Evaluation and Program Planning 58 (2016) 160–170
Contents lists available at ScienceDirect
Evaluation and Program Planning
journal homepage: www.elsevier.com/locate/evalprogplan
Studying the teaching of kindness: A conceptual model for evaluating
kindness education programs in schools
Deanna M. Kaplana,* , Madeleine deBloisb , Violeta Dominguezb , Michele E. Walshb
a
Department of Psychology, University of Arizona, 1503 E. University Blvd., P.O. Box 210068, Tucson, AZ 85721-0068, United States
Community Research, Evaluation and Development, Frances McClelland Institute for Children, Youth and Families, Norton School of Family and Consumer
Sciences, University of Arizona, 650 N. Park Ave., P.O. Box 210078, Tucson, AZ 85721-0078, United States
b
A R T I C L E I N F O
Article history:
Received 22 December 2015
Received in revised form 24 May 2016
Accepted 1 June 2016
Available online 14 June 2016
Keywords:
Concept mapping
Mixed-method
Community-based participatory research
Kindness
School climate
Social-emotional learning
School program
School-based prevention
A B S T R A C T
Recent research suggests that school-based kindness education programs may benefit the learning and
social-emotional development of youth and may improve school climate and school safety outcomes.
However, how and to what extent kindness education programming influences positive outcomes in
schools is poorly understood, and such programs are difficult to evaluate in the absence of a conceptual
model for studying their effectiveness. In partnership with Kind Campus, a widely adopted school-based
kindness education program that uses a bottom-up program framework, a methodology called concept
mapping was used to develop a conceptual model for evaluating school-based kindness education
programs from the input of 123 middle school students and approximately 150 educators, school
professionals, and academic scholars. From the basis of this model, recommendations for processes and
outcomes that would be useful to assess in evaluations of kindness education programs are made, and
areas where additional instrument development may be necessary are highlighted. The utility of the
concept mapping method as an initial step in evaluating other grassroots or non-traditional educational
programming is also discussed.
ã 2016 Elsevier Ltd. All rights reserved.
1. Introduction
Strong social-emotional skills and a positive school climate are
widely regarded as important ingredients for students’ academic
success. Research has found that programming that aims to
develop students’ social-emotional skills positively influences a
range of academic, health and social outcomes. Social-emotional
skills are broadly defined as the skillset which enables one to
regulate emotions, have positive relationships, and engage in
effective goal-setting (for reviews, see Durlak, Weissberg,
Dymnicki, Taylor, & Schellinger, 2011; Greenberg et al., 2003;
Tough, 2013). Relatedly, school climate, defined as “the quality and
character of school life” (Cohen, McCabe, Michelli, & Pickeral,
2009), is also known to have a measurable impact on the academic
success and wellbeing of students. Past research has demonstrated
that positive school climate is associated with positive emotional
and mental health outcomes, increased self-esteem and
self-concept, increased motivation to learn, decreased bullying
and violence, decreased student absenteeism, and also mitigates
* Corresponding author.
E-mail address: [email protected] (D.M. Kaplan).
http://dx.doi.org/10.1016/j.evalprogplan.2016.06.001
0149-7189/ã 2016 Elsevier Ltd. All rights reserved.
the impact of socioeconomic risk on academic performance (for a
review, see Thapa, Cohen, Guffey, & Higgins-D’Alessandro, 2013).
Moreover, improving the social environment of a school has been
identified as one of the top strategies for preventing and
intervening in school safety concerns such as bullying and peer
aggression (U.S. Department of Health and Human Services, 2014).
In the context of this research, there is growing momentum for
the idea that kindness is a valuable social-emotional skill and
component of school climate that should be explicitly taught in
schools. A number of school-based programs focusing specifically
on “kindness education” have been developed to meet this
demand, such as the Random Acts of Kindness Foundation
Kindness in the Classroom Program (The Random Acts of Kindness
Foundation, 2015); the Center for Investigating Healthy Mind’s
mindfulness-based Kindness Curriculum (Flook, Goldberg, Pinger,
& Davidson, 2015); Think Kindness (Think Kindness, 2015); and the
Ben’s Bells Kind Campus Program (Ben’s Bells Project, 2015).
School-based kindness education is a burgeoning – and consequently loosely defined – category of programming, but typical
components among programs include dedicated time for exploring
what kindness is and why it is important, mindfulness or
self-awareness exercises, gratitude reflections and exercises, and
a programmatic structure for acknowledging and performing acts
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
of kindness at the student, classroom, school and/or community
level.
Preliminary research about the mindfulness-based Kindness
Curriculum, a program for preschool children based largely in
secular mindfulness instruction, found that the program had
positive effects on social competence, social-emotional development, cognitive flexibility and delay of gratification (Flook et al.,
2015). Beyond this study, very little research has focused on
school-based kindness education programs, and other existing
programs have not been systematically evaluated. There is a
paucity of research focusing on kindness interventions in youth
even beyond the school setting, although developmental literature
suggests that compassion and mindfulness training in adolescence
may promote pro-social behavior, empathy, perspective taking,
compassionate evaluations of self and other, and enhance
self-regulation and emotional awareness (Roeser & Pinela,
2014). Separately, research on a gratitude intervention for
adolescents found that gratitude practice had positive immediate
and longer-term effects, and was related to optimism, wellbeing,
life satisfaction, and school experience satisfaction (Froh, Sefick, &
Emmons, 2008). Recent research on one Positive Psychology
Intervention (PPI) designed for middle school students that
included gratitude and kindness as part of the intervention also
found that the curriculum increased life satisfaction among
participants (Suldo, Savage, & Mercer, 2014).
Relatedly, previous studies in adult populations demonstrated
that performing intentional acts of kindness and recognizing
kindness in others has positive mental health outcomes, including
reducing depressive symptoms and increasing subjective happiness and life satisfaction (Buchanan & Bardi, 2010; Otake, Shimai,
Tanaka-Matsumi, Otsui, & Fredrickson, 2006; Post, 2005). Compassion cultivation practices have been associated with decreased
stress response and negative affect, as well as increased positive
affect, feelings of social connectedness, and increases in personal
resources such as physical health, sense of purpose in life,
self-acceptance, mindfulness, and positive relations with others
(Fredrickson, Cohn, Coffey, Pek, & Finkel, 2008; Hofmann, Grossman, & Hinton, 2011; Pace et al., 2009). Compassion is defined as
“the feeling that arises in witnessing another’s suffering and that
motivates a subsequent desire to help” (Goetz, Keltner, & SimonThomas, 2010, p. 352). Therefore, although compassion is a related
construct, compassion-based interventions may be distinct from
kindness-based interventions (which typically aim to increase
kind, pro-social behaviors even in the absence of witnessing
suffering).
