Qualitative research methods were applied to a set of interviews to produce a grounded theory about how adolescent boys and girls reason about challenging problems in their lives. A metasynthesis is a technique for drawing inferences from similar or related studies; bringing together and examining datasets, discovering essential features, and articulating findings representative of all data. Synthesizing is a methodology that increases the credibility and trustworthiness of qualitative research by data triangulation (Meadows-Oliver, 2003). As such, it is a valuable tool for generating theory about the phenomenon and producing new and integrative interpretation of findings about the constructs that emerge.

Data of interviews from three studies will be synthesized to produce a grounded theory about problem solving thinking among adolescent boys and girls. Interviews with adolescent boys and girls; and adults (n =337) with proximal knowledge of the lived experience of urban youth (e.g. teachers, parents, community center staff, security guards, cafeteria staff, peer mediators, and bus drivers) were completed as part of a larger study to improve the ecological validity of violence prevention intervention research. Respondents lived in the city of Richmond, Virginia and were interviewed over the course of two years (2003-2005). Of these subjects, 97 % were African American, and all lived in an urban environment. Teens and adults were asked to identify the range of problems youth experience, and the perceived barriers and supports to engaging in various responses. These responses ranged on a continuum from prosocial to antisocial and from effective to ineffective in coping with the problem. Subjects represented a heterogeneous group of adolescents, including students in regular, learning disabled and gifted academic classes; designated aggressive, non-aggressive and socially competent youth; and males and females.

The data were collected in Richmond, Virginia (population 197,790), a community with a large population of low-income African American adolescents living in neighborhoods with high rates of crime and violence. Rapid decline in the City and growth in surrounding counties has resulted in a high concentration of poor people, especially minorities, within the City’s boundaries. In terms of crime, Richmond was ranked the ninth most dangerous city among all U.S. cities with populations of over 75,000 based on 2003 FBI violent crime statistics (Nolan, 2004). In 2001, the violent crime rate of 39 per 10,000 for youth under 18 was almost four times the state average. In 2002, the number of intake cases involving juvenile delinquency was 103 per 1,000 students, more than twice the state average (Kids Count, 2004).

Three sets of interviews
The Essential Skills for Violence Prevention study was a sequentially designed set of studies to guide the development of effective strategies to reduce violence among youth, particularly those in urban settings. These studies were modeled after the behavior-analytic model proposed by Goldfried and D’Zurilla (1969), involving the following steps: (a) situational analysis, (b) response enumeration, (c) response evaluation, (d) development of a measurement format, and (e) evaluation.

Institutional approval was obtained and researchers followed approved policies on protection of safety, rights and dignity of human subjects. Interviews were conducted by trained research staff, graduate students, and one faculty member between April, 2002 and December, 2005. Interviewers included women and men. Forty percent were African American and 60% were Caucasian. Interviewers completed ten hours of training on the purpose of the research and interview procedures; completed relevant readings, and participated in weekly meetings during data collection to discuss and improve interviewing strategies. A portion of each interviewer’s audiotapes was reviewed to evaluate interviewing skills and consistency with the protocol. Data for the metasynthesis come from the following three studies:

Problem Identification Study: Interviews (n = 110) on identifying problem situations and the context in which they occur were conducted with 60 adolescents including students and peer mediators; and with 50 adults, including parents of middle school students, community center staff, and middle school staff including bus drivers, cafeteria workers, teachers, security guards, and school administrators. The sample was purposively selected (Lincoln & Guba; 1985) to include individuals with proximal knowledge of the lived experience of urban youth.

