Understanding Student Sense of Peer Belonging at the
University of Wyoming
Carmen Elliott
University of Wyoming
Spring 2020
Elliott 2
Abstract:
Sense of peer belonging has been found to be an essential aspect of university life and academics
in general. My project focused on using statistical analysis to determine what factors contribute
to sense of peer belonging in students attending the University of Wyoming. Using anonymized
data from the 2018 National Survey of Student Engagement graciously provided by the Office of
Institutional Analysis, I was able to identify significant factors that contribute to students’ sense
of peer belonging for 2018 through method of multiple regression analysis. Through a
combination of R and Minitab programs while meeting key statistical assumptions, I was able to
determine which Engagement Indicators and High Impact Practices are most effective at UW.
Post hoc analysis and analysis of variance tests concluded there are several other Engagement
Indicators and High Impact Practices outside of my regression model that affect sense of peer
belonging as well. Information provided by my capstone/thesis project can provide useful insight
for the university in efforts to increase students sense of belonging and therefore their academics.
Introduction:
The University of Wyoming utilizes the annual National Survey of Student Engagement
(NSSE) to better understand the undergraduate student experience. This report summarizes the
findings of the university’s 2018 results in effort to better understand student sense of peer
belonging. The 2018 NSSE results for UW were graciously provided by the Department of
Institutional Analysis and were anonymized before the dataset was received as to not reveal any
students’ personal information. Hundreds of four-year colleges and universities employ the
results of the NSSE survey to recognize how different programs and activities contribute to
Elliott 3
students’ academic and personal development. These different student opportunities and
programs can be quantified at Engagement Indicators and High Impact Practices (“NSSE”).
Using this valuable NSSE dataset from 2018, this report aims to understand how Engagement
Indicators, High Impact Practices, and other factors contribute to a student’s sense of peer
belonging at the University of Wyoming. After a brief introduction, this report will explore the
statistical methods, results, and then conclusions, there will be an appendix with tables and
figures followed by a works cited page.
Acknowledging and understanding factors that contribute to students’ sense of peer
belonging is increasingly important at institutions of higher learning. Several pieces of literature
have found that sense of peer belonging is incredibly important for academics and especially at
universities. In higher education, this can be defined by Terrel Strayhorn, a professor of urban
education and VP for Academic and Student Affairs at LeMoyne-Owen College as
In terms of colleges, sense of belonging refers to students’ perceived social support on campus, a
feeling or sensation of connectedness, and the experience of mattering or feeling cared about,
accepted, respected, valued by, and important to campus community or others on campus such as
faculty, staff, and peers (Terrel 3).
This indicates that sense of peer belonging can be attributed to a multitude of feelings by a
student. It requires that a student not only feels connected to their peers, but that they have a
sense of mattering or being important to their surroundings. Dr. Terrel also describes this as
being a matter of feeling respected by one’s peers.
These feelings of validation that exist as aspects of sense of peer belonging have an
important impact on students’ academics. Professor and social psychologist Roy Baumeister and
professor of psychology and neuroscience Mark Leary suggest that having this feeling of being
Elliott 4
socially connected or socially belonging is a fundamental human motivation (Baumeister and
Leary). A motivating factor for students (whether academic or otherwise) is when they feel they
are connected or that they belong at their university. Furthermore, Gregory Walton and Geoffrey
Cohen members of Department of Psychology at Stanford University and have found that
students lacking sense of peer belonging can undermine their academic success (Walton and
Cohen 82-96). To encourage good grades and success academically, it is important to encourage
students to have a positive sense of belonging with their peers. Finally, it has been found in the
Journal of College Student Development by Joseph Berger that sense of peer belonging can be a
motivation for retention (Berger). When a student feels connected or as though they belong at
their university, they are more likely to be motivated to return the following academic year or
semester. Thus, it is crucial to understand the components that influence sense of belonging at
UW as they are essential in students’ academic success and motivation.
Methods:
To complete this project in understanding student sense of peer belonging at the
University of Wyoming by utilizing the 2018 NSSE survey, several initial procedures had to
occur. As the dataset initially contained sensitive student information, the Department of
Institutional Analysis had to anonymize the results to ensure student privacy. A combination of
both R and Minitab software was employed for computer output and helpful built in tools and
packages. Before any statistical analysis could occur, the data also needed cleaned up. All rows
of data that were incomplete or existed where a student quit the survey prematurely were
removed. Furthermore, rows were removed where students had ‘completed’ the survey or
answered until the end but skipped large sections in the middle. It was necessary to remove all
these rows because not only were they incomplete, but they were not true representations of the
Elliott 5
data and had an influential impact on the results. Only one outlier was removed as the computer
output flagged it in a table of “Fits and Diagnostics for Unusual Observations” as both an
unusually large leverage value and an extreme standardized residual (> 2).
