• No se han encontrado resultados

Análisis de resultados para un modelo de canal de dos rayos

3.5 Análisis de resultados de los métodos de combinación

3.5.1 Análisis de resultados para un modelo de canal de dos rayos

This chapter of the dissertation marks a transition from quantitative methods to qualitative methods by means of case study analysis for neighborhoods located with the four cities focused upon in Chapter 6. In part, this reflects a transition from an urban economics perspective towards grounded research methods related to city planning and urban design character. While the quantitative phases identified to some extent the key indicators of economic outcomes, there remains a gap in addressing the non-economic outcomes related to neighborhood stability. This scale of analysis provides valuable insight about urbanism and policy efficacy to the extent that indicators of neighborhood are revealed.

The overarching purpose of the third phase of research is to assess the less tangible qualities of urban stability—specifically with respect to that which cannot be readily measured or understood using administrative or proprietary data alone. Ascertaining the extent to which there exists a relationship between neighborhood character and stability outcomes is the key focus of this third phase. As such, two key questions emerge. The primary question is what extent is it possible to understand the influential factors related to stability that are not specified in the models from Chapters 5 and 6; that is, how one might explain places of high that do not necessarily conform to the models’ explanatory value. A secondary and important set of questions is whether city planning efficacy is evident and whether urban character is associated effective with respect to greater stability outcome measures.

What is clear from the quantitative analysis, beyond an inability to model the influential variables associated with non-economic outcomes, is two pronged: first, cities matter to the extent that the introduction of city-level controls within the quantitative models enhanced the explanatory value of each model; second, from the fine-grained analysis in Chapter 6, it is possible to

differentiate between the significance of indicators present in auto-oriented cities (i.e. Atlanta and Salt Lake City) versus cities that strike a balance among mode choices (i.e. Philadelphia and San Francisco). In simple terms, there are observable differences in terms of factors for housing choice in less-walkable cities compared with more-walkable cities. The case study approach provides greater insight as to the degree of difference both within each city as well as across these two types of cities.

The structure of this chapter first details the methods of investigation, including the process for identifying target neighborhoods within each city as well as the qualitative methods employed in the case study approach. The second part of the chapter discusses each city separately and takes a deep dive approach to each neighborhood. Lastly, the discussion section provides context to the individual case study neighborhood by examining patterns within individual cities as well as across cities.

Neighborhood Selection

The first step in generating a set of case study neighborhoods is to establish specific controls. The use of matched-set analysis (a subset of multi-case study methods) of each city features two neighborhoods, each exhibiting high economic and non-economic stability factors relative to other units within each city. This approach is helpful in addressing the secondary question about the efficacy of planning; that is, the research can control for planning capacity and activism by identifying two neighborhoods from the same city.

It should be noted that there are several ways in which one can approach the case study methods. Prior to the selections of cases, for example, one must reach a determination as to whether to engage in preliminary research, analysis, or other forms of scoping. Central to the decision making process is the question as to whether the researcher should know anything about the potential cases to be studied prior to the actual investigation. This type of blind-identification creates the potential for randomized collections of case studies, which has its benefits and potential drawbacks. In terms of the former, the primary benefit is this does inherently involve any source of bias—either on the part of the researcher or the data collection methods. That the cases are random suggests a need to develop a broad set of evaluation criteria as well, since there are no baselines of comparison or means of controlling for variability or intervening variables. This does not, however, reduce the potential for observer bias,

particularly in instances where it is necessary to have multiple observers or the conditions in which the observations are conducted change—in the case of neighborhood research, time of day, day of the week, and weather conditions can have significant impacts on both the observer and that which is being observed. In terms of the latter, however, there are potential drawbacks. One specific such example is that the type of research design and set of hypotheses may require a substantial number of randomized case studies in order to generate internal validity around a

particular topic. Along this vein, it follows that research questions and related hypotheses that are very specific and target a narrow set of outcomes may not be well-served by the randomized selection approach if the measurement tools and processes are not capable of providing

sufficient data for analysis.

In the context of this research, the case study selection process is at the opposite end of the spectrum as is described above. The process, however, is not as sequential as the phase numbers might suggest. While Phase I (i.e. tract level analysis) helps to set the stage for what to focus upon in subsequent phases of analysis, the results from Phase I are highly informative with respect to the staging and sequencing of research. As is detailed in Chapter 5 (Urban Stability at 30,000 Feet), the regression models specified for the economic and non-economic outcome variables clearly indicate that the former is well-explained by the models while the latter outcome variables are not well-explained by these models—even when controlling for inter-city variability. While these outcomes were highly informative in terms of the research process for fine-grained analysis, it provides a strong foundation upon which to conduct case studies at the neighborhood scale. Specifically, in the absence of having regression models that are capable of explaining the variation in the non-economic outcome variables specified, this suggests one of two things: either (1) the underlying theory supporting these models’ construction and specification of independent variables is wrong, or (2) these outcomes have the potential to be understood through

observational studies as opposed to a purely-quantitative focus.

