ways to do what is needed to solve problems and manage difficulties (Tyler, 2007). For example, the battlefield advantage of democratic armies is linked to the superior ability of soldiers to be innovative in adapting to battlefield conditions, as well as their greater motivation to sacrifice on behalf of their country (Tyler, Callahan, and Frost, 2007). The Rule of Law and Legitimacy Our conceptual model posits that legitimacy has four antecedent normative conditions: procedural justice; trust/confidence in authorities; government performance; and, administrative competence. Procedural justice, that is the commitment of government to uphold the laws fairly and to apply them equally to all, should enhance
6 deference and willing obedience. Trust and confidence reflects the judgment that the authorities are motivated to do what is right for the people they serve, seeking policies that truly benefit their societies. This reflects the confidence that government officials can be relied upon to deliver on their promises. To the extent citizens have confidence that relevant government agencies have the competence to deliver on promises they are also are likely to obey rules (Cook, Hardin, and Levi 2005; Levi 1988, 1997; Rothstein 2005). For example, people want to believe that the government can effectively enforce laws by punishing those, themselves or others, who break them. Governments that provide services and protections that bolster its citizens welfare or are quickly developing the capacity to do so should be more likely to elicit the willing deference of citizens than ineffective and poorly performing governments. These two propositions form the framework of this paper. Study one Our first analysis explores the antecedents of compliance within the United States. It is based upon a study of 1,653 New Yorkers, interviewed in 2002 concerning their views about the NYPD, as well as their law related behavior. In this assessment legitimacy is assessed in three ways. First, as the obligation to obey the law. Respondents were asked seven items of the form: If the police make a decision you should obey it. In addition, respondents were asked three questions drawn from the Ewick and Silbey (1998) legitimacy framework. These items ask people to agree or disagree that the law represents the interests of the rich, rather than the concerns of people like myself.
7 Results Proposition 1: Previous analyses demonstrate that legitimacy shapes self-reported compliance with the law (Tyler, 2006a; Sunshine and Tyler, 2003; Tyler and Fagan, in press). Equations 1 and 2 in Table 1 replicate this analysis. Consistent with the argument we are making, those people who say the authorities are legitimate indicate more frequent compliance with the law. This finding accords with other studies that rely upon people s self-reported behavior (Tyler, 2006b), as well as studies that assess behavior via official arrest records (Tyler, Sherman, Strang, Barnes, and Woods, 2007). Legitimacy is consistently found to enhance compliance. Our goal is to move beyond the prior analysis of individual level compliance. We want to examine whether a climate of legitimacy leads to a climate of law abidingness. To do this analysis we used the objective crime rates reported for each of the 75 police precincts of the city of New York. We predict that those respondents who viewed the law as legitimate, as randomly sampled respondents from their precinct, would reflect the general level of legitimacy among members of that neighborhood. And, those neighborhoods in which the law was viewed as legitimate should have lower rates of crime. The results of Equation 3 supported this hypothesis. They suggested that legitimacy was linked to the crime rate within the respondent s neighborhood. This provides evidence that the climate of legitimacy has objective consequences for a policy because it influences the general rate of rule following within it. Table 1 shows the linkage between legitimacy and compliance. The equation includes a control for the population size within each precinct. However, we also need to
8 take account of the socio-economic character of each neighborhood. We do that using multilevel modeling (Raudenbush and Bryk, 2002; Raudenbush, Bryk, Choeng, Congdon, and du Toit, 2004). On the aggregate level we include three objective indices of neighborhood, coded on the FISC level using census data. Those indices are: the percentage of families living in poverty; the percentage of families without a wage earner; and the median value of homes. These aggregate factors were included in a multilevel model, shown in Table 2, and are found to shape the precinct crime rate. However, introducing these controls did not change the basic finding. Legitimacy continued to shape crime rate. Taken together these findings support the argument that legitimacy promotes compliance. People comply, in other words, because they feel they have an obligation to comply. So, a climate of legitimacy encourages a climate of compliance. And, of course, people also comply because of the risk of punishment, as is indicated in equations 1 and 2. This deterrence effect was also found in the multilevel model shown in Table 2. Hence, the competence of the police is important. Proposition 2: Our second question is whether legitimacy fits the parameters of the rule of law. As we noted, we identify four antecedents of legitimacy: performance, competence, procedural justice, and trust. Of these our focus is on two elements that are central to the rule of law procedural justice and trust. Table 3 indicates that legitimacy is affected by both procedural justice and trust. In other words, legitimacy does reflect the features associated with the rule of law. In fact, at least in this analysis these rule of law elements are very strong, while both
9 performance and competence have no significant influence. Among Americans the elements of the rule of law are central to judging the legitimacy of the police. Study Two Our second analysis explores the antecedents of compliance within a wide variety of sub-saharan African states. Africa is an especially good place to examine these issues because of the large amount of variation both within and across African countries in the extent to which governments are perceived as legitimate. Afrobarometer data surveys Africans views towards democracy, economics, and civil society with random, stratified, nationally representative samples. Specifically, we model the relationship between the existence of legitimating beliefs and low crime rates at the primary sampling unit (PSU) using the second and third rounds of Afrobarometer survey data. 1 Round 2 surveys were conducted between 2002 and 2003 in fifteen sub-saharan African countries. Round 3 surveys were conducted in 2005 in eighteen sub-saharan African countries. 2 The dataset used for this paper has a multilevel structure; individuals are nested within PSUS, which are in turn nested within countries. The PSUs are the smallest, welldefined geographic units for which reliable population data are available and they tend to be socially homogenous, thereby producing highly clustered data. In most countries, these will be Census Enumeration Areas (Afrobarometer 2005, 37-38). Ignoring the multilevel structure of our data can generate a number of statistical problems (Guo and Zhao 2000; Snijders and Bosker 1999). We discuss some of these problems and the advantages of using multilevel modeling in an earlier paper (Levi and Sacks 2007). Thus, taking into account the multilevel nature of our data, we estimate a two level model
10 with random intercepts for countries. Since we are concerned with crime levels at the community-level, we aggregated the individual-level data up to the PSU level. Results Proposition 1: First, we ask whether communities with higher average levels of legitimating beliefs have lower average crime rates. Our dependent variable is the average of the number of times respondents within the same PSU report having had something stolen from their home, been physically assaulted, or feared crime in one s home within the past year. Our main explanatory variable, legitimacy, is the average level of willingness respondents within the same community report of accepting the following: the tax department s right to make people pay taxes; the court s right to impose binding decisions; and, the police s right to make people obey the law. Table 4 reports the results from the hierarchical linear regression on the average level of crime in 1,472 PSUs for the 2002-2003 Afrobarometer data. We include controls for the socio-demographic characteristics of communities including the averages of the following: the number of female respondents; the number of children under eighteen living in respondents households; the frequency respondents report having gone with a sufficient amount of food within the past year; the level of respondents education; and, the number of rural respondents. We control for the presence of police within the PSU by including two variables indicating whether survey enumerators observed a police station and police within each PSU. We also include a variable indicating the average ease respondents within the same community face in obtaining help from the police.
11 Table 5 reports the results from the hierarchical linear regression on the average level of crime in 2,368 PSUs for the 2005 Afrobarometer survey. We use the same variables as the regression on the 2003 data with a few exceptions for the sociodemographic variables. The 2005 survey asked respondents whether they own a number of household items including a car, television, book, radio, and bike. We find that those geographical units with higher average legitimacy have a statistically discernable lower average crime. The coefficient estimate for legitimacy is.05 (p<.05) for Round 2. An increase in the average level of legitimacy present within a community corresponds to an increase of.05 in the safety of a neighborhood. The coefficient estimate for legitimacy is.06 (p<.05) for Round 3. These findings suggest both that there is a clear impact of the climate of legitimacy upon the crime rate and that this influence is weak in terms of the amount of variance explained. Of our socio-demographic controls for round 2, the average level of food security and the average number of rural households within communities are each positive and significant at the p<.05 level. None of our socio-demographic indicators are significant in our analysis of the round 3 data. Two of our indicators of police presence are significant at the p<.05 level in the round 2 analysis. We find that the relationship between the presence of police in the PSU and the safety of a community is negative. Also, the relationship between the average ease of obtaining help from the police and the average safety of a community is also negative. One plausible explanation for these negative relationships is that police are present where there are higher levels of pre-existing crime. In our round 3 analysis, we find that the relationship between the presence of a police station in a PSU and the
12 average safety of a community is again negative (p<.05). The relationship between the average ease of obtaining help from the police and the average safety of a community is now positive. As we anticipated, we find that a positive but weak relationship between the presence of legitimacy within a community and the average level of crime in that community while accounting for police presence and the socio-demographic wealth of a community. These data are cross sectional, reflecting influences measured at one time. It would further advance our case if we could show that changes in legitimacy lead to changes in the rate of crime. We would need longitudinal data over a long period of time to be able to assess whether the strengthening or weakening of legitimacy in a community corresponds to a reduction or increase in crime. We hope to find such data and to use it to extend our analysis. Proposition 2: We measure the extent to which legitimacy is related to the rule of law, especially procedural justice and trust and confidence in authority, using the third round of Afrobarometer. We model three separate dependent variables that capture citizens legitimating beliefs in terms of their willingness to obey the police, courts, and the tax department. These three items tap value-based legitimacy: the sense of obligation or willingness to obey. Specifically, our dependent variables are whether respondents agree with the statements: the tax department always has the right to make people pay taxes ; the courts always have the right to make decisions people abide by ; and, the police always have the right to make people obey the law. Respondents answers were originally coded as strongly disagree, disagree, neither disagree nor agree, agree, strongly
13 agree, don t know and refused to answer. We dichotomized this variable from a five point scale because we do not believe there is a substantive difference between the various categories. To examine whether a link exists between citizens perception of government fairness and their acceptance of governmental authority, we include variables indicating whether respondents believe the government treats citizens fairly and whether they believe the government treats members of their own ethnic group fairly. Afrobarometer includes questions that tap respondents trust and confidence in the police and courts. Although Afrobarometer does not include a question probing citizens confidence in the tax department, the survey asks respondents whether they trust their local government council. In areas with poor transportation infrastructure, inadequate media, and sparse settlement patterns, citizens interact more with local than national government authorities and are often subjected to more local than national taxes We also include measures of our two other antecedents of legitimacy: administrative competence and government performance. Citizens perception of the honesty of tax officials, police, and judges is likely to affect an individual s willingness to grant government the right to make and impose laws and regulations. We include measures of whether citizens believe that a large portion of tax administrators, police, and judges are corrupt. Willingness to comply with authorities is also related to the government s ability to effectively detect and punish crimes and tax evaders. Although an imperfect measure of states deterrent capacity, Afrobarometer includes survey questions on perceptions of government enforcement and monitoring capacities. One set of questions probes respondents on whether they believe that the state will enforce the
14 law if a citizen does not pay taxes or commits a serious crime. Another set of questions asks respondents about the likelihood that government authorities would enforce the law if a top government official committed a serious crime or did not pay taxes. Our measure of government performance is a dummy variable indicating whether a respondent or a household member ever went without sufficient food in the year preceding the survey. This should capture the extent to which citizens believe their government is meeting its end of the fiscal contract. We control for standard socio-demographic variables that can affect citizens legitimating beliefs. We include reasonably good proxies for income including whether respondents own a television, radio, car and bike and other demographic factors that affect household resources: health, age, employment, and urban or rural residence. We also control for standard country-level indicators for 2004: logged GDP per capita; logged aid per capita; logged population size; and, Freedom House s measures of civil liberties and political rights. Because none of these country-level indicators are significant at the p<.05 level, we left these results out to conserve space. Results We estimate the effects of procedural justice, trust, administrative competence, and government performance on legitimacy using multilevel logistic regression with random intercepts for the PSU and country levels. Because of the difficulty of interpreting multilevel logistic parameters, we focus our discussion on the point estimates of the first differences and the confidence intervals surrounding them (see table 6). 3 In support of our hypothesis, we also find considerable evidence of a link between procedural justice and deference to government authority. We model two indicators of