In sum, the limited existing research suggests that kindness
education programs may improve student social-emotional skills
and the overall social environment of schools, thereby having
positive impacts on students’ wellbeing, achievement, and success.
Kindness education has emerged as a promising but underevaluated area of school-based programming.
161
in action to benefit others). Program materials are provided to
participating schools at no cost and include age-appropriate
classroom activities, school-wide activities, take-home activities,
and exercises for school faculty and staff. These materials are
distributed to participating schools on a monthly basis, and each
month addresses a different theme related to kindness. In the
context of other kindness education school programs, a unique
feature of Kind Campus is its inclusivity of the adults on school
campuses. This stems from the organization’s philosophy that the
working environment perceived and created by adults is an
important component of overall school climate and social learning.
Kind Campus uses a bottom-up, non-traditional program
framework. In contrast to more traditional school-based programs
that proscribe a set curriculum, Kind Campus invites schools to
adapt the program to meet the unique needs of their individual
school environment. Schools are encouraged to incorporate
activities and materials into existing school and classroom
routines, such that they become a “way of life” for the campus
community. This type of grassroots, bottom-up approach is known
to be valuable in helping to assure that programs can fit the
complexities and needs of diverse schools (Astor, Meyer, Benbenishty, Marachi, & Rosemond, 2005).
Evaluating Kind Campus presents two key design challenges.
First, the grassroots nature of Kind Campus leads to both dosage
and practical differences in how participating schools are
implementing the program. An evaluation of a grassroots program
like Kind Campus therefore also entails consideration of how the
teaching and practice of kindness naturally unfolds from the
bottom-up in diverse school types and demographics. Second, as
little research has been done in the area of school-based kindness
education, the field presently lacks a conceptual framework for this
type of programming. It is unclear how the construct of kindness
concretely manifests at the classroom and school levels or across
the developmental trajectory. What outcomes could kindness
education programming reasonably be expected to change, and
through what mechanisms? How can these outcomes and
mechanisms be meaningfully measured? Establishing a conceptual model for kindness education is crucial to designing evaluations
that can answer questions such as to what extent these programs
are effective, where and under what conditions they work best,
which program components are most effective, and the degree to
which dosage matters in program implementation.
In light of the growing number of both grassroots and more
traditional kindness education programs and the rising number of
schools that have adopted them, rigorous evaluations of such
programming are important for helping schools allocate their
limited resources towards empirically-based efforts that are
known to keep students safe and help them thrive. In partnership
with Ben’s Bells and Kind Campus schools, the present study used a
community-based participatory research method called concept
mapping to develop a conceptual model for researching and
evaluating kindness education programming in schools.
1.1. Kind Campus and the present study
2. Method
Kind Campus is a widely adopted kindness education program
currently reaching more than 200,000 students in over 300 schools
in the United States. The program was developed by Ben’s Bells, a
non-profit organization headquartered in Tucson, Arizona. The
goal of Kind Campus is to educate entire school communities
(students, families, faculty, and staff) about the positive impacts of
kindness and empower them to cultivate a culture of kindness in
their schools and beyond. The program is based on four tenets: the
Kind Mind (becoming aware of the nature of mental experience),
Self-Kindness (becoming self-aware and practicing self-compassion), Social-Kindness (understanding interdependency and
cultivating empathy), and Kindness in Action (putting kindness
With the goal of developing a conceptual model for evaluating
school-based kindness education programs, a mixed method,
community-based participatory research (CBPR) approach to
construct conceptualization called concept mapping was selected
for this project. Previous reviews have noted the substantive
benefits of CBPR methods in intervention, health and community
research: minimization of the gap between research and translation in program implementation, buy-in and trust that enhances
data quality, and an authentic framework for developing and
testing culturally appropriate instruments (Viswanathan et al.,
2004; Wallerstein & Duran, 2010).
162
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
Concept mapping was selected because it uniquely offers the
ability to bring diverse stakeholders together in order to build a
common framework (Kane & Trochim, 2007). Additionally, concept
mapping offers a mixed method approach to construct conceptualization, in contrast to other approaches that are primarily
qualitative (e.g., focus groups, phenomenological approaches,
grounded theory). This method has been successfully employed
by social scientists in areas such as health and education in order to
articulate theory and provide a basis for measurement (Anderson
et al., 2006; Kane & Trochim, 2007; Markham, Mintzes, & Jones,
1994; Rosas & Camphausen, 2007).
Concept mapping is a group process of generating ideas and
articulating the relationship between these ideas, and is commonly
conducted across five phases: 1) brainstorming, 2) idea synthesis,
3) sorting and rating, 4) production of concept maps by the project
investigators using multidimensional scaling and cluster analysis
techniques, and 5) group interpretation of the resultant concept
maps (Kane & Trochim, 2007). The Concept System Global MAX
software (Concept Systems, Inc. Copyright 2000–2015), a
web-based program developed for facilitating concept mapping
studies, was used for data collection, management, and analysis.
This study was not considered human subjects research by
the Human Subjects Protection Program at the University of
Arizona and so did not require review by its Institutional
Review Board.
2.1. Participants
Recruitment for this study was done in partnership with Ben’s
Bells staff in Tucson, Arizona. Three participant panels were
recruited in an effort to capture diverse perspectives across the
various phases of the study: 1) a student panel comprised of youth
in sixth through eighth grades, 2) a school professional panel
comprised of teachers, administrators, staff, and others working in
the school setting, and 3) a scholar panel comprised of scholars
with academic expertise relevant to kindness education. Table 1
provides an overview of how the five steps of the concept mapping
method were carried out across the three participant groups.
Student panel participants (n = 123) were recruited for the
brainstorming phase of the project from four public middle schools
(two charter and two district schools) participating in the Kind
Campus program in Tucson, Arizona. Middle school students were
selected because elementary and middle schools are the primary
users of Kind Campus. Student panel participants were in sixth
through eighth grades, and the percentage of students eligible for
free/reduced price lunch in participating schools ranged from 47 to
85%. Due to the cognitive complexity of the task, 42 public high
school students, rather than middle school students, were
additionally recruited to participate in the sorting and rating
project phase. However, these data were ultimately excluded from
analysis due to concerns about data quality, and so the student
panel contributed to the brainstorming phase of the project only.