Middle school students were interviewed during the school day in the library or a vacant classroom. Students were asked to identify three recent problems that were difficult. A problem was defined as “any situation that doesn’t feel good or comfortable, or makes you feel angry, sad, scared, disappointed or worried.” Participants were then prompted to identify a fourth problem that differed from the ones previously discussed. The following cues were used to probe for details: (a) “What makes this a problem?” (b) “How did you feel?” (c) “What is your relationship to the person/people” (d) “What led up to the problem?” (e) “Where did it happen?” (f) “Who was there?” and (g) “What was the outcome?” Interviewers summarized the interview and asked students to clarify any inaccuracies in their summary. Peer mediators were asked to identify up to six problems that were brought to peer mediation. Two situations were explored using probes similar to those used with the middle school students. Community center adolescents were interviewed at their center. The initial interview followed the same format used with middle school students. Follow-up interviews were then conducted on four consecutive weekdays following the initial interview to identify problems that had occurred during the previous 24 hours. Adolescents completed a worksheet during the follow-up interviews to facilitate their memory of the previous day. Parents were interviewed in their homes; school and community center staff were interviewed at work. Parent interviews asked about problems their child experienced during this transitional period. School and community center staff members were asked more generally about their perceptions of the problems middle school aged students experience internally and within their social networks.

Response Enumeration Study: Interviews about responses to the problems generated in the first study were done with 122 adolescent boys and girls attending one of three middle schools. Twenty five problem situations were selected and divided into 6 protocols with three problems each. Subjects were randomly assigned a protocol with 3 problem situations. Using this procedure resulted in between 12 to 16 responses to each problem. Semi-structured interviews were conducted at participants’ schools or homes. Interviewers received training and supervision. Problem situations represented internal struggles, and conflicts within family, school, peer and neighborhood networks. Interviewers read the problem and gave each participant a written description of the situation. Participants were asked to imagine the problem was happening to them; and then to respond to the following series of questions: (a) What are your thoughts, concerns, feelings or worries about the situation?, (b) What would be your goal?, (c) What would you most likely do?, (d) What would help you or keep you from responding in the way you identified?, (e) What would most kids do and why?, (f) What should a teen do and why?, (g) How likely are you to do your response (rating 1-4 on a Likert scale) and why?, and (h) What would help you and keep you from responding in the way you think you should?

Barriers and Supports Study: Interviews were done about paired problem and response set to explore barriers and supports for engaging in the particular response. Responses were either prosocial or aggressive. Students (n = 106) were sixth and seventh grader boys (n = 41) and girls (n = 65) from three public middle schools. In order to obtain a broad sample, students were randomly sampled from classrooms serving students with learning disabilities, advanced curriculum classrooms, and general classrooms. Interviewees were given a written description of a problem situation and response, instructed to imagine that the problem was happening to them and that they were going to make the response. They were then asked a series of questions about how easy it would be to make the response, how likely they would be to make the response, and why they would or would not make the response. Researchers then asked about factors that would make it easier and factors that would make it harder to make the response in that situation in the following domains: (a) personal characteristics (e.g., the type of person they are, things important or hard for them, ways they think or feel about things), (b) family, (c) school, (d) neighborhood and community, (e) friends, and (f) other people. Participants were then asked to respond spontaneously to eight fill-in-the-blank statements about the domains: (e.g., “Kids need to be [blank] to be able to do this [response],” “Kids’ parents need to be [blank] to be able to help them do this”). These were followed up with prompts for clarification by interviewers.

Procedures

Transcribed interviews were imported into NVivo, a computer software program designed for qualitative analysis (QSR International, 2005). An a priori list of themes was developed based on analyses of the individual datasets, unanswered questions from the individual data sets, and from triangulation with literature on adolescent coping and problem solving behavior. This list and the definitions were reviewed by the researchers.
The meta-ethnographic approach of Noblit and Hare (1988; 2003) and grounded theory approach of Strauss and Corbin ( ) will be used to analyze the three sets of interviews. These approaches emphasize the importance of context, or understanding behavior in context and both focus on research questions that seek meaning of ‘lived experiences’ of groups and individuals.
Coding

The first level of coding consisted of identifying distinct problems experienced by urban youth and their associated responses (prosocial or aggressive). A problem was defined as a situation, event, or experience that may have happened directly to the young person or that they may have some knowledge of through interactions with others in the social environment. In the first of the three sets of interviews, problems were the focus; in the second and third, responses were the focus. All problems that respondents talked about were coded and labeled. Responses were also coded, as they represented the solution. Only clearly prosocial and aggressive responses were coded, using the definitions listed below.

