**3. Methodology**

By identifying factors contributing to crime prevention and fear of crime, this article aims to provide empirical evidence regarding a safe city program launched in Malaysia. A quantitative survey questionnaire methodology was employed, in which 400 pedestrians were asked about their perceptions of Kuala Lumpur's central business district. Survey research is probably the optimal method available to the social researcher concerned with collecting any original data that describes a population too large to observe directly [64] (p. 118), in this case, the 1.73 million inhabitants of Kuala Lumpur. The following subsections outline the site, population, and sample size; research framework, variables and research instruments; reliability test and data analysis; and research ethics.

### *3.1. Site, Population, and Sample Size*

The site selection was based on the significance and importance of Kuala Lumpur to Malaysia, accounting for more than 40% of the country's urban population [59]. Kuala Lumpur has been selected as a policing model under Malaysia's new economic model. The police force intends to make Kuala Lumpur a safer city for locals and tourists alike [65]. Besides that, the greater Kuala Lumpur conurbation [66] has also been identified as a National Key Economic Area (NKEA) with projects and initiatives to enhance the city's competitiveness [67].

Recently, Kuala Lumpur has been ranked 35th in the Safe City Index 2019 [57], dropping four places from the year 2017, and most of the category ratings are below average compared to the other 60 cities in the world. Kuala Lumpur recorded approximately 24.37% of the national street crimes for the year 2016, and the Dang Wangi District is home to the main hot spots for street crimes, with about 37.86% of the overall street crime in Kuala Lumpur [68]. Due to such hot sport for street crimes, the Dang Wangi District was selected as the survey area, encompassing four streets: Sultan Ismail Street, Ampang Street, Raja Chulan Street, and Bukit Bintang Street and Imbi Street (Figure 3).

**Figure 3.** The central business district in Kuala Lumpur and pedestrian survey area (source: authors).

Survey personnel were stationed along these four streets with three survey points on each street during both day and night time and throughout weekdays and weekends. This method followed [69], which identified "day" interviews as occurring from 9 a.m. to 6 p.m. and "night" interviews as occurring from 7 p.m. and 10 p.m. In this study, "weekdays" was defined from Monday to Friday, and "weekends" from Saturday to Sunday. Sunday to Thursday nights were defined as "weeknights", while Friday and Saturday nights were "weekend nights".

According to Table 2, 400 samples were collected using proportionate stratified random sampling from June to December in 2017. The purpose of stratification is to categorize a population into

subgroups based on homogeneity. Different subsets are heterogeneous. Next, a suitable number of units from each subset is chosen for sampling. The representative character of each sample is increased through this process, at least where stratification variables are concerned [64]. For each street, 100 samples were collected, with equal samples of 25 each during weekday and weekend, by day and night. In order to prevent bias, every 10th person in the stratified arrangemen<sup>t</sup> was selected for the sample. Those who confirmed that they had not visited the city previously were disqualified from the survey. Four hundred samples are adequate for study areas like Kuala Lumpur. A population of 1.73 million people requires a sample size of 386 people to represent the population at a 95% confidence level and a 5% error [70].


**Table 2.** Distribution of survey respondents (source: authors).

### *3.2. Research Framework, Variables, and Research Instrument*

The survey questionnaire was designed mainly upon the effectiveness of the prevention items towards street crime and the fear of street crime.

Acknowledging that a cross-sectional survey can only measure the items at the occasion factor (i.e., at one point in time) [71], to measure the effectiveness (cause and effect), the items were designed such that respondents compared their experiences at the point of survey (the year 2017) with the previous year (2016 as the control variable). All variables which involve perceptions of fear or street crime in the Kuala Lumpur's central business district (KL CBD) are compared to the previous year (Table 3).

Each survey had an average duration of 20 to 30 min, including the time taken to explain the purpose of the survey to the respondent. The survey questionnaire was developed based on studies by [69,72], which are concerned with a safer city and have been proven to be effective.

The design of the questionnaire covered firstly the respondent's background. Five basic demographic items were included: gender, age, monthly income (in Ringgit Malaysia, RM), educational level, as well as their usual reason for visiting KL CBD.

Second, for the main research items, two dependent variables (DV) identified for securing a safe city program are Street Crime and Fear of Street Crime. For the first dependent variable, street crime was measured through the average perceived security level in the central city following time from 6 a.m. morning to after 10 p.m. for both weekdays and weekends, compared to last year. The 5-point Likert scale used ranged from 1 as being very dangerous to 5 as being very safe. While the second dependent variable, the fear of crime, was measured through the average perceived fear of crime compared to last year. The 5-point Likert scale used ranged from 1 as being much worse now to 5 as being much safer now.

