Liveblog of Patrick Sharkey’s presentation to the Inequality & Social Policy Seminar Series at Harvard on September 23, 2013.

Patrick Sharkey is an associate professor of sociology at New York University and affiliate of the Robert F. Wagner School of Public Service. His research looks at stratification and mobility with a specialized interest in neighborhoods and cities, and he is the author of the recently published book Stuck in Place: Urban Neighborhoods and the End of Progress toward Racial Equality.

His talk entitled “Where, when, why, and for whom do neighborhoods matter?” was based on a just completed review paper, co-authored with Jacob Faber, updating the literature on neighborhood effects (specifically the research looking at cognitive / developmental outcomes), looking at how this literature has been interpreted and evaluated over time, and proposing what is or should be the future of this work.

Sharkey focused on three examples from his own research that suggest ways in which the literature can get beyond what has been long-term hangup in the field in terms of ignoring various contextual factors and nuance in order to answer the question, “do neighborhoods matter?” I found that his insights offered a helpful critique not only to his own sociological subdiscipline but for social scientific research in general, pushing us to never blindly disregard the particular for the universal or play into dominant research narratives.

Do Neighborhoods Matter?
Sharkey arguest that the two previous major literature reviews in this area set the tone for how research on neighborhood effects was evaluated, covered in the media, and ultimately carried out going forward. He cites their titles: Mayer and Jencks’ “Growing up in poor neighborhoods: How much does it matter?” and Ellen and Turner’s “Does neighborhood matter? Assessing recent evidence.”

The titles reflect a trend in the field to condense literature to a single question of whether or not neighborhoods matter. The hope is that it’s possible to reach such a black and white answer. Sharkey stresses that the content of the studies do not necessarily suggest this but the titles and subtitles of studies often do. Media representations of this literature follow this theme, and so this has become a dominant way to interpret the literature, which he calls the “dichotomous perspective.”

Academic literature hasn’t ignored the nuances, but the literature is still interpreted according to this dominant perspective. This has pushed the literature in certain directions and away from others.

Sharkey outlined four consequences of the dichotomous approach:

Emphasis on methods as opposed to mechanism
Focus on standardization of definitions, measurement of neighborhoods
Attempt to isolate neighborhood effects from school/family effects
Simplification of theory and evidence on context effects

The first consequence has been helpful in some ways by focusing researchers n methodological rigor to understand potential, causal mechanisms. But generally, this has led to a distorted view of what we have learned. His goal of his review paper and this presentation is to make a case for how the literature looks different when we move away from question of “Do neighborhoods matter?”

When Do Neighborhoods Matter?
Asking, “When do neighborhoods matter?” allows us to consider factors of duration, timing, and historical periods. Sharkey’s book includes an example of a multi-generational study investigating how time matters.

His 2008 paper, “The inter-generational transmission of context” [pdf], demonstrated how neighborhood effects on development outcomes actually change over generations. This should change how we think about the effects; the model becomes much more complex when doing a multi-generational study. Sharkey argued that standard regression or variation decomposition approaches look at one factor while controlling for all of these family effects over time, rather than looking at how the trends might be interrupted from one generation to another.

Sharkey’s 2011 paper with Elwert, “The Legacy of Disadvantage: Multigenerational Neighborhood Effects on Cognitive Ability” [pdf], used James Robins’ marginal structural models appraoch to analyze multi-generational data in terms of cumulative exposure, and found that neighborhood effects are not just varied by also cumulative across generations.

These studies suggest one of the limitations to research posed by consequence #3: there is a complex interaction between family effects and neighborhood effects rather than separate factors they need both be considered.

Where Do Neighborhoods Matter?
Asking, “Where do neighborhoods matter?” allows us to consider factors of scale, boundaries, and geographies. From the co-authored chapter, “Converging evidence for neighborhood effects on children’s test scores: an experimental, quasi-experiemental, and observational comparison,” in the 2011 book Whither Opportunity? Sharkey talked about the findings of three different studies with different methodologies which each investigated the effect of moving from high poverty to low poverty neighborhoods.

One study found an enormous effect of moving out of disadvantaged neighborhoods in Chicago, another found a smaller effect from random voucher assignment to Chicago families trying to move from high poverty to low poverty neighborhoods, and a third found no effects when no effects when pooling outcomes from multiple cities using a voucher program. But when isolating just the Chicago sample in the last study, the estimated effect size was almost identical to the first study. After analyzing all the differences among contexts and findings, the only two potentially explanatory factors were local violence and nonlinearity (in terms of the intensity of disadvantaged neighborhood from where they started).

The meta-analysis suggested that it was not methodological variation but geographic variation, i.e. it makes a bigger difference to move from high to low poverty neighborhoods within Chicago than within another city.

This demonstrates the problems leading from consequences #1 and #4. The default assumption is that methodological definitions are the source of conflicted findings in neighborhood effects literature, rather than emphasizing the mechanisms at work, which might help us explain what is going on. As per #4, there is always this effort to come to a single answer about whether it makes a difference to come from a disadvantaged neighborhood to a less disadvantaged rather than look more closely on the variations among contexts.

The last point results in things like the NBER statement: “Improved Neighborhoods don’t raise academic achievements,” a condensation of very complex literature, which is not even accurate for Chicago and Baltimore, where an effect was definitely observed.

Why Do Contexts Matter?
Asking “Why do contexts matter?” allow us to consider factors like institutions, interactions, resources, and exposures. Sharkey argued that few studies are methodologically aligned with exploring these factors.

Sharkey’s work has attempt to use census tracts to examine some of these. Rather than look at abstract qualities of communities like poverty rates, he has designed studies to look at specific, discrete events like homicides in correlation with testing dates.

His 2010 paper, “The acute effect of local homicides on children’s cognitive performance” [pdf], found that homicides that occurred within a week of a cognitive interview were correlated with students scoring a standard deviation lower than peers interviewed at another time.

Another paper co-authored with Strayer, Papachristos, and Raver in 2012, “The effect of local violence on children’s attention and impulse control,” fond that kids exhibited lower abilities to focus on a task after recent exposure to violence. (Sharkey mentioned that his current research is looking at acute effects of short-term exposure effect to see what cumulative, long-term effects might be.)

These studies demonstrate problems emanating from consequences #1, #2, and #3. In particular, #3 is problematic because the findings clearly indicate, beyond the anecdotal, that what happens in a child’s residential environment comes into the school environment, affecting performance in high stakes tests.

Moving Beyond the Dichotomous Perspective
Sharkey posited that two claims motivate the study of context effects:

American inequality is organized, in part, by place, and
The spatial dimensions of american inequality plays an important role in the maintenance and reproduction of inequality across multiple dimensions,

and the literature should be organized around these two claims.

In his review article, he suggests that the field of neighborhood effects can move forward by using a/an:

Flexible definition and measurement of residential contexts
Flexible timeframe for the study of context effects
Assumption of heterogeneity
Alignment of methods with theoretical mechanisms

Sharkey ended by quoting Robert Sampson’s ASC presidential address: “Relentlessly focus on context.” If it was up to him, this would be at the heart of the literature, i.e. “We need to know more; we don’t need to condense what we already have.”