Public Trust in Governing Institutions: A Study on the Bangladesh Police

Abstract

Trust in the police is an important issue in any country because of its role and responsibility in ensuring safety, security, and overall quality of governance. If the public has low trust in the police, they will be reluctant to report crimes or call for services and assistance from the police, which can result in serious consequences. Trust in the police is the most contemporary issue in the context of Bangladesh. Although studies have been conducted in this field by various researchers from different countries and few surveys and studies were conducted, most of these studies evaluated either satisfaction with the performance of police or corruption of police, only a single study (Hossain and Rahman, 2017) focused on the level of trust of urban, preferably educated citizens in the police. The study attempted to evaluate the level and extent of public trust in the police. The study also evaluated the differences in public perceptions of trust in the police between the rural and urban areas in Sylhet and Sunamganj and suggested policy measures to uplift the trustworthiness of the Bangladesh Police. The study has been conducted based on integration of both descriptive and explanatory research using the social survey method which followed a mixed research approach in gathering both quantitative and qualitative and data for the topic under study. Face -to- face interview was then employed by using a structured interview schedule with both closed and open-ended questions to gather primary data from respondents. The data was analyzed using statistical software such as Statistical Package for Social Sciences (SPSS), Stata and Microsoft Excel where descriptive statistics were used to describe the characteristics of the sample and data nature via graphs, tables, and charts in accordance with the objectives of the study and inferential statistics (correlation, chi-square tests, binary logistic regression) have been used to test hypothesis about the relationship between variables. This was supplemented by secondary data collected from different published and unpublished sources. Cronbach’s alpha reliability test was also conducted for multi-itemed 5-point Likert scales of dependent variable of the interview schedule to ensure reliability and validity of the used items in the study. While conducting the study, the ethical issues of social research such as anonymity, confidentiality and informed consent have been maintained. A total of three binary logistic regression models were conducted to determine if various demographic, contact, and contextual variables influence citizens’ perception of trust in the vi police. In model 1, effects of the demographic variables were analyzed only. The model revealed that respondents of urban area (Sylhet)had less trust in the police than those who were from rural area, at 1% level of significance. (AOR= 0.18, 99% CI: 0.13, 0.27). The other three variables were categorized into different categories like respondents' age was measured in actual years during data collection, but later on, it was categorized under five categories (<25, 25-34, 35-44, 45-54, and 55>). In the case of age, the Adjusted Odds Ratio of the other four categories was higher than that of the first category, and one of them 25-34 years was significant (AOR= 2.10, 95% CI: 1.19, 3.72). Education was also measured in actual years of schooling during data collection and then categorized into four groups (Up to Primary, Secondary, Higher Secondary, and above Higher Secondary). In the case of education, the Adjusted Odds Ratio of the other three categories was less than the first category, and two of them secondary (AOR= 0.47, 99% CI: 1.19, 3.72), above higher secondary (AOR= 0.45, 99% CI: 0.26, 0.79) were statistically significant, respondents with increased education are less likely to have trust in the police. So, education is negatively associated with trust in the police. Males had less trust than females, and respondents with higher income had less trust than those of having lower income, though these did not have statistical significance. In the second model, two contact variables, harassment by police and being asked to pay a bribe were added with the previously used demographic variables. Effects of residence continue to be significant. Effects of gender, which was not significant in the first model, became significant (AOR= 0.65, 95% CI: 0.43, 0.98) in this model. Effects of age also continue to be significant positively in this model 25-34 years (AOR= 2.53, 99% CI: 1.37-4.67), 35 44years (AOR= 1.90, 90% CI: 0.97-3.71), 45-54 years (AOR= 2.12, 90% CI: 0.98-4.57), >54 years (AOR= 2.55, 95% CI: 1.06-6.11). Association of higher income category and trust became significant negatively in this model (AOR= 0.60, 90% CI: 0.32, 1.10), which means, those having higher income was 40 times less likely to have trust in the police than those who are in the lower income category. Those who confronted harassment by police (AOR= 0.20, 99% CI: 0.11, 0.38) which means, those who confronted harassment with the police are 80 times less likely to have trust in the police compared to those who did not confront harassment with the police, and it was statistically significant at 1% level. The second contact variable, being asked to pay a bribe also significantly lowered citizens’ trust in police (AOR= 0.22, 99% CI: 0.13-0.37). In model 3, contextual variables– feel safe at the walk after dark, trust in neighbours, someone robbed you (victimization against the person), and belief in negative news of police were vii included with the second model. Effects of harassment and being asked to pay a bribe continue to have significant negative effects, and effects of higher income also continue to be significant negatively. Safety feeling of walking alone after dark had a positive effect on trust in the police (AOR= 3.03, 99% CI: 2.07, 4.42). Those who had ‘trust in neighbours in calling police after seeing a suspicious person in the neighbourhood’ (AOR= 1.62, 99% CI: 1.23-2.14), and ‘trust in neighbours in returning favour’ (AOR= 1.39, 90% CI: 1.00, 1.93) expressed more trust in the police than those who had not trust in their neighbours in these two variables. Believing negative news of police misconduct, brutality, or abusive language (AOR= 0.68, 99% CI: 0.51, 0.90) which means, those who believe in negative media news about police were 32 times less likely to have trust in the police than those who do not believe in negative media news about the police. Being robbed by someone lowered citizens’ trust in the police, though it did not have statistical significance. Four, out of the five contextual variables, appear to have both statistical and substantive significance for the explanation of public trust in the police. Model three is the ultimate model. The Pseudo r2 value and loglikelihood value demonstrate that among the three models, model 3 is the best. The results were compared and consistent with the evidence from similar other studies. Therefore, to make police more trustworthy and based on the findings, some policy suggestions are proposed, like strengthening the performance of community policing, arranging training on police-citizen relationship, etiquette etc., and ensuring non-interference by political parties Besides, reforming the Police Act of 1861 and establishing a Police Commission were also suggested by the researcher. To sum up, it is expected that the implementation of these suggestions is likely to improve the practice of policing style in reducing crime through a safer Bangladesh for citizens to live in peace and security.