With the decision to conduct preliminary analysis that informs the means for case selection, there is a second set of choices; namely, whether the case selection should include counter-factual cases or not. If counter-factual cases are included, it can serve as a “control group” and provide a basis for comparison to the cases of interest. This may be particularly well-suited for exploratory cases in which it is possible to observe or control for intervening variables (to an extent). In the context of this research, however, there is a level of specificity about such intervening variables that may not be possible. Specifically, because this research involves different cities, each of which contains its own macro-level intervening variables, it would be difficult to determine whether a counter-factual case would be a product of the variables of interest or a response to a different set of conditions occurring at the municipal level. As there is the potential to draw conclusions from comparable sets of neighborhoods across cities, this may not be a good approach as the counter-factual would have to apply to each case uniformly. In addition, the need to focus upon a

set of cases that are ultimately not important to the outcomes of this research shifts the attention from generalizable observations across study areas (i.e. similarities) to a focus upon differences. One must first identify those places of interest that have the ability to serve well not only with respect to single-case observations but also help to address the larger set of research questions overall. That is, the intent is to generate knowledge across the cases that provide a deeper and richer understanding of urban stability at the neighborhood level. Developing the criteria for identifying these neighborhoods is based upon the dependent variables used in the first phase of quantitative analysis: (1) change in median self-reported home values, (2) change in percent of households living in each tract less than 10 years as a proxy for household turnover, and (3) age diversity, constructed using the Simpson Diversity Index.

If any of these are taken separately to identify neighborhoods, they lead to an exploration of questions that are particularly well-suited for the sole criterion, but this may fail to advance any knowledge about the other outcome variables of interest. To overcome this perceived obstacle, each the data for each of the outcome variables at the tract level was ranked against every other tract within the study cities. This rank-order approach provides a basis for comparison within specific indicators. To generate a comprehensive understanding of urban stability, however, requires some cross-comparison of outcomes. To this end, a mean rank-order score was generated for each tract, and the lowest mean scores (i.e. those ranking the highest in terms of order) helped to identify potential areas of interest. Then, by examining the variation within the mean scores, the research identified those tracts for which there were indications of some consistency across outcome measures.

It is important to note that there are other means by which the research could select cases in a “pre-evaluative” manner. What is described above is a simple means for selecting cases, but it does not include and pre-selection controls for demographics or urban characteristics, for example. The position of this research is that this creates a potential source of bias as it precludes an examination of evidence outside of such restrictions. A more salient point, at least in the context of this research, is that the subset of cities selected for fine-grained analysis are not comparable in many respects, and to apply a set of pre-conditions to reduce the set of potential cases could have significant unintended consequences.

the rank-order scores were highly correlated among the highest overall ranking tracts. For example, tracts that experienced significant increases in terms of median home values were often associated with large resident turnover (i.e. possible gentrification) or increasing rates of age diversity (i.e. related to resident turnover, but aligned not only with income accessibility but also with individual household preferences). In this example, it is relatively easy to understand that rational homeowners, perceiving increasing value in the housing market, may capitalize on this by monetizing their investment; one might also expect that some households are not solely motivated by capital interests, and perhaps those households’ original intent was to buy into a place that was “on the rise.” In either event, this is not a disappointing outcome for this research; rather, it relates well to the complexity of housing choice within competitive markets.

Overall, this approach to case selection provides the basis to conduct observational studies that have the potential to inform broadly in two ways: first, there is the potential to understand the neighborhood stability through an economic and non-economic lens within each cities; second, and perhaps more importantly, it is may be possible to make inter-city comparisons if there are patterns in demographics or urban form that reveal potential matched sets within the cases. To put this succinctly, there is an inductive question to be explored through this process: are there observable commonalities among case neighborhoods that appear to reinforce or bolster their economic and/or social outcomes? To the extent that this can be understood both within each city but across cities as well will be particularly informative for the research outcomes.

