UBRAhY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 2/05 c:/CIRCIDateDuo.indd-p. 15 mu‘i’apet” .. ' - P oflotsky, Greg F. 1998 The Effect of Educational Expenditures on Crime & Juvenile Arrest Rates in Michigan Cities Plan “B” Thesis by Greg F. Orlofsky Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Master in Urban & Regional Planning - Urban Studies Spring 1998 ACKNOWLEDGMENTS I wish to extend my gratitude to the following groups and individuals for their assistance in compiling and analyzing the necessary data: Greg Winter; Christina Dejon g, Ph.D;; Glenda Rader, Michigan Department of Education, Office of Administrative Services; Bernadette Scarborough, Michigan Department of State Police, Central Records Division (Uniform Crime Reporting); and the Michigan Information Center. 7 . 'j‘ ”1;. Special thanks to Zenia Kotval, AICP, PhD. and James Mertes, PhD. for their help in preparing this document. And finally, I wish to express my sincerest appreciation to my friends and family for their continuing support. TABLE OF CONTENTS EXECUTIVE SUMMARY I. Introduction 11. Purpose III. Methodology IV. Background on “Proposal A” V. Education, the Community and Crime VI. VII. VIII. IX. A. Educational Indicators B. Education and Economic Activity in States C. Involving the Public More in Education D. Education as a Deterrent to Criminal Activity Data Collection Data Adjustment Correlations A. Results from Correlations B. Discussion Regressions A. Results from Regressions B. Discussion Conclusions Page Nmber 11 11 11 12 13 17 21 30 33 53 55 56 59 60 Bibliography APPENDIX A - list of Variables APPENDIX B — Population Estimates APPENDIX C - Descriptive Statistics APPENDIX D - Distributions APPENDIX E - Correlation Output APPENDIX F - Regression Output Page Number 61 A1 A4 A6 A16 A32 EXECUTIVE SUMMARY In 1994 the Michigan legislature passed “Proposal A” which abolished property taxes as the primary source of funding for public education and raised the State Sales Tax by 2%. The change in the tax structure means that all consumers are now contributing to the generation of funds for public education. The purpose of this paper was to discover whether or not communities, in general, benefit from increased levels of spending on local education through a reduction in crime and juvenile arrest rates. It has been argued that changes in educational spending per student is not an adequate means of measuring student performance and that testing scores (i.e. ACT, SAT, etc.) would be better indicators, however, for the purpose of this paper, spending per student is the appropriate variable, as student performance is not in question. Previous efforts to show linkages between crime and education have been fairly unsuccessful, still, no attempts have been made to link the actual amount of spending on education to crime. The unit of analysis in past studies seems to have always been the individual, rather than the community. Moreover, recent studies do imply that education increases the productivity of workers and that productivity can, in turn, lead to a reduction in crime. Data for the study was obtained from the Michigan Department of Education and Michigan Department of State Police. Crime/juvenile arrest variables were adjusted using population estimates to develop rates. Expenditure data was adjusted for inflation. There were five variables used in the study. Current Operating Expenditures per Pupil was the independent variable. Index Crime Rate, Non-Index Crime Rate, Total Crime Rate and Juvenile Arrest Rate represented the dependent variables. Scatterplots accompanied by a fit—line provided a visual indication as to whether or not these variables were, in fact, related. These inferences were confirmed using statistical analysis (i.e. Pearson’s correlation coefficient r and a t-test). The level of significance for the null hypothesis to be rejected was < .05 (95% confident that there was not a Type I error). Analysis for those correlations found to be statistically significant, proceeded further with regressions to determine the extent of the relationship between the two variables (coefficient of determination r’) and the regression equation to see exactly how crime rates vary with increases in operating expenditures. . The results seemed to indicate that communities with large amounts of expenditures are just as susceptible to crime and juvenile arrests as are communities with relatively low expenditures on education. Comparing % change in expenditures with % change in crime/juvenile arrest rates also produced insignificant results. When a “five-year lag” was incorporated into expenditures there appeared to be a clear relationship between the two variables. Although, still statistically insignificant, the likelihood of a Type I error had fell considerably. Finally, when a “ten-year lag’ was incorporated there were three statistically significant correlations between % change in current operating expenditures and % change in the index crime rate, non-index crime rate and total crime rate. This implies that there is a relationship between educational expenditures and crime/juvenile arrest rates, but only in the long term. There was no relationship between changes in the juvenile arrest rate and expenditures, even with regard to the longest lagged correlation. This was attributed to the fact that increased expenditures help students to form a more law-abiding community when they are older. After all, juveniles, would cease to be juveniles after ten years. The fact that there was no relationship in the non-lagged correlations would seem to indicate that a community doesn’t benefit from increased expenditures on education until juveniles have grown and become part of the adult community. Changes in the Non-Index Crime Rate appears to be affected most by changes in expenditures, followed by the Total Crime Rate and finally the Index Crime Rate. The coefficient of determination (r’) for the Non-Index Crime Rate was 16.6% (i.e. explained variation). While this may not seem significant , the goal here was not to “explain” crime rates. The purpose was to determine whether or not, educational expenditures alone can have any impact on crime rates. The results from the regressions indicate that there is, in fact, a relationship between the two. Whether the relationship between the variables found to be related is truly a causal one may remain suspect. Since the unit of observation had to be consistent for the study, the sample was in effect, non-random. It may be inappropriate to generalize the findings of this study to cities that did not meet the criteria for the sample. Nonetheless, the very fact that a relationship was found to exist between the variables, even if only in lagged correlations, serves as a step in the right direction with regard to showing the way in which public education is, in fact, a benefit to all. I. Introduction Traditionally, local property owners have contributed the most revenue to public education. In 1994 the Michigan legislature passed “Proposal A” which abolished property taxes as the primary source of funding for public education instead raised the State Sales Tax by 2%. This was done to equalize both the amount of revenue that is going to each of the school systems in the State and the way in Which revenue is generated (provide property tax relief and make revenue generation more equitable). With the addition of the 2% increase in the Sales Tax, which goes towards the public schools, all consumers are in effect contributing to the generation of funds for public education. While education has, for the most part, stayed out of the planning arena, this Legislative act has implications for not only students and parents, but for the rest of the community as well. The purpose of education, as most see it, is to provide children with an adequate knowledge base so that they may become willfully employed. It seems safe to assume that the benefits parents derive in paying for the school system is that their children are educated. Very often non-parent property tax-payers contend that they derive no benefit from having other people’s children educated and therefore should not be held responsible for funding schools. With the introduction of Proposal A, all members of the community are now contributing, to a certain extent, revenue for the public school system. It is my contention that there does exist a benefit to the community from public education. 1]. Purpose The purpose of the paper is to discover whether or not communities, in general, benefit from increased levels of spending on local education through a reduction in crime and juvenile arrest rates. Crime data will be correlated with figures on educational expenditures to determine which, if any, are correlated with one another, if the relationship is positive or negative, and finally which indicator is most affected by changes in spending per pupil. For those variables that do exhibit a correlation, regressions will be run to determine to what extent they are related and their regression equation. There are many implications for this research. If, for instance, education is found to contribute to the rest of the community through a reduction in crime rates, then equity issues about the way in which school’s are funded (i.e. whether there should be less of a reliance on property owners) could be addressed by future legislation. Although the focus here is the benefit to a community from public education, if there is in fact a relationship between expenditures and crime rates, that variable could be used in modeling crime rates for multiple regressions. III. Methodology The study began with a more in-depth investigation into .“Proposal A” followed by a review of existing literature to demonstrate the linkages between education, the community and crime. Next, data on school expenditures, crimes and juvenile arrests were collected from various state agencies. Once compiled, crime data was adjusted using population estimates so that “rates” could be developed. Educational expenditures data was also adjusted to reflect inflation. Finally, the data was aggregated into five-year time periods so that “lags” could be incorporated into the study. Analysis of the data began with correlations between crime/juvenile arrest rates and inflation adjusted expenditures. The following general hypotheses were tested: Ho: The crime/juvenile arrest variable (independent) and the variable on educational expenditures (dependent) are INDEPENDENT events (crime/juvenile arrests (10—th depend on the educational expenditures). HA: The crime/juvenile arrest variable (independent) and the variable on educational expenditures (dependent) are INDEPENDENT events (crime/juvenile arrests d_o depend on the educational expenditures). Hypotheses were “non-directional” to account for the possibility of a positive relationship between the variables. Scatterplots accompanied by a fit-line provided a visual indication as to whether or not the variables were, in fact, related. These inferences were confirmed using statistical analysis (i.e. Pearson’s correlation coefficient rand a t-test). The level of significance for the null hypothesis to be rejected was < .05 (95% confident that there was not a Type I error). Analysis for those correlations found to be statistically significant, proceeded further with regressions to determine the extent of the relationship between the two variables (coefficient of determination r’) and the regression equation to see exactly how crime rates vary with increases in operating expenditures. The final section suggests directions for future research. The methodology and remaining sections of the paper are then ordered as follows: “Proposal A” L Education, the Community and Crime L Data Collection L Data Adjustment L Correlations (discussion) L Regressions (discussion) L Conclusions (future research) IV. Background on “Proposal A” Prior to 1973, Michigan used a “Minimum Foundation Approach” in which the State guaranteed a per pupil cost to local school districts which depended upon the local districts levying the state-determined tax rate. The State paid the difference between what was generated under this tax rate and the guaranteed per-pupil cost. If they levied less that the State tax rate, then local districts received less than the guaranteed amount. In 1973, the State switched to the “District Power Equalization Approach” which provided a guaranteed revenue yield and paid each district below this yield the difference in the form of State formula aid payments. If a district’s local revenue per pupil per mill exceeded the State’s guaranteed revenue under this formula, the district was “out-of-formula” and thus received no state aid. The benefit of this system was that school districts had the power to raise more revenue than was guaranteed by the State by levying a higher millage rate. This gave school districts more “control” over the amount of funds that their schools received, and the only disadvantage to them was that they were unable to receive state aid. The drawback of this system, however, was that it produced great wealth disparities between districts. Wealthier districts that enjoyed a high property tax base could raise large revenues with a low property tax rate, while poorer districts with a low tax base raised less money even after levying high rates. In other words, where a family lived largely determined the quality of its children’s education. Between 1972-1987, Michigan voters rejected nine of ten proposals to alter taxes, three of which involved education. One election in particular in 1981 involving a plan called Proposal A— a property tax cut tied to a sales tax increase was defeated by a 3-1 margin. For fifteen years people were debating about property taxes and education and yet nothing had been resolved (Christoff, 9/27/87). On February 10, 1987 the Michigan Citizens PrOperty Tax Commission released a 38 page report recommending long-term changes in the property tax system, especially with regard to the financing of elementary and secondary education. It recommended an increase in the sales tax to 6% and advocated that the state should guarantee a $3,000—per—pupil minimum (similar to the recommendations made in 1981). The report said that the minimum amount would ensure that all students in Michigan receive a basic quality education regardless of theeconomic wealth of their jurisdiction. The report suggested that local property taxes in support of schools be lowered from the current average of 32 mills to an average of less than 20 mills (1/ 1000 of the taxable property value) (Jones, 2/10/87). The purpose of this proposed legislation was not only to reduce property taxes for homeowners and business owners but to reduce the enormous gap in funding that existed between school districts. Whitefish School District in Chippewa County had the highest level of current operating expenditures per pupil for the 1985-86 school year with $6,208. Kingsley Area School District in Grand Traverse County ranked lowest (525“) with $2,107. This amounted to almost a 3-1 ratio between the highest and lowest funded schools. A key recommendation in the Commissions’ report was the increase in the statewide Sales Tax from 4% to 6%. The increase in sales tax revenue would be pulled and redistributed to all school districts by the state so that wealthier districts would no longer be able to create such disparities (DFP, 9/24/87). Wealthier districts were against the proposal, arguing that their schools were being punished just because they had a natural advantage. State Treasurer Robert Bowman also objected to the proposal, saying it could cause a huge flow of tax revenues to Washington, since property taxes are deductible on federal income tax returns and sales taxes are not. Others claimed that sales taxes are regressive, and consequently, the poor would suffer the greatest burden. Since many of the poor do not own their own homes, property tax cuts do not provide any kind of relief, conversely the poor do spend a relatively large amount on consumer goods that are charged sales tax (Stroud, 5/9/93). Proposal A, first voted on in 1981, reemerged as a special issue twelve years later in 1993 and was defeated. In July 1993, without regard to the consequences, the Michigan State House and Senate overwhelmingly approved a $5.6 Billion property tax cut (PA 145 of 1993) without identifying replacement funds for the school system (Andrews, 7/23/93). Many schools were left wondering how they would be able to function the following school year. In Eaton Rapids, property taxes made up about 58% of the school budget—approximately $7.8 million. In East Lansing, 94% or nearly $23 million of the school budget was generated through property taxes all of which was reduced as a result of the cuts (Iorio, 7/23/93). On March 15, 1994, voters were asked once again to choose between alternative plans to fill the hole in funding left by the property tax cut eight months earlier, Proposal A was passed by a 69-31margin (Kearney, 1994). The following, outlines the specific changes made in the taxing system as a result of Proposal A: Sales Tax Income Tax Property tax (mills): Homestead Second homes Comm. & Ind. Enhancement ISD’s Assessment cap Property transfer tax Single business tax Cigarette tax Out-of-state calls Personal income Tax exemption Pre-propgsal A 4% 4.6% 34 (average) 34 (average) 34 (average) N/A 3 (average) N/A .0011% 2.35% 25 cents 4% $2,100 Proposal A 6% 4.4% 6 24 24 3 3 (average) 5% or CPI 2.0% 2.35% 75 cents 6% $2,100 Source: A Primer on Michigan School Finance, C. Philip Kearney 10 V. Education, the Community and Crime A. Educational Indicators It could be argued that changes in educational spending per student is not an adequate means of measuring student performance and that testing scores (i.e. ACI', SAT, etc.) would be better indicators when trying to establish a correlation between education and other variables (in this case crime and juvenile arrest rates). Still, many schools are often accused of excluding low scoring children from testing and focusing instruction on the skills measured by tests only ('Murname, 1988). This would make the level of spending a more attractive variable. For the purpose of this paper, spending per student is the appropriate variable, as student performance is not in question. Students are not the unit of observation, rather it is communities and the benefit each derives from the level of spending. If spending levels do not mirror performance, then students, in turn, will not be productive in the future, which will only prove that spending on education has no positive effect on communities. Despite the comments of Former U.S. Secretary of Education , William Bennett who contended that there was not a strong correlation between school spending and achievement referring to earlier research in an article “The Economics of Schooling: Production and Efficiency in Schools” (Hanushek, 1986); more recent findings suggest the relationship between spending and achievement was much higher than previously thought. Educational levels not “gains” were the focus of the earlier study which did not address what students had learned (Baker, 1991). B. Education and Economic Activity in States According to one study (Quan et al., 1987), the level of educational services may affect economic growth in a state in two ways. Potential migrants may choose to locate in states that 11 have a higher quality of education. This, in turn, contributes to a growing population, increasing both the supply of labor and the demand for local goods. They have termed this “parental migration effect” (Quan, 1987, 361). And secondly, as noted by other studies (Mandi, 1981, Teng, 1991, and Psacharoopoulos, 1993) with regard to the effects of education on nations, education may increase the productivity of workers. This productivity effect should raise wage rates in a state, however the studies revealed that increases (or decreases) in the quality of education can only be expected to have an impact on nations after 4 to 9 years and used lags in their regression to adjust for this contingency. Their findings seemed to indicate that the effects of eduCational expenditures on the levels of wages and employment differ in the Northeast and the Sunbelt. Education expenditures have positive and significant effects on the levels of wage and employment in the Northeast, while the reverse is true in the Sunbelt. These “Northeast states” included, among others, Illinois, Ohio and Wisconsin, but not Michigan (Quan, 1987). C. Involving the public more in Education There is a fear that schools are becoming disconnected from the public. One of the main criticisms of Proposal A is that it has taken away “local control” of the schools and put it in the hands of the State. State dollars now provide 75% of the revenue schools receive with local revenues making up approximately 19% and the remaining 6% from the national government (Kearney, 1994). As part of Proposal A’s mission, wealthier districts are less capable of raising their millage rates and increasing the amount of local funds available to schools. For some, problems with schools seem to be a question of “legitimacy” rather than effectiveness of the public schools. There are a great many people who don’t believe that the 12 public schools are their agents, who don’t believe that the public schools are responsive to their concerns. Consequently, the public needs to get more involved (Danzberger et al., 1994). Social problems that affect student achievement can be addressed only if schools, families, and communities work together. And yet budget battles and voucher movements attest to the public’s growing disenchantment with the public school system. The issue of legitimacy lies in the fact that people don’t recognize that schools serve a public purpose, that everyone benefits not just parents. The solutions they suggested involve focus groups and town meetings to discuss the state of local education (Mathews, 1997). If linkages between community and spending on public education exist, it could be a catalyst in increasing the amount of community involvement the authors above believe are necessary to increase student achievement because communities will have a proven stake in the effectiveness or “legitimacy” of schools. D. Education as a deterrent to criminal activity For years, the widespread assumption about the role of schools is that they function as a positive form of social control and act as a deterrent to delinquent behavior. Schools provide an important environment in which children learn to be law-abiding, and acquire the qualifications that give them the opportunity to have a stake in society (Gilling, 1997). Those who do not perform well, and perhaps even drop out, are presumed to be less likely to obtain employment, which, in turn, could lead to criminal behavior. Assumptions, such as these, have led to policy making to increase educational opportunities for young people and encourage those who have already “dropped out” to return to school, all in an effort to reduce delinquency and crime (Wolfgang et al., 1987). 