15 procedural justice: whether a respondent perceives that the government treats members of his or her ethnic group fairly; and, whether a respondent believes that the government treats citizens fairly. All of our indicators of procedural justice are significant at the p<.01 level in each of the three regression models. A belief that the government is procedurally fair corresponds to an average 5.84 percentage point increase in the probability that a citizen will defer to the authority of the tax department, police and courts. 4 As we expected, citizens trust and confidence in the police, courts and their local government councilors correlate with their acceptance of government authority. Trust and confidence in authority corresponds to an average 16.13 percentage point increase in the probability that a respondent will defer to the authority of the courts, tax department and police. 5 Our results suggest that citizens assessment of administrative competence is positively related to their acceptance of the authority of the tax department, courts and police. There are two dimensions of administrative competency, its honesty and the extent to which it can monitor and enforce laws and regulations among citizens and elites. A perception that the government is competent, as opposed to believing that the government is corrupt and unlikely to enforce its regulations and laws, translates into an average 11.45 percentage point increase in the probability that a respondent will accept the court s, tax department s, and police s authority, respectively. 6 Our indicator of government performance, whether a respondent and household members went without sufficient food in the year preceding the survey, is significant at the p<.001 for our regressions on acceptance of the tax department s and court s
16 authority. Enjoying food security translates into an average 1.52 percentage point increase in the probability that a respondent will accept the authority of the courts, tax department and police. 7 A disinclination to pay taxes or to abide by court s decisions on the part of the hungry may reflect poverty and an inability to pay or to adhere to a court s decision. It may also reflect a belief that government is not fulfilling its end of the fiscal contract. In summary, we find considerable evidence of a link between the extent of procedural justice and trust and confidence in authority and citizens perceptions of legitimacy in a wide range of developing societies in Africa. We also find evidence that administrative competence and government performance are related to legitimacy. Of these four factors, the most important is trust and confidence (16.13%), followed by competence (11.45%), procedural justice (5.84%), and performance (1.52%). We did not find any evidence that country-level indicators including GDP (per capita), civil liberties and political rights explain variation in citizens acceptance of government authority. Discussion Our goal is to present and test a framework for thinking about the rule of law. That framework is organized around two propositions. The first is that people s law related behavior is related to their normative judgments about the legitimacy of government. Operationalizing legitimacy as obligation to defer to authorities we show that the level of legitimacy found in a particular community is linked to the level of compliance as reflected in indices of crime. High legitimacy areas have lower rates of crime. This is true both in an American sample and in two multicountry samples within Africa.
17 The approach we use to testing our model is a daunting one for any model rooted in the judgments of particular individuals. We use the respondents within particular communities as indicators of the general level of legitimacy within that community. We then link that generalized judgment to the level of crime within that community. Because communities differ in many ways, we adjust for factors such as the economic status of the community, the presence of effective policing (or even any policing). Of course, we recognize that these adjustments can never be complete. As a result, we think it is striking that all of our sample suggest that legitimacy does shape the rate of crime. This demonstration is perhaps most striking in the case of the New York City sample, where crime rates are drawn from police statistics. Those statistics are not linked in any way to the judgments of the respondents about legitimacy. Still in those communities in which respondents generally report higher legitimacy, the police report lower crime. In the African samples we use the respondents to tell us about the level of crime. The number of respondents who report having experienced crime or having fear of crime within a given geographical area is used to estimate the rate of crime. Again, we find that higher legitimacy is associated with lower crime rates. Finally, we test our model concerning the influence of the rule of law on legitimacy. That model finds support within each of our samples. In all cases we find that issues of trust and of procedural justice shape judgments of legitimacy. In the New York sample these issues are the key to legitimacy. In the African sample both rule of law and performance/competence shape legitimacy. In each case, however, our basic finding is that issues associated with the rule of law are important when legitimacy is being judged. While in both New York City and Africa people evaluate the legitimacy of
18 their government based upon its ability to provide basic services or upon its competence, equally and sometimes more important is their judgment about the trustworthiness and fairness of government.