The school professional panel, also recruited by Ben’s Bells, was
comprised of over 125 school-based professionals from approximately 50 schools in the Tucson area participating in the Kind
Campus program. School professionals were not asked to disclose
their job title during the brainstorming phase to assure anonymity,
but those who participated in the sorting and rating phases of the
project were asked to provide this information. Of these 16 school
professionals, seven were classroom teachers, three were parents
or parent-teacher organization members, two were learning
supports coordinators, one was a school administrator (school
principal or vice principal), one provided other school administrative support (school secretary or administrative assistant), and two
indicated being “other school faculty/staff.” Fourteen school
professionals participated in both sorting and rating; two
participated in one or the other.
The scholar panel was recruited by project investigators.
Research faculty at the University of Arizona with relevant
expertise (n = 37) were invited to participate, and participation
invitations were also extended to 37 scholars outside the
University of Arizona (selected from publications between 2003
and 2013 in the areas of school climate, social-emotional learning,
kindness/compassion interventions, or the evaluation and
measurement of these constructs).
Based on recommendations from Kane & Trochim (2007), the
number of individuals allowed to participate in the brainstorming
phase of this project was not capped, but sorting and rating phases
were restricted to smaller groups. Past research suggests that a
minimum of 10 sorters and raters are needed, and the reliability of
the results increases with more sorters and raters (Rosas &
Camphausen, 2007). Table 2 shows a breakdown of participants in
each phase of the project by panel.
2.2. Brainstorming
In the brainstorming phase of concept mapping, participants
generate ideas in response to a prompt. The objective of brainstorming is to yield a list of statements central to the project; in this case,
kindness in the school setting. Brainstorming is typically facilitated
either as an in-person group process or remotely online (Kane &
Trochim, 2007). Investigators facilitated in-person brainstorming
sessions at each school in the student panel. For the school
professional and scholar panels, investigators held a joint in-person
group brainstorming session at an annual conference hosted by Ben’s
Bells that aims to bring professionals at Kind Campus schools
together with University scholars. School professional and scholar
panels additionally had the opportunity to participate online.
The brainstorming prompt, developed based upon guidelines
provided by Kane & Trochim (2007) was:
Describe for us (in words or phrases) what results we should expect
to see from a program designed to increase kindness in schools, if it
were effective. Please be as specific as possible.
Table 1
Overview of project phases and participant groups.
Project Phase
Description of Phase
Generating ideas in response to a prompt
Brainstorming
Idea Synthesis
Editing brainstormed statement set for length and clarity
Sorting and Rating Sorting and rating of the statement set to collect information about how the statements are related to
one another
Multidimensional scaling and hierarchical cluster analysis of sorting and rating data
Production of
Maps
Group
Collaborative interpretation of concept maps
Interpretation
Actors
Students, School Professionals, Scholars
Project Investigators
School Professionals, Scholars
Project Investigators
School Professionals, Scholars, Project
Investigators
163
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
Table 2
Participants by project phase.
Panel
Brainstorming
Sorting
Importance Rating
Feasibility Rating
Data Interpretation
Scholar
School Professional
Student
Total
12–24a
125–150a
123
180a
11
15
–b
42
11
14
–b
49
10
13
–b
47
5
5
0
10
Note: Participation in the brainstorming phase was anonymous in order to encourage participation among those concerned about being personally linked with their
brainstormed statements.
a
These ranges are conservative estimates based on the number of individuals who attended in-person brainstorming sessions and participated anonymously online.
b
Although 42 high school students were recruited for the sorting and rating task, these data were excluded from analysis due to concerns about data quality and the
appropriateness of this cognitively complex task for youth.
“One result of an effective program designed to increase kindness in
schools would be . . . ”
Student panels received an adapted brainstorming prompt
developed from feedback provided by a group of middle school
students who participated in an initial brainstorming session:
“Let’s pretend that you visited another school for a day and you
were trying to decide if that school was a kind school or not. What
things might you see, hear, or notice that would help you know
whether or not the school is a kind place?”
The brainstorming phase of the project yielded a total of 899
statements from the three participant groups.
2.3. Idea synthesis
Next, investigators edited and collated the brainstormed
statements based on recommended concept mapping guidelines
that a) each statement expresses only one distinct idea, b) each
statement is unique and non-redundant, c) each statement is
relevant to the focus of the project, d) each statement is clear and
comprehendible (Kane & Trochim, 2007). After removing duplicative statements, the final statement set contained 108 unique
statements specifically related to the brainstorming prompt. The
final statement set is provided in Appendix A.
2.4. Structured sorting and rating
During structured sorting and rating, a subset of participants
are asked to sort the statements in the final statement set into
“piles” of items that they perceive as thematically similar, and are
then asked to rate each statement on one or more specified
dimensions (Kane & Trochim, 2007; Trochim, 1989).
Adult participants (school professionals and scholars) completed this phase of participation online individually and at a time and
location of their choice using the Concept Systems Global MAX
software, which offers an interface designed to facilitate this
process. They were instructed to sort statements into piles such
that statements in each pile have more in common with each other
than statements in other piles, and then to label each pile with a
name that describes its theme. Detailed directions for this task,
including examples, were provided. Participants were also
instructed to contact project investigators by email or by phone
with any questions about the task. After completing the sorting
task, participants rated each statement on importance relative to
all other statements using a Likert-type scale (1 = relatively
unimportant, 5 = extremely important). Next they were prompted
to rate each statement on feasibility using a Likert-type scale
(1 = relatively unfeasible, 5 = relatively feasible) with the prompt,
“Which results do you believe would be the most feasible for a
school-based intervention to accomplish, relative to the others?”
Adult participants were compensated with a $5 Starbucks eGift
card upon completion of the sorting and rating tasks.
Student participants completed the sorting and rating tasks
during in-person sessions conducted at their schools. For the
sorting task, students were each provided with a set of 108 cards
with one statement printed per card and given the same
instructions as adult participants for sorting their card deck into
physical piles. Student participants then completed the rating task
using a paper-pencil version of the Likert-type scale described
above. Students were provided with snacks during the sessions.