13 Increasing amounts of delinquent behavior, both in and outside of schools, has been attributed by some to impersonal atmosphere created there as a result of budget crunches which have led to the consolidation of school districts, large classes, and other factors which have reduced opportunities for positive social interaction between school personnel and students (Kratcoski, 1990). The majority of young men and young women participate in some kind of delinquent acts during their juvenile years. As one researcher noted, “By the age of 18 possibly over 90 percent of young males have participated in delinquent acts. . .50 to 60 percent of young females have been involved in delinquent acts by the time they are 18” (W itte, 1997, 219). Still, for most individuals, criminal activity is usually restricted to the teen years and those individuals who began offending as juveniles have usually stopped by their mid-twenties. According to the National Crime Survey, the level of crime today is lower than it was in the late 1970’s and early 1980’s for crimes such as rape, aggravated assault, burglary, and larceny, as well as less-serious (non-index) offenses (W itte, 1997). Juvenile arrests, however, have been on the rise, particularly in Michigan. Between 1990 and 1994 juvenile arrests in the state increased by nearly 8%. Of these approximately 47% were index crimes1 (Stoetzer et al., 1997). In 1996, the Institute for Public Policy and Social Research (IPPSR) at Michigan State University conducted its State of the State Survey (SOSS) of adult residents in Michigan. One of its goals was to gather information on perceptions of the causes of juvenile delinquency”. A listing of index crimes appears on page 19. 2 “Juvenile” for this study meant persons under the age of 18. 14 Irresponsible Parents Gangs No Punishment Crinind Parents Poor Justice System Poor Schools Not Enough Jobs Single Parents Factors which Michigan residents believe contribute “a great deal” or “quite a bit” to why Perceived Causes of Delinquency 0% _, 79% 70% '* 53% 50% . -— 53% - «.1 53% 20% 40% 50% 80% .A Great Deal .Quite a Bit some teenagers are juvenile delinquents. MSU State of the State Survey (SOSS) Spring 1996; N = 1133; Sampling Error = 2.9% Michigan State University, lPPSR As the figure shows, people seemed to believe that the principle reasons for delinquency were irresponsible parents (79%) and gangs (70%) (these factors contributed “a great deal” or “quite a bit” ). While it was not at the top of the list, “poor schools” tied for fifth as a perceived influence of juvenile delinquency (53%) (Stoetzer et al., 1997). Despite all the assumptions and perceived influences, there is little evidence that education and criminal activity are related. Education has apparently not been analyzed in any great detail in correlational studies of crime. When studies did include education as a variable (measured in terms of grade completed, possession of a high school diploma, or scores on tests), researchers 15 concluded there was so significant relationship to crime. Some correlational studies, however, did find a statistically significant inverse relationship between the amount of time spent and the level of criminality. . These results have been interpreted as indicating the importance keeping young people in school and off the streets (Witte, 1997). While efforts to show linkages between crime and education have been less than startling, it appears that no attempts have been made to link the actual amount of spending on education to crime. Moreover, the unit of analysis in past studies seem to have always been the individual, rather than the community. Recent studies do imply that education increases the productivity of workers and that productivity can, in turn, lead to a reduction in crime. Increases Reduction Education “'9 Productivity _) in Crime It is with this assertion, that this study shall continue. 16 VI. Data Collection The purpose of the study is to compare data on school expenditures per student with data on various crimes and juvenile arrests. Crime/juvenile arrest data is available on an annual basis at the city level; while annual data on expenditures per student is available only for each school district. School districts may cut across city boundaries and may include more than one city. Cities may, in fact, have more than one school district operating within their city limits. The unit of analysis must remain consistent across variables, therefore cities were chosen based on the following criteria: 1) Current Operating Expenditures per Student (COES) data exists for each city/school district selected for years 1974-93 (i.e. the school district has operated since at least 1974). 2) There is only one school district for each city. 3) The school district entirely surrounds the city boundaries so that the figures to which it will be compared (crime rates) are consistent with the data on expenditures. 4) Crime data is available for cities for years 1982-1993. While this does represent a non-probability (purposive) sample, and therefore has certain drawbacks with regard to the generalization of the results, using a probability sample (i.e. simple random sample, systematic random sample, stratified random sample, etc.) would undoubtedly produce cities in the sample population that would not fit the criteria and not be consistent with the unit of analysis. The data had to be available for the range of years mentioned above so that time-periods and lags could be developed. This will be outlined in greater detail below. The following is a list of thirty-nine cities with their corresponding school district chosen for the study: 17 CITY Allen Park Alpena1 Benton Harbor Berkley Birmingham Cadillac‘ Clawson Dearbom East Lansing Ecorse Escanaba1 Femdale Garden City Hamtramck Harper Woods Hazel Park Highland Park Kentwood Lincoln Park Menominee‘ Mt. Clemens Muskegon Heights Norton Shores Novi Oak Park Owosso Portage Romulus Roseville Royal Oak Southfield Southgate Sterling Heights Taylor Traverse City1 Trenton Troy ‘ Westland Wyandotte 1 DIST'RI NAME Allen Park Public Schools Alpena Public Schools Benton Harbor Area Schools Berkley School District Birmingham City School District Cadillac Area Public Schools Clawson City School District Dearbom City School District East Lansing School District Ecorse Public School District Escanaba Area Public Schools Femdale City School District Garden City School District Hamtramck Public Schools City of Harper Woods Schools Hazel Park City School District Highland Park City Schools Kentwood Public Schools Lincoln Park Public Schools Menominee Area Public Schools Mt. Clemens Community Schools Muskegon Heights School District Mona Shores Public School District Novi Community School District Oak Park City School District Owosso Public Schools Portage Public Schools Romulus Community Schools Roseville Community Schools School District City of Royal Oak Southfield Public School District Southgate Community School District Utica Community Schools Taylor School District Traverse City Area Public Schools Trenton Public Schools Troy School District Wayne-Westland Community School District Wyandotte City School District An intermediate school district is also located in these cities which directs expenditures for other services (special education, etc.). All public schools have intermediate school districts like these that direct these operations for a region. 18 Data on school expenditures was obtained from the Michigan Department of Education’s, “Ranking of Michigan Public School Districts by Selected Financial Data: Bulletin 1012” for years 1974-75 through 1992-93. The publication includes several general fund expenditure categories for each of the school districts in the state. The category chosen for this study was Current Operating Expenditures per Pupil (COEP). This category represents the closest approximation of the dollar amount devoted to each pupil in each school district and does not include payments for community services and capital outlay. Crime and Juvenile arrest variables were taken from the Michigan Department of State Police “Crime in Michigan: Uniform Crime Reports” for the years 1982-93. Four variables were used to compare to educational expenditures: Index Crimes, Non-Index Crimes, Grand Total Crimes, and Juvenile Arrests. Index crimes include the following: Murder & Non-negligent Manslaughter Rape Robbery Burglary larceny Motor Vehicle Theft Arson ‘ In 1980 the subheading of the publication was changed from Bulletin 1012 to Bulletin 1014. 19 Non-index crimes include the following: Negligent Manslaughter Assault (Non-aggravated) Forgery & Counterfeiting Fraud Embezzlement Stolen Property Vandalism Weapons (carry, possession, etc.) Prostitution & Common Law Vice Sex Offenses (except rape & prostitution) Narcotic Laws Gambling Family & Children Driving Under Influence Alcohol or Narcotics Liquor Laws Disorderly Conduct All Other (includes drunkenness & vagrancy) Total crimes are the sum of all index and non-index crimes and juvenile arrests includes arrests of all individuals ages 16 & under. VII. Data Adjustment Expenditure data was adjusted for inflation using annual % change rates from a Consumer Price Index (CPI) for all urban consumers. CONSUMER PRICES - ALL URBAN CONSUMERS 1970 THROUGH 1998 (1982-84 a 100) CALENDEFI u.s. PERCENT YEAR CONSUMER CHANGE PRICE INDEX 1970 38.8 5 9 1971 40.5 4 3 1972 41.8 3 3 1973 44.4 8 2 1974 49.3 11 o 1975 53.8 9 1 1978 58.9 5 8 1977 80.8 8 5 1978 85.2 7 8 1979 72.8 11 4 1980 82.4 13 5 1981 90.9 103 1982 98.5 82 1983 99.8 3 2 1984 108.9 43 1985 107.8 3 8 1988 109.8 1 9 1987 113.8 3 8 1988 118.3 4 1 1989 124.0 4 8 1990 130.7 5 4 1991 138.2 4 2 1992 140.3 3 o 1993 144.5 3 0 1994 148.2 2.8 1995 152.4 2.8 1998 158.9 3.0 Source: U.S.DeparunentotComrnerce,BweeuothorStatlstlca 21 Formulas for adjustment of the data are shown below: Adjusted cores 1975 = (COEP 1975)/(1 + % Change in CPI from 1974 to 1975) = (COEP1975)/(1+0.091) -.- (COEP1976)/1.091 Adjusted COES 1976 = (COEP1976)/{(1.091)*(1 + 0.058)} Adjusted C0E51977 = (COEP1977)/{(1.091)*(1.058)*(1 + 0.085)} And so forth, through 1993 in which the adjusted figure is: Adjusted COEP1993 = (COEP1998)/{ (1.091)*(1.058)*(1.065)*(1.076)*(1.114)* (1 .135)*(1 .103)*(1 .062)*(1.032)*(1 .043)* (1 .036)*(1 .041 )*(1 .048)*(1.054)*(1 .042)* (1 .030)*(1 .030) } There is a significant difference between adjusted and unadjusted expenditures. Below is the average expenditures for years 1974-93 for the thirty-nine cities Chosen for study. 22 Current Operating Expenditures per Pupil 1974-1993 —— Man-Adjust — Adjust 8 Level i 1974 1979 1984 1989 1994 Yes Only inflation adjusted expenditures will be used in the study. Graphing average inflation adjusted expenditures alone enables us assess the amount of volatility that exists over the 20 year span. Inflation Adjusted Current Operating Expenditures per Pupil 1974-1993 1800 161114? 1400 ’ 12m 1. , . 