Criteria for Assessing Walkability

A distinct advantage of the case study approach is the ability to understand data using

methods that extend beyond the limitations of quantitative methods. One area of methods that is particularly useful with respect to the assessment of walkability is spatial observation and analysis. To define what this is, it helps to define what it is not; quantitative spatial analytic tools are exceptional for providing accurate measurements of quantities, especially with respect to the spatial density of quantities. However, such tools are not particularly useful for delimiting between types of spatial configurations, at least not to the extent that is employed here in assessing walkability.

While other methods of observation in the following case studies will assess the degree to which walkable environments are comfortable, safe, and interesting, this method focusing on the

spatial configuration of amenities provides a meaningful description of the usefulness of a walk. Specifically, there are four typologies of spatial configurations that are original contributions from this research: the node, the corridor, the field, and the void.

These typologies of spatial configurations provide a meaningful interpretation of the use of a walk within each of the study areas. With the node, there is a focal point that can serve as the destination for a walk, or (in the case of multiple nodes) there are terminal or anchor points that one might choose to walk between or towards. With the corridor, regardless of where one is located within the area, the shortest distance to the points of interest is the perpendicular distance to the corridor along which these points are aligned. One is able to then maximize the usefulness of the walk by first reaching the corridor and then traveling along it. It is important to note as well that the corridor can exist at the periphery or bisect the neighborhood, and there may be more than one that provides context to urban form. In the field there is an overall lack of centrality insofar that one can travel in almost any direction to reach a potential destination point. To this end, the experiential quality of the field is one that is more akin to discovery than oriented towards a singular destination. The void is the antithesis of all others—as the name implies, there is a sufficient lack of amenities as to convey the perception of a void within an otherwise Figure 7.1: Walkability Typologies—Node, Corridor, Field, and Void

defined neighborhood space. One should note as well that, with each of these spatial types, one should note that they are not mutually exclusive. That is, one might find combinations of this configurations—a “field” bordered or intersected by a “corridor,” or a “field” populated by a number of “nodes.” To an extent, such combinations may imply greater flexibility within zoning regulations as well as possible evidence of adaptive reuse occurring over time.

Fieldwork Protocols

The case study fieldwork occurred between June 9, 2014 and June 20, 2014 in each of the four cities studied. Providing consistency to the observation periods, the hours of 9 AM to 4 PM during weekdays were spent in the neighborhoods—for each neighborhood, the research allocated one day. By observing the neighborhoods during midweek days, this offers the perspective of the neighborhood as it exists during the typical workweek.

There are several advantages to this approach and one key disadvantage. Turning first to the advantages, an important observation that can be ascertained through this approach is determining how the neighborhood functions during the typical workday. Specifically, a key question is whether there is pedestrian activity—particularly with respect to businesses and other amenities—that occurs during the workday. On the one hand, if there is evidence of pedestrian activity, this would suggest that the area may be a destination for visitors from outside of the community; on the other hand, a lack of pedestrian activity would suggest that residents are the primary users of the area. Second, this provides an opportunity to observe the extent to which these areas are served by transit as this is the primary means for accessing these areas for observational study. A third observation point is an assessment of the vehicular traffic moving through the area during the workday, which provides an understanding of whether the area serves more than resident uses.

There are some disadvantages to this approach, which are detailed here. First, there is some difficulty in measuring community as might be observed through the social interaction of residents. This acknowledges that it is unlikely to reveal all forms of interpersonal relationships as the observations are conducted during traditional full-time employment hours. What is not clear, however, is when the most appropriate time to explore evidence of these interactions. As a control for this, preliminary research of each neighborhood for community organizations did not reveal any considerable web presence for such organizations within any of the neighborhoods.

On-the-ground fieldwork examines this in greater depth, seeking evidence of community-based organizations during the observation periods.

In terms of the fieldwork itself, the observational study of each neighborhood combines three methods. In addition to these on-the-ground fieldwork methods, two preliminary research tools were generated: first, the data points for neighborhood stability (i.e. those used to score and rank each neighborhood relative to others within the city) were summarized as a means for providing advance indications of the quantitative data; second, maps with amenity destinations were prepared for orientation as well as for managing the walking protocols. The first on-the- ground method is the walking tour of the neighborhood. These walking studies begin and end at the same transit node and the walk is directed to follow along roads towards potential destination points (i.e. business nodes or public recreation amenities) while also paying attention to evidence of a street design hierarchy. In the discussion of each neighborhood case study, there is a graphic providing a map illustrating the walking path through the neighborhood as well as the amenities that were noted in the analysis. The second method is photographic documentation of urban design elements, which include typical street scenes, notable public amenities such as parks or iconic wayfinding signage, as well as evidence of resident-initiated improvements that exist either on private property or within the public realm. The third method is the physical inventory of urban

Documento similar