As the NSSE survey is completed on an opt-in basis, it is important to understand the
survey respondents when considering the results. While the NSSE survey aims to collect data on
first year and senior students, this dataset contained students of all class levels. Most students
were first year (total of 300), followed closely by seniors (233), and a fair number of sophomores
(137) and juniors (124), with the remaining six students being unclassified (table 1). The average
age of students was 19.911 with the majority being 23 years old or younger (total of 771) and
only a small portion of students older than 23 (table 2, figure 1). Furthermore, the students that
opted-in to completing the survey came from diverse backgrounds. Several students reported
being international students, first generation students, veterans, athletes, or members of Greek
life (table 3). Table 3 also demonstrates how most of the students reported UW as being their
first attended college. More students identified their gender as women (482) than as men (304)
(table 4). Furthermore, students of all different majors also took this survey, the most being
engineering students, followed by biological sciences, agriculture, and natural resources.
However, most majors had representation to some degree, which can be shown in table 5. The
majority of students who completed this survey reported that they have received mostly grades of
A’s and B’s so far in their academic career, as displayed in table 6 and figure 2. Further
descriptives can be found in table 7.
According to the NSSE, there are several factors that can be used to measure a student’s
sense of peer belonging. Some of the factors are a student’s indicated quality of interaction with
other students, academic advisors, faculty members, student service staff, and other
Elliott 6
administrative staff and offices, all measured on a seven-point Likert scale, a response of 1
indicating poor quality and 7 indicating excellent quality. Furthermore, factors of three questions
on a scale from 1-4 with 1 indicating very little and 4 indicating very much on how much the
institution emphasizes the following: providing support to help students succeed academically,
using learning support services (including tutoring and writing centers), and providing support
for students overall well-being (including through health care, recreation, and counseling).
Another question that was used to determine sense of peer belonging was on how a student
would evaluate their entire educational experience at the institution (1=poor, 4=excellent).
Finally, the last question used to determine a student’s sense of peer belonging was if a student
had the option to start over, would they do so at the same institution they are attending now
(1=definitely no, 4=definitely yes). A score for a student’s sense of peer belonging was found by
adding up the responses to each of these contributing questions and was used as a continuous
response or dependent variable.
Statistical analysis in the form of multiple regression of the student sense of peer
belonging are based on NSSE defined Engagement Indicators (EI’s) and High Impact Practices
(HIP’s) in combination with other miscellaneous survey questions including demographics.
Engagement Indicators represent the multi-dimensional nature of student engagement and can be
broken down into four themes that include Academic Challenge, Learning with Peers,
Experiences with Faculty, and Campus Environment each with multiple subthemes (“NSSE”).
Each subtheme contains multiple questions with most response options on a four-point Likert
scale with a response of 1 indicating the student has never participated in the EI and a 4
indicating they have participated very often throughout the current academic year. For the
questions that had more than four response options, they increased similarly from ‘poor’ to
Elliott 7
‘excellent’. The question responses within each subtheme were added up to create a score for
each student on each Engagement Indicator. The ten categories of scores produced by
Engagement Indicators (subthemes) that were used for analysis were: Higher Order Learning,
Reflective & Integrative Learning, Learning Strategies, Quantitative Reasoning, Collaborative
Learning, Discussions with Diverse Others, Student-Faculty Interaction, Effective Teaching
Practices, Quality of Interactions, and Supportive Environment. Each score for the ten variables
represented continuous independent variables in the multiple regression analysis.
The High Impact Practices established by the NSSE are activities that require
considerable time and effort on the students’ part outside of the classroom. These High Impact
Practices were measured as they were found to be positively associated with academics and
student retention. High Impact Practices could have occurred at any time during students’
academics and require meaningful interactions with others (“NSSE”). Unlike the Engagement
Indicators, High Impact Practices only consist of one question each. Similarly, the six questions
were also scored on a four-point Likert scale with 1=have not decided to, 2=do not plan to do,
3=plan to do, 4=done or in progress. The following are High Impact Practices used and their
‘short’ title: LearnCom or participation in a formal program where groups of students take
multiple classes together, Leader or a formal leadership role in a student organization, Intern or
an internship or field experience (including co-op, student teaching, or clinical placement),
Abroad or a study abroad experience, Research or work with a faculty mentor on a research
project, and Capstone or a culminating senior experience. Each of these six High Impact Practice
variables were used as categorical independent variables in the multiple regression analysis.
