Reston At The Tipping Point--violent Crime & Increased Population Density

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1)

Reston at the Tipping Point: Violent Crime and Increased Population Density (Paper No. 1) Terrill D. Maynard Reston, Virginia January 15, 2007

Preface Reston is at a tipping point, a point of no return in its maturation as one of the nation’s premier planned communities. Within weeks, the Fairfax County Department of Planning will present proposed changes in the county’s Planned Residential Community (PRC) zoning code to the Board of Supervisors that, most importantly, would allow the community’s 64,000 or so population to mushroom by more than 25 percent to nearly 81,000. This change would be accomplished by simply changing three numbers: the household population factors for single-family detached (houses), single-family attached (townhouses), and multi-family (condominium) dwelling units. By reducing these factors, established in the mid-1970s, to reflect the reality of smaller household sizes in the 21st century, the County would allow developers to build as many as 5,545 single-family detached or 6,161 single-family attached or 7,921 multi-family dwelling units. The proposed density increase comes at a time when the County and Reston are anticipating the 2012 arrival of Metrorail at Wiehle Avenue and, several years later, at Reston Parkway as part of the Dulles Corridor Metrorail extension. To take advantage of the new access to public transit and rationalize development around existing Metrorail stations, the County is developing a set of guidelines for Transit-Oriented Development (TOD) within a half-mile radius of existing and proposed Metrorail stations. These “urban villages” at station areas and in nearby Reston Town Center will feature highdensity multi-family units, high-density office space, and a variety of retail businesses linked to the nearby Metrorail stations. This is the first of a series of papers to examine the implications of these proposed changes for Reston’s quality of life. 1 The core argument of this series is that the Fairfax County Board of Supervisors must recognize the full reality it would impose on Reston and the County before it decides to change only one element of that reality—household population factors.

1

These papers build on an earlier overview by this author called “Future Reston Development: A First Look at Some Implications for Quality of Life,” November 13, 2006. The paper is available upon e-mail request to [email protected].

1

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1)

Introduction Crime grows as population grows. It grows for many reasons, but the growth is largely a function of a growing number of opportunities, criminals, and potential victims. Of the categories of crime, violent crime—crimes against the person, including murder, rape, robbery, and aggravated assault as defined by the FBI in its Uniform Crime Reporting (UCR) system—most seriously affects quality of life. It is the sometimes literally the sharp edge of crime and warrants special consideration in understanding the implications for Reston’s quality of life of the potential increase in Reston’s density by more than a quarter. The most salient questions in looking at violent crime are whether population density is a determinant of violent crime and, if so, the extent to which violent crime rates grow as population density increases. This brief paper tries to capture the essence of independent academic research on these key issues and apply it to the situation Reston—and the County Board of Supervisors—faces as it contemplates changes in the community’s authorized population density. In particular, it summarizes the findings of key credible research—those that follow rigorous analytical techniques and are not driven by particular views (e.g.— “smart” growth) and provides limited statistical information about violent crime in northern Virginia counties to test the applicability of the broader results to Fairfax County. Most importantly, it presents some original research based on analysis of a sample of wealthy and populous counties (such as Fairfax County) to examine the specific implications of increased density for violent crime in Reston.

Key Academic Research Results Volumes of detailed academic, government, think tank, and other research have been conducted over decades on the relationship between population density and deviant social behavior of all kinds, including violent crime. In wrapping up the multitude of judgments about the link between population density and violent crime, a 1994 National Academy of Sciences report stated: “Finally, several studies report a significant and large association between population density and violent crime. The average correlation across neighborhoods between population density and violent crime rates is .68 in the studies of Beasley and Antunes (1974), Mladenka and Hill (1976), and Smith and Jarjoura (1988). In the latter multivariate study, density also had one of the strongest effects on violent crime, despite controlling for a host of social and economic variables.” 2 2

Robert J. Sampson and Janet L. Lauritsen, “Violent Victimization and Offending: Individual-, Situational-, and Community-level Risk Factors,” Understanding and Preventing Violence, Volume 3: Social Influences, Albert J. Reiss, Jr., and Jeffrey A. Roth, eds., National Academy Press, Washington, DC, 1994, p. 55.