11!!) t ‘ . : ' 1974 1979 1984 1989 1994 3 Level Year 23 Crime/juvenile arrest variables were also adjusted with regard to the population of the city to develop a rate per 100,000 people. Census figures can be used to adjust for population, however, the decennial census would provide only one change in population over the twelve-year span. To adjust the data more accurately, population estimates were obtained from the Michigan Information Center. ‘ Population estimates were available annually at the sub-county level for the year 1990-93 (1990 = Census data). For the years 1982-89, estimates were only available on a biannual basis (even years), therefore, uneven years were adjusted using the most recent estimate. ' A complete listing of population estimates for each of the thirty-nine cities appears in Appendix B. Example using the City of Alpena: Crime/Juvenile Arrest Rate per 100,000 = {(Annual # of Index Crimes)/(Population)} * 100.000 1984 1985 1988 1987 Index Index Index Index Crime Crime Crime Crime Total Total Total Total City of Alpena 537 567 621 601 Population Population Population Population Estimate Estimate Estimate Census 1984 1988 1988 1990 CityofAlpena 11535 11290 11350 11354 1984 Alpena Index Crime Rate per 100,000 people = {(537)/(11535)}*100000 = £5.15 1989 Alpena Index Crime Rate per 100,000 people = {(606)/(11350)}*100000 5332.21 1988 Index Crime Total 680 1989 Index Crime Total 606 1990 Index Crime Total 607 {(1984 # of Index Crimes)/(1984 Population» * 100,000 {(1989 11 of Index Crimes)/(1988 Population» * 100,000 Rates were developed for non-index crimes, total crimes and juvenile arrests in the same fashion. The following graphs show average trends for the thirty-nine cities with regard to these rates. Rate per 100,000 people Average Index Crime Rate 1982-1993 7700 7500 7300 7100 6900 6700 6500 t t ' 1982 1986 1990 1994 Year Rate per 100,000 people Ave rage Non-Index Crim e-Rate 1982-1993 Year 26 Average Total Crime Rate 1982-1993 18000.. 17000.. 16°00 —9 _ i 15000 _ , 14000 I f f 1982 1986 1990 1994 Rate per 100,000 people Year Rate per100,000 people Average Juvenile Arrest Rate 1982-1993 880 640 ;4 ' w 620 6°° 1 “ 58o ' 560 . . . .. _, _ .‘ , 1982 1986 1990 1994 Year 27 There is a lot more volatility in the average crime/juvenile arrest rates than in inflation adjusted expenditures. This proved difficult in finding a relationship between each of the variables, particularly the juvenile arrest rate; however, since the graphs above only show the “average” trends, they say nothing about the way in which inflation adjusted expenditures and crime/juvenile arrest rates are correlated with one another for each individual city. In order to correlate the variables, data on expenditures and crime/juvenile arrest rates were placed into four time periods for each of the cities in the study. The four periods are as follows: Period Years Data required Period 1 1974-1978 Expenditure data only Period 2 1979-1983 Expenditure data only Period 3 1984-1988 Expenditure & Crime data Period 4 1989-1993 Expenditure & Crime data In order to make use of “lags” in the study, percentage change rates were developed using the procedure below: 1) Sum the inflation adjusted expenditures and crime/juvenile arrest rates for each of the applicable periods to develop four five-year time periods. 2) Develop a % change rate for expenditure data between the first and second period, the second and third period and the third and fourth period. 3) Develop a % change rate for crime/juvenile arrest rates between the third and fourth periods. Current Operating Expenditure per Pupil (COEP) data actually cut across two different years (as does the school year) so the latter year was arbitrarily Chosen as the year to which crime/juvenile arrest rates will be compared. Example using the City of Alpena: 1974 1975 1976 1977 1978 Adjusted Adjusted Adjusted Adjusted COEP 73-74 COEP 75-76 COEP 75-76 COEP 76-77 COEP 77-78 Alpena 978 1018 1090 1062 1054 Alpena Adjusted COEP for Period 1 (1974-78) = 5202 This represents the entire amount of dollars spent on each student for the years-1974 through 1978 in the City of Alpena. Figures were computed for each of the other three periods and a % change computed from one period to the next. Alpena Adjusted COEP for Period 2 (1979-83) 5150 % Change in COEP from Period 1 to Period 2 {(Adjusted COEP for Period 2 - Adjusted COEP for Period 1 ) / (Adjusted COEP for Period 1 )} * 100 = {(5150 -5202)/5202}* 100 0.999616 Crime/juvenile arrest rates were adjusted in the same fashion to develop % change rates for the applicable periods. ‘ ‘ A complete listing of all variables is located in Appendix A. 29 VIII. Correlations There were twenty individual correlations of variables (4 sets). Below are a list of the independent and dependent variables in each as well as a description of what each was to accomplish. Set 1 Dependent variable Independent variable Crime/Juvenile Arrest Expenditures Period 3 Rates for Period 3: Index Non-Index Total Arrest Crime/Juvenile Arrest Expenditures Period 4 Rates for Period 4: Index Non-Index Total Arrest The eight correlations above will only indicate whether or not cities that have higher expenditures on education have correspondingly lower (or higher) crime rates. It would say nothing about whether or not a change in expenditures would potentially change crime rates. To find out whether or not a change in the level of expenditures has any effect on crime/juvenile arrest rates, data on the % change from one period to the next must be utilized. 30 Set 2 Dependent variable Independent variable % A in Crime/Juvenile % A in Expenditures from Arrest Rates from Period 3 to Period 4 Period 3 to Period 4: Index Non-Index Total Arrest The four c0rrelations above will show whether or not the change in expenditures from the third period to fourth produced any effect on the change in crime/juvenile arrest rates from the third period to the fourth. Set 3 Dependent variable Independent variable % A in Crime/Juvenile % A in Expenditures from Arrest Rates from Period 2 to Period 3 Period 3 to Period 4: Index Non-Index Total Arrest These next four correlations allow a five-year “lag” in the comparison of variables by comparing the change in expenditures from the second to third period with change in crime/juvenile arrest rates from the third to fourth period. 31 Set 4 Demndent variable Indemndent variable % A in Crime/Juvenile % A in Expenditures from Arrest Rates from Period 1 to Period 2 Period 3 to Period 4: Index Non-Index Total Arrest The final four correlations compare the change in expenditures from the first to the second period with the change in crime/juvenile arrest rates from the third to the fourth period to allow a longer (IO-year) “lag” to be used. Data from each of the four sets of correlations was analyzed to determine whether or not the relationship between the variables is statistically significant. ‘ ' Data was analyzed using Statistical Programming for Social Science (SPSS) software. Correlation output with matrices with Pearson’s (r), two-tailed significance and number of cases are located in Appendix E. 32 A. Results from Correlations Set 1 H0: Current Operating Expenditures per Pupil for period 3 (1984-88) and Index Crime Rate for Period 3 are independent events (r is equal to 0). Ha: Current Operating Expenditures per Pupil for period 3 (1984-88) and Index Crime Rate for Period 3 are related. (r is not equal to 0). 100000 0 D 80000 . D D m D 3 80000 - ' D E a ° E I: a :1 I 6 40000 E o .—. a n a D D D D D D ch a D c1 :1 Q: B u I: 20000 1 o D D D D o H I r ’r 4000 8000 8000 10000 12000 14000 COEP for Period 3 The data points are dispersed evenly about the graph. The slope of the fit line is nearly horizontal, perhaps even positive, indicating there is almost no relationship between these two variables. Statistical Analysis Pearson’s Correlation (r) = .002 Significance (2-tailed) = .992 As expected, the correlation is not significant (.992 > .05) and therefore the null hypothesis can not be rejected. 33 Ho: Current Operating Expenditures per Pupil for period 3 (1984-88) and Non-Index Crime Rate for Period 3 are independent events (r is equal to 0). HA: Current Operating Expenditures per Pupil for period 3 (1984-88) and Non-Index Crime Rate for Period 3 are related (r is not equal to 0). 120000 100000‘ D 80000. D 60000' a a NICRforPen'od3 40000 . '3 20000I a a TIF’T TITT’ - I 4000 6000 8000 10000 12000 14000 COEP for Period 3 The data points again seem to have a good dispersion. The slope of the fit line is negative, however it doesn’t seem great enough to be significant. Statistical Analysis Pearson’s Correlation (r) = -. 148 Significance (2-tailed) = .369 The correlation is not significant (.369 > .05) and therefore the null hypothesis is accepted. 34 H0: Current Operating Expenditures per Pupil for period 3 (1984-88) and Total Crime Rate for Period 3 are independent events (r is equal to 0). HA: Current Operating Expenditures per Pupil for period 3 (1984-88) and Total Crime Rate for Period 3 are related (r is not equal to 0). 180000 160000I 140000'l 0 120000“ 100000! a 80000! a TCFtbrPaiod3 0 60000! a D D 40000' a 20000 r 4 _ r 4000 8000 8000 10000 12000 14000 COEP for Period 3 There is a wide dispersion of data points on both sides of the fit line. A negative slope indicates a modest negative relationship, but most likely one that is statistically insignificant. Statistical Analysis Pearson’s Correlation (r) = -.088 Significance (2-tailed) = .594 The relationship is statistically insignificant (.594 > .05) and the null hypothesis is accepted. 35 H0: Current Operating Expenditures per Pupil for period 3 (1984-88) and Juvenile Arrest Rate for Period 3 are independent events (r is equal to 0). HA: Current Operating Expenditures per Pupil for period 3 (1984-88) and Juvenile Arrest Rate for Period 3 are related (r is not equal to 0). 20000 a a I: ca 10000 I i 0.. a r: a 5 n a n U D l I: u D fl D u a D g 1:: I: :13 D O I a I: an m0 u a a a '1 000° 1 I T r 4000 6000 8000 10000 12000 14000 COEP for Period 3 The data points are much more grouped around the fit line than in previous correlations, with the exception of three outliers. Apparently there is a much lower variance for the cities studied with regard to Juvenile Arrest Rate for Period 3. The fit line has a modest downward slope, but there is clearly no significant relationship between these two variables. Statistical Analysis Pearson’s Correlation (r) = -.042 Significance (2-tailed) = .800 The relationship is indeed insignificant (.800 > .05) and the null hypothesis is accepted. 36 H0: Current Operating Expenditures per Pupil for period 4 (1989-93) and Index Crime Rate for Period 3 are independent events (r is equal to 0). HA: Current Operating Expenditures per Pupil for period 4 (1989-93) and Index Crime Rate for Period 3 are related (r is not equal to 0). 120000 0 100000 I 80000 - r3 ‘- D 8 a .. '5 E 80000 I E 5’ I: D D 40000 1 u g, D o a D D can on Dan D n D D 20000 I a I31:: 9 P c, D D I: o t i I 1 i 4000 6000 8000 10000 12000 14000 18000 COEP for Period 4 The data points are scattered nicely about the fit line with one notable outlier (Benton Harbor ICR = 111,211.0). Apparently there is a very slight negative relationship, but nothing significant. Statistical Analysis Pearson’s Correlation (r) = -.057 Significance (2-tailed) = .729 The relationship is not significant (.729 > .05), therefore the null hypothesis can not be rejected. 37 Ho: Current Operating Expenditures per Pupil for period 4 (1989-93) and Non-Index Crime Rate for Period 4 are independent events (r is equal to 0). HA: Current Operating Expenditures per Pupil for period 4 (1989-93) and Non-Index Crime Rate for Period 4 are related (r is not equal to 0). 160000 I: 140000 1 120000 " V' a '8 100000 ‘ n a E a 80000 i D 0 DD § 60000 ' an” 1:: an '3 :1 c: I: 40000 . Tb ° on c, ° a an a: ”D 20000 - 8" ° 0 Ifi t 1 - t 4000 6000 8000 1 0000 1 2000 14000 1 6000 COEP for Period 4 The data points are nicely dispersed with one notable outlier (Mt. Clemens NICR = 155,038.1). The slope of the fit line is slightly negative, but obviously insignificant. Statistical Analysis Pearson’s Correlation (r) = -.095 Significance (2-tailed) = .564 The relationship is indwd insignificant (.564 > .05) and the null hypothesis is accepted. 38 H0: Current Operating Expenditures per Pupil for period 4 (1989-93) and Total Crime Rate for Period 4 are independent events (r is equal to 0). HA: Current Operating Expenditures per Pupil for period 4 (1989-93) and Total Crime Rate for Period 4 are related (r is not equal to 0). 