Although all participants were instructed to “not create categories
according to priority or value (e.g., ‘Important’, ‘Hard to Do’),” many
students did not follow this instruction, suggesting that they did
not understand the task. Additionally, many students were unable
to finish the task; although adults had the ability to complete the
task online without a time limit, student sorting and rating
sessions were restricted to the length of one class period
(35–60 min) due to time constraints of the school day. Because
of these concerns about the quality of their data, student sorting
and rating data were excluded prior to data analysis. The concept
maps therefore reflect adult (school professional and scholar
panel) sorting and rating data only. Although the students’
perspectives are still represented through their contributions to
the statement set, the resulting model may be biased towards the
adult perspective, and so would benefit from efforts to validate its
interpretation with students. Approaches for doing so are
discussed further in the Discussion section of this paper.
2.5. Production of concept maps
The software chosen for this project allows investigators to
use multidimensional scaling (MDS) and hierarchical cluster
analysis (HCA) to analyze the data. First, a similarity matrix is
generated in which all 108 statements are paired with one
another and assigned a numerical value indicating the number
of people who put that pair of statements in the same pile
(Kane & Trochim, 2007). Next, a two-dimensional MDS solution
is used to produce X-Y coordinates for each statement in twodimensional space following a bivariate distribution (Fig. 1).
Then, HCA is used to mathematically group each statement into
clusters that reflect similar themes. The software uses an
agglomerative cluster analysis approach using Ward’s algorithm
(Ward, 1963), which functions to minimize total within-cluster
variance based on the sum of squared distances of each point
from the centroid of its cluster (stepwise). In short, the MDS X-Y
coordinate matrix is used as input by the algorithm, and the
output hierarchically arranges clusters by beginning with each
statement and combining them sequentially until the number
of clusters specified by investigators has been created (Everitt,
Landau, Leese, & Stahl, 2011; Kane & Trochim, 2007). There is
164
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
Fig. 1. Statement point map (MDS analysis of sorting data).
no mathematical formula for choosing a number of clusters;
more clusters yield a finer-grained and more detailed concept
map (Kane & Trochim, 2007). In the present study, a 10 cluster
solution was selected because it balances a high level of detail
with a sufficient number of statements in each cluster for
facilitating interpretation.
2.6. Group interpretation
The concept mapping method commonly includes group data
interpretation as a final step, in which project investigators meet
with project participants to collaboratively interpret the cluster
maps and assign meaningful labels to the clusters based on the
statements they contain (Kane & Trochim, 2007). Investigators
invited all adult sorting and rating panel participants to a 90minute data interpretation session. Of those invited, the session
was attended by five school professionals and five scholars. Two
project investigators and two Ben’s Bells staff members also
attended.
2.7. Development of conceptual model
Kane & Trochim (2007) note that after concept maps are
produced, clusters can be further grouped into meaningful regions
in order to simplify the map and aid interpretation. Following the
group interpretation session, investigators therefore further
aggregated spatially related, thematically similar clusters. As a
final step, investigators created a conceptual model based on the
concept map.
3. Results
MDS analysis was first used to locate each statement in twodimensional space as shown in the Point Map in Fig. 1. Each dot on
the point map represents one statement, and points on the map
that are closer together were sorted together by participants more
commonly than points on the map that are further away from one
another. In MDS analysis, the stress index, which functions as a
measure of the goodness of fit between the values of the similarity
matrix and the X-Y coordinates of each point assigned by MDS, is
the key diagnostic statistic. The stress index value for this project
was 0.31, which falls into the normal and acceptable range for a
concept mapping study (Kane & Trochim, 2007; Miller & Jones,
2015; Trochim, 1989).
Agglomerative HCA was then used to mathematically group
points into ten clusters. Clusters were named as follows by
participants in the group data interpretation session (names are
intended to represent the perceived conceptual theme of the
statements in each cluster): Inviting Environment; Positive
Interactions; Positive Student Community; Intentionality and
Awareness; School Operational Outcomes; Staff Impact; Family
and Community/Extramural; Student Social-Emotional Skills;
Empathy/Acceptance/Caring; and Emotional Well-Being and Civility.
Following the group data interpretation, the investigators
further aggregated spatially related, thematically similar clusters
into six regions in order to simplify the map: Student SocialEmotional Skills; School Climate; Kindness Focus; Work Environment; School Operational Outcomes; and Family and Community.
This final cluster map is shown in Fig. 2.
Bridging values are the key statistic for hierarchical cluster
analysis. Bridging values range from zero to one and indicate how
frequently each statement was sorted with the other statements in
its own cluster, as compared to statements in other clusters. Lower
values indicate that a statement was commonly sorted with the
other statements in its cluster, and higher values indicate that a
statement was commonly sorted with statements on either side of
its cluster and therefore connects more distant areas of the concept
map (Kane & Trochim, 2007). Bridging values can serve as a
quantitative guide for interpreting the concept map (Kane &
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
165
Fig. 2. Cluster map.
Trochim, 2007). At the statement level, low bridging values may
indicate statements that are most central to their cluster. At the
cluster level, higher bridging values suggest that a cluster is more
interrelated with other parts of the map, whereas clusters with low
bridging values may be more conceptually independent. In the
present study cluster bridging values were low, ranging from 0.06
to 0.48, with the exception of the Family/Community cluster (0.86).
This suggests that with one exception, the clusters were relatively
conceptually independent. Appendix A shows the bridging value
for each statement and for each cluster overall.
Figs. 3 and 4 show the mean importance and feasibility ratings
assigned to each cluster by scholar and school professional
participants. Variance among ratings was very low. Mean
importance ratings for the clusters ranged from 3.79 to 4.40 on
a five-point scale and mean feasibility ratings ranged from 3.32 to
3.92 on a five-point scale, suggesting that participants felt that
creating change in all clusters was highly important but only
moderately feasible. However, clusters that were rated as relatively
most important were also those rated as relatively most feasible.
One exception to this was the “Intentionality and Awareness”
cluster, which was rated as the most feasible, but relatively less
important.
Concept mapping is often used to generate a conceptual model
in order to articulate program theory and provide a basis for
measurement (e.g., Sundra et al., 2006; Yampolskaya, Nesman,
Hernandez, & Koch, 2004). Investigators therefore next created a
conceptual model (Fig. 5) using the spatial relationships inherent
in the cluster map as a guide by beginning in the center of the map
with the most central and highly feasible cluster (Cluster 4) and
then rotating outwards. The model suggests that kindness
Fig. 3. Importance cluster rating map.