2

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1)

To be sure, population density is a top-level demographic determinant of violent crime and only one of a multitude of individual, situational, and community risk factors addressed in this one hundred-plus page chapter by Robert J. Sampson, one of America’s most respected and prolific writers on urban issues, and Janet L. Lauritsen. . A research report published in the British Journal of Sociology the following year that examined drivers of violent and property crimes in 98 percent of all U.S. counties (3,076 counties) corroborated the results of the National Academy study. 3 The focus of the research was on testing the “subculture of violence” theory and its possible drivers, especially in the South and among African-Americans. 4 Few of the study’s findings supported the subculture of violence theory prevalent at the time, largely putting the concept to rest, but—by examining a series of variables that might explain criminal activity—it provided valuable insights into the importance of population density (and population change) as a determinant of violent crime. In summarizing their results, the authors note: “Urbanity is the main determinant of property crime, urbanity and population density are important factors in violent crime, and poverty, divorce and density figure strongly in homicide.” These results are captured in Table 1 below taken from this study, which shows the variables in their order of importance as determinants of violent crime as reflected in the absolute value of their “t” scores. Table 1: Regression Results for Determinants of Violent Crime In a County-level Study of U.S. Counties Variable

b

Beta

t

Signif.

% Black 6.941 .369 15.1 .0000 % Urban 3.031 .317 15.0 .0000 Pop. Density .031 .196 12.4 .0000 % Hispanic 4.399 .161 8.9 .0000 % Church members -1.550 -.101 -5.3 .0000 % Nat. American 4.299 .086 5.2 .0000 % Unemployed 16.939 .077 4.4 .0000 % Pop. Change 1.116 .093 4.3 .0000 % Age 5-17 -10.391 -.095 -4.0 .0001 Divorce Rate 6.342 .053 3.1 .0025 ___________________________________________________________ Source: Kposowa, et al., p. 93.

3

Augustine J. Kposowa, Kevin D. Breault, & Beatrice M. Harrison, “Reassessing the Structural CoVariates of Violent and Property Crimes in the USA: A County Level Analysis,” The British Journal of Sociology, V. 46, No. 1, March 1995, pp. 79-105. 4 To help minimize analytical bias in using others’ research results, I have found it useful to use the results—where the results stem from rigorous analysis—in a manner not intended by the original research. In this case, I have shifted the focus from the subculture of violence to the impact of population density.

3

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1) Violent Crime in Northern Virginia A look at the situation in northern Virginia using 2004 data suggests that the judgments reported a decade ago by the National Academy and Kposowa, et al., remain true today. U.S. Bureau of Justice Statistics for the eight northern Virginia local police agencies in 2004 corroborates that, as the Academy study states, violent crime rates increase with population density—except for rural localities. (See Figure 1 below.) The correlation between density and violent crime rates in this small northern Virginia sample is .64, similar to the rates recorded in the much larger studies noted by the National Academy. Fairfax County, with some 59.3 violent crimes per 100,000 population in 2004, is right on the trend line for northern Virginia counties.5 Moreover, the “J-curve” shape of the trend line is consistent with a broad range of research studies showing rural violent crime rates somewhat higher than suburban rates with violent crime rates increasing sharply as density increases in urban areas.

Figure 1: Northern Virginia Counties/Cities: Violent Crime Rate by Population Density, 2004 350

Violent Crimes per 100,000 People

300

Alexandria

250 Arlington County

200 Prince WilliamCounty Spotsylvania County

150

Stafford County

100

Loudon County Fairfax County

50

Current Reston Density

Proposed Reston Density

Fairfax City

0 0

2

4

6 8 People per Acre

5

10

12

14

Data from the Fairfax County Police Department available on the County’s website indicates that the County had a much higher violent crime rate in 2004—99.77 violent crimes per 100,000 people. The BJS data is used here to ensure comparability across multiple jurisdictions that may have different reporting standards and policies.