300000 13 * 200000 I o '8 o '5 n- D a , e a j . . ; 100000 __ c, e D U D c: a U a db D D D DO U D D DE D U D :0 n o I_ r F, T 4000 6000 8000 10000 12000 14000 16000 COEP for Period 4 The data points are dispersed nicely with no major outliers. The slope of the fit line is negative, but the relationship is insignificant. Statistical Analysis Pearson’s Correlation (r) = -.088 Significance (2-tailed) = .593 The correlation does not produce a significant relationship (.593 > .05), therefore the null hypothesis can not be rejected. 39 H0: Current Operating Expenditures per Pupil for period 4 (1989-93) and Juvenile Arrest Rate for Period 4 are independent events (r is equal to 0). HA: Current Operating Expenditures per Pupil for period 4 (1989-93) and Juvenile Arrest Rate for Period 4 are related (r is not equal to 0). 30000 20000 A a * .3 5 ‘3- 10000 n 5 u g I: a 81:1 I: a 1: :5 a ‘5 on n U :1" u 0 I, a 1:1 1:1 dE'I no ”a D D 4000 6000 8000 10000 12000 14000 16000 COEP for Period 4 The data points are grouped around the fit line producing a slightly negative relationship with one notable outlier (Harper Woods JAR = 20,137.51). The relationship between these two variables is obviously insignificant. Statistical Analysis Pearson’s Correlation (r) = -.177 Significance (2-tailed) = .480 Once again the relationship is statistically insignificant (.480 > .05) and the null hypothesis is accepted. 40 Set 2 H0: % change in Current Operating Expenditures per Pupil from period 3 (1984-88) to period 4 (1989-93) and % change in Index Crime Rate fi'om period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). HA: % change in Current Operating Expenditures per Pupil fi'om period 3 (1984-88) to period 4 (1989-93) and % change in Index Crime Rate fi'om period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to 0). 20 ‘- u D D D I: 8 '5 10I 0. I: O (‘0 D D .8 0‘ D D D D E 1:1 1:1 :1 1:1 1:1 :1 i . . '3 D 1:1 I: g -10'l 1:1 1:1 c: .E I: D a c1 c: E -20' a a a a 33 D D 30 i i t -10 0 10 20 30 % Change in COEP from Period 3 to Period 4 This is the first correlation to show percentage change rates between periods and has perhaps the greatest dispersion of data points thus far. This represents a high variance between both variables. It appears that several Cities actually fell in the amount of expenditures allocated fiom period 3 to period 4 which proves that adjusting for inflation was, indeed, a worthwhile effort. The slope of the fit line is negative, but the relationship is most likely insignificant. Statistical Analysis Pearson’s Correlation (r) = -.131 Significance (2-tailed) = .427 The relationship is insignificant (.427 > .05) and the null hypothesis is accepted. 41 H0: % Change in Current Operating Expenditures per Pupil from period 3 (1984-88) to period 4 (1989-93) and % change in Non-Index Crime Rate fi'om period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). HA: % change in Current Operating Expenditures per Pupil from period 3 (1984-88) to period 4 (1989-93) and % change in Non-Index Crime Rate fi'om period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to O). 80 601 40I 20‘ DD .20 l %ChmgeinNCRiomPeriod3to Period4 40 -10 6 10 20 % Change in COEP from Period 3 to Period 4 30 Unlike the former correlations, this one Clearly exhibits a positive relationship; however is it still appears to be statistically insignificant. Once city, in particular, seemed to have both a reduction in COEP and NICR (City of Allen Park -7 .4466 and -23.6005 respectively). Statistical Analysis Pearson’s Correlation (r) = .059 Significance (2-tailed) = .721 The relationship is not significant (.721 > .05) and therefore the null hypothesis can not be rejected. 42 Ho: % change in Current Operating Expenditures per Pupil from period 3 (1984-88) to period 4 (1989-93) and % change in Total Crime Rate from period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). Ha: % change in Current Operating Expenditures per Pupil from period 3 (1984-88) to period 4 (1989-93) and % change in Total Crime Rate from period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to 0). 60 I: V B i 8 4O . a (0 i .. ° :1 O. a a E 20 I a 1:1 5 ° ° 1: D n .E U D 1:: D 1: a D :1 c1 a D “.5 5’ c: a .20 D i 1' fit -10 0 10 20 30 % Change in COEP from Period 3 to Period 4 There appears to many cities from the sample that exhibited both a positive % change in COEP and a negative % change in TCR and yet, the fit line is almost horizontal, perhaps even positive. Statistical Analysis Pearson’s Correlation (r) = .006 Significance (2-tailed) = .959 The relationship is, in fact, positive, but nonetheless insignificant (.969 > .05), therefore the null hypothesis must be accepted. 43 Ho: % change in Current Operating Expenditures per Pupil fiom period 3 (1984-88) to period 4 (1989-93) and % change in Juvenile Arrest Rate fi'om period 3 (1984-88) to period 4 (1989-93) are independent events (I is equal to 0). 11.4: % change in Current Operating Expenditures per Pupil fi'om period 3 (1984-88) to period 4 (1989-93) and % change in Juvenile Arrest Rate from period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to 0). 3000 V‘ .‘3 i 2 2000- D t") E s 1000- .E :1 ° :9 U .0 E 0' '3 5p 58:11:: ch agaumcb EDIE—“5* u - a? -1000 F— l r -10 0 10 20 30 % Change in COEP from Period 3 to Period 4 The data points appear to be Situated closely around the fit line, however, this is most likely a result of the wide range required for the graph to include the one major outlier which represents the City of Norton Shores. This city displayed a more than 2000% increase in the juvenile arrest rate fi'om period 3 to period 4 (period 3 = 14.04; period 4 = 248.68). The negative slope of the fit line appears to be insignificant. Statistigfl Aggysis Pearson’s Correlation (r) = -. 121 Significance (2-tailed) = .463 The relationship is insignificant (.463 > .05) and the null hypothesis is accepted. Set 3 H0: % change in Current Operating Expenditures per Pupil from period 2 (1979-83) to period 3 (1984-88) and % change in Index Crime Rate from period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). HA: % change in Current Operating Expenditures per Pupil fiom period 2 (1979-83) to period 3 (1984-88) and % change in Index Crime Rate from period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to 0). 20 a D g g e '5 10'l 0. c1 .9. 4"! .3 8 O. .E e\° a D -30 T r I I -10 0 10 20 30 40 % Change in COEP from Period 2 to Period 3 Clearly, a negative relationship exists between these two variables. The data points are dispersed nicely and it appears that the relationship could be Significant. Statistical Analysis Pearson’s Correlation (r) = -.220 Significance (2-tailed) = .178 Despite the downward slope of the fit line, the chances of making a type I error are still too great (, 17 8 > .05). Therefore the null hypothesis cannot comfortably be rejected. 45 H0: % change in Current Operating Expenditures per Pupil from period 2 (1979-83) to period 3 (1984-88) and % change in Non-Index Crime Rate from period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). HA2 % change in Current Operating Expenditures per Pupil from period 2 (1979-83) to period 3 (1984-88) and % change in Non-Index Crime Rate fi'om period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to 0). 80 60I a 40' DD %CImgeinNICRIromPeIiod3toPeriod4 a D .20 I D .40 i F ‘- ‘r -10 0 10 20 30 40 % Change in COEP from Period 2 to Period 3 Again there seems to be a Clear negative relationship between the two variables and the data points are dispersed evenly about the graph. Judging by the last correlation, however, the relationship here would appear to be insignificant as well. Statistical Analysis Pearson’s Correlation (r) = -.175 Significance (2-tailed) = .288 The relationship is not Significant enough to comfortably reject the null hypothesis (.288 > .05) therefore it must be accepted. Ho: % change in Current Operating Expenditures per Pupil from period 2 (1979-83) to period 3 (1984-88) and % change in Total Crime Rate from period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). Ha: % change in Current Operating Expenditures per Pupil from period 2 (1979-83) to period 3 (1984-88) and % change in Total Crime Rate from period 3 (1984-88) to period 4 (1989-93) are related (I is not equal to O). so D 1’ B E 2 40- a ('0 3 ° a 5 :1 o. a D g 201 h .s 39 a D D .20 _ _ fi r ~10 o 10 20 30 4o % Change in COEP from Period 2 to Period 3 The downward slope of the fit line indicates a negative relationship between the two variables. The data points are grouped around the center of the graph with a few exceptions. Statistical Analysis Pearson’s Correlation (r) = -.207 Significance (2-tailed) = .205 Although there is a clear negative relationship, it is not significant enough to comfortably reject the null hypothesis (.205 > .05). 47 Ho: % change in Current Operating Expenditures per Pupil fi'om period 2 (1979-83) to period 3 (1984-88) and % change in Juvenile Arrest Rate from period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). HA: % change in Current Operating Expenditures per Pupil from period 2 (1979-83) to period 3 (1984-88) and % change in Juvenile Arrest Rate from period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to O). 3000 1- T8 8 % 2000 I» a co.- co % 0.. g 1000 I a. .E c: r: n D n n g, § 0 ‘ g D D % D #0” Bu 0 D U U u a? -1000 _ _ I 40 Ol tn. 0 N O r» O -10 % Change in COEP from Period 2 to Period 3 Again the data points appear to be closely positioned around the fit line because of the graphs wide y-axis range to include the large outlier (Norton Shores). There appears to be a slight negative relationship, but nothing significant. Statistical Analysis Pearson’s Correlation (r) = -.114 Significance (2-tailed) = .489 The relationship is, indeed, insignificant (.489 > .05) and therefore the null hypothesis is accepted. 48 Ho: % change in Current Operating Expenditures per Pupil from period 1 (1974-7 8) to period 2 (1979-83) and % change in Index Crime Rate from period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). Ha: % change in Current Operating Expenditures per Pupil from period 1 (1974-78) to period 2 (1979-83) and % change in Index Crime Rate fi'om period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to 0). 20 a 5’: D E a "5 10* 0. 2 (0 E 8 5 -10- .E E -20‘ I: D a D Q 0 D D -20 -10 o 10 20 30 4o % Change in COEP from Period 1 to Period 2 The first of the correlations in the longest time lag produced a nice scattering of data points about the graph and a clear negative relationship between the two variables. The relationship appears as though it could be a significant one. StatisticalAnalysis Pearson’s Correlation (r) = -.385 Significance (2-tailed) = .015 The relationship between the two variables is, in fact, statistically significant (.015 < .05). This is the first correlation in which the null hypothesis can comfortably be rejected. 49 Ho: % change in Current Operating Expenditures per Pupil fiom period 1 (1974-78) to period 2 (1979-83) and % change in Non-Index Crime Rate fi'om period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). HA: % change in Current Operating Expenditures per Pupil fi'om period 1 (1974-78) to period 2 (1979-83) and % change in Non-Index Crime Rate fi'om period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to 0). 80 %GimgeinNiCRimmPeriod3toPeriod4 .40 i t I t -20 -10 '0 10 20 30 40 % Change in COEP from Period 1 to Period 2 Again there is a nice scattering of the data points and a clear negative relationship between these two variables. Statistical Analysis Pearson’s Correlation (r) = -.407 Significance (2-tailed) = .010 This correlation is statistically significant as well (.010 < .05), therefore the null hypothesis can be rejected for its alternative. 