In preparation for the multiple regression analysis process, each variable was observed on
its own. Each Engagement Indicator and High Impact Practice was observed as a predictor of
Elliott 8
sense of peer belonging individually through fitted line plots and analysis of variance. This
action was preformed multiples times for each to check for any variables that were not linear
(quadratic, logarithmic, etc.) and if there were any interactions. Figures 3-12 display the Fitted
Line Plots of the ten Engagement Indicators. While I also tested for all options besides linear
terms for these variables, the linear form of each term was the best option. Figure 11 of the Fitted
Line Plot of the Quality of Interactions EI alerted to me to a possible interaction with another
variable. An interaction between Quality of Interactions and Supportive Environment resulted in
being significant in all of the regression models I tried. Figure 13-18 display the Fitted Line Plots
of the High Impact Practices. This tool is designed to demonstrate the relationship between the
response variable (sense of peer belonging) and each continuous independent variable. The plots
of the HIP’s solidified that these variables should be used as categorical variables in the
regression model. When finding the Fitted Line Plots, analysis of variance tests were also
computed for each variable. Using the common standard across the statistics field of α = 0.05 to
indicate a p-value being significant (p-value < 0.05), this initial analysis of variance tests found
consistent results for the EIs. Every Engagement Indicator in this preliminary test resulted in
being a significant predictor of sense of peer belonging.
With a better understanding of the relationship with each EI and HIP, several regression
models were created. Using the Best Subsets tool in Minitab with every EI, HIP, and extra
demographic variables of class level, age, gender identity, and full-time students three contender
models were chosen. These three models were selected as a result of their r-squared and
Mallows’ Cp values. The r-squared values ensured the amount of variance in sense of peer
belonging that was explained by the independent variables. Mallows’ Cp was important in
selecting an unbiased model that estimated the true regression coefficients and future sense of
Elliott 9
peer belonging responses. As Best Subsets only compares models of linear terms, each contender
model was evaluated individually with the added suspected interaction terms. Maintaining the
standard of significant predictors as a result of a p-value < 0.05, the largest p-values in each
model were removed one at a time until all that was left in the contender models were significant
predictors of sense of peer belonging. Coincidently, two of the three contender models broke
down into the same regression model. These two met several assumptions and were ultimately
chosen as the final regression model.
Several precautions were taken to ensure the final regression model was valid including
checks for appropriate residuals, lack of multicollinearity and independence, lack of overfitting,
and outliers. The residuals produced by the final regression model can be seen in figure 19. The
histogram and Normal Probability Plot in figure 19 show the assumption of normality is met.
Homoscedasticity is the idea that there is equal variance across samples, or the error term is same
across all of the independent (predictor) variables and is important to ensure that standard errors
are not biased. This assumption is met by looking at the Versus Fit (Residuals vs Fitted) and
Versus Order (Residuals vs Order) plots that show equal scatter in figure 19. As described earlier
in this report, the only serious outlier was removed and no longer has an influential effect on the
regression model. Multicollinearity occurs when there are high correlations between predictor
variables that violate the assumption of independence. To check this was not an issue, the
Variance Inflation Factor (VIF) for each significant factor was evaluated with a VIF score over
two indicating multicollinearity. Table 8 displays the VIF results for the significant predictors
and shows low VIF scores for all of the predictors except the ones involved in the interaction
term. However, the interaction term is causing the VIF to be higher for these variables rather
than these variables having an issue with multicollinearity, and the assumption is met. Finally, to
Elliott 10
check that the model was not overfit, the model summary and predicted r-squared were used
(table 9). Overfitting occurs when the regression model is too complex and describes the random
error in the data rather than the relationship with the significant predictors. However, since there
is not a large difference between predicted r-squared (a cross-validation method) and r-squared
in table 9, we can conclude this model is not overfit.
Post hoc analysis was also performed on this dataset to understand how all of the
Engagement Indicators and High Impact Practices contribute to student sense of peer belonging
outside of the regression model. Post Hoc analysis was completed through test of one-way
analysis of variance for each variable against sense of peer belonging. In order to do this, the
variables were transformed from continuous to two level categorical variables. This testing was
designed to capture and separate students who generally scored low or high in each EI and HIP
and the effect on their sense of peer belonging. The low/high levels were created for the
Engagement Indicators by taking the average of the highest and lowest possible scores for each
variable as a cut off. The High Impact Practices existed as four level categorical variables but
were also reduced to two level (low/high) with a score of low for those who indicated they had
not decided to/had no plans of participating and high for those who have plans of participating or
are already in the progress of completing on of the HIPs.
Certain assumptions were also required for the post hoc analysis to determine if there was
a significant difference in sense of peer belonging between the levels of each variable including
continuous response variable, multiple levels of the predictors, independent observations, and
equal variance. As no levels were assigned to the dependent variable sense of peer belonging, it
was still the sum of multiple questions and a continuous variable. Because of the separation into
low/high for EI’s and HIP’s, the assumption of multiple levels was also met. Each observation
Elliott 11
was independent as these were different students were taking the survey (each observation) and
one student’s response does not affect another student’s response. Finally, equal variances are
important for both levels of each variable to properly run the conventional one way of analysis of
variance test. To ensure that this assumption was met, Levene’s test was used on every variable
with the consistent standard of p-value < 0.05. For any predictor variables that did not meet this
assumption through Levene’s Test, Welch’s Test of Unequal Variance was used to determine if
there was a significant difference in the means of each level.