Table: Definitions for categorizing pro-social and aggressive responses
Responses Preliminary/Emergent Definition
Pro-social Behaviors, thoughts, attitudes, values, beliefs, and adaptive coping efforts that alleviate stress associated with the problem or focus directly on solving the problem. These efforts do not violate the rights of others. Responses were coded pro-social if they were not ambiguous, irrelevant, and clearly demonstrated pro-social motives (e.g. respect, compassion).
Aggressive Behaviors, thoughts, attitudes, motivations, values, beliefs, and maladaptive coping efforts that increases the likelihood of violence or results in a violation of the rights of others. Aggressive responses include physical (e.g., hitting, pushing, threatening with physical violence), relational (e.g., rumor, gossip, social exclusion), stealing or destruction of property, and more general verbal abuse (e.g., teasing).

Transcripts were also coded for barriers and supports. Barriers were defined as factors that decreased the likelihood of engaging in a particular response and supports were defined as factors that increased the likelihood of engaging in a particular response. Once a section of text was determined to be a problem, barrier or support, it was coded and labeled. Sixty transcripts, representing equal numbers of boys and girls and a representative sample of adult transcripts were coded by one researcher using this strategy. Common themes were collapsed, the literature was reviewed and additional themes added, resulting in a list of 40 problem types and 57 barriers and support themes.

The 337 interviews were divided and independently coded by two researchers (primary and secondary) using the a priori list of constructs and definitions resulting from the above procedure. An inter rater reliability procedure was designed to increase consistency in coding by multiple researchers. As the primary researcher finished coding 30 transcripts, two were randomly selected and coded by the secondary researcher. A report was run in NVivo comparing the coded and uncoded text to determine areas of agreement (e.g. with theme name). A coefficient of .90 was set. Any randomly selected transcript not achieving .90 agreement was discussed by the two coders until there was full agreement in coding. Any time the random selection of two transcripts did not achieve .90 agreement, an additional two transcripts were selected and recoded before any further coding was done. Through this close monitoring of coding, decision rules were developed, theme names and definitions were revised through constant comparison with the data, and a high degree of inter rater reliability was achieved and maintained.

Axial Coding

Axial coding will begin after open coding is completed. Axial coding will consist of within and across theme coding and collating and merging in order to create hierarchical structures and explanatory models (see Strauss & Corbin, 1990). Coding will consist of looking at relationships or patterns among themes, to determine where the elements of grounded theory can be applied to the data:

Element Description

Phenomenon The outcome of interest or the subject; including barriers for the use of aggression, supports for the use of aggression, barriers for prosocial responses, supports for prosocial responses, boys and girls.
Causal Conditions Events or variables that directly precede and lead to an occurrence of the phenomenon. It is a set of causes and their properties or relevant dimensions.
Context The context is a set of conditions, influencing the phenomenon, or moderator variables. It includes factors in the person and the environment that influence the self system and future development.
Intervening conditions Intervening conditions are mediator variables that influence the phenomenon.
Action strategies The purposeful, goal oriented activities that respondents perform in response to the phenomenon and intervening conditions.
Consequences The intended or unintended result of implementing action strategies.

An excel file with demographics including gender and type of respondent will be imported into the dataset. Emergent hypotheses will be generated. Examples of possible questions are:
1. What are the factors that were associated with endorsing the use of a prosocial response?
2. What are the factors that were associated with using an aggressive response?
3. What factors intervened to increase or decrease the likelihood of using an aggressive response?
4. What factors intervened to increase or decrease the likelihood of using a prosocial response?
5. What are the conditions under which teens use aggressive responses to solve
a. peer problems
b. school problems
c. family problems
d. internal problems
e. neighborhood/community problems?

This study is ongoing, and it anticipated to be completed by May, 2007