50 Ho: % change in Current Operating Expenditures per Pupil from period 1 (1974-78) to period 2 (1979-83) and % change in Total Crime Rate from period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). Ha: % change in Current Operating Expenditures per Pupil from period 1 (1974-78) to period 2 (1979-83) and % change in Total Crime Rate from period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to 0). 60 a 1' .‘3 E 2 40 I D m g g 20 1 oh .5 E o I $ '20 r I - f I— i -20 -10 0 1O 20 30 40 % Change in COEP from Period 1 to Period 2 This correlation, the third in the set of longest lags, produces another fit line with a negative slope which implies that these two variables are inversely related as well. Statistfisal Analysis Pearson’s Correlation (r) = -.400 Significance (2-tailed) = .012 This correlation, like the latter two, is significant (.012 < .05) and the null hypothesis is rejected. 51 H0: % change in Current Operating Expenditures per Pupil from period 1 (1974-78) to period 2 (1979-83) and % change in Juvenile Arrest Rate from period 3 (1984-88) to period 4 (1989-93) are independent events (r is equal to 0). Ha: % change in Current Operating Expenditures per Pupil from period 1 (1974-78) to period 2 (1979-83) and % change in Juvenile Arrest Rate from period 3 (1984-88) to period 4 (1989-93) are related (r is not equal to O). 3000 V B 8 % 2000 1» D 0-0 . 0') E a i § 10000 oh .9: c: a g 0- a e a DB 3‘5 go‘bé’b 5; guess :1 “a... 39 '1000 - I— - - I - -20 -10 0 10 20 30 40 % Change in COEP from Period 1 to Period 2 The fourth correlation, in the longest lagged series, does not appear to have a significant relationship. It resembles the two previous correlations that used 3 change in JAR from period 3 to period 4 as its y-axis. Statistical Analysis Pearson’s Correlation (r) = -.133 Significance (2-tailed) = .419 The final correlation is not a significant one (-419 > .05) and therefore there is no other choice but to accept the null hypothesis. 52 B. Discussion While the ultimate goal of this study is to determine whether or not w in school expenditures have any effect on crime/juvenile arrest rates, the aim of the first set of correlations was only to show whether or not cities with higher expenditures on education enjoy lower crime/juvenile arrest rates, but said nothing about increases or decreases. The correlations for both period 3 and period 4 produced no statistically significant results. This would seem to indicate that communities with large amounts of expenditures are just as susceptible to crime and juvenile arrests as are communities with relatively low expenditures on education. The second set of correlations used the % change in expenditures from period 3 to period 4 for both expenditures and crime/juvenile arrest rates as dependent and independent variables, respectively. Again there was so significant relationship to be found between these variables. It was not until the third set of correlations in which the “five-year lag” was incorporated into expenditures that there appeared to be a clear relationship between the two variables. Although, still statistically insignificant, the likelihood of a Type I error had fell considerably. It was in the fourth set (IO-year lag) that there were three statistically significant correlations between % change in current operating expenditures per pupil from period 1 to period 2 and % change variables from period 3 to period 4 (i.e. index crime rate, non-index crime rate and total crime rate). This implies that there is a relationship between educational expenditures and crime/juvenile arrest rates, but only in the long term. Surprisingly enough, there was no relationship between changes in the juvenile arrest rate and expenditures, even with regard to the longest lagged correlation. This could be attributed to the enormous volatility in the juvenile arrest rate as shown in the graph on page 27, or perhaps it was the more than 2000% 53 increase in the City of Norton Shores juvenile arrest rate that skewed the data. There is, perhaps, another explanation. Educational expenditures were, in fact, found to be related to crime rates, but only in the long term. This could suggest that increased expenditures help students to form a more law-abiding community when they are older. After all, juveniles, would cease to be juveniles after ten years. The fact that there was no relationship in the non-lagged correlations would seem to indicate that a community doesn’t benefit from increased expenditures on education until juveniles have grown and become part of the adult community. 54 IX. Regressions The three sets of correlated variables found to be statistically significant were further analyzed using Ordinary Least Squares (OLS) regressions‘: Y=b0+b1Xr+e where Y = dependent variable X1 = independent variable b0 = constant (y-intercept) bi = x-coefficient (slope) e = standard error term Regressional analysis determines the extent to which the variables are related (i.e. explained variation in the dependent variable). It will also indicate how the independent variable (crime/juvenile arrest rate) responds to per-unit changes in the independent variable (educational expenditures). Assumptions in linear regression are as follows: 1) The relationship between x (the independent variable) and y (the dependent variable) is linear. 2) The values of x are fixed (y varies as a function of x). 3) The data points are evenly distributed about the line: a. Error term has constant variance across values of x (homoskedasticity). b. The errors are uncorrelated across observations. c. The error is normally distributed. ' Data was analyzed using Statistical Programming for Social Science (SPSS) software. Regression output with model summaries, AN OVA and coefficients appears in Appendix F. 55 A. Results from Regressions The three regressions were carried out are as follows: Dgpendent variable Indeggndgnt variable 1. % A in Index Crime Rate (ICR) % A in Current Operating from period 3 (1984-88) to Expenditures per Pupil (COEP) period 4 (1989-93) from period 1 (1974-78) to period 2 (1979-83) (ICR % A from period 3 to period 4) = b0 + br(% A in COEP from period 1 to period 2) + e Statistical Analysis Coefficient of Determination (r’) = .148 Standard Error of the Estimate = .1127 Constant (B) = -3.6E-03 = -0.0036 X-coefficient = -.452 The regression equation that describes the relationship between the two variables: (ICR % A from period 3 to period 4) = -.0036 - (.452)(% A in COEP from period 1 to period 2) The proportion of squared deviations from the mean that are “explained” by the regression equation and the degree to which a change in expenditures can affect a change in the Index Crime Rate in a city in the next ten years 14.8%. For each 1% increase in Current Operating Expenditures per Pupil from period 1 to period 2 there is a corresponding decrease in the Index Crime Rate from the third to fourth period by 0.452%. 56 Deeendent variable Indemndent variable 2. % A in Non-Index Crime Rate % A in Current Operating from period 3 (1984-88) to Expenditures per Pupil from period 4 (1989-93) period 1 (1974-78) to period 2 (1979—83) (NICR % A from period 1 to period 2) = bo + br(% A in COEP from period 1 to period 2) + e Statistical Analysis Coefficient of Determination (r’) = .166 Standard Error of the Estimate = .2068 ’ Constant (B) = .268 X-coefficient = -.884 The regression equation that describes the relationship between the two variables: (NICR % A from period 1 to period 2) = .268 - (.884)(% A in COEP from period 1 to period 2) The proportion of squared deviations from the mean that are “explained” by the regression equation and the degree to which a change in expenditures can affect the Non-Index Crime rate in a city in the next ten years is 16.6%. For each 1% increase in Current Operating Expenditures per Pupil from period 1 to period 2 there is a corresponding degease in the Non-Index Crime Rate from the third to fourth period by 0.884%. 57 Qegendent variable Indemndent variable 3. % A in Total Crime Rate % A in Current Operating from period 3 (1984-88) to Expenditures per Pupil from period 4 (1989-93) period 1 (1974-78) to period 2 (1979-83) (TCR % A from period 1 to period 2) = b0 + br(% A in COEP from period 1 to period 2) + e Statistical Analysis Coefficient of Determination (r1) = .160 Standard Error of the Estimate = .1425 Constant (8) = .129 X-coefficient = -.597 The regression equation that describes the relationship between the two variables: (TCR % A from period 1 to period 2) = .129 - (.597)(% A in COEP from period 1 to period 2) The proportion of squared deviations from the mean that are “explained” by the regression equation and the degree to which a change in current operating expenditures per pupil can affect the Total Crime Rate in a city in the next ten years is 16.0%. For each 1% increase in Current Operating Expenditures per Pupil from period 1 to period 2 there is a corresponding decrease in the Total Crime Rate from the third to fourth period by 0.597%. 58 B. Discussion Changes in the Non-Index Crime Rate appears to be affected most by changes in expenditures, followed by the Total Crime Rate and finally the Index Crime Rate. While 16.6% explained variation may not seem significant , the goal here was not to “explain” crime rates. If that were the case, then adding additional independent variables, already proven to be related to crime, to a multiple regression would undoubtedly produce a more significant explanation. The purpose here, however was to determine whether or not, educational expenditures alone can have any impact on crime rates. The results from the regressions indicate that there is, in fact, a relationship between the two. Whether the relationship between the variables found to be related is truly a causal one may remain suspect. Since the unit of observation had to be consistent for the study, the sample was in effect, non-random. It may be inappropriate to generalize the findings of this study to cities that did not meet the criteria for the sample. Nonetheless, the very fact that a relationship was found to exist between the variables, even if only in lagged correlations, serves as a step in the right direction with regard to showing the way in which public education is, in fact, a benefit to all. 59 X. Conclusion A reduction in crime is but one of many potential ways in which changes in educational expenditures could affect communities. Originally, it was my intention to include other dependent variables in this study, in addition to crime/juvenile arrest rates, to determine whether or not they exhibited any relationship to changes in educational expenditures. The potential variables to be studied include median income, wages and salaries and unemployment rates. Unfortunately, this data was not available annually at the sub-county level. Proposal A in Michigan, which raised the sales tax by 2% for educational expenditures, prompted this study. The thought being that if non-property owner, non-parent citizens of Michigan are expected to pay for provision of education, then research should demonstrate that these citizens benefit from public education. For future study, the inclusion of the other potential dependent variables would be a worthwhile effort. That is, of course if those variables could be obtained. Perhaps other states in the US do have access to this information annually. The fact that this study was, in essence, directed by the availability of the data, makes it difficult to generalize the results. Developing a method by which cities with more than one school district could be added to the model, could make the study more representative and add to the strength of the results. If this were done, a similar study could even