Measuring Crime: Crime Forecasting System - Dissertation Example

Measuring Crime: Crime Forecasting System Introduction: The department of police is responsible to safeguard the security of the residents of an area. Keeping an eye over any of the suspected activities in the vicinity is one of the responsibilities of the police. The current report revolves around the data collected by the police during their stop and search operations. There are a number of reasons why the analyzing of data collected by the police could be of integral importance. The most important aspect is that it may be used to assess the trend of crime within the area concerned. The collection of data and its proper analysis could be for the benefit to the policy and the community as a whole owing to the following reasons: 1. The increasing or decreasing trend in crime can be assessed and recorded. 2. Finding out and assessing the types of crimes taking place and the areas in which their concentrations converge/ disperse. 3. The identification and quantification of the victims of crimes trending in specific areas. 4. The identification of specific graphical regions with their ratio of crime happenings and the possibility for the same can also be assessed by the collection of data. Surveying of the possible victims of crime is one of the means of data collection. This is exactly the means that has been deployed for the conduction of the intended analysis. Background and Purpose of Research: The increasing rates of crime and the sense of insecurity prevailing among the residents of Lynfield provoked the police of the area to conduct stop and search operations. The objective and purpose was to analyze the stop and search activity that the police had conducted. They wanted to evaluate the data they had collected so that it may be found out whether it was any good for the detection and prevention of crime within the area. Data Description: The data currently at hand is obtained via a generic stop and search operation conducted. The data comprise of more than 10,500 records. Each record contains 22 variables. The variables are chosen to record a comprehensive picture of stop/search operation. One of the variables i.e. SusAge (Age of Suspect) is considered as one of the independent variables. Another important data variables that may also be considered as independent variables are OffAge (Age of duty Officer that conducted Stop/Search operation). A stop/search record gives the overall picture of the recorded activity like time and day of search and profile of suspect etc. Missing values are indicated as -9.SPSS is used to conduct the statistical analysis of data. There are several research questions the answers to which can be obtained from the collected data. The explanatory picture of data variables/parameters used in this research is presented in Table 1.0. Table 1.0: Explanation of Data Variables Variable Name Explanation Measure - Nominal/Ordinal 1 StopType The variable denotes whether the individual whose data is recorded was only stopped or was searched as well. This shows that some individuals may also have been left without being searched. Nominal 2 StopDay The day of the week on which the individual was stopped/ searched. Nominal 3 StopTime Time of the day at which the stop or search was conducted. Nominal 4 StopHours The time duration taken for the conduction of stop and search (each time slot was recorded in time recoded into 4-hour periods) Nominal 5 StopLocation Location at which the stop/search operation was conducted. Scale 6 SusGender This variable is used to identify the gender of the suspected person. Nominal 7 SusAge This variable is used to identify the age of the person being suspected. Ordinal 8 SusWork The employment of the person being suspected is being shown via this variable. Scale 9 PrevRecord This variable is used to denote whether the suspect being stop/searched has a previous history/record with the police or not. Nominal 10 SusActivity The activity in which the suspect was involved prior to being subject to stop/search. Nominal 11 Reason This variable denotes the reason for which the suspect for stop/search was held. Scale 12 Authority This variable denotes the Authority for stop/search Nominal 13 IC_Code This variable denotes the