Cities, Citizens, and Crime: The Ecological/Nonecological Debate Reconsidered

  • James M. Byrne
Part of the Research in Criminology book series (RESEARCH CRIM.)


Over the past 50 years, a great deal of thought, not to mention a considerable body of empirical research, has been directed at a deceptively simple question: Why do some cities have higher rates of crime and other forms of “unconventionality” (e.g., conflict, suicide) than others? Not surprisingly, most researchers have begun their examination of intercity variation in crime rates by looking at the gross variation in index crime rates by city size. What we invariably find is “a monotonic relationship between crime rates and the size of the reporting jurisdiction” (Chaiken & Chaiken, 1983, P.17). In 1980, for example, the index crime rate in very large cities (i.e., cities with populations over 250,000) was almost twice as high as the comparable small city (i.e., cities in the 10,000–25,000 population range) rate (9,356 vs. 5,346 per 100,000 pop.).1 How can we explain these differences? Is there something (inherently bad) about the structure of larger cities or the individuals who reside in them which could explain these findings? Or does the answer lie elsewhere?


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    See Jan M. Chaiken and Marcia R. Chaiken, “Crime Rates and the Active Criminal,” p. 17, Table 2, in James Q. Wilson (1983).Google Scholar
  2. 2.
    In particular, see Louis Wirth, “Urbanism as a Way of Life,” in Albert J. Reiss, Jr. (1964). See also, Wirth (1938).Google Scholar
  3. 3.
    See Herbert Gans (1968).Google Scholar
  4. 4.
    See Fischer (1975). Fischer emphasized that, while there is little empirical evidence to confirm Wirth’s urban alienation hypothesis, “the association between urban residence and unconventionality is pervasive” (p. 1322).Google Scholar
  5. 5.
    Gottfredson and Taylor (chapter 8, this volume) take a similar approach (using neighborhood) in their study of parolees in Baltimore. In particular, see their discussion of the Olweus’ person-environment integrity model. Of course, the interactionist perspective is exemplified by Lewin (1936).Google Scholar
  6. 6.
    Clearly, this is not what most criminologists would describe as the ecological approach. Most of us still link ecology and crime to work of the Chicago School of Social Ecology which included Burgess, Park, and Shaw and McKay, or perhaps more specifically, the zonal model of cities and crime. However, after reviewing these early efforts (and the subsequent critiques) Bottoms concluded that “ecology in the full sense as delineated by Park is justly discredited” (Baldwin & Bottoms, 1976, p.14).Google Scholar
  7. 7.
    For a critique of this approach, see Gideon Sjoberg, “Theory and Research in Urban Sociology,” Chapter 5, in Hauser and Schnore (1965).Google Scholar
  8. 8.
    A more detailed discussion of the ecological complex is found in Chapter 2 in Byrne (1983).Google Scholar
  9. 9.
    Cohen (1981) recently suggested that criminal opportunities are affected by more general technological changes. For example, the increased number of lightweight goods (e.g., TV sets, stereos) available to steal has, in essence, made burglary easier and more profitable.Google Scholar
  10. 10.
    Because two elements of the ecological complex—technology and social organization—were measured inadequately, any statements about the applicability of a general ecological model are equivocal, underscoring the preliminary nature of the analysis.Google Scholar
  11. 12.
    See Chapter 2 in Byrne (1983). Harries (1980) has reported similar variations in violent and property crime.Google Scholar
  12. 13.
    These findings may not be accurate, however. Briefly, the authors’ initial factor analysis included all cities; but, at one point, they used the extracted factors as independent variables in separate regression analyses of large and small cities. They should have redone their factor analysis for each subsample of cities (i.e., those over and under 100,000). The same problem biased their finding of regional variation in crime correlates.Google Scholar
  13. 15.
    It is interesting to note that the 5 boroughs of New York City were inadvertantly included as separate cities (along with N.Y.C.) in the County and City Data Book. The boroughs were removed from this study. For a discussion of the issue see Byrne (1983: Chapter 4). The comparability of the three randomly selected subsamples can be directly assessed by examining Appendix A, Table 1, in Byrne (1983).Google Scholar
  14. 16.
    For a fuller discussion of these issues, see Chapter 4 Byrne (1983).Google Scholar
  15. 17.
    See Kowalski et al. (1980) for a discussion of regional crime patterns. In addition, see Flango and Sherbenou’s (1976) analysis of the culture of poverty in Southern cities.Google Scholar
  16. 18.
    The selection of the city as the “appropriate unit” of analysis in a macro-environmental study of crime is not discussed here. (See Byrne, 1983, for a discussion of this issue.) However, a note on the “definition” of cities is in order. Walton and Cams (1973) have observed that:Google Scholar
  17. 18a.