4

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1) Testing a Sample of Populous and Wealthy Counties Although scientific research and contemporary anecdotal information on northern Virginia counties suggest that density plays an important role in violent criminal behavior, these results do not adequately control for two vital features of Fairfax County: its economic vitality and its large population. To control for these factors, a data set containing pertinent information combining the 100 wealthiest U.S. counties (as measured by median household income) and the 217 counties with a population of 250,000 or more based on U.S. Census 2000 data. The result is a sample set of the 42 wealthiest and most populous counties in the United States (see Figure 2 below). Even in looking at this group of its populous and wealthy peers, it is important to note that Fairfax County had the highest median household income in the country in 2000. Fairfax was also the fourth most densely populated county in the 42-county sample behind Nassau County, NY, Orange County, CA, and DuPage County, IL. In short, Fairfax County is a true leading edge county among its peers as well as in the nation. Figure 2: Median Household Income & Population Density, Populous & Wealthy US Counties, 2000 $85,000 Fairfax County

Median Household Income

$80,000 Morris & Somerset Counties, NJ

$75,000 Nassau County, NY

$70,000

DuPage County, IL Orange County, CA

$65,000 $60,000 $55,000 $50,000 0

1,000

2,000

3,000

4,000

5,000

People per Square Mile

Of these 42 elite counties, the FBI has complete UCR data on twenty-eight counties for 2002, the year closest to the 2000 census for which count-level violent crimes data is publicly available. Even so, two of these counties are extreme outliers: Anne Arundel County, MD, with a reported violent crime rate of 650 per 100,000 population and Rockland County, NY, with a rate of 1.4 per 100,000. These appear to be

5

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1) artifacts and these two counties were dropped from the sample, leaving a sample size of 26 counties. There appear to be other issues with the FBI Uniform Crime Reporting (UCR) data, in particular, the FBI data and Fairfax County Police Department are not always consistent, reporting is voluntary and some states (including Illinois) and many counties (including New Jersey) do not report this information, etc. Nonetheless, it is the most complete and uniform set of nationwide serious crime data available and sufficient for this preliminary assessment. Population Density As shown below (Figure 3), a linear regression analysis of the FBI violent crime rates in these counties against their population shows that, even among the 26 wealthiest American counties, a positive relationship exists between density and violent crime rates. 6 From a statistical analysis perspective, the highly dispersed nature of this small sample, reflected in the extremely small coefficient of determination (r2=.0113) and broad standard deviation (82.0), undercuts the explanatory power of the relationship, however. Fairfax County fares extremely well in comparison with its peers in this regression analysis. The violent crime rate in Fairfax County in 2002 (51.9 per 100,000 according to the FBI Crime in the United States (CIUS) report) was much less than the average for this elite set of counties (an average violent crime rate of 77.3 per 100,000 people). It was also only about three-fifths (59.1%) of the rate estimated by the regression for a wealthy and populous county of its density in 2000 (2,449 people per square mile). More broadly, the County rate was less than one-ninth the violent crime rate for the nation as a whole in 2002 (494.6 violent crimes per 100,000 people) near the end of a decade of annual declines in the national violent crime rate. The specific causes for this exceptional result are beyond the scope of this research, but no doubt include a culture of strong family and personal values, a vibrant local economy, a highly effective police force, and a strong social services program.

6

This paper uses linear regression analysis using the least squares method throughout. Non-linear logarithmic analysis better fits the entire sample range with the curve bending sharply toward the origin on the left and the slope flattening some on the right. Nonetheless, over the range of values and sample size explored here, the linear regression is suitable and easier to comprehend.

6

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1)

Figure 3 : Violent Crime Rate and Population Density, 2000/2002: Populous & Wealthy US Counties 300 y = 0.0086x + 66.042 2