ethnicity of the person being stopped/searched as assessed by the Officer Nominal 14 SDIC_Code This variable denotes the ethnicity of the person being stopped/searched as stated by the person himself. Nominal 15 Weapon The variable denotes whether the suspected person was carrying a weapon or not. Scale 16 Arrested This variable denotes whether the suspect was arrested or not. Nominal 17 OffAge The age of the officer conducting the stop/search. Scale 18 Worry The variable denotes the degree of worry of the suspect regarding crime in their area. Scale 19 Complaint This variable denotes whether any complaint was made or not. Scale 20 Satisfied This variable denotes that if a complaint was made was the complainant satisfied with the response received? Scale 21 Trust This variable is used to record the response to the following query: Police can be trusted to deal fairly with all sections of community? Scale 22 AutoRecode The first step of Time Recoding to 4-hour periods Scale In order to achieve the optimal analysis of given data some of the variables are encoded as new variables. The encoding scheme used for “SusAge” variable is presented in Table 1.1. The name of new encoded variable is given as “SAgeCoded”. Table 1.1 : Encoding scheme used for Age of Suspect Code Class Interval 1 10 – 15 2 15.1 – 20 3 20.1 – 25 4 25.1 – 30 5 30.1 – 35 6 35.1 – 40 7 40.1 – 45 8 45.1 – 50 9 50.1 – 55 10 55.1 – 60 11 60.1 – 65 12 65.1 – 70 13 70.1 – 75 14 75.1 – 80 In order to establish two groups on the basis of age of officers involved in stop/search activity the variable “OffAge” is further encoded. The respective encoding scheme is presented in Table 1.2. Table 1.2: Encoding criterion for classification of officers on the basis of their ages Code Criterion Description 1 1 – 2 Officer under the age of 30 2 2.1 – 4 Officer over the age of 30 Research Questions and related Hypotheses: There can be several research questions that may arise or formulated on the basis of given data. For example at what times of the day/night were the suspects were stopped or being searched the most? Whether the suspects are stopped and searched without any bias like gender discrimination or racism? Etc. The profile analysis of Figure 1.0 clears the answers to the former. The two consecutive slots of 4 hours period each i.e. from 8:00 AM – 3:00 PM witnessed the highest number of stop/search. Figure 1.0 The three questions that are included in this report in order to formulate the respective hypotheses are stated below, Research Question 1: What is the age group in which most of the suspects fall? Is there a significant difference between the mean value of the ages of male suspects and the mean value of the ages of female suspects? Hypothesis 1 (Null and Alternate): Considering µ1 and µ2 to be the means of ages of male and female suspects the null and alternate hypotheses will be given as, H0: µ1 = µ2 H1: µ1 ≠ µ2 Rationale: The reason why this hypothesis is adopted is that the knowledge about significant difference between the age groups of male and female suspects may improve the focus of stop/search activity. Research Question 2: The young aged officers are generally proved to be more energetic but comparatively inexperienced as compared to their senior colleagues. Whether this is the case with this study or not? Hypothesis 2 (Null and Alternate): The null and alternate hypotheses in this case can be given as, H0: The frequency of stop/search is dependent on the age of duty officer. H1: The frequency of stop/search is independent of duty officer’s age. Rationale: In order to optimize the performance of young officers in identifying the suspects intermittent learning sessions coached by senior officers may prove to be effective. The analysis of this hypothesis may help to understand the need of such sessions. Research Question 3: Another research question that may be formulated here is whether the possession of weapon was linked to the suspect’s arrest or not. Hypothesis 3 (Null and Alternate): The null and alternate hypotheses in this case can be given as, H0: The arrest measures are totally independent of weapon possession. H1: There is a strong dependence of arrest on possession of weapon. Rationale: Testing this hypothesis would assist in figuring out the correct