Results:
The multiple regression analysis that resulted from the three contender models consisted
of seven predictors terms: Higher-Order Learning, Effective Teaching, Quality of Interactions,
Supportive Environment, interaction of Quality of Interactions and Supportive Environment,
student class level, and whether a student participated in a research project. Table 10 displays the
Analysis of Variance results from this final regression model, including F-values, and significant
P-values. Since research and class level were four-level categorical variables, sixteen regression
equations to predict sense of peer belonging resulted (table 11). For simplicity and to provide a
more interpretable model, research and class were split up as they were for Post Hoc analysis.
Research was recoded so answers of 1 or 2 (undecided or do not plan to participate) were
grouped together and answers of 3 or 4 (plan to or are in the process of participation) were
grouped together. Class levels were recoded so that one group consisted of underclassmen and
one of upperclassmen. Table 11 and 12 show the new Analysis of Variance and equations for the
multiple regression model.
Some of the predictors in the regression model have a larger effect than others on sense
of peer belonging. Figures 20-25 showed the effect of each individual variable on the response
Elliott 12
variable of sense of belonging with low, medium, and high levels of every other variable. Figure
20 focuses on the effect of Higher-Order Learning (HO) and does not show a very steep slope as
HO increases for sense of peer belonging. However, it does show that has HO increases, so does
sense of peer belonging slightly. Similarly, figure 21 shows the effect of Effective Teaching
Practices (ET) with a slight increase in sense of peer belonging as ET increases. The effect of
Quality of Interactions (QI) on student sense of belonging is much more apparent. The slope is
much steeper and student belonging increases significantly as the QI score increases. Figure 23
shows the effect of Supportive Environment (SE) as being one of the more influential terms on
sense of peer belonging, like QI. Figure 23 and 24 for research and class show only a small effect
of each term. Participating in the HIP of research slightly increases sense of belonging, while an
increase in class level actually decreases this feeling for students. Regardless of the amount of
effect each term had on sense of belonging, they coexist to create a significant model to predict
sense of peer belonging for students at UW. Together, these terms (and the interaction term)
explain 94.76% of the variation in student sense of belonging as per the computed r-squared
(table 9).
Even though not all of the Engagement Indicators and High Impact Practices were
included as significant terms in the regression model, this doesn’t mean they do not contribute to
sense of peer belonging, as explored by post hoc analysis. In this section of statistical analysis,
some EIs and HIPs had similar counts for high and low scores while others did not as shown in
figures 26-41. The EIs are figures 26-35 and for Collaborative Learning, Reflective & Integrative
Learning, Effective Teaching Practices, Higher Order Learning, Learning Strategies, Discussions
with Diverse Others, Quality of Interactions, and Supportive Environment there are far more
responses with high scores, indicating that for the majority students in this survey are
Elliott 13
experiencing high levels of these Engagement Indicators. However, for Quantitative Reasoning
and Student-Faculty Interactions, the majority of students are reporting low scores. HIPs are
shown in figure 36-41 and the number of students scoring low/high are more even such as in
leadership role, and research projects. More students are scoring higher for internships and
capstones while more are scoring lower for study abroad and participation in a learning
community.
Regardless of the count of students for each level, the one-way analysis of variance and
post hoc tests indicated that there is a significant difference for most Engagement Indicators and
High Impact Practices. Table 13 and 14 display the results for the EIs and include the Levene F
statistic, Levene p-value, whether the Welch Test for Unequal Variances was used, the Analysis
of Variance F statistic and p-value, and whether there is a significant difference between the
levels. For almost all of the EI’s, there is a significant difference in the average of sense of peer
belonging for the low and high levels with the exception of Quantitative Reasoning, and all of
the variables met the assumption of equal variances. Table 15 displays the same results for the
HIPs and four of the six met the assumption of equal variances. Only one of the HIPs,
participation in study abroad, concluded there was no significant difference in the average of
sense of belonging between the low and high level of study abroad. For the rest of the HIPs, it
was found that there is a significant difference in the average sense of peer belonging for the
high and low levels.
Conclusion:
Student sense of belonging has been shown to be a key component in academic life and
the results of this report aim to understand what contributes to this at the University of
Wyoming. Utilizing the NSSE 2018 survey several Engagement Indicators, High Impact
Elliott 14
Practices, and other demographic questions have been analyzed to understand their collective
and individual impact on student sense of belonging. Through methods of multiple regression
analysis, it