166
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
Fig. 4. Feasibility cluster rating map.
education programs focus on developing awareness about
kindness and its positive impacts, thereby fostering intention
among members of the school community to try to practice
kindness and be aware of kind acts happening around them. This
process promotes a positive school climate and supports student’s
social-emotional skills. School climate is also closely related to the
quality of the work environment perceived by adults at the school,
and positive changes in one are likely to enhance the other.
Improvements in the overall school climate and the socialemotional skills within the school community may have positive
downstream effects on school operational and school achievement
outcomes as well as outcomes in the broader community affiliated
with the school.
Fig. 5. Processes and outcomes of kindness education programming.
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
4. Discussion
The aim of this study was to develop a conceptual model that
can be used to inform the research and evaluation of kindness
education programs in schools. Through the use of concept
mapping, a model was derived from data provided by school
students, teachers, administrators, staff, and university scholars—
namely, those who can speak from their direct, lived experience as
well as those who also bring an empirical perspective to kindness
education.
This model proposes that kindness education programs yield
student-level and school-level impacts in large part through
making the idea of kindness salient. This suggests that effective
kindness education programs would offer a common language for
school students and staff to talk about kindness and its positive
impacts, and provide a framework that encourages members of the
school community to acknowledge acts of kindness happening
around them and to practice kindness towards themselves and
others. This fundamental foundation is expected to give rise to a
more positive school climate for students and a positive work
environment for adults, and supports the development of student’s
social-emotional skills. Downstream effects of improved school
climate and student social-emotional skills may include positive
school operational outcomes (e.g., achievement, disciplinary,
health) as well as outcomes in the greater school community
(e.g., within school families and the local community at large).
School improvement efforts such as kindness education are
highly complex. As reflected in this model, the social ecology of
school life includes variables that happen at the level of each
individual student and educator, the level of each classroom, the
level of the school overall, as well as the level of the school’s
families and the surrounding community. Over the course of
school improvement efforts, each of these variables may influence
one another in an ongoing manner, prompting many scholars to
call for school research designs using methods such as hierarchical
linear modeling and multilevel modeling that aim to capture some
aspects of this complexity (e.g., National School Climate Council,
2015). Guided by the model developed from the present work, we
offer suggestions for variables at multiple levels of school life that
would be useful to include in the evaluation of kindness education
programs.
First, the model suggests two key outcome variables that should
be assessed: school climate (including the work environment
perceived by adults on school campus), and student socialemotional skills. Several well-validated and grade-appropriate
measures for these constructs are available, as are measures for
closely related constructs such as social-emotional knowledge and
student dispositions. Second, this model suggests two additional
downstream outcomes of kindness education programming that
would be useful to assess in longitudinal designs: improvements in
school operational outcomes, and outcomes occurring within
school families and the surrounding community.
Third, this model points to an important area that requires
instrument development: “Kindness Focus” (originally named
“Intentionality and Awareness” by participants), which emerged in
the conceptual model as unique and distinctive from the
well-established domains of school climate and social-emotional
skills. This cluster refers to the emergence of an increased focus on
kindness throughout the school community, and an increased
awareness of the nuances of kindness and its positive impacts. This
is a likely mechanism through which kindness education programs
create change in schools—and notably, participants rated the
statements in this cluster as the most feasible results of kindness
education programs (Fig. 4). Measuring change in this area
therefore seems necessary for assessing the extent and success
of program implementation and for unpacking whether kindness
167
education adds something above and beyond components of a
program that may also directly influence school climate or student
social-emotional skills. The development and validation of instruments that capture change in this process variable are key next
steps to facilitating a better understanding of the mechanisms by
which kindness education programs have an effect and accounting
for variability in their outcomes. The statements in this cluster
(Cluster 4, Appendix A) contain examples of some of the
perceptions and concrete behaviors that could comprise this
domain.
Finally, the “Family/Community Outcomes” cluster generated
during concept mapping warrants discussion. The value of
thinking about schools as a component of their greater community
is well supported by previous research. Integrated strategies for
improving physical and social environments within schools and
neighborhoods have been found to promote optimal child health
and well-being, especially among children living in high-poverty
neighborhoods (Komro, Flay, Biglan, & Promise Neighborhoods
Research Consortium, 2011). Further, family support and
community involvement positively impact students’ academic
performance, and children who are at high risk for struggling
academically experience particular gains from family and
community engagement (Henderson & Mapp, 2002). For program
developers, it would therefore be useful to consider how
school-based programming could be designed with integrative
strategies in mind. For researchers, family and community
outcomes that are relevant to the intervention (e.g., kindness
focus at home, family engagement in school) may be useful to
include in research designs.
Although students’ voices were represented in the brainstorming phase and therefore in the final statement list, student sorting
and rating data were ultimately excluded from MDS and HCA
analysis due to concerns about the quality of these data. The
conceptual model may be biased towards the adult perspective,
and future research could attempt to validate this model with
students through the use of focus groups or similar methods. There
may also be discrepancies between what adults view as important
and feasible components of kindness education programs and
what students feel is most important and most feasible. Broadly,
little is known about what students perceive to be the most
important and feasible elements of school-based programs. This is
an important topic that warrants its own systematic investigation
and subsequent integration into a refined model.
4.1. Lessons learned
Concept mapping is a recommended methodology for uniting
diverse stakeholders in building a common framework, articulating theory, and providing a basis for measurement. A unique
feature of this study is that it represents one of the first attempts to
apply this method to the field of education by unifying the
perspectives of school students, faculty, staff, and administrators,
as well as academic scholars with relevant expertise. This is also
one of the first published studies that has attempted to use the
concept mapping method with youth.
Concept mapping does not yield a definitive prescription for
designing an evaluation; it is still necessary to subsequently think
through a theory-driven and rigorous design. However, this
method was a valuable starting place in designing an evaluation
of a program with a grassroots, non-traditional program structure,
which by design offers schools freedom to creatively implement
the program to fit their unique needs. The concept maps produced
in this project yielded an easily interpretable model, and although
the model is limited by only partial inclusion of student data, it
provides a conceptual framework for future evaluation of Kind
Campus and other school-based kindness education programs.
168
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
Additionally, engaging in the concept mapping process
generated enthusiasm among adult participants. Following completion of this project, many of the schools, teachers, and
administrators who participated in the project reached out to
investigators to express interest in contributing to further research
and evaluation efforts about kindness education programming in
schools. Given that buy-in is often a limiting factor in program
evaluation, particularly in heavily evaluated places like schools,
this is another notable strength of this method. The concept
mapping method is likely to be useful to other questions in
educational research, particularly for bottom-up or otherwise nontraditional programming, or in situations where buy-in is an
obstacle.
The largest methodological challenge encountered in this
project was adapting the concept mapping process to working
with students, rather than adults. Two modifications to the
procedure were made for students. In the brainstorming phase, the
language of the brainstorming prompt was adapted with the aim of
making it more concrete and relatable to youth. This seemed to
work well, with youth appearing engaged and enthusiastically
providing examples. In the sorting and rating phase, investigators
facilitated in-person card sorting sessions for students (rather than
asking them to participate online) so that they had the option to
ask the investigators clarification questions. Students were also
told that some of the statements were written by university
professors or other adults, and were encouraged to ask questions if
they didn’t understand a statement. Despite these precautions, as
previously described, student sorting and rating data were
ultimately excluded from analysis because students did not
complete the task as designed. One obstacle may have been that
students had limited time to complete the task due to time
constraints associated with facilitating the task during the school
day. In the future, investigators who wish to conduct this process
with youth should assure that students can take all the time they
need, and may also wish to conduct a pilot sorting and rating
session with a small group of youth to solicit feedback about how
this phase of the process can be made more accessible.
prosocial education more broadly to focus on identifying specific
practices that are most successful in achieving these common
aims, taking into account that the most effective practices may
vary according to the diversity of schools and their surrounding
communities. The findings from the present work also suggest that
program elements that promote positive workplace climate and
support the ongoing social-emotional needs of educators may be
additional, important aspects of prosocial educational efforts.
Acknowledgments
We would like to thank Ben’s Bells for their support of this
project, especially Jeannette Maré, Founder and Executive Director,
and Cassie Williams, former Education Program Manager. We
would also like to express our appreciation for the students, school
faculty, and scholars who volunteered their time to participate in
this project, and to the University of Arizona Foundation and The
Office of the Senior Vice President for Research for providing seed
funding for this work. We also thank John Daws and Kara Tanoue
for their assistance in facilitating this project, and Maggie Van Dop
and Ayana Blackey for their help with data entry.
Appendix A. Final statement set by cluster with bridging values
Final statement set by cluster with bridging values
Cluster
1. Inviting
Environment
4.2. Recommendations and future directions
There are genres of school programming other than kindness
education that focus on the social, emotional, and civic elements of
youth development and school life. A recent paper attempting to
identify the shared components of “prosocial education” programs
(including character education, whole child, school climate, socialemotional learning, and positive psychology school programs
among others) found that although each of these programs have
unique features, they also share many common goals and elements
(National School Climate Council, 2015). Although the review did
not include kindness education, many of the anticipated results of
kindness education that emerged in the present study align with
the common goals outlined by the National School Climate Council.
These include the promotion of safe, supportive, and engaging
school climates; meaningful relationships among youth and
adults; indicators of success that capture social, emotional, and
civic outcomes; pro-social education for students; an emphasis on
adult modeling; and professional development opportunities that
support student safety, support, and engagement.
Kindness education is unique among prosocial educational
efforts in that its explicit focus is on kindness, and increasing
kindness is the means through which this programming aims to
positively impact other student and school-level outcomes. Yet,
the preliminary conceptual model presented here suggests that the
overarching goals and desired outcomes of kindness education are
similar to those of other prosocial education programs. It may
therefore be useful for future research in kindness education and in
2. Positive
Interactions
Statement
increased school pride
a friendly and nurturing environment for
learning
a school culture that makes kindness socially
enhancing AND meanness socially damaging
that school is a “happy place”
less graffiti and vandalism
a cleaner campus
happier students, teachers, parents,
administrators, and school staff
a “greener”, more environmentally friendly
campus
a more relaxed and inviting school
atmosphere
a greater sense of connectedness to school
less emphasis on the word “bullying”
a greater sense of safety at school
creating a culture where students can feel
safe to take educational and social risks
a sense of belonging for all involved/
everyone feels “a part of” the school
community
increased participation in clubs, activities,
events and after school groups
seeing students greeting each other and
adults
teachers and students working together to
create a kind and positive environment
finding subtle kind messages and empathy
lessons being worked into the classroom
that adults are more approachable
increased trust among students & staff,
students & students, and staff & staff
physical evidence in the school showing
support of the program
positive framing of expectations (e.g., going
from “don’t forgot homework” to “remember
to do your homework”)
teachers and staff being more kind in
behavior and speech
all members of the school community treat
each other with respect as shown through
both language and behavior
Bridging
Value
0.18
0.05
0.05
0.08
0.1
0.14
0.16
0.16
0.17
0.19
0.23
0.24
0.24
0.25
0.39
0.21
0.06
0.08
0.1
0.11
0.15
0.2
0.22
0.23
0.27
0.27
169
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
(Continued)
(Continued)
Cluster
Statement
Bridging
Value
Cluster
all campus members supporting and
0.33
participating in the program
a stronger sense of community within class, 0.33
school and locally
0.4
more positive recognition of students and
staff
3. Positive Student
Community
a decrease in behavioral issues and difficult
student behavior
a reduction in disciplinary referrals and
disciplinary action
greater cooperation in the classroom and on
school grounds (vs. competing with or
disrupting others)
fewer cliques and less tension between
groups of students
students going out of their way to find ways
to be kind inside and outside of school
fewer students feeling alienated/alone at
school
more evidence of students helping others
voluntarily (e.g., with school work, pick up
others’ stuff, tying shoes, carry books,
holding the door open, taking someone who’s
injured to nurse's office)
higher, consistent levels of student
engagement
more cross-cultural interactions between
students
more smiles and laughter from students and
adults
4. Intentionality and
Awareness
students speaking well of teachers and the
school (e.g, saying things like “the teachers at
my school are all great/nice/etc.”)
more positive, caring interactions between
students and staff
that a visitor would notice that all students
use their manners
that all members of the school community
(staff, teachers, students, parents) have a
shared vocabulary for kindness and use it in
their everyday life
more creative and collaborative problemsolving
staff, students, and parents taking more time
to recognize kindness in others
deeper and more meaningful conversations
5. School
Operational
Outcomes
students return to that school the following
year at a higher rate
a decrease in physical violence between
students (pushing, kicking, hitting, fighting)
increased attendance
more academic success/improved grades
decreased tardiness
more sharing of resources among students
children acting in a caring way to students
not at their age level (more cross-age
interactions)
better physical health
reduced nurse/health office visits
compassion, patience, and understanding
toward animals
6. Staff Impact
more collaboration between staff members
staff modelling a positive attitude
greater camaraderie and better relationships
among teachers and staff at school
support for student leadership from school
administration and staff
Statement
more staff and teacher energy and
engagement in events (e.g. family events at
school)
increased communication across the school
community (between teachers, staff,
administration, parents and students)
less teacher turnover
fewer teacher absences
staff and teachers speaking with students
instead of at students
0.06
0
Bridging
Value
0.38
0.45
0.48
0.57
0.02
0.03
7. Family and
Community/
Extramural
0.06
0.06
0.07
0.08
0.1
8. Student SocialEmotional Skills
0.1
0.12
0.25
0.17
0.2
0.22
0.23
0.27
0.29
0.36
0.48
0.23
0.24
0.25
0.42
0.46
0.46
0.55
0.56
0.74
0.84
0.4
0.32
0.33
0.34
0.36
0.38
9. Empathy/
Acceptance/
Caring
more positive phone calls and
communications with parents
more interest in community service
more positive sibling relations
increased parental involvement
increased community member involvement
in the school
evidence of the community outside of school
recognizing student positive behavior
parents are able to model kindness to their
child in words & actions
that parents feel school staff are truly
watching out for their child’s well being
students resolving conflict on their own
that students display more self-confidence
more diversity in students’ social groupings
(i.e., those with varied interests/tastes
socializing and “hanging out” together)
students would be able to voice their
opinions without speaking hate or anger
students know that even when they make
mistakes, they still have an opportunity to
improve
students feel like they can be themselves
students are equipped with vocabulary and
other tools to stop unkind behavior and
reduce bystanding
less abusive language
students demonstrating more consideration
for the thoughts and feelings of others
students having a desire to be kind
the ability to recognize, identify, and express
one’s emotions
more apologizing and forgiving
an increase in self-kindness
that students are better able to concentrate
increased hope and a positive outlook
a decrease in the number of students
experiencing problems with stress and
coping
increased creativity and flexibility
students looking forward to going to school
everyday
that students have more fun in class
people feel loved
promoting a “think before you act” mentality
a change in the thought pattern that being
kind is being “soft”
acceptance of differences in culture, abilities,
body, beliefs, etc.
genuine caring and concern for others
increased empathy in staff and students (i.e.,
expressed positive emotions toward others in
need or distress)
students being encouraging and supportive
of each other (of everyone, not just their
friends)
more gratitude
0.86
0.78
0.79
0.82
0.84
0.85
0.87
0.95
1
0.26
0.08
0.13
0.13
0.16
0.19
0.2
0.21
0.22
0.22
0.25
0.27
0.3
0.36
0.36
0.44
0.46
0.52
0.28
0.11
0.23
0.25
0.26
0.27
0.29
0.32
0.34
0.35
0.36
0.18
170
D.M. Kaplan et al. / Evaluation and Program Planning 58 (2016) 160–170
(Continued)
Cluster
Statement
Bridging
Value
10. Emotional Wellbeing and Civility
students showing kindness without
reminders or encouragement
a decrease in negative or disrespectful
behaviors (criticizing/blaming/
condescending/gossiping/making fun/
teasing/interrupting)
students connecting with one another and
making new friends
an increased sense of wellbeing (less
depression and anxiety)
a decrease in students being emotionally hurt
during school
students taking pride in their actions
students are more aware of kind acts
happening around them
increased patience with others
an awareness of the impact one has on others
(both positive and negative)
a wider array of expression of individuality
0.06
0.09
0.1
0.13
0.13
0.14
0.27
0.3
0.31
0.31
Note: The authors can provide the original brainstormed list of
899 statements upon request.
References
Anderson, L., Gwaltny, M., Sundra, D., Brownson, R., Kane, M., Cross, A., Mack, R., &
White, C. (2006). Using concept mapping to develop a logic model for the
Prevention Research Centers Program. Preventing Chronic Disease, 3(1)
Retrieved from http://www.cdc.gov/pcd/issues/2006/jan/05_0153.htm.
Astor, R. A., Meyer, H. A., Benbenishty, R., Marachi, R., & Rosemond, M. (2005). School
safety interventions: best practices and programs. Children & Schools, 27(1), 17–
32. http://dx.doi.org/10.1093/cs/27.1.17.
Ben’s Bells Project (2015). Kind Campus, from https://bensbells.org/kindnesseducation/kind-schools.
Buchanan, K. E., & Bardi, A. (2010). Acts of kindness and acts of novelty affect life
satisfaction. The Journal of Social Psychology, 150(3), 235–237. http://dx.doi.org/
10.1080/00224540903365554.
Cohen, J., McCabe, L., Michelli, N. M., & Pickeral, T. (2009). School climate: research,
policy, practice, and teacher education. The Teachers College Record, 111(1), 180–
213 Retrieved from http://www.tcrecord.org.
Durlak, J. A., Weissberg, R. P., Dymnicki, A. B., Taylor, R. D., & Schellinger, K. B. (2011).
The impact of enhancing students’ social and emotional learning: a metaanalysis of school-based universal interventions. Child Development, 82(1), 405–
432. http://dx.doi.org/10.1111/j.1467-8624.2010.01564.x.
Everitt, B., Landau, S., Leese, M., & Stahl, D. (2011). Cluster analysis (Wiley series in
probability and statistics). Chichester, UK: Wiley.
Flook, L., Goldberg, S. B., Pinger, L., & Davidson, R. J. (2015). Promoting prosocial
behavior and self-regulatory skills in preschool children through a mindfulnessbased kindness curriculum. Developmental Psychology, 51(1), 44. http://dx.doi.
org/10.1037/a0038256.
Fredrickson, B. L., Cohn, M. A., Coffey, K. A., Pek, J., & Finkel, S. M. (2008). Open hearts
build lives: positive emotions, induced through loving-kindness meditation,
build consequential personal resources. Journal of Personality and Social
Psychology, 95(5), 1045. http://dx.doi.org/10.1037/a0013262.