or main reasons of arrest of a suspect. Analyses and Discussions: Analysis 1: The coding scheme for suspects’ ages is presented in Table 1.0. The descriptive statistics created for variable “SAgeCoded” (Table 1.2) highlights the fact that mean value of suspects’ ages lies between 30 – 35 years. Whereas the mode indicates that the number of suspects aged between 25 – 30 years is the highest. The histogram presented in Figure 1.1 indicates a left skewed normal distribution with assertion of the above mentioned facts. Table 1.4 (Appendix A) presents the frequency distribution of suspect ages. The mean ages of female and male suspects are indicated through Table 1.3 labeled as “Group Statistics for ages of male and female suspects”. Table 1.2: Descriptive Statistics of Suspects’ Ages N Valid 10409 Missing 200 Mean 5.2766 Std. Error of Mean .02227 Median 5.0000 Mode 4.00 Std. Deviation 2.27257 Range 13.00 Table 1.3: Group Statistics Gender of suspect N Mean Std. Deviation Std. Error Mean Age of suspect Male 9769 34.60 11.356 .115 Female 645 31.01 9.951 .392 In order to analyze and test Hypothesis 1, the means ages of male and female suspects are compared under Independent Sample t – test. With 95% level of significance it was found that the p-value (0.000) < α. Hence the null hypothesis is rejected and it is concluded that there is a significant difference between the mean ages of male and female suspects. The test values are presented in Table 1.5. Figure 1.1 Table 1.5: Independent Samples Test Levene's Test for Equality of Variances t-test for Equality of Means F Sig. t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower Upper Age of suspect Equal var assumed 20.309 .000 7.829 10412 .000 3.588 .458 2.690 4.487 Equal var not assumed 8.788 759.129 .000 3.588 .408 2.787 4.390 Analysis 2: Table 1.7 assists in the analysis of 2nd hypothesis. It contains the frequency distribution for the reasons of stop/search. The apparent analysis of Table 1.7 and profile analysis of Figure 1.2 reveal that the major reasons behind the stop/search activity were “Officer Intuition” and “Suspicious actions of suspects”. Table 1.7: Frequency Distribution for Reasons for stop/search Frequency Percent Valid Percent Cumulative Percent Valid Officer Intuition 3325 31.3 31.3 31.3 Suspect acting suspiciously 3005 28.3 28.3 59.7 Called to Scene 1374 13.0 13.0 72.6 Prior Information 1464 13.8 13.8 86.4 Public Complaint 1441 13.6 13.6 100.0 Total 10609 100.0 100.0 Figure 1.2 A general analysis of officers’ ages involved in stop/search activity reveals that there had been less stop/search activity by officers that were in the age group higher than 30 years. Table 1.8 highlights the figures. The subsequent Figure 1.3 indicates the percentage share of both the groups as per the encoding mentioned in Table 1.2. Figure 1.3: The percentage share of both the age groups of officers in stop/search With chi square statistic χ2 = 10609 > critical value (7.8), level of significalnce (0.05), and degrees of freedom (df =3) and p – value 0.00 the null hypothesis that the stop/search frequency is dependent on the age of duty officer is rejected. The contingency table (Table 1.9) chi-squared test results (Table 1.10) indicate the rejection of null hypothesis clearing that the frequency of stop/search is not dependent on the age of officers. The lesser frequency for higher age officer can be due to the less number of respective duties. Table 1.9: Age of Officer involved in stop * CodeOffAge Crosstabulation Count CodeOffAge Total 1.00 2.00 Age of Officer involved in stop 20-25 4650 0 4650 26-30 4176 0 4176 31-35 0 1733 1733 Over 35 0 50 50 Total 8826 1783 10609 Table 1.10: Chi-Square Tests Value df Asymp. Sig. (2-sided) Exact Sig. (2-sided) Exact Sig. (1-sided) Point Probability Pearson Chi-Square 10609.000a 3 .000 .000 Likelihood Ratio 9607.605 3 .000 .000 Fisher's Exact Test 9579.972 .000 Linear-by-Linear Association 6519.816b 1 .000 .000 .000 .000 N of Valid Cases 10609 a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 8.40. b. The standardized statistic is 80.745. Analysis 3: With chi square statistic χ2 = 6.909 < critical value (9.48), level of significalnce (0.05), and degrees of freedom (df =4) and p – value 0.14 the null hypothesis that the The arrest measures