Violent Crimes per 100,000 People

R = 0.0113 250 Current Reston Density

Proposed Reston Density

200

150

100

Fairfax County

50

9,000

8,000

7,000

6,000

5,000

4,000

3,000

2,000

1,000

0

0

People per Square Mile Sources: US Census 2000, FBI CIUS 2002

A serious concern arises, however, when one applies this model to Reston’s current and prospective density, which are both literally “off the chart” in density in this sample of wealthy and populous counties. Even now, at 10.33 persons per acre (6,611 per square mile) Reston’s density is 42% greater than Nassau County, NY, and moving to a density of 13 people per acre would make Reston 79% more densely populated than Nassau County. Using the model alone suggests that the 2002 violent crime rate in Reston ought to have been about 123 violent crimes per 100,000 people. Discounting that rate for the overall low violent crime rate in Fairfax County suggests that Reston would have had a violent crime rate of 88 per 100,000 people, or 57 violent crimes in 2002. Although data on Reston’s violent crimes is not available (the Reston Police District data incorporates most of northwest Fairfax County, excluding Herndon), these results intuitively seem close to the mark. Increasing Reston’s population density to 13 people per acre (8,320 per square mile) and using the same regression and county adjustment in this sample suggests: (a) Reston’s projected violent crime rate is likely to rise 17 percent to about 102 per 100,000 people and (b) its total violent crimes are likely to rise to about 83 per year. Thus, by looking simply at the relationship between population density and violent crime, the model—adjusted for the County’s exceptionally low violent crime rate—suggests that violent crime in Reston would climb some 47% as the community reaches its maximum allowable population density.

7

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1) Increased Density over Time More important than a relationship between population density and violent crime at a point in time is the extent to which increasing population density affects violent crime at the end of the growth period. Using changes in population from 1990 and 2000 reported in U.S. census data for the same 26-county sample of wealthy and populous counties provides some insight in to the relationship. Three of the top four fastest growing counties were in Texas (Collin, Denton, and Fort Bend counties) and Gwinnett County, GA, was the fourth. Collin and Denton counties are noteworthy in that they experienced huge percentage increases in population density over the decade (86% and 56% respectively) while maintaining exceptionally low violent crime rates in 2002 (19.9 and 26.8 per 100,000). Still, these two large Texas counties had a population density less than one-quarter of that in Fairfax County in 2000. Results for rapidly growing Gwinnett County, GA, most illustrate the link between density growth rate and violent crime rate, and nearby Montgomery County, MD, show that violent crime rates may be high even if density increases are relatively low. (See Figure 4 below.) Also, as with the earlier regression, the relationship between the dispersed data and the small sample size suggest a modest explanatory value in the relationship as reflected in the moderate correlation (corr. = .1742), small coefficient of determination (r2 = .0303), and other statistical measures.

Figure 4: Violent Crime Rate and Increased Density, 1990-2000: Wealthy and Populous U.S. Counties 300

Montgomery

y = 69.942x + 59.361 2 R = 0.0303 Gwinnett

200

150

100 Denton

Fairfax County

Collin

50

Growth in Population Density

8

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0 0%

Violent Crime Rate per 100,000

250

Proposed Increase in Reston Density

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1)

In contrast to its leading role in wealth and population density in this 26-county sample, Fairfax County’s population density grew at a below average rate (18.5%) compared with the 25.6 percent growth rate for this peer group in the 1990s. Probably the key reason for Fairfax’s comparatively low density growth rate over the decade is the fact that, among the counties in this sample, it was already the third most densely populated county in 1990 behind Nassau County, NY, and Orange County, CA. As in the earlier results, its 2002 violent crime rate (51.2 per 100,000) was still significantly below that suggested by a linear regression of the 2002 rate against the change in population (72.3 per 100,000 population). As with the earlier regression on density alone, the stronger relationship between increased density and increased violent crime is worrisome when applied to the situation in Reston. Using the results of the previous regression, we estimate that Reston has a violent crime rate of 88 per 100,000 population or about 57 violent crimes per year consistent with its current estimated population of 64,277, i.e.—consistent with the County’s low violent crime rate and intuitively close to actual results for 2006. If Reston nears its maximum allowable population density by growing 25 percent in the next ten years, the regression indicates Reston’s violent crime rate would grow 23.8 percent to 105.5 violent crimes per 100,000 people, and its number of violent crimes would grow by 59.8 percent to 91 violent crimes per year. (See Figure 5 below.) A possible methodological concern in this calculation is whether Reston will achieve its full proposed allowable population increase of more than 25 percent in a decade. In the 1990s, counties among this peer group averaged better than 25 percent increases in density and, now in Reston, developers already have applications pending for 3,570 multi-family units that could add nearly 7,500 people—about 47% of the potentially allowable population growth—within the next 3-5 years. Thus, it is not unreasonable to expect Reston to reach its maximum allowed population density within a decade.