Froh, J. J., Sefick, W. J., & Emmons, R. A. (2008). Counting blessings in early
adolescents: an experimental study of gratitude and subjective well-being.
Journal of School Psychology, 46(2), 213–233. http://dx.doi.org/10.1016/j.
jsp.2007.03.005.
Goetz, J. L., Keltner, D., & Simon-Thomas, E. (2010). Compassion: an evolutionary
analysis and empirical review. Psychological Bulletin, 136(3), 351. http://dx.doi.
org/10.1037/a0018807.
Greenberg, M. T., Weissberg, R. P., O’Brien, M. U., Zins, J. E., Fredericks, L., Resnik, H.,
et al. (2003). Enhancing school-based prevention and youth development
through coordinated social, emotional, and academic learning. American
Psychologist, 58(6–7), 466. http://dx.doi.org/10.1037/0003-066X.58.6-7.466.
Henderson, A. T., & Mapp, K. L. (2002). A new wave of evidence: the impact of school,
family, and community connections on student achievement. Annual synthesis
2002. National Center for Family and Community Connections with Schools
Retrieved from http://www.sedl.org/pubs/catalog/items/fam33.html.
Hofmann, S. G., Grossman, P., & Hinton, D. E. (2011). Loving-kindness and
compassion meditation: potential for psychological interventions. Clinical
Psychology Review, 31(7), 1126–1132. http://dx.doi.org/10.1016/j.
cpr.2011.07.003.
Kane, M., & Trochim, W. M. (2007). Concept mapping for planning and evaluation, Vol.
50, Sage.
Komro, K. A., Flay, B. R., Biglan, A., & Promise Neighborhoods Research Consortium
(2011). Creating nurturing environments: a science-based framework for
promoting child health and development within high-poverty neighborhoods.
Clinical Child and Family Psychology Review, 14(2), 111–134. http://dx.doi.org/
10.1007/s10567-011-0095-2.
Markham, K. M., Mintzes, J. J., & Jones, M. G. (1994). The concept map as a research
and evaluation tool: further evidence of validity. Journal of Research in Science
Teaching, 31(1), 91–101. http://dx.doi.org/10.1002/tea.3660310109.
Miller, J. J., & Jones, B. (2015). Using concept mapping as a planning tool: child
welfare citizen review panels. Evaluation and Program Planning, 53, 99–106.
http://dx.doi.org/10.1016/j.evalprogplan.2015.08.001.
National School Climate Council (2015). School climate and prosocial educational
improvement: essential goals and processes that support student success for all.
Teachers College Record Date Published: 05.05.15 http://www.tcrecord.org ID
Number: 17954, Date Accessed: 04.06.16.
Otake, K., Shimai, S., Tanaka-Matsumi, J., Otsui, K., & Fredrickson, B. L. (2006). Happy
people become happier through kindness: a counting kindnesses intervention.
Journal of Happiness Studies, 7(3), 361–375. http://dx.doi.org/10.1007/s10902005-3650-z.
Pace, T. W., Negi, L. T., Adame, D. D., Cole, S. P., Sivilli, T. I., Brown, T. D., Issa, M. J., &
Raison, C. L. (2009). Effect of compassion meditation on neuroendocrine, innate
immune and behavioral responses to psychosocial stress.
Psychoneuroendocrinology, 34(1), 87–98. http://dx.doi.org/10.1016/j.
psyneuen.2008.08.011.
Post, S. G. (2005). Altruism, happiness, and health: it’s good to be good. International
Journal of Behavioral Medicine, 12(2), 66–77 Retrieved from http://link.springer.
com/journal/12529.
Roeser, R. W., & Pinela, C. (2014). Mindfulness and compassion training in
adolescence: a developmental contemplative science perspective. New
Directions for Youth Development, 2014(142), 9–30. http://dx.doi.org/10.1002/
yd.20094.
Rosas, S. R., & Camphausen, L. C. (2007). The use of concept mapping for scale
development and validation in evaluation. Evaluation and Program Planning, 30
(2), 125–135 Retrieved from www.elsevier.com/locate/evalprogplan.
Suldo, S. M., Savage, J. A., & Mercer, S. H. (2014). Increasing middle school students’
life satisfaction: efficacy of a positive psychology group intervention. Journal of
Happiness Studies, 15(1), 19–42. http://dx.doi.org/10.1007/s10902-013-9414-2.
Sundra, D. L., Anderson, L. A., Gwaltney, M. K., Brownson, M., Kane, M., Cross, A.,
Mack, R., & White, C. (2006). Using concept mapping to develop a logic model
for the prevention research centers program. Preventing Chronic Disease, 3(1) .
Thapa, A., Cohen, J., Guffey, S., & Higgins-D’Alessandro, A. (2013). A review of school
climate research. Review of Educational Research, 83(3), 357–385. http://dx.doi.
org/10.3102/0034654313483907.
The Random Acts of Kindness Foundation (2015). Kindness in the Classroom, from
https://www.randomactsofkindness.org.
Think Kindness (2015). Think Kindness, from http://www.thinkkindness.org/.
Tough, P. (2013). How children succeed. Random House.
Trochim, W. M. (1989). An introduction to concept mapping for planning and
evaluation. Evaluation and Program Planning, 12(1), 1–16. http://dx.doi.org/
10.1016/0149-7189(89)90016-5.
U.S. Department of Health and Human Services (2014). Prevention at School
Retrieved July 1, 2014, from http://www.stopbullying.gov/prevention/atschool/index.html.
Viswanathan, M., Ammerman, A., Eng, E., Garlehner, G., Lohr, K. N., Griffith, D., . . . &
Lux, L. (2004). Community-based participatory research: assessing the
evidence. Summary.
Wallerstein, N., & Duran, B. (2010). Community-based participatory research
contributions to intervention research: the intersection of science and practice
to improve health equity. American Journal of Public Health, 100(S1), S40–S46.
http://dx.doi.org/10.2105/AJPH.;1;2009.184036.
Ward, J. H. Jr. (1963). Hierarchical grouping to optimize an objective function.
Journal of the American Statistical Association, 58(301), 236–244. http://dx.doi.
org/10.1080/01621459.1963.10500845.
Yampolskaya, S., Nesman, T. M., Hernandez, M., & Koch, D. (2004). Using concept
mapping to develop a logic model and articulate a program theory: a case
example. American Journal of Evaluation, 25(2), 191–207. http://dx.doi.org/
10.1177/109821400402500204.