are totally independent of weapon possession is not rejected. The contingency table (Table 1.11) chi-squared test results (Table 1.12) indicate the acceptance of null hypothesis indicating that the possession of weapon did not surely ended up in suspect’s arrest. This means that the arrest action for a suspect is not dependent on the possession of weapon. Table 1.11: Was suspect carrying a weapon? * Was suspect arrested? Crosstabulation Count Was suspect arrested? Total -9 No Yes Was suspect carrying a weapon? No Weapon 498 9177 771 10446 Knife 0 48 0 48 Gun 0 2 0 2 Total 498 9227 771 10496 Table 1.12: Chi-Square Tests Value df Asymp. Sig. (2-sided) Pearson Chi-Square 6.909a 4 .141 Likelihood Ratio 12.919 4 .012 Linear-by-Linear Association 1.675 1 .196 N of Valid Cases 10496 a. 5 cells (55.6%) have expected count less than 5. The minimum expected count is .09. Further Discussion, Conclusions and Recommendations: There are other facts that are to be considered as well along with the proven results. Following is the discussion regarding those facts. It is observed that the number of professionals stopped or searched is very low as compared to unskilled, semi skilled and skilled workers (Figure 1.4). Only 707 suspects were professionals among more than 10500 suspects that were stopped and searched. Figure 1.4 The gender of the duty officer is not mentioned in the data. Therefore it is difficult to observe that whether the stop/search activity was gender biased or not. However the figures told that there were just 6.1% female suspects. Assuming that the majority of the duty officers were male the scenario seems to have contained a bias regarding gender (Table 1.13). Table 1.13: Frequency of Gender of suspect Frequency Percent Valid Percent Cumulative Percent Valid Male 9811 92.5 93.8 93.8 Female 647 6.1 6.2 100.0 Total 10458 98.6 100.0 Missing -9 151 1.4 Total 10609 100.0 Following may be concluded after the analysis, There is a significant difference between the mean ages of male and female suspects. This asks for a different focus during observation of different genders. The age of officer does not have any significant affect on suspects’ stop/search. The senior officers however must share their experiences as the major reason of stop/search is appeared to be the duty officer’s intuition. The possession of weapon is not the reason of arrest in most of the cases. The findings of this report recommend that The time slot of 8:00 AM – 3:00 PM must be duly addressed for stop/search activity due to the high number of suspects. The data must include the reasons of arrest and acquittal of suspects. The data must mention the gender of duty officer in order to properly figure out the gender biases in the process. The ethnic background of duty officer must also be recorded. There is a very weak relationship between suspect having previous criminal record and arrest figures/data. This needs a thorough emphasis on maintain a stop/search protocol without dependence on suspects previous criminal record. References: MEENAI, Y, MEENAI, T, TEHSIN, A, & ILYAS, M 2011, 'Crime Forecasting System (An exploratory web-based approach)', Journal Of Systemics, Cybernetics & Informatics, 9, 4, pp. 87-90, Academic Search Complete, EBSCOhost, viewed 18 January 2013. MOORE, D. S., & MCCABE, G. P. (1993). Introduction to the practice of statistics. New York, Freeman. WAGNER, S. F. (1992). Introduction to statistics. New York, HarperPerennial SALKIND, N. J. (2000). Statistics for people who (think they) hate statistics. Thousand Oaks, Calif, Sage Publications, Inc. BAUM, S., GABLE, R. K., & LIST, K. (1987). Chi square, pie charts and me. Monroe, N.Y., Trillium Press, Inc. FOX, J. A. (1978). Forecasting crime data: an econometric analysis. Lexington, Mass, Lexington Books. Appendix A Table 1.4: Frequency Distribution SAgeCoded Frequency Percent Valid Percent Cumulative Percent Valid 1.00 71 .7 .7 .7 2.00 803 7.6 7.7 8.4 3.00 1690 15.9 16.2 24.6 4.00 1819 17.1 17.5 42.1 5.00 1658 15.6 15.9 58.0 6.00 1578 14.9 15.2 73.2 7.00 1104 10.4 10.6 83.8 8.00 678 6.4 6.5 90.3 9.00 462 4.4 4.4 94.8 10.00 307 2.9 2.9 97.7 11.00 144 1.4 1.4 99.1 12.00 66 .6 .6 99.7 13.00 19 .2 .2 99.9 14.00 10 .1 .1 100.0 Total 10409 98.1 100.0 Missing System 200 1.9 Total 10609 100.0 Read More