    The concept “city” has not always meant what it does today, and probably will not mean the same thing in the future. As we use the term now, a city consists of a relatively dense population living off an agricultural hinterland, with or without manufacturing, but with some form of interdependence and specialization of functions. (p.11)Google Scholar
  18. 18c.
    Briefly, use of such concepts as Gibbs and Erickson’s (1976) “ecological position” allows us to explore this interdependence without using the entire SMSA as the unit of analysis. For a somewhat different perspective, see Hoch (1974) or Stafford and Gibbs (1980).Google Scholar
  19. 19.
    Only the number of cities used by Flango and Sherbenou (1976) is comparable (n = 840 cities).Google Scholar
  20. 20.
    Other population compositional variables which should be included in subsequent analyses (at least in the initial phases) are population change, unemployment stability, residential mobility, family composition, and female laborforce participation rate. (For a complete review of these variables, see Chapter 4 in Byrne, 1983.)Google Scholar
  21. 21.
    Other physical characteristic variables which have been employed in previous studies include the employment/residence ratio, structural density, technology, community age, occupational diversity, climate, and seasonality. (For a complete review, see Chapter 4 in Byrne, 1983.)Google Scholar
  22. 23.
    Michael Lewis-Beck (1980) outlines the common solutions to this problem. See also Holmes (1979, 1981); Belsley, Kuh, and Welsh (1980); Farrar and Glauber (1967); and Rockwell (1975).Google Scholar
  23. 24.
    Indeed, when the cities identified as outliers are dropped from the analysis, the corresponding correlations often change dramatically. See Chapter 4 in Byrne (1983).Google Scholar
  24. 25.
    For example, marked differences in the size of the correlations can be identified for robbery. The bivariate correlation between percent youth (18-to 25-year-olds) and robbery is —.47 for the larger cities, but.04 for the smaller cities.Google Scholar
  25. 26.
    This hypothesis uses somewhat misleading terminology. “Relative importance” simply refers to the size of the standardized regression coefficient (Beta) for a particular variable in each of the four crime-specific all-city regression analyses. Here we are comparing the same sample of cases using the construction sample (n = 356).Google Scholar
  26. 27.
    No specific tests were conducted to establish the significance of these differences. This approach has been criticized by some. See, for example, Chapter 2 in Rich, Sutton, Clear, and Saks (1982.)Google Scholar
  27. 28.
    See Sampson (1983) for a discussion of this issue.Google Scholar
  28. 29.
    For example, there are striking differences in the impact of the SMSA/city population ratio on burglary in larger versus smaller cities. The unstandardized coefficient (b) is -34.2 for cities over 100,000 but b = 1.7 for cities under 100,000. (See the discussion in Byrne, 1983, pp. 366–368.)Google Scholar
  29. 30.
    Clearly, we have no reason to expect such regional variation from the nonecological perspective. But perhaps the distribution of larger and smaller cities in the North Central region affected the size-robbery correlation.Google Scholar
  30. 31.
    Flango and Sherbenou’s (1976) urbanization variable was one of six extracted factors the authors used as independent variables in a multiple regression analysis. They found that urbanization explained 40.8% of the variance in robbery in larger cities (n = 153), but only 12.6% of the variance in smaller cities (n = 687).Google Scholar
  31. 32.
    The black/white dichotomy presented in this section ignores other minorities, in particular citizens of Spanish descent The data on other groups were not available from the 1977 County and City Data File. Thus, the white category is actually a residual category including all other posssible minorities. Clearly, this limits our understanding of ethnicity in areas of the country (e.g., percent Mexican-American in Houston) where blacks are not the dominant minority group. See Chilton, Chapter 6 for a discussion of this volume.Google Scholar
  32. 33.
    See Christopher S. Dunn, “Crime Area Research,” Chapter 1, in Georges-Abeyie and Harries (1980, p.5–25). According to Dunn, we cannot explain away the ethnicity effect in terms of socioeconomic status.Google Scholar
  33. 34.
    In particular, Georges-Abeyie (1981) provides a discussion of the following criminological explanations of this correlation (e.g., labelling theorists; conflict theorists; conservative criminologists, such as control theorists; the structural functionalists; and social ecologists). This is the author’s typology of theory.Google Scholar
  34. 35.
    The compositional (nonecological) position holds that this relationship will disappear once relevant population-compositional variables are controlled. One alternative explanation here is that inadequate measures of population composition preclude this type of comparison.Google Scholar
  35. 36.
    For an excellent compendium of research on cities and the change process, see Walton and Carns (1973).Google Scholar

Copyright information

© Springer Science+Business Media New York 1986

Authors and Affiliations

  • James M. Byrne

There are no affiliations available

Personalised recommendations