Sociation Today Fall/Winter 2013

At the start of the journal Sociation Today, one of the goals was the replace book reviews with reviews of important articles addressing issues of interest to sociologists. We now continue that tradition by looking at an empirically-based article which addresses the issue of how density affects human behavior.

Human behavior is examined in the social sciences in general, but today many of the theories developed by sociologists are being tested and updated not by sociologists, but by economists who use different theories but often the same data available publicly to all researchers. Brueckner and Largey (2007), economists interested in urban economics, have recently tested the concept of how urban density affects our relationships with each other.

Older Sociological Theory and Urban Behavior

    How density affects behavior in the city has long been a core concept in sociology. The sociologist Simmel, who coined the term "sociation," is well-known for his comparisons of urban and rural life. His approach, found in the work "The Metropolis and Mental Life" (1903) has been widely cited for over 100 years. Part of the original text is available here.

    Simmel's 100-year-old argument is that urban life tends to overload the senses, so people have to strive to limit their interactions with neighbors and others in order to preserve their energy. He uses the word blasé to characterize life in cities, suggesting that urbanites are dominated by social interactions which are brief, segmented and quite superficial.   Of course, urban life also gives the individual vastly greater freedom compared to rural areas where your life is known to all and the rules of the village apply.

    Louis Wirth's (1938) essay entitled "Urbanism as a Way of Life" also presents urban life as dominated by rational thought designed to preserve time. Thus he also characterized urban life as one in which our interactions are brief, transitory and segmented. The properties of the city which cause the changes in behavior are also well-known, being size, density and heterogeneity. Brueckner and Largey accept the density argument about human interactions, but do not bother to mention that size and density as key variables have long been associated with the Chicago School. Rather. they focus on the revisionist theory, mentioned below.

Revisionist Theory

Yet despite well-known theory of why urban life would be expected to be dominated by interactions which would be brief, segmented and superficial, books which dominate popular sociology state quite the opposite. According to Brueckner and Largey (2007)*, the well-known book Bowling Alone (Putnam 2000) argues the opposite. Putnam believes that the process of suburbanization has lowered urban densities and brought on a sense of being lonely and isolated. Putnam thus basically reverses the theories of Simmel and Wirth to state that as population density goes up, meaningful social interactions would also increase. Suburbanization is blamed for lowering population density and is thus branded as "bad." While Brueckner and Largey cite Putnam as their straw man, it is also true that the anti urban sprawl movement also argues that as density goes up, social interactions increase. After all, it is somewhat logical to think that if more people are around you, a person might have the ability to make more friends.

The Testing of the Revised Theories

Bruckner and Largey set out to test empirically the assumption that as population density goes up, social interactions would increase. As they say,

...(T)he main goal of the paper is to appraise the empirical relevance of an anti-sprawl argument based on social interaction. This task requires an empirical test of the underlying hypothesis, which asserts that social interaction is an increasing function of population density. If this hypothesis is validated, then the existence of a density externality follows naturally, leading to the conclusion that the spatial expansion of cities is excessive. (Brueckner and Largey, p. 2 of the free edition, link below.)

The data set is from the Social Capital Benchmark Survey, which was carried out in 2000 by the Saguaro Seminar at Harvard's Kennedy School of Government and later distributed by Roper. The sample size for those living in urban areas is 14,823.

The authors examine many variables, but are able to summarize the effects of reducing density from 11,591 per square mile to 799 per square mile as follows:

The results show that the given decline in density raises the likelihood of talking to neighbors at least once a weak (sic) by 7%, increases the number of times the respondent socializes with friends in a public place by 11%, raises the likelihood of belonging to a hobby club by 24%, and raises the number of club meetings attended, or the number of group memberships by 26% and 8%, respectively. While none of the effects is overwhelming in size, each is appreciable. It is also interesting to note that if density were to experience a larger drop, falling from 163,000 per square mile (from a log value of 12, near the sample maximum) to 779, then log value would fall by almost 5.4 instead of 2.7, and percentage declines in the last column of Table 5 would nearly double (Brueckner and Largey, pp. 13-14 from the Science Direct download).

Here is Table 5, explained above:

Table 5
Effect of decline in TRACT DENSITY from one std. deviation above mean to one std. deviation below mean*

	Marginal Effect	Change in level or probability	Mean	Percent
#NEITALK	-0.014	0.04 (prob)	0.538	7%
SOCPUBLIC	-0.690	1.14 (level)	16.8	11%
HOBBYCLUB	-0.022	0.06 (prob)	0.252	24%
CLUBMTGS	-0.570	1.54 (level)	6.0	26%
#GROUPS	-0.092	0.25 (level)	3.2	8%

*Change in TRACT DENSITY is -2.7, corresponding to a decline from 11,591 to 779 people per square mile.

The results in Table 5 have had controls introduced for several dozen control variables. As density goes up, social interaction goes down. In short, the old social theories are correct and the revisions proposed by authors such as Putnam are proven to be incorrect. For a detailed explanation of variables, please consult the footnote below.

The article considered 10 dependent variables. The results shown above were similar to the findings for the other variables considered.

Conclusions

Brueckner and Largey summarize their findings as follows:

Various authors, most notably Putnam (2000), have argued that low-density living reduces social interaction, and this argument has been used to buttress criticisms of urban sprawl. But urban expansion must involve market failures if it is to be inefficient, and this paper shows that such a distortion indeed arises if low density depresses social interaction. Then, in appraising the gains from greater individual consumption of living space, consumers fail to consider reduced interaction benefits for their neighbors, which arise through lower neighborhood density. Space consumption is then too high, and cities are excessively spread out.

The key element in this argument is a positive link between social interaction and neighborhood density, and the paper tests empirically for such a link. The results are unfavorable: whether the focus is friendship-oriented social interaction or measures of group involvement, the empirical results show a negative, rather than positive, effect of density on interaction.

The paper’s findings therefore imply that social-interaction effects cannot be credibly included in the panoply of criticisms directed toward urban sprawl. In fact, the results suggest an opposite line of argument (Page 16 in the free edition).

In attempting to explain why increased density seems to reduce social interaction, the authors, as economists, seem to ignore the theories of Wirth and Simmel, at least by name. But they do seem to think that the need for privacy in a crowded urban environment might one valid interpretation of the findings. It is possible that, as sociologists, we might even have gone farther to state that the authors really should have developed this line of argument and given credit the theories of earlier scholars who are well-known for their theories of how size and density would affect human interaction and behavior.

Nevertheless, Brueckner and Largey have presented us with a very sophisticated and complex analysis of an important data source. So often these days it seems that the core concepts of sociology are incorporated into other disciplines where they have been found useful. I can only wonder why this article was not written by a sociologist!

*Footnotes

The Brueckner and Largey article is available as a free manuscript at
<http://www.economics.uci.edu/files/economics/docs/workingpapers/2006-07/Brueckner-07.pdf>
The published version is available from Science Direct if your institution is a member of this service at:
<http://www.sciencedirect.com/science/article/pii/S0094119007000939>
The downloaded versions have different page numbers from the published version, also available for a considerable fee from the original journal.

Here is how each variable in Table 5 was coded: