What Differences Do They See? Using Mixed Methods to Capture Adolescent Perceptions of Neighborhood Contexts

Abstract

Recent neighborhood research emphasizes the importance of individuals’ perceptions of their neighborhoods, as well as expands the definition of “neighborhood” to include the different contexts encountered during routine daily activities (Coulton et al., 2013; Kwan, 2012). The present study uses qualitative interviews, sketch mapping, and survey data to explore adolescents’ experiences of different neighborhoods within their activity space. Participants included 55 racially diverse youth aged 11–19 (M = 14.64, SD = 2.33) who resided in low-income neighborhoods in a small city in the Midwest. The majority reported spending time in multiple self-defined neighborhoods, noting significant differences between neighborhoods on collective efficacy, street code, and on participant-generated dimensions. Self-defined neighborhoods did not correspond to Census tracts, and Census indicators were not associated with youth’s perceptions (e.g., collective efficacy, street code). Youth spent time in neighborhoods that differed significantly on multiple Census indicators of structural disadvantage, though within-individual differences tended to be small in magnitude. Type of routine activity was largely not predictive of distance traveled from home, though some findings suggest youth were more likely to cross neighborhood boundaries to engage in structured activities compared to different unstructured activities. Implications for neighborhood research and interventions are discussed.

Introduction

Research on neighborhoods and youth has established that structural characteristics such as concentrated poverty and racial segregation contribute to negative individual outcomes including delinquency, reduced academic achievement, and poorer mental health (Leventhal & Brooks-Gunn, 2000; Leventhal, Dupéré, & A. Shuey, 2015). Most of this research has conceptualized the neighborhood as an “environment” (Nicotera, 2007, p. 27), using objective measures such as percent of households in poverty within an administrative boundary such as a Census tract. However, more recent research has shifted to conceiving of the neighborhood as a “place” (Nicotera, 2007, p. 28). As a place, a neighborhood is experienced and perceived differently by different residents, and is more of a social construct than geographic. Research combining environment (e.g., Census indicators) and place measures (e.g., resident self-reports) often shows a disconnect between these sources, suggesting that residents do not perceive environmental influences as might be expected; even resident self-reported neighborhood boundaries tend to differ from Census-defined boundaries (Coulton, Jennings, & Chan, 2013). As neighborhood research moves away from a uniform reliance on Census measures, researchers from diverse disciplines are grappling with choices on how to bound, define, and evaluate neighborhoods (Browning & Soller, 2014; Hipp & Boessen, 2013; Kwan, 2012; Nicotera, 2007). Mixed-methods approaches have the potential to shed light on the connections between environmental characteristics and resident perceptions, building on the body of research on neighborhood environmental influences on youth well-being to illuminate the mechanisms by which youth experience these influences (Leventhal, Dupere, & Brooks-Gunn, 2009).

Adolescents are often the focus of neighborhood research, with a growing emphasis on understanding their “routine activity space” beyond simply their home neighborhood. For the first time in their lives, teens are able to independently travel to school, shopping malls, work sites, and hangout spots with decreasing supervision from adults (Csikszentmihalyi & Larson, 1986; Furstenberg, Cook, Eccles, & Elder, 2000). Most adolescents cross Census tract boundaries in the course of their daily activities and spend a substantial proportion of their time outside of their immediate home neighborhoods (Basta, Richmond, & Wiebe, 2010; Wikström, Ceccato, Hardie, & Treiber, 2010). This increased freedom to choose their own contexts can be associated with risk—unstructured, unsupervised time with peers increases adolescents’ risk for engaging in substance use or delinquent behaviors, particularly when the unstructured time is spent in public and/or open spaces (Hoeben & Weerman, 2014; Osgood, Wilson, O’Malley, Bachman, & Johnston, 1996). Understanding neighborhood influences on adolescents requires study not only of the neighborhood of residence but of the neighborhood contexts adolescents encounter during their routine daily activities. Neighborhoods where adolescents choose to spend time may be just as influential as neighborhoods where they reside, and it is important to explore the interplay between neighborhood characteristics and adolescents’ movement.

Youth often perceive their neighborhoods differently than adults, and their neighborhood perceptions tend to be more closely associated with their individual outcomes than are other measures of neighborhoods (Byrnes, Chen, Miller, & Maguin, 2007; Goldman-Mellor, Margerison-Zilko, Allen, & Cerda, 2016; Spilsbury, Korbin, & Coulton, 2009). Most current literature on youth within neighborhoods focuses on either the home neighborhood or the routine activity space, with relatively less attention given to youths’ perceptions of distinct spaces within their activity space. Some exceptions are studies that ask youth to identify distinct types of spaces, such as risky or safe, familiar or avoided; these studies begin to demonstrate youth’s ability to discriminate between different areas within their activity spaces using researcher-generated categories (Mennis & Mason, 2011; Pearsall, Hawthorne, Block, Walker, & Masucci, 2015). Given that most youth cross Census boundaries during their routine activities, it seems likely that they perceive distinctions between different spaces they encounter, including across self-defined neighborhood boundaries. Alternatively, it is possible that youth will define their entire activity space as a single “neighborhood,” if they are using their own familiarity with an area as their index of neighborhood boundaries. It is also possible that youth’s different contexts are objectively similar, such as when an individual residing in a disadvantaged neighborhood tends to spend time in a similarly disadvantaged neighborhood (Krivo et al., 2013). Understanding how youth define and contrast different spaces, as well as the degree to which they encounter neighborhoods with varying structural characteristics, is an important next step in person-centered neighborhood research. Such research also has the potential to guide neighborhood-based interventions for youth, as community revitalization efforts often seek to improve public spaces for children and youth (Dennis, 2006; Patton-López et al., 2015). Knowing how youth view programs as within or outside their home neighborhoods, including their willingness to travel to a program located within a particular neighborhood, can help engage youth and shape program outreach.

The Current Study

The current study addresses three questions regarding adolescents’ perceptions of different neighborhoods within their activity spaces: (a) To what extent do adolescents perceive distinct neighborhoods within their activity spaces, as opposed to defining “neighborhood” as all familiar space? (b) To what extent do adolescents’ activity spaces encompass areas that differ on characteristics commonly studied in neighborhood literature? (c) How do adolescents define distinctions between neighborhoods? In addition, the study explores whether the structure of daily activities are associated with distance traveled from the home address, or with the likelihood of crossing either Census tract or self-defined neighborhood boundaries. These questions center the adolescent’s unique perspectives of their neighborhood context and are best explored using mixed methods. A convergent parallel design is used to integrate administrative data with different types of data generated during intensive individual interviews with adolescents: sketch maps of different activity sites and neighborhoods, surveys of perceptions of different neighborhoods, and open-ended questions about adolescents’ views of neighborhoods. The integration of these different sources of data, with a particular focus on adolescent perspectives, facilitates a more thorough and nuanced examination of adolescents’ experiences of neighborhood contexts.

Method

Participants

Participants included 55 adolescents aged 11–19 (M = 14.64, SD = 2.33) who resided and/or attended community activities in a low-income neighborhood in a mid-sized city in the Midwest. This city has approximately 275,000 residents within city limits (600,000 within the metropolitan area), and relatively few historic named neighborhoods. Participants were recruited from a variety of settings within one mile of the local high school, including a local community center, after school programs at the local high school, the library, YMCA, a neighborhood organizer, and Facebook. The majority of participants were male (58.2%, n = 32). Most participants self-identified as African American (78.2%, n = 43), Caucasian (7.3%, n = 4), or Latino/a (5.5%, n = 3). Two participants identified as multiracial, two identified as Other, and one identified as Native American. Twenty-nine participants (53%) indicated that within the past two years, they had resided at more than one home address—including moving from one location to another, or having multiple concurrent residences (such as with divorced parents).

Procedure

Parents were contacted through the agency assisting in recruitment or directly by the researchers to provide written informed consent. Parents were provided with a $5 gift card for completing and returning a consent form, regardless of whether they granted permission for their child to participate in the study. Adolescents were then contacted by telephone and/or with the assistance of cooperating agencies, to arrange the first interview. Adolescents met individually with a researcher in a private space at a public location, such as a study room at the library or a quiet room at the neighborhood organization. After providing written assent, participants completed a 90-minute interview. Interviews were audio-recorded and later transcribed (n = 42); in 13 cases, audio-recording was not available (early in the study, a second interviewer transcribed interviews live), and interviewer notes were coded. After completing the interview, participants received $30 gift cards to a local store or restaurant. The institutional review board at the university conducting the study approved the procedures before the project began recruitment.

Measures

Mapping Neighborhood Spaces

Adolescents gave their home address and were asked if they considered their current place of residence to be their “home neighborhood.” Adolescents were oriented to their home address or other self-identified home on a paper map. A series of local maps (approximately 1 square mile) were available, in addition to a large city map for those adolescents not residing near their interview location. Adolescents were then prompted to “draw your home neighborhood” on the paper map. Those participants requiring additional prompts were asked, “what are the boundaries of your home neighborhood?” or “how do you know when you cross a boundary and are outside of your home neighborhood?” In some cases, the interviewer used Google Maps on a smartphone to help locate landmarks on the paper map.

A similar procedure was used following the time diary portion of the interview (described below). Adolescents were prompted, “Of the places we have talked about so far, are any of them outside what you would consider your home neighborhood?” Follow-up prompts asked the adolescent to give each non-home neighborhood a label (such as “girlfriend’s house” or “neighborhood near the mall”). The same paper maps, including participant drawings of home neighborhoods made earlier in the interview, were then used to draw boundaries of non-home neighborhoods. These procedures are consistent with a growing literature on qualitative GIS methods (Boschmann & Cubbon, 2014; Dennis, 2006; Kwan & Ding, 2008).

Routine Activity Time Diaries

A time diary (also called a space-time budget; Wikström et al., 2010) was used to identify common activities, social contacts, and locations of activities. Respondents were asked to choose a “typical weekday” within the past week and then recounted every place they went throughout the day. For each location, they were asked the times they were present, the ages and gender of other people who were present, and whether any supervising adults were present. Respondents were also asked to briefly summarize their activities at each location (such as “hung out” and “cheerleading practice”). The participant then identified each site on a map. In cases where they were uncertain of precise location, they were prompted to describe nearby landmarks to assist in approximating the location. They were then asked to describe the route they took between each site. Once these procedures were complete for a typical weekday, they were repeated for the most recent typical weekend day. Figure 1 depicts a simplified rendering of a participant map, with a home neighborhood, routine activity neighborhood, and path to one activity site.

Fig. 1.

Example participant map. This figure includes an example of a participant’s home and routine activity neighborhood. Points A and B represent routine activity locations or could represent a participant’s home. The dotted line indicates the path traveled between sites.

Activities were then coded as structured (e.g., school, church) or unstructured (e.g., going to see friends) and whether the activity involved visiting family or friends.

Neighborhood Collective Efficacy

Perceptions of home neighborhood collective efficacy were assessed using the scale developed by Sampson, Raudenbush, and Earls (1997). Five items assessed the subscale of neighborhood social cohesion on a 5-point Likert-type scale from “strongly agree” to “strongly disagree,” including items such as “People in my neighborhood are willing to help their neighbors.” An additional five items assessed the subscale of neighborhood informal social control on a 5-point Likert-type scale from “very likely” to “very unlikely,” including items such as “If some children were spray-painting graffiti on a local building, how likely is it that your neighbors would do something about it?” The mean across all ten items was used to measure perceived collective efficacy in the home neighborhood, and internal consistency was acceptable (α = .71). Youth who identified a neighborhood separate from their home neighborhood were re-administered the collective efficacy scale referring to that neighborhood. Internal consistency for collective efficacy of non-home neighborhoods was also acceptable (α = .79).

Neighborhood Street Code

Perceptions of home neighborhood street code were assessed using items developed by Stewart and Simons (2010). Participants responded to eight items on a 4-point Likert scale ranging from “strongly agree” to “strongly disagree,” including items such as “In my neighborhood, sometimes you need to threaten people in order to get them to treat you fairly.” The mean across items was used for the total score, and internal consistency was good (α = .90). Youth who identified a neighborhood separate from their home neighborhood were re-administered the street code scale referring to that neighborhood. Internal consistency for street code of non-home neighborhoods was also good (α = .92).

Sense of Community

Sense of community was assessed using the Brief Sense of Community scale (Peterson, Speer, & McMillan, 2008). Eight items on a 5-point Likert scale assess the participant’s sense of belonging to their home neighborhood, their belief that their home neighborhood meets their needs, their perceived ability to influence their home neighborhood, and their emotional connection to their home neighborhood. Internal consistency for sense of community was good (α = .83).

Participant Demographics

Participants self-reported their race (using Census-derived categories), gender (using binary male/female options), and age. Participants also reported whether they had “regular access to a car,” which could include having their own car or a close family member with a car.

Neighborhood Demographics

Census information was used to determine neighborhood demographics, including Census tract boundaries, total population, percent Black population, median household income, percent owner-occupied structures, and percent renter-occupied structures (U.S. Census Bureau, 2016).

Qualitative Neighborhood Perceptions

Two open-ended questions, and one section of the neighborhood mapping exercise, were coded to explore dimensions on which participants compared and contrasted neighborhoods. Prior to mapping the home neighborhood space, participants were asked if they considered the area around their current address to be their “home neighborhood” or if they considered another space to be their home neighborhood. They were then asked, “what makes that your home neighborhood?” and participant responses were transcribed verbatim. Only responses to this item that included comparison or contrast with other neighborhoods were analyzed for the current study. Later in the interview, those participants that mapped at least one routine activity neighborhood in addition to the home neighborhood were asked, “what makes these neighborhoods similar or different?” and participant responses were transcribed verbatim. For both items, follow-up prompts were not used, except when participants sought clarification they were encouraged to “just say in your own words what you think.” In addition, when participants were mapping routine activity (non-home) neighborhoods, spontaneous descriptions of the neighborhoods were also transcribed and analyzed.

Data Analysis Plan

A mixed-methods convergent parallel design informed the analysis of GIS, survey, and qualitative data. Each source was first analyzed separately as described below, and then, results are integrated at the conceptual level in the Discussion.

Geographic Information System (GIS)

Paper sketch maps were entered into ESRI GIS ArcMap 10.3. Polygons were coded over defined spaces, and open polygon shapes were closed by the researcher. Figure 2 depicts a simplified rendering of a participant map, with an open polygon shape of a neighborhood that would have been closed by the researcher. Composite Census variables were created for participants’ self-defined neighborhoods weighted by the percent of the neighborhood encompassed within each Census tract. Street distance was calculated between activities, as well as the shortest street distance between home address and each activity point.

Fig. 2.

Example participant map. This figure includes an example of an open polygon home or routine activity neighborhood. Such open figures were closed by researchers prior to analysis.

Quantitative Analyses

Correlations were used to examine associations between participant perceptions of home neighborhood and Census variables. HLM 7.0 software was used to nest multiple neighborhoods within participant (two-level modeling) and examine associations between perceptions and Census data across neighborhoods. Separate models nested activity trips within participant and using structure of activity to predict street distance of activity from the participant’s home. Two-level logistic regression models used structure of activity to predict crossing a self-defined neighborhood boundary, as well as crossing a Census tract boundary.

Qualitative Content Analysis

Interviews were transcribed into ATLAS.ti 8 software. The portions of the interviews in which participants were asked to describe their home neighborhood, other neighborhoods that they visited, and their perceptions of neighborhoods as similar or different were coded for neighborhood characteristics. After meeting to discuss the initial set of codes, the third and fourth authors independently coded the same 10 transcripts. Following a second meeting to revise and add codes, the third and fourth authors coded remaining transcripts independently, meeting periodically to discuss ambiguous quotations. Participant answers were initially coded using specific codes describing perceived neighborhood characteristics, which were then categorized into four broader themes. Once all transcripts were coded, 10 were independently coded by the first author to examine inter-rater reliability. Following a final coding meeting to discuss discrepancies, Krippendorf’s alpha was 0.80 (84% agreement) for specific codes and 0.81 (90% agreement) for themes.

Results

Descriptive Analyses

Home Neighborhood

Across participants, self-defined home neighborhoods overlapped between 1 and 15 Census tracts (M = 4.94, SD = 3.76). Home neighborhood area ranged from <0.01 square miles to 13.2 square miles (M = 1.21, SD = 2.07). Descriptive data on home neighborhood are reported in Table 1. All Census indicators of neighborhood disadvantage were intercorrelated. Perceptions of collective efficacy and street code were significantly correlated with each other, but not with Census indicators. Sense of community was significantly correlated with median household income, percent owner-occupied residences, percent renter-occupied residences (negative correlation), and perceived collective efficacy.

Table 1.

Descriptive data for self-reported home neighborhood

	Black percent	Median household income	% owner-occupied	% renter-occupied	Perceived collective efficacy	Perceived street code	Sense of community
Black percent	–
Median household income	−0.71***	–
Owner-occupied percent	−0.44***	0.80***	–
Renter-occupied percent	0.52***	−0.72***	−0.82***	–
Perceived collective efficacy	−0.04	0.20	0.21	−0.19	–
Perceived street code	−0.12	0.06	0.02	0.06	−0.33*	–
Sense of community	−0.19	0.32*	0.29*	−0.36**	0.54**	−0.16	–
Range	<0.01%–92%	$16,500–$54,181.33	7%–36%	6%–30%	0.50–3.90	0–2.88	0–3.88
M (SD)	51.13% (26.08%)	$31,954 ($10,220.53)	20.17% (5.38%)	18.10% (4.26%)	2.39 (0.65)	2.52 (0.70)	2.37 (0.65)

^*p < .05

^**p < .01

^***p < .001.

Routine Activities and Distance Traveled

Over the two days chosen for the time diary, participants reported 4–17 daily activities (M = 8.48, SD = 2.94) with between 0 and 4 structured activities (M = 1.27, SD = 1.09). One participant reported one unusually distant trip (to visit a grandparent in a neighboring state), and this distance was trimmed to 0.01 miles farther than the next-farthest distance. The average summed distance of trips traveled in a day ranged from 1.04 to 25.19 miles (M = 8.23 miles, SD = 6.27 miles), and the average distance traveled from home in a day was 2.01 miles (SD = 2.13 miles). The farthest distance participants traveled from home for an activity ranged from 0.72 to 19 miles; (M = 6.01 miles, SD = 4.37 miles). Table 2 presents the distance traveled from home by type of activity. Due to the significant positive skew of distances traveled, remaining analyses used a log-transformed distance as the outcome.

Table 2.

Distance from home in miles by type of activity

Type of activity (n)	Mean (SD)
Visit family (61)	3.71 (3.47)
Visit friends (37)	4.11 (3.62)
School (36)	2.61 (2.05)
Shopping (32)	3.16 (4.09)
Park (14)	5.67 (17.04)
Extracurricular activity (12)	2.97 (1.92)
Church (12)	6.32 (4.22)
Library (8)	1.45 (1.38)
Restaurant (7)	6.13 (5.63)
Community center (6)	2.11 (1.56)
Walking/bike ride (4)	0.81 (0.59)
Work (3)	6.16 (2.24)

No significant differences were found in mean or farthest distance traveled for youth who reported having access to a car (n = 37, 66%). Neither gender nor age were associated with farthest or mean distance traveled. Participant self-reported race was categorized into African American (n = 43, 77%), White (n = 4, 7%), and all others (n = 8, 14%). Although the ANOVA indicates a significant difference in farthest distance from home (F(2, 52) = 3.64, p < .05), Tukey’s post hoc did not identify specific group differences.

Modeling Associations Between Types of Activity and Travel

Multilevel modeling was used to nest activity trips within participant, using street distance from home as the outcome variable. The average number of trips per participant was 8.48 (SD = 2.94). No significant difference was found for structured versus unstructured activities in predicting distance traveled from home. Dummy coding was used to break down unstructured activities into trips to visit family, visiting friends’ homes, and other (e.g., shopping, restaurants), and to compare each of these categories with each other as well as with structured activities. Out of six pairwise comparisons, only one was significant, with youth traveling farther from home for structured activities versus unstructured “other” activities (t(200) = 2.40, p < .05). Adding age and access to a car as between-individual (Level 2) predictors did not change patterns of results.

Two-level binomial logistic regression models tested whether youth were more likely to cross neighborhood boundaries—whether self-defined or as Census tracts—depending on structure of activity. Most activity trips (not including return trips home) crossed a self-defined neighborhood boundary (67%), with 48 participants (87%) crossing a self-defined neighborhood boundary at least once. Most trips crossed a Census tract boundary (78%), with all participants crossing a Census tract boundary at least once. Out of six pairwise comparisons predicting crossing a self-defined boundary, only one was significant, with youth being more likely to cross a boundary to engage in a structured activity than an “other” unstructured activity (t(83) = −4.76, p < .001). Youth were more likely to cross Census tracts when engaging in structured activities when compared with visiting friends (t(46) = −3.25, p < .01), visiting family (t(58) = −2.52, p < .05), or other unstructured activities (t(87) = −3.20, p < .01). When age was added as a between-individual (Level 2) moderator, older (but not younger) youth were more likely to cross a self-defined neighborhood for “other” unstructured activities than to spend time with friends. Older youth showed a stronger likelihood to cross Census tracts for structured activities compared with visiting friends (t(45) = 2.80, p < .01), as well as for structured activities compared with other unstructured activities (t(131) = −2.66, p < .01).

Quantitative Examination of Multiple Neighborhoods

Forty-five participants (82%) completed collective efficacy and street code perception measures for at least one neighborhood in addition to their home neighborhood. Paired-samples t-tests were used to explore absolute differences in perceptions of the home and routine activity neighborhoods. Participants reported significantly different levels of collective efficacy between neighborhoods (t(44) = 8.38, p < .001), as well as neighborhood street code (t(44) = 12.59, p < .001), with about 54% of participants indicating their home neighborhood had higher collective efficacy and 53% indicating higher street code scores for their home neighborhood. Twenty-nine participants (53%) identified at least two distinct routine activity neighborhoods outside of their home neighborhood, and reported significant distinctions in collective efficacy (t(27) = 7.84, p < .001) and street code (t(27) = 5.16, p < .001) between the routine activity neighborhoods.

Absolute differences in composite Census data were contrasted for the self-defined home neighborhood with the self-defined first routine activity neighborhood, for the thirty-six participants (65%) who were able to map both a home neighborhood and at least one other neighborhood. Across the sample, youth spent time in neighborhoods with markedly different Census characteristics; however, statistically significant within-individual differences between home and routine activity neighborhoods tended to be relatively small in magnitude. Participants spent time in neighborhoods that ranged from 0% Black to 92% Black (mean within-individual difference between highest and lowest percent Black neighborhoods = 27%, SD = 23%, t (35) = 6.99, p < .001) with median income ranging from $11,535 to $75,351 (mean within-individual difference = $11,377, SD = $9503, t (35) = 7.18, p < .001). Percent of owner-occupied structures ranged from 2% to 46% (mean within-individual difference = 7%, SD = 5%, t (35) = 7.59, p < .001), and percent renter-occupied ranged from 4% to 49% (mean within-individual difference = 6%, SD = 5%, t (35) = 6.22, p < .001).

Multilevel analyses tested associations of Census variables with participant perceptions of collective efficacy and street code. Due to high collinearity between Census variables and the small sample size, separate models were run for each of four predictors (% Black, median household income, % owner-occupied, % renter-occupied) with each self-reported outcome (collective efficacy and street code). Out of eight models run, only two findings were significant at p < .05, with percent owner-occupied predicting youth perceptions of collective efficacy (t(40) = 2.08, p < .05) as well as youth perceptions of street code (t(40) = −2.23, p < .05).

Qualitative Content Analysis of Neighborhood Comparisons and Contrasts

A total of 110 utterances were coded across 55 interviews. All participants had at least one utterance where they were describing differences between neighborhoods (total number of utterances = 71), but only 12 participants described similarities between distinct neighborhoods (total utterances = 12). Content analysis yielded 32 specific codes which were categorized into four overarching themes. Three of those themes represented distinct dimensions on which participants compared and contrasted neighborhoods: Familiarity or Personal Experiences within the neighborhood, Perceived Social Characteristics of the neighborhood, and Observed Structural Characteristics of the neighborhood. A fourth theme, Subjective Appraisal (positive or negative) of a neighborhood, was interwoven with the other themes. Participants often described a neighborhood as simply “good” or “bad,” which would be coded within the Subjective Appraisal theme. However, when their appraisal was more specific, as in, “When I lived there the people were nice to me,” a code was used that included the positive/negative appraisal but was categorized within the more specific theme (in this example, the Familiarity or Personal Experiences theme).

Familiarity or Personal Experiences

The theme of Familiarity or Personal Experiences was used by 24 participants in describing neighborhoods. Codes within this theme share a reference to personal familiarity, comfort level, or specific personal experiences within a neighborhood—all indicators of a subjective or personal dimension rather than an objective indicator that could be observed by others. Some respondents provided specific examples of experiences they had had in a neighborhood as part of how they described the neighborhood, such as:

When we moved into the Green Street1 house we used to have, like, campfires—we do campfires and stuff now, and when we put our tent outside in the backyard and we can sleep in the backyard. But when we lived over here, we didn’t do that.

Participants also described identifying the boundary of a neighborhood as corresponding to the personal experience of feeling uncomfortable, as in, “I can’t really name places, it’s just when I feel uncomfortable. When I get that feeling, that’s how I know it’s time to go.” A similar theme appeared to be familiarity with people in a neighborhood, such as:

You would meet different people [in neighborhood #1]. Towards [neighborhood #2] you see like people you went to elementary school with people in the neighborhood. Towards [neighborhood #1] you see people you don’t know, stuff like that.

Other participants characterized neighborhoods by whether they were allowed to go there by their parents or guardians, how much time they spend there, or the presence of friends or relatives.

Perceived Social Characteristics

The theme of Perceived Social Characteristics encompassed more general, objective observations about people within a neighborhood, and was used by 25 participants. Some participants remarked on the visibility or density of residents in a neighborhood, such as “This one has kids running around…This one, not much kids not like a lot of kids.” Several participants described positive or negative social characteristics, as in “some people [in that neighborhood] are rude” or “most neighbors are friendly,” in which case codes encompassing both the social observation and the positive/negative appraisal were applied.

“Busy-ness” was a higher-order code within this theme encompassing the use of the specific terms, “busy,” “quiet,” or “loud” to describe a neighborhood (12 participants)—these terms were used in contexts describing the presence of people and/or social interactions within the neighborhood, as in “it’s crowded, busy.” Five participants explicitly tied in loudness or quietness with perceptions of crime or safety, as in “It’s a lot more quiet and the police don’t- they don’t really drive down the street.” or “[the neighborhoods are] different because the south is quieter. I say safer. More peaceful.” Though the term “safe” or “safety” was only used by two participants, “busy” codes were often used in contexts implying an awareness of safety or crime.

Within the Perceived Social Characteristics theme, only three participants explicitly referred to common indicators of social or physical disorder. One participant mentioned house fires in describing a neighborhood, another described, “[people] literally sleep outside downtown. They would just sit there right on the curb and go to sleep,” and another referred to a neighborhood as “a lot less taken care of by those in the actual neighborhood and the city.” The term “neighbor” was not coded in the current study, though this may be in part due to excluding quotations referring only to the home neighborhood.

Observed Structural Characteristics

Observed Structural Characteristics was a theme coded for 25 participants and primarily included the use of landmarks or geographical characteristics such as distance to differentiate between neighborhoods or to describe neighborhood boundaries. Schools were the most common landmarks cited, with the local mall, parks, stores, specific street names, and churches also being used to identify neighborhoods. Schools and malls were used in two distinct ways to define neighborhoods: either as a landmark within a larger neighborhood, or as being in itself the entire “neighborhood” for a participant—as in, a participant described being dropped off and picked up directly from school (or the mall), thereby having no knowledge of the characteristics of the surrounding streets. Ten participants indicated this latter definition of a school or mall as a neighborhood itself; in these cases, the “neighborhood” was mapped as simply the point of interest, and follow-up survey questions about neighborhood characteristics were not administered. Three participants referred to observable socioeconomic indicators, including the size of houses, or their perception of a neighborhood as “upper class.” More commonly, structural observations about a neighborhood were more general, including the presence of tall or short buildings, colorful murals, or density of homes versus businesses.

Subjective Appraisal

The fourth theme of Subjective Appraisal encompassed both nonspecific positive/negative codes (10 participants), as well as codes within the Familiarity or Personal Experiences or the Perceived Social Characteristics theme that included a positive/negative appraisal (8 participants). Nonspecific Subjective Appraisal codes encompassed vague impressions such as referring to a neighborhood as “a bad area,” “decent,” or “pretty cool,” or in one case simply referring to one neighborhood as “better” than another. Examples of more specific codes that blended Appraisal within another theme include, “I didn’t like that experience” (negative Personal Experience) or as noted above “people are rude” (negative Perceived Social Characteristic).

Discussion

The current study used several types of data to examine youth experiences and perceptions of their neighborhoods; together, the data demonstrate that youth cross Census and self-defined neighborhood boundaries in their daily routine activities, identify distinct neighborhoods within their activity space, and describe similarities and differences between neighborhoods using several dimensions. Results integrating Census data and youth perceptions were mixed: While Census-derived indicators of disadvantage did not consistently align with youth perceptions, Census data support youth perceptions of encountering “different” neighborhood spaces. These findings support the move in neighborhood literature away from defining neighborhoods through home Census tracts to incorporating person-centered measures of neighborhood, including a person’s routine activity space beyond their home neighborhood (Browning & Soller, 2014; Kwan, 2009). Large studies using administrative data have established important findings about neighborhood influences on youth, but studies using qualitative GIS and other mixed methods facilitate finer-grained examinations of youth experiences within neighborhoods.

Unlike younger children, adolescents have increasing agency in selecting their own neighborhood contexts—while they have little choice in their place of residence, teens have more freedom to choose where they spend their time. These choices shape their routine activity space and can lead them through different types of neighborhood environments. The current sample of teens engaged in diverse activities such as visiting family and friends, attending school or church, going shopping, or hanging out at the park, very often engaging in these activities outside of their self-defined home neighborhoods. They were more likely to cross Census or self-defined neighborhood boundaries for structured activities, such as attending church or other events, than for general unstructured activities, though for the most part the type of activity did not predict actual distance traveled away from home. The effect of activity structure was stronger for older youth, consistent with literature suggesting that as youth age they tend to travel farther distances for their activities (Andresen, Frank, & Felson, 2014; Cope & Lee, 2016). In the current sample, access to a car did not predict distance traveled; all participants resided in low-income neighborhoods and while none reported having their own car, most reported having easy access to a ride via a family member. Rather, it seems that older youth were more likely to be engaged in diverse structured activities that drew them to different neighborhoods. These results begin to shed light on reasons why a youth might travel between neighborhoods, contributing to other studies finding youth travel to hangout spots such as malls and fast food restaurants (Bichler, Christie-Merrall, & Sechrest, 2011). However, future research might expand focus on reasons why youth travel between particular types of neighborhoods—such as leaving the home neighborhood to enter a higher-crime area—that could inform models of youth exposure to different environmental contexts.

Both Census and self-reported data confirmed that in the current sample, travel to daily activities exposed youth to different types of neighborhoods. For the majority of youth who self-identified multiple neighborhoods within their routine activity space, these neighborhoods significantly varied on Census indicators of neighborhood disadvantage—though no participants spent time in wealthy neighborhoods and the magnitude of differences on Census indicators tended to be small. As all youth were recruited from organizations serving low-income neighborhoods, the small magnitude of differences on disadvantage is perhaps not surprising and may reflect the circumscription of youths’ movement through neighborhoods by social forces such as historic segregation (Krivo et al., 2013; Sampson, 2012). However, youth also perceived significant differences between neighborhoods on survey measures of collective efficacy and street code. Together with the qualitative data discussed below, these results highlight youth perceptions of distinctions between neighborhood spaces, consistent with studies that find teens identify distinct “safe” or “risky” areas within their activity space (Mennis & Mason, 2011; Teitelman et al., 2010). While these fine-grained distinctions between small areas may not be apparent to outside observers or administrative data sources, youth are reporting that they experience multiple distinct neighborhoods in their daily lives.

Youth responses to open-ended questions about their neighborhoods expand on their maps and survey data, filling in gaps and clarifying their experiences of multiple spaces. Qualitative analyses identified four different dimensions youth used to compare and contrast neighborhoods. Given their developmental stage, it might be expected that teens would exclusively describe neighborhoods in terms of their personal experiences—while this did occur, teens were just as likely to define neighborhoods using observable structural or social characteristics. Teens did not tend to use terms commonly found in neighborhood literature; they generated few references to physical disorder, no references to social control (such as willingness of neighbors to intervene), and few references to neighborliness. Instead, teens tended to describe people in neighborhoods as “nice” or “rude,” the streets as “quiet” or “loud,” and their own personal history within a neighborhood as positive or negative. Of particular interest, was the limited use of the terms “safety” or “crime”; instead, youth tended to use “quiet” or “loud” in contexts that suggest an appraisal of safety. This may just be due to differences in local youth vernacular, as some youth appeared to be using their sense of safety or comfort in part to identify neighborhood boundaries. Teens were much less likely to identify similarities between two distinct neighborhoods than differences; it may be that for adolescents, it is easier to distinguish neighborhoods where they can identify personally experienced or observed differences, and geographically proximate neighborhoods with similar characteristics may simply be perceived as a single neighborhood. It is possible that youth residing in cities with more clearly defined, named historic neighborhoods might be more likely to identify neighborhoods as distinct spaces sharing similar characteristics. It is important to note that teen responses to qualitative prompts received little follow-up probing in the current study; these results depict teens’ initial, immediate responses to simple questions about neighborhoods, and it is possible that on deeper reflection they would generate terms more similar to existing neighborhood literature.

While multiple sources of data confirmed that youth encounter distinct neighborhoods within their routine activity space, characterization of those environments is less consistent. Across analyses, youth perceptions of neighborhoods appeared to have little association with Census boundaries or variables. Consistent with prior studies, youth-defined neighborhoods tended to encompass multiple Census tracts and youth regularly crossed Census boundaries in their routine activities (Basta et al., 2010; Coulton, Korbin, Chan, & Su, 2001). In addition, Census indicators of neighborhood disadvantage did not consistently correlate with youth self-reported perceptions of neighborhood characteristics: neither collective efficacy nor street code correlated with any Census variables. Other researchers have also identified a lack of association between Census indicators and resident perceptions of their neighborhood, despite a contrasting body of literature linking these constructs, and call for more research explicating connections between structural characteristics such as poverty and resident perceptions (Duncan, Duncan, Okut, Strycker, & Hix-Small, 2003; Lin & Reich, 2016; Sampson, 2012). The current study did find that youth sense of community within the home neighborhood correlated with some structural indicators of disadvantage. It may be that the dimensions of sense of community—sense of belongingness, feeling of making a difference in the community, needs fulfillment, and emotional connection—were more immediate to the experiences of youth participants in their home neighborhood and therefore more closely connected to neighborhood structural characteristics than other types of neighborhood perceptions. However, given that this finding was unexpected and to our knowledge no prior studies have found different associations for sense of community versus other types of neighborhood perceptions, future research is needed to explore this variable.