19
20 Tables Note. Multiple regression analysis. Compliance measured via self-report. Precinct level felony arrests via police statistics.
Note. HML multilevel modeling. Weighted data. *p<.05; **p <.01; ***p<.001. 21
22
23
24
Afrobarometer. 2005. "Survey Manuel: Round 3 Surveys, March 2005 - February 2006." Cook, Karen S, Russell Hardin, and Margaret Levi. 2005. Cooperation Without Trust? New York: Russell Sage Foundation. Guo, Guang, and Hongxin Zhao. 2000. "Multilevel Modeling for Binary Data." Annual Review of Sociology 26:441-62. Hetherington, Marc. 2005. Why trust matters: Declining political trust and the demise of American liberalism. Princeton: Princeton University Press. Leff, Mark H. 1991. "The Politics of Sacrifice on the American Home Front in World War II." The Journal of American History 77 (4):1296-318. Levi, Margaret. 1988. Of Rule and Revenue. Berkeley: Los Angeles: London: University of California Press.. 1997. Consent, Dissent and Patriotism. New York: Cambridge University Press. Levi, Margaret, and Audrey Sacks. 2007. "Legitimating Beliefs: Concepts and Indicators." In Afrobarometer Working Paper Series. Cape Town, South Africa: Afrobarometer. Levi, Margaret, Audrey Sacks, and Tom R. Tyler. Forthcoming. "Conceptualizing Legitimacy, Measuring Legitimating Beliefs." American Behavioral Scientist Special issue on Legitimacy in the Modern World. Lieberman, Evan S. 2007. "Ethnic Politics, Risk, and Policy-Making: A Cross-National Statistical Analysis of Government Responses to HIV/AIDS." Comparative Political Studies 40 (12):1407-32. Rothstein, Bo. 2005. Social Traps and the Problem of Trust. New York: Cambridge University Press. Snijders, Tom, and Roel Bosker. 1999. Multilevel analysis : an introduction to basic and advanced multilevel modeling London; Thousand Oaks: Sage Publications. Tyler, Tom R. 1990. Why People Obey the Law. Princeton and Oxford: Princeton University Press.. 2006. "Psychological Perspectives on Legitimacy and Legitimation." Annual review of Psychology 57:375-400. Weatherford, M. Stephen. 1992. "Measuring political legitimacy." American Political Science Review 86:149-66. 1 The Afrobarometer is a joint enterprise of Michigan State University (MSU), the Institute for Democracy in South Africa (IDASA), and the Centre for Democracy and Development (CDD, Ghana). For more on the Afrobarometer, see www.afrobarometer.org.
26 2 We excluded Zimbabwe from our sample for both rounds because of concerns with data quality. We also excluded Uganda from our sample for round 3 since the dataset did not come with PSU codes. Fieldwork was conducted by national research institutions affiliated with the Afrobarometer project. Samples were designed using a common multi-stage, stratified, area-cluster approach. Random selection methods were used at each stage, with probability proportional to population size where appropriate. Sampling frames were constructed in the first stages from the most up-to-date census figures or projections available, and thereafter from census maps, systematic walk patterns, and project-generated lists of household members. 3 In each case, we repeated the first differences algorithm 10,000 times to approximate a 95-percent confidence interval around the probability of accepting the tax department s, court s and the police s authority. 4 A belief that the government is procedurally fair corresponds to a percentage point increase of 9.34, 4.65 and 3.54 in the probability of accepting the authority of the tax department, police and courts, respectively. 5 Trust and confidence in authority corresponds to a percentage point increase of 15.78, 14.80 and 17.80 in the probability that a respondent will defer to the authority of the tax department, police and courts, respectively. 6 A positive assessment of government competence corresponds to a percentage point increase of 11.98, 9.49, and 12.90 in the probability of accepting the authority of the tax department, police and courts, respectively.
27 7 Enjoying food security corresponds to a percentage point increase of 2.61,.35 and 1.6 in the acceptance of tax department s, police s and courts authority, respectively.