Third, for the independent variables (IV), 27 items/factors, which are summarized in Table 1, were applied in this study, which also mainly comes from two main categories, namely the CPTED with 18 items, and the CPSD with nine items (Figure 4).




**Table 3.** *Cont.*

Note: CBD stands for central business district, CCTV stands for closed-circuit television, CPTED stands for crime prevention through environmental design, CPSD stands for crime prevention methods through social development, GIS stands for geographic information system, ICT stands for information and communication technology, SCP stands for safe city program, and S.D. stands for standard deviation.

**Figure 4.** Relationships among variables (source: authors). Note: CPTED represents crime prevention through environmental design, CPSD represents crime prevention through social development, IV represents independent variables, and DV represents dependent variables.

For independent variables, a 5-point Likert scale was used to measure the levels of effectiveness of the 27 items, ranging from 1 as being not at all effective to 5 as being extremely effective. The mean score range values of the scale ratings were used for the descriptive analysis and to identify the effectiveness of the crime prevention steps stated in the safe city program. From the sample of 400 respondents, all mean score for the effectiveness of initiatives in reducing street crime and fear of crime are above 3, which indicated that all respondents have agreed that all twenty-seven initiatives under the safe city program are effective (Table 3). Besides, most of the data showed a skewness or kurtosis value of less than 1.0. If the skewness or kurtosis value is between −1.0 and 1.0, the distribution is considered normal, and indicate that are almost preferred to be symmetrical [73]. Two exceptional cases of CCTV installation with kurtosis 1.051, and the existence of police post surveillance with kurtosis 1.108, were assumed to contain normal data since their skewness values were in the normal range. Furthermore, the authors intended to measure both of these items, as they are important features in CPTED, as highlighted in [16,33]. Assuming the sample size is sufficient in number, which in this study amounts to 400 samples, no serious issues should arise even if the normality assumption is violated. The implication is that parametric procedures can be used in this study, despite a lack of normal distribution of data [74].

### *3.3. Reliability Test and Data Analysis*

All the collected data from the public survey involving 400 respondents were checked for quality using the Statistical Package for the Social Sciences (SPSS) software by conducting reliability tests to verify the consistency of all the measurements.

Cronbach's alpha (α) was utilized in running the reliability test. A Cronbach's alpha value < 0.6 is considered poor, 0.61–0.7 is considered questionable, 0.71–0.8 is acceptable, 0.81–0.9 is considered good, while more than 0.9 represents an excellent level with high reliability [75]. Based on reliability tests, all data used for measuring the effectiveness of Safe City initiatives in reducing street crime and fear of crime are excellent, with Cronbach's alpha values ranging from 0.912 to 0.959 (see Table 4). These values indicate that all the scales contain internal consistency, and are therefore considered reliable for use.


**Table 4.** Reliability and internal consistency of data in reducing street crime and fear (source: authors).

Note: CBD stands for central business district, CCTV stands for closed-circuit television, CPTED stands for crime prevention through environmental design, CPSD stands for crime prevention methods through social development, GIS stands for geographic information system, ICT stands for information and communication technology, and SCP stands for safe city program.

Subsequently, Pearson's Correlation through Bivariate Analysis was generated to test the relationship between the variables [75]. A value that measures the strength of a relationship is the correlation coefficient, r, otherwise known as Pearson's r. The r values between 0.3 and −0.3 indicate a weak relationship, values from 0.3 to 0.7 and from −0.3 to −0.7 indicate a moderate linear relationship, and values from 0.7 to 0.9 and from −0.7 to −0.9 are considered strong.

Next, inferential analysis in the form of regression analysis was performed to test the significant contributing factors of preventing crime and the fear of crime. The statistical significance was examined at the traditional *p*-value of less than 0.05. Multiple correlation coefficient, R, measures the quality of the prediction of the dependent variable. In other words, simple linear regression was used to estimate the dependent variables or outcomes of reducing street crime and fear of crime based on the Safe City initiatives as the independent or predictor variables. This analysis also determined the overall fit or variance explained of the model and the relative contribution of each Safe City initiative.

### *3.4. Research Ethics and Survey Research*

In the majority of cases, survey research incorporates requests for respondents to offer personal information which is probably otherwise unavailable [64]. Furthermore, underage respondents were also involved in this study. Therefore, all the data collected from respondents will not be publicly disclosed in order to protect their privacy, and to avoid any possible psychological distress. Moreover, respondents were given an information sheet to read, and signed a consent form guaranteeing the confidentiality of data provided to this survey. The study received ethical approval from the Royal Malaysia Police.