9

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1)

Figure 5: Reston: Violent Crime & Violent Crime Rate Increases from Increased Density over a Decade 120

Violent Crime

Proposed Reston Density Increase

Violent Crime Rate 100

Count

80

88.0

91.5

95.0

98.5

102.0

105.5

109.0

60

40 57

62

0%

5%

73

67

79

85

91

20

0 10%

15%

20%

25%

30%

Percent 10-year Density Increase

Multivariate Analysis Combining Density and Density Increase Effects Examining this 26-point data set of wealthy and populous counties nationwide as a whole using more rigorous linear multivariate analysis techniques paints a bleaker picture of the potential impact of greater density on violent crime while increasing the explanatory power of the regression. In this case, both the population density in 2000 and the change in population in the decade 1990-2000 are solved simultaneously to assess the effect on violent crime. The addition of the density change variable raises the explanatory value of the model (to r2 = .0656) although the coefficient is still too small for full-scale statistical analytical purposes. One can use the regression results (Table 2 below) to project the impact different of changes in density over a decade. As reflected in Figure 6 below, the impact of those changes grows dramatically density increases. If Reston were to “fill out” to near its maximum density—a 25 percent increase in population density—over the next decade, the regression suggests that the violent crime rate in Reston would rise by 60 percent and the number of violent crimes would virtually double from 57 to 113.

10

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1)

Table 2: Multivariate Linear Regression Results for Density and Changes in Density (LSM) Factor Coefficient = Sex = R2 = SEy = Std Dev (y) = N=

d(Density) 101.33 87.66

Density 0.0165 0.0177

Intercept 29.88 41.14

0.0656 82.59 81.95 26

Figure 6: Reston: Impact of Current Population Density & Increased Density over a Decade on Violent Crime & Violent Crime Rate 160 Proposed Increase in Reston Density

Violent Crimes per Year 140

Violent Crime Rate 140.5 130.0

120 119.5

100

98.5 80

113

109.0

100

88

88.0 77 66

60 57 40 20 0 0%

5%

10%

15%

Percent 10-year Density Increase

11

20%

25%

Reston at the Tipping Point: Violent Crime and Increased Population Density (No. 1) Conclusion The preceding regression analyses of a sample of wealthy and populous peer counties nationwide suggest that increasing Reston’s authorized population density by reducing the household population factors alone may have a profound adverse impact on violent crime—and the quality of life—in the community. In particular, the results suggest that Reston’s violent crime rate per 100,000 people could escalate some 17-60 percent if the community’s density were increased by the 25 percent proposed in the PRC zoning law changes. The concrete result of that density increase is the likelihood that violent crime in Reston would increase by 47-98 percent, from an estimated 57 violent crimes per year to somewhere between 83 and 113 violent crimes per year. Although there can be little doubt that violent crime rates and, hence, violent crime, tend to increase with increased population density, the results presented here are not without their methodological issues. The key methodological issues relate to the small size and highly-dispersed results of the sample in each of the three regressions. These characteristics undermine the explanatory power of the regression analysis. In addition, other potential determinants of violent crime may need to be factored into the analysis. Of value in making a decision to allow increased density would be an assessment of the impact of potential strategies to mitigate increases in violent crime with density increases and their associated costs. In the end, the Fairfax County Board of Supervisors may decide that it is worth the social and political cost to change the PRC zoning ordinance to accommodate more intensive development in Reston. Before it does that, however, it ought to conduct a thorough, systematic analysis of the impact of population density increase on the full range of economic and quality of life values, not just violent crime. From a Restonian’s perspective, this preliminary assessment suggests a decision to increase the community’s population allowed density by some 25 percent will lead to a substantial deterioration in a critical element of Reston’s quality of life—the risk of being a victim of violent crime. In short, Reston would become a more dangerous place to live, work, and play.

12