take place using samples from various states. Bibliography Anderson, Robert C., “The Interorganizational Community,” Lewiston, NY, Oueenston, Ontario and Lapeter, Wales, The Edwin Mellen Press, 1993. Andrews, Chris and Greg J. Borowski, “State slashes tax, no plan for schools,” Lansing State Journal, pp. 1A, July 23, 1993. Andrews, Chris, “Citizens’ commission proposes property tax levy cut, sales tax hike,” Lansing State Journal, pp. 3A, Feb. 10, 1987. Babbie, Earl, “The Practice of Social Research,” Wadsworth Publishing Company, 1995. Bailey, Kenneth D., “Methods of Social Research,” New York: Free Press; Toronto: Maxwell Macmillan Canada; New York: Macmillan International, 1994. Baker, Keith, “Yes, Throw Money at Schools,” Phi Delta Kappan, April 1991, v72, n8, pp. 628- 31. Christoff, Chris, “School tax reform plan may face rocky road,” Detroit Free Press, September 27, 1987. Clark, W . A. V. and P. L. Hosking, “Statistical Methods for Geographers,” New York, John Wiley & Sons, Inc., 1986. Danzberger, Jacqueline P. and Michael D. Usdan, “Local education governance: perspectives on problems and strategies for change,” Phi Delta Kappan, Jan. 1994, v75, n5, p. 366. “Data Collection and Analysis,” London; Thousand Oaks, CA, Sage in association with Open University, 1996. Decker, Paul T., “Findings From Education and the Economy: An Indicators report,” Washington, DC: National Center for Education Statistics, US Dept. of Education, Office of Educational Research and Improvement. Gilling, Daniel, “Crime Prevention: Theory, Policy and Politics,” London and Bristol, PA, UCL Press, 1997. “GOP calls property tax funding for schools unfair, wants changes: Republican leaders say that less affluent districts are losing out,” Detroit Free Press, February 3, 1987. Iorio, Mary E., “Schools stunned by revenue loss,” Lansing State Journal, July 23, 1993. 61 Jones, Tim, “Report says sales tax of 6% would aid schools,” Detroit Free Press, February 10, 1987. Kearney, C Philip, “A Primer on Michigan School Finance Third Edition,” The University of Michigan, 1994. Kratcoski, Peter C. and Lucille Dunn Kratcoski, “Juvenile Delinquency: Third Edition,” Englewood Cliffs, NJ, Prentice-Hall, Inc., 1990. Mandi, Peter, “Education and Economic Growth in the Developing Countries,” Budapest, Hungary, Akademiai, 1981. Mathews, David, “The Lack of Public for Public Schools,” Phi Delta Kappan, June 1997, v78, n10, pp. 740-55. Mumane, Richard J. and Edward W. Pauly, “Lessons from Comparing Educational and Economic Indicators,” Phi Delta Kappan, March 1988, , v69, n7, pp. 509-13. Nation, Jack R., “Research Methods,” Upper Saddle River, NJ, Prentice Hall, 1997. Pierce, Hank, “Fiscal Focus: Proposal A and Pupil Equity,” prepared for the House Fiscal Agency, December 1996. Psacharopoulous, George and Eduardo Velez, “Educational quality and labor market outcomes: evidence from Bogota, Colombia,” Sociology of Education, April 1993, v66, n2, pp. 130- 45. Psacharopoulous, George assisted by Keith Hinchliffe, “Returns to Education: An International Comparison,” San Francisco; Washington, Jossey-Bass Inc., 1973. Quan, Nguyen T. and John H. Beck, “Public Education Expenditures and State Economic Growth: Northeast and Sunbelt Regions,” Southern Economic Journal, Oct. 1987, v54, n2, pp. 361-66. “Questions, answers on plan to overhaul school finance,” Detroit Free Press, September 24, 1987. Stroud, Joe H., “Prop A offers meaningful improvements,” Detroit Free Press, May 9, 1993. Stoetzer, Karin E. and Merry Morash, “Juvenile Crime in Michigan: Evidence and Public Perceptions,” East Lansing, MI, Institute for Public Policy & Social Research, 1997. Teng, Teng, “Education Spurs Economic and Social Progress,” Beijing Review, Dec. 2-8, 1991, v72, n8, pp. 28-31. 62 Whorton, Joseph W. Jr. and David R. Morgan, “Measuring Community Performance: A Handbook of Indicators,” Bureau of Government Research, 1975. Witte, Ann Dryden, “Crime,” from The Social Benefits of Education, ed. Jere R. Behrman and Nevzer Stacey, Ann Arbor, MI, The University of Michigan Press, 1997. Wolfgang, Marvin E., Terence P. Thomberry and Robert M. Figlio, “From Boy to Man, from Delinquency to Crime,” Chicago and London, The University of Chicago Press, 1987. 63 IPPENDIX A — List of Variables DONOW&UN-| «0..—00. 0.050. 200.8 0:2.0 «0.60.0 0030: 1032 )3. «0.60.0 00030.8 )3. 0:2.0 00:00.0 43.330 05 >30 0:2.0 00300.0 000. P0323 00.60. 0.0.10. 003000 0:2.0 00:00.0 x3250. 0:28 00.60.. i. 0.030:- 0033334 00.60.. 3002.... 00335.2 00300.0 5.00 0033334 «0.60.0 {0:03.300 )3. 0:2.0 «0.60.0 35.600: 10.030 00:00. 0.030. 20.... 0.630 0:2.0 00.60. 0.050. 02:04 00.80. 0.050. 0.43.3033 0.? 90:02 0.0.10. 03530: 03. 00:00. 0.0.10. 0033.0 0.2 00.60. 0.0.10. 132 00:. 0.2 00.52 0.0.4.0. Zo<. 00333.? 00.62 0.0.10. 00.. 0o} 0.? 00:02 0.030. 00:00. 0.0.10. 0.2 0. 3040. Cox M3530... 0:20 00.60. 0.0.10. 434 00.60. 0.0.10. 020000 0:2.0 «0300.0 2.0... 0!... 0:2.0 00:00.0 002.003 0.... 00.60. 0.030. 00030 0:2.0 «0.50. 0.050. 0030: 0:4 00.60. 0.0.10. 103.330.. 0:2.0 00.60.. 0.... 0. 10304 5.0000 00:00.0 1.03.02. 02.. 0.2 00:00.: 2:00.: 02.. 0:2.0 00:00.0 0032:- 003353 00:00.. 00:303. 00335.2 «0.60. 0.0.10. 43.2 «0.60. 0.3.0. 4330: 0:26 00:00.. $0395.33... 00335.2 «0.60. 0.0.10. 5.42.0030 0.2 00.60. 0.0.4.0. 02....00 >30 0:2.0 00:00.0 OOM0 00m... 00m0 00m- 0030... 00:00» 003000 00:00. u»0» 000. .00» .400 4»uu 008 0.».. 0000 0.00 0040 0.00 0040 00.» .000 4... 0400 0.00 00.0 0000 $0 00.0 0400 40.0 0400 0004 00.0 0.00 40». 0000 040. .0».4 OOM0 OOM0 8M0 .03 #03500 $03300 #0330. 02.000 .04...30.040..§.8f.8¢.§..08 00:00.8» 03000.00 02.2.0.0. .8...8¢ 000008.00 0.00.0800. 0.00.3»00. 0.3040 0.80.4»40» 0.3mm»u400 0.83000. 0.. 30000.0 0.. 00.00440 0.00004»: 0.80400»: 0.80. 00»04 0.8..00000 00.33000 0.0800430 0.. .4008. . 0..»...0400 0.00.0430 0.». .»uu». . 0.04. 0000. 0 0.0.000»! 0.»000.4000 0.00uu.»u. 0.8»..0000 00.80.00 0.. 380000 0.00». 4040 0.».004.4»0 0.0000. 440. . 0.0. .8008 . .0. 30400.4 . 98.404000 . 0.0000000. . 0.. 40800»4 . 0.00.0800 . 0.0.».0000 0.0.0.0004 0.»0.8».»m 0.308»... 0.049.000. 0.0003000 0..u.».»u.0 0...900».u. 0.49.0200 0.4.4008“ 0.3040000. Pig.“ 0.8»0004. 4 0.»000008. 0.. 44.04004 0.4002004 0.30.0000» 0.2.30.3 . 00000.88 0.0.0». 00 0.0000030 0...04004»4 0.. 43400.0 0.00.380 0.»..4»000o 0.»00.»0.04 0.8.80.0. 0.. . 4»0»w.u 0.0.000». 0.80.0000» 004000040» 0.8048.» 0..»..0.»00 0...».4.0u 0.. 3.30.0 0.830000 0.0080000. 0.00.40». 0.40.0044. 9:804: 0.0500000 0.. .8384 0..»uuo»»»o 0.84.0000 0.00.880» 0.. u.0..0$ 0.». 04»». 40 0.0.0». .00» 00048.2. 0..0.00.»0. 0.00». .0..4 0.. .04004». 0.0.0.8. 0.»4.084.» 0.88.0.0 0.049.000. 0.0»0000400 0.»..80..u 0..04048»u 0.80.00.“ 0.300083 0.0.»00000 0.03.04 0.04. .400» 0.80.0034 0.. 5004.». »800.40 Al DONOGbUN-I 882388288388 03 2...... 00:30:03.! 00000000 43330 0.2 000. .0013 003000 .9330... 3. 0.03000 0009...? 0.00.00 10.0...- 2303.300 5.0.800: 10.0...- 228: 0.630 0010.! 0.3.3.33 0.02000 10300.0 I000. 00} 20¢. 00.. 00:. 30.0. 00.. 09050.0 43.. 020000 2.00 00.... 000303 00030 00.00: 05 103.330.. 10.02 .2800 1.02000 00:. 2:00.: 00... 003200 005.60.- 4010. 4830: 5000000 5.100030 000...00 00:22 $0.58. 0380 0322 .8938 02.808. .8..38.80..08 02.808. 330838.88 028008. .8308.80..80 03808.. »4000.0» 0.00.04.00 ...».. 0.0034». ».040.00 .0.»...00»00 »4404.00 .0.»o»004000 »..00.40 0..0000.40 »0..0.»0 0.4.00.40. »000..0» b.00..8»00 ..3».0. .0.0.44..00 .000.»» 0.0»».0..00 »0..0.00 .0.0.0»00..0 »430.0. 0.0440..04 400.4. 0.0004000. 3.44.00 0.04»»040 .0400.00 0.00040».0. .0440». .0.»40»00... .00»..0 0.000.084. 04404.»0 00.040030 .0000.00 0.40.4»040 »0..0.»0 .0..00.»00.0 0403.0» 0.38.0. »044000 0.340.000 00.0.0. 0.»404.0040 30.04 .0.»000.004» »0».0.04 .0....0400. »800 0.30.30 3000.4. 000000.03 00.4.00 0.3.0.004 ».0...04 00040.0.0 00»...00 0.00.0300 40.000 0.000.40..0 4000..»0 b.0..80»00 0.00.04 .0..»00...0» 30.0 .0.0.»00.0 800.00 0.04.0.080 030.00 0.0..000.0. .3400. 0.0.0000 »0000.4. 0.0000000»4 .00»0.0 .0.»»00.0»40 0030..» 0.0004030 0»0»0.00 4.004.. 00... .0 4000. .0. 00.00.04 ». 3.00 ».0»0.00 30000.0 0»0...0» .. 30.00 0.4.0.04 00.0.04 3.4» .8.0».0 .300» . 0»004.00 0.400.»» 0030.40 »0.0. .04 0.04000 .00». .00 $030300 004.000 0010... 0.0. .00. 0. 0 b.0»».»0. .» 0... 400004» 0.. 4000. .00 0.8000404. 0.»00400400 0.»8000400 0.40. 0»04. » P30400000 0.004. 300 0.. ..40.. .4 0.0003000 0.00. 004. 0 0.»000. .00. 0.».400»4.» 0.30. 00. 0.». .00000. 0.300.. »04 0.32.0. 00000003 0.0.0. 0000. 93003.00 0.00»000».0 0.»04040.00 .0. 30000.0 0.00. 0. .00 0.300000. . 0.04. »»04. 0 0.00.. 0.00 . 0.»0000..40 0.04.004. .0 0.000040» 0.0.3000.» b.0..0»00»0 403 004. 0.00 .0400..0 04000.4. .000. 0.0 0. 040.00 .0»0..00 0. 80.8 .0008. 0000.00 0.0400» 00000.04 ..0.04.0 .0040. 4 ..0000» 4800... 00000.». 4.00. .0. 30040.0 000. 0.00 00.0. .00 00»00.04 3.00.0 4.3.00 3000.00 0»000.0» 40400.04 44»04.0. 00000.0» . .0000. ...000.» 3000..» 0840.00 ..00.4.0 43.0.00 40000.. . 0000» .4. 3000.4. 03..»0 .004.0.0 401408 3.00.0. 0.»0.40. 00. ». .0400 0.»0.0300 4080.00 0.0... 00. 4. 00000... .0. 0. 30400 0300.0» 0.0040000»4 3040.0. 0.0440000.» .00.0..0 0.0003». 4» .00000. 0.»00»00.00 .00444 .0 0.0300000 0.044.40 .000. 0000»» 4.0.0.4. 0.. 44000000 .40004 0.»»..00»00 443030 000000000. ...00.0 00.008000 040.0». 0.000.433 01.00.40 0.0400. 08» 4.00004 0.0040000.4 04040.00 0040000000 00000.00 0.0000000.0 30.0.0. 0...00440»4 030»... 0.0...»400» 0.»0».00 004000400 38034 .0.. »0000044 00004 .. 4 0.0000.»»» .0000... 0.040000. 0 44004.0 0.0. »000.00 3.0000 0.0. .»0.0.0 000...00 0.0.0.0400» . .00. 0.. 0.004..0440 30000.0 00.00030 ..0000.0 0.. 4.0.0000 0008.00 .0000. 3». » ..»004.0 0.300.000 4.000.04 0.3008400 00000.00 0.0.04.0.. 00040.00 no.0»».00»40 00000.. 0.30. 04000 00000.00 0.8404430 .0.404.. 0.0000.»4.0 0)) $030300 08.000 300.» .040.»4 00.0.0 004... ...; 000.00. 33.0 3».»» ».00..» ». 4. .00 .0300 000. .0» .400.00 000.40 0000.00 ...0.0 . .4800 . 300.04 »00.00 000... 040.00 .»00.»4 ..44.0. »000.». 334.0 »400.00 44. .0. .0000 »000.». .0..00 0»0.. 3.0. .0004 .000.00 ».00.»4 »0.04.0. 0.00.04 0.0. .»4 »08 »000.04 040. .00 ».. .40 »040.4. 000.0 0000.0 0).. #090000 0.004040..0 0.044004». 0.00000. 00. ... 404»44»0 0.0. .000. .» ...4».. 0040 900000800 0.0. ..0».04 .94. 000.040 0.34.. 040 »0.»4000.»4 ».000000.0. .0...0400»00 0.0400.»; 0000000040 0.000.340 ..0..00..40.0 0.. .3300 000000800 .0.. 330» 0.04.0008» 0.34000 000000000» 0004400000 0.0040030 0.00.0000. 0.0030000 0.»..»00N0. .0.»0.. 0. ».0 0.0020003 0.00»00000. 003400.00 0.00.40000 b.0.00000»0 .0.»00»000.0 0.00.0080 .»008300 APPENDIX B — Population Estimates SBBfiBBBBBBSEGififi36”°““m““”‘ 883882889888 1962 538. 16226 to} £38.: to} .3355 «9.598 43.2 4:58: 58.33 3:53. 03...!" t§t§§t§t§t§v§§ 0328 mango manage macs-3 macs-8 mg mango macs-8 .8. ..Wm .33 .0: ..N8 3.8 .8 .85 8.8 g 848 .3 9V8 ...g 88 .590 ”.30 ..33 M88 .38 803 MOMS NS 9.8 8.8 .00. £3 .QB .88 .83 Q3 $3 $3 .83 fig :§8 38 E8 fig .da .88 .38 $8 33 Qfiu QB. 38 . V. 03 Bus .mg g 8.8 .MAS 8.8 .38 .520 g bug 3 R80 4.8 N63 33 830 .03 .8: 38¢ .83 .g 5.8 .00. ....3 .3. 3.8 .58 .34 2% 8.3 93.0 83A .0. ma .0.» .03 ..m8 3.8 .