The discrepancy between youth perceptions and Census data highlights the need for mixed-methods approaches in exploring how neighborhoods influence adolescents. Relying solely on Census indicators clearly does not capture the lived experiences of young neighborhood residents, highlighting the importance of their self-reported perceptions. Youth perceptions can be assessed through the integration of multiple methods such as their self-generated maps, responses to surveys derived from existing research, and analysis of their narratives about their spaces. The current study provides one example of a qualitative GIS approach—a set of methods bringing together administrative GIS data with participant perceptions, which has the potential not only to provide more precise measurement of experiences of contexts, but can lend the power and perceived legitimacy of GIS to amplify the voices of young residents (Dennis, 2006; Kwan & Ding, 2008). By moving beyond the study of the area immediately around a youth’s home to include their self-reported routine activity space, the current study uses a person-centered approach to defining neighborhood contexts. Future research on neighborhoods should take into account the specific aspects of neighborhoods that must be measured to answer the questions of interest; when studying neighborhoods as objective environments, Census data may be given more weight, while studies of neighborhoods as spaces will attend more to the perceptions of those who live in and move through them (Nicotera, 2007). Given the importance of neighborhood influences on adolescent development, mixed methods can provide more accurate and nuanced information on risk and protective factors as well as on the variability of youth experiences of neighborhoods.

Strengths and Limitations

The current study integrated data from quantitative surveys, participant-drawn maps recorded in GIS, and qualitative prompts to yield a multifaceted depiction of adolescent perceptions of neighborhoods. Responding to an increased call for qualitative GIS and person-centered definitions of spaces, we expand on the routine activity space literature to identify distinct self-defined neighborhoods within a youth’s activity space.

A number of limitations of the current study are important to consider in interpretation of our findings. The time diary method for identifying routine activity space provides an open-ended approach to identifying spaces a youth visits in a typical day, in contrast to methods that ask youth to identify particular spaces from a general list such as sites where they socialize or work. The primary limitation of this method for generating activity spaces is that by only describing two specific days, we likely missed spaces that youth consider important but do not visit regularly. Youth sometimes struggled to locate specific sites on a map, and in particular, youth who received rides from others had difficulty tracing their route on a map. While we made every effort to work with youth during the interview to locate sites, and later used transcriptions in further effort to pinpoint locations, it was impossible to avoid measurement error in self-report of locations. Future studies may benefit from integrating current technology such as Google Street View to help youth define neighborhood spaces.

The small sample size, including sampling from a single city, limits the generalizability of current findings to other populations. The city where the study took place has approximately 275,000 residents within city limits and has relatively few historically consistent or named neighborhoods. By contrast, the bulk of existing neighborhood research has been conducted in larger cities such as Chicago, Philadelphia, or Baltimore, which have historic neighborhoods recognized by generations of residents. It seems likely that youth perceptions of neighborhood boundaries are at least in part shaped by adult recognition of historic neighborhoods, and youth familiar with historic boundaries may rely on different dimensions of neighborhood characteristics such as landmarks or large streets to define neighborhoods. Limited variability in the current sample may have attenuated quantitative findings, though youth did report on diverse Census tracts within the city. Finally, by recruiting youth through neighborhood-based organizations, our findings may be more representative of some teens than others. Teens who are heavily involved in criminal or “street” activity may be less likely to be involved in prosocial afterschool programs, and their perceptions of neighborhoods may be shaped by their unique experiences. Finally, while the study design incorporated diverse methods, qualitative prompts and follow-up probing were limited. We have sought to present our interpretations in the context of existing literature on youth within their neighborhoods and encourage future research to incorporate more intensive qualitative methods.

Conclusions and Future Directions

The current study demonstrates one approach to integrating Census data with several youth-reported measures: maps of their home neighborhoods and routine activity spaces, responses to surveys, and responses to open-ended questions about their perceptions of neighborhoods. Together these measures demonstrated that youth clearly distinguished between multiple neighborhoods in their activity spaces, and used both researcher-provided (collective efficacy, street code) and youth-generated (“busy,” “quiet”) dimensions to describe neighborhoods. Future research on youth within neighborhoods can further explore how youth contrast different self-defined neighborhoods, and how these contrasts might be related to other youth outcomes. For example, youth appear more likely to engage in criminal activity in some neighborhoods rather than others, and travel between neighborhoods with different risk characteristics during routine activities (Wikström et al., 2010). Mixed-methods designs have potential to inform new insights into the mechanisms of neighborhood influences, through the integration of methods designed to assess different aspects of neighborhood contexts. As youth are so often the population of interest in studies of urban neighborhoods, it is particularly important to consider their unique perspectives on their contexts and experiences.

Highlights.

• Adolescents experience different neighborhoods within their activity space.

• Neighborhoods differ in terms of Census indicators and youth perceptions.

• The current study demonstrates one approach to integrating Census data with youth-reported measures.