§ .98 .0: APPENDIX C - Descriptive Statistics Range Minimum Maximum Variance % Change in COEP from Period 1 to Period 2 96 Change in COEP from Period 2 to Period 3 % Change in COEP from Period 3 to Period 4 COEP for Period 1 COEP for Period 2 COEP for Period 3 COEP for Period 4 % Change in ICR from Period 3 to Period 4 ICR for Period 3 iCR for Period 4 39 39 39 39 39 39 39 39 39 39 .47 .40 .37 4512.00 4739.00 7077.00 8424.00 .45 81203.88 98937.99 -.12 -.07 -.07 4411.00 4895.00 5444.00 6075.00 -.28 14643.65 12273.01 .35 .33 .29 8923.00 9634.00 12521.00 14499.00 .17 95847.53 111211.0 .1113 .1489 .1168 6053.82 6735.79 7753.38 8645.10 -5.E-02 37524.4 35805.3 1.1E-02 6.1 E—03 6.4E-03 10281 52 1779146 2997836 3946804 1 .5E-02 3.6E+08 4.2E+08 A4 Range Minimum Maximum Variance % Change In JAR from Period 3 to Period 4 JAR for Period 3 JAR for Period 4 % Change in NICR from Period 3 to Period 4 NICR for Period 3 NICR for Period 4 % Change in TCR from Period 3 to Period 4 TCR for Period 3 TCR for Period 4 Valid N (listwise) 39 39 39 39 39 39 39 39 39 39 20.99 16570.85 1 9895.78 1.00 94897.81 135517 .73 1 30972 1 76097 -.72 14.04 241.73 -.24 14468.05 19521.59 -.17 36382.71 35576.35 20.27 16584.89 20137.51 .76 109365.9 1 55038.1 .57 167354.6 211673.3 .8105 3217.25 3302.96 .1701 44658.5 51927.0 6.3E-02 82182.9 87732.3 10.964 1 .2E+07 1.2E+07 5.0E-02 4.2E+08 6.8E+08 2.4E-02 1 .ZE+09 1 .7E+09 APPENDIX D — Distributions 12 101 81 4| an MW =ms szm 0 £00 5110 no mo mo 31110 mo mm amo mo CIB’b’Fbia‘H 31110 81130 mo 8113.0 amo sano mo 75110 no mo CIE’thich A6 qukmgsfifiso Wmswsmm mumm4 A7 -.10 2(5 .0) .6 .10 .15 Z) .5 .fl .5 *Omhmmw1bw2 81 4d 205405.fi.075.1fi.17532755 awwmmmimmzom %Omhmmm2bm3 A8 -_075 {m .025 .075 .15 .175 .E .275 %mmmm9mmmmmmm4 amassmsmmsm ooooooooooooooooo imumma A9 SiDFMQ ”8&3 o N=$CD 101110 311110 sumo mo 911110 1101110 211110 mo 811110 811110 mmo KRb'FBiaM a!) -5 -Z) -.15 -.10 ..05 .CD .(5 .10 .15 %GamhmiunPakxi3tthm4 A10 o _ 7 , , ,. I ma armo sumo mno sumo 11111110 211110 moo mo mo mo NCRUWS wwsmwmé All 21 sum/=2 this-.17 0 M30) -Z -.13 am .13 z .3 .5) .3 .75 -.19 ~13 .(B .19 .31 .44 .$ .$ %Gmhhmiunm3tom4 6| -$CD sasszasgmm. A12 8018410475 “18877.23 0 N830 mo 811110 1211110 mo ammo mo mmo 1mm m0 m0 ‘KRb’PBiuM *O'I'ph'lmiImWSbmli A13 at 4 2 8 0183013 Wkkfi'bfi3fi%%§%%§% m I! W3 - JN-m no mo mnfirzfno'umo mo mm mm mo mo mo mm4 A14 3) Z) 10 8:01:33! has 0 imam 00 20 40 6'0 50 100150110 150 15.0 210 %Olwhmmm3bm4 A15 \PPENDIX E - Correlation Output Correlations COEP for iCR for Period 3 Period 3 Pearson CEEP Correlation for 1.000 .002 Period 3 ICR for Period 3 .002 1.000 Sig. COEP (2-talled) for . .992 Period 3 iCR for Period 3 '992 N COEP for 39 39 Period 3 ICR for Period 3 39 39 Correlations COEP NICR for for Period 3 Period 3 Pearson 55§P Correlation for 1.000 -.148 Period 3 NICR for -.148 1.000 Period 3 Sig. COEP (2-talled) for . .369 Period 3 NICR for .369 Period 3 N COEP for 39 39 Period 3 NICR for 39 39 Period 3 A16 COEP for TCR for Period 3 Period 3 Pearson 056P Correlation for 1.000 -.088 Period 3 TCR for P erlod 3 -.088 1.000 Sig. COEP (2-tailed) for .594 Period 3 TCR for Period 3 '59“ N COEP for 39 39 Period 3 TCR for Period 3 39 39 Correlations COEP for JAR for Period 3 Period 3 Pearson 0551r Correlation for 1.000 -.042 Period 3 JAR l'or P erlod 3 -.042 1.000 Sig. COEP (2-talled) for .800 Period 3 JAR for Period 3 30° N COEP for 39 39 Period 3 JAR for Period 3 39 39 A17 Correlations COEP for iCR for Period 4 Period 4 Pearson COEP Correlation for 1.000 -.057 Period 4 ICR for Period 4 -.057 1.000 Sig. COEP (24ailed) for .729 Period 4 iCR for Period 4 '729 N COEP for 39 39 Period 4 iCR for Period 4 39 39 Correlations COEP NICR for for Period 4 Period 4 Pearson CCEP Correlation for 1.000 -.O95 Period 4 NICR for -.095 1.000 Period 4 Sig. COEP (2-tailed) for .564 Period 4 NICR for .564 Period 4 N COEP for 39 39 Period 4 NICR for 39 39 Period 4 A18 Correlations COEP for TCR for Period 4 Period 4 Pearson COEP Correlation for 1.000 -.088 Period 4 TCR for Period 4 -.088 1.000 Sig. COEP (2-tailed) for .593 Period 4 ’ TCR for Period 4 '593 N COEP for 39 39 Period 4 TCR for Period 4 39 39 Correlations COEP for JAR for Period 4 Period 4 Pearson COEP Correlation for 1.000 -.117 Period 4 JAR for P eriod 4 -.117 1.000 Sig. COEP (2-tailed) for .480 Period 4 JAR for Period 4 "80 N COEP for 39 39 Period 4 JAR for Period 4 39 39 A19 Cbnehmows Change COEP hem Pmmm 3k) Pakd4 cmmme hHCR flom Paku 3hr Ruhd4 1fimnwn Cbmfldhn 96 cmmme m COEP flom Paw” 3k) lfiuhd4 ‘% cmmme hHCR flom Pmmm 3k: Huhd4 1xmo -J31 -J31 rxmo Sig. (2-taiied) 96 Chane m COEP ham Pmmm 3k) me04 ‘% Change uHCR flom Pawn 3k) Puhd4 427 A27 cmmme COEP flom Pmmm 3k) Rmhd4 cmmme hHCR flmm Pumm 3n: Rmhd4 39 39 39 39 A20 Correlations Change COEP from Period 3 to Period 4 % Change in NICR from Period 3 to Period 4 earson Correlation % Change in COEP from Period 3 to Period 4 % Change in NICR from Period 3 to Period 4 1.000 .059 .059 1 .000 Sig. (2-talied) % Change in COEP from Period 3 to Period 4 % Change in NICR from Period 3 to Period 4 .721 .721 % Change in COEP from Period 3 to Period 4 96 Change in NICR from Period 3 to Period 4 39 39 39 39 1 -i Correlations % Change in COEP from Period 3 to Period 4 Change in TCR from Period 3 to Period 4 'Pearson Correlation % Change in COEP from Period 3 to Period 4 % Change in TCR from Period 3 to Period 4 1.000 .006 1.000 Sig. (2-tailed) % Change in COEP from Period 3 to Period 4 % Change in TCR from Period 3 to Period 4 .969 .969 % Change in COEP from Period 3 to Period 4 96 Change in TCR from Period 3 to Period 4 39 39 39 39 A22 Correlations Change COEP from Period 3 to Period 4 Change in JAR from Period 3 to Period 4 58318011 Correlation % Change COEP from Period 3 to Period 4 Change in JAR from Period 3 to Period 4 1.000 -.121 -.121 1 .000 Sig. (2-taiied) % Change COEP from Period 3 to Period 4 Change in JAR from Period 3 to Period 4 .463 .463 Change COEP from Period 3 to Period 4 Change in JAR from Period 3 to Period 4 39 39 39 39 A23 Correlations Change COEP from Period 2 to Period 3 96 Change in ICR from Period 3 to Period 4 'Pearson Correlation 96 Change in COEP from Period 2 to Period 3 96 Change in iCR from Period 3 to Period 4 1.000 -.220 -.220 1.000 Sig. (2-talled) 96 Change in COEP from Period 2 to Period 3 96 Change in iCR from Period 3 to Period 4 .178 .178 96 Change in COEP from Period 2 to Period 3 96 Change in iCR from Period 3 to Period 4 39 39 39 39 A24 Correlations 96 Change in COEP from Period 2 to Period 3 96 Change in NICR from Period 3 to Period 4 'Pearson Correlation 96 Change in COEP from Period 2 to Period 3 96 Change in NICR from Period 3 to Period 4 1.000 -.175 -.175 -1.000 Sig. (2-talled) 96 Change in COEP from Period 2 to Period 3 96 Change in NICR from Period 3 to Period 4 .288 .288 96 Change in COEP from Period 2 to Period 3 96 Change in NICR from Period 3 to Period 4 39 39 39 39 iJthfl'“-q 1 . v a I 1...“? L Conehfiows ‘% Change m COEP hom Pmku 2k: Puwd3 cmmme lnTCR hem Pamm 3h) Pakd4 lflmmmn Correlation 96 CMmme m COEP flom Pawn 2h) Fauna 96 Change lnTCR flom Faun 3k) lfiubd4 10m) -207 -207 11m0 Sig. (2-taiied) '% Change m COEP flom Pmkm 2k) Puwd3 ‘% Chane lnTCR hem Pmkm 3k) med4 205 205 96 CNmme m COEP flom Pmkfl 2k) lfinbd3 96 cmmme lnTCR flom Fahd 3k) Pmmd4 39 39 39 39 A26 Correlations Change COEP from Period 2 to Period 3 Change in JAR from Period 3 to Period 4 ‘Pearson Correlation 96 Change in COEP from Period 2 to Period 3 % Change in JAR from Period 3 to Period 4 1.000 -.114 -.114 1 .000 Sig. (2—tailed) 96 Change in COEP from Period 2 to Period 3 96 Change in JAR from Period 3 to Period 4 .489 .489 96 Change 39 39 39 39 A27 Correlations Change COEP from Period 1 to Period 2 Change in lCR from Period 3 to Period 4 Pearson Correlation 96 Change in COEP from Period 1 to Period 2 96 Change in iCR from Period 3 to Period 4 1.000 -.385' -.385" 1.000 Sig. (2-taiied) 96 Change in COEP from Period 1 to Period 2 96 Change in iCR from Period 3 to Period 4 .015 .015 96 Change in COEP from Period 1 to Period 2 96 Change in iCR from Period 3 to Period 4 39 39 39 39 '. Correlation is significant at the 0.05 level (2-tailed). A28 Correlations 96 Change in COEP from Period 1 to Period 2 96 Change in NICR from Period 3 to Period 4 'Pearson Correlation 96 Change in COEP from Period 1 to Period 2 96 Change in NICR from Period 3 to Period 4 1.000 -.407" -.407' 1.000 Sig. (2-taiied) 96 Change in COEP from Period 1 to Period 2 96 Change in NICR from Period 3 to Period 4 .010 .010 96 Change in COEP from Period 1 to Period 2 96 Change in NICR from Period 3 to Period 4 39 39 39 39 '. Correlation is significant at the 0.05 level (2-taiied). A29 Correlations Change COEP from Period 1 to Period 2 Change lnTCR from Period 3 to Period 4 'Pearson Correlation 96 Change in COEP from Period 1 to Period 2 96 Change in TCR from Period 3 to Period 4 1 .000 -.400* -.400" 1 .000 Sig. (2—talied) 96 Change in COEP from Period 1 to Period 2 96 Change lnTCR from Period 3 to Period 4 .012 .012 96 Change in COEP from Period 1 to Period 2 96 Change in TCR from Period 3 to Period 4 39 39 39 39 *. Correlation is Significant at the 0.05 level (2-tailed). A30 Correlations Change COEP from Period 1 to Period 2 Change in JAR from Period 3 to Period 4 'Pearson Correlation 96 Change in COEP from Period 1 to Period 2 96 Change in JAR from Period 3 to Period 4 1.000 -.133 -.133 1.000 Sig. (2-tailed) 96 Change in COEP from Period 1 to Period 2 96 Change in JAR from Period 3 to Period 4 .419 .419 96 Change in COEP from Period 1 to Period 2 96 Change in JAR from Period 3 to Period 4 39 39 39 39 APPENDIX F - Regression Output Model Summary” Model Variables Entered Removed R Square Adjusted R Square Std. Error of the Estimate 96 Change in COEP from Period 1 to Period 2°" .385 .148 .125 .1127 a. Dependent Variable: 96 Change in iCR from Period 3 to Period 4 b. Method: Enter c. independent Variables: (Constant), 96 Change in COEP from Period 1 to Period 2 d. All requested variables entered. ANOVA‘ Sum of Mean Model Squares df Square F SE. 1 Regression 8.2E-02 1 825-02 6.444 .015 Residual .470 37 1.3E-02 Total .552 38 a. Dependent Variable: 96 Change in iCR from Period 3 to Period 4 b. independent Variables: (Constant), 96 Change in COEP from Period 1 to Period 2 Coefficients' Standar dized Unstandardized Coefilci Coefficients ents Std. Model 8 Error Beta t Sig. 1 ( onstant) -3.6E-03 .027 -.136 .893 96 Change in COEP from -.452 .178 -.385 -2.538 .015 Period 1 to Period 2 a. Dependent Variable: 96 Change in iCR from Period 3 to Period 4 A32 Model Summary“ Std. Error Variables Adjusted or the Model Entered Removed R R Square R Square Estimate 1 96 Change in $05? .407 .166 .143 .2068 rom Period 1 to c ‘ Period g ' a. Dependent Variable: 96 Change in NICR from Period 3 to Period 4 b. Method: Enter 6. Independent Variables: (Constant). 96 Change in COEP from Period 1 to Period 2 d. All requested variables entered. ANOVA' Sum of Mean Model Sgares dt Square F Sig. 1 Pegression .314 1 .314 7.343 .010 Residual 1.582 37 4.3E-02 Total 1.896 38 a. Dependent Variable: 96 Change in NICR from Period 3 to Period 4 b. independent Variables: (Constant), 96 Change in COEP from Period 1 to Period 2 Coefficients‘I Standar dized Unstandardized Coefflci Coefficients ' ents Std. Model 8 Error Beta 1 Sig. 1 (Constant) .268 .049 5.463 .000 96 Change in COEP from -.884 .326 -.407 -2.710 .010 Period 1 to Period 2 a. Dependent Variable : 96 Change in NICR from Period 3 to Period 4 A33 Model Summary“ Std. Error Variables Adjusted of the Model Entered Removed R R Square R Square Estimate 1 96 Change in COEP .400 .160 .137 .1425 from Period 1to 6 ‘ Period 2 ' a. Dependent Variable: 96 Change in TCR from Period 3 to Period 4 b. Method: Enter c. independent Variables: (Constant). 96 Change in COEP from Period 1 to Period 2 d. All requested variables entered. ANOVA' Sum of Mean Model Squares dt Square F SE. 1 Regression .143 1 .143 7.052 .012 Residual ' .751 37 2.0E-02 Total .894 38 a. Dependent Variable: 96 Change in TCR from Period 3 to Period 4 b. independent Variables: (Constant). 96 Change in COEP from Period 1 to Period 2 Coefficients‘ Standar dized Unstandardlzed Coeffici Coefficients ents Std. Model B Error Beta t Sig. 1 (Constant) .129 .034 3.809 .001 96 Change in COEP from -.597 .225 -.400 -2.655 .012 Period 1 to Period 2 a. Dependent Variable: 96 Change in TCR from Period 3 to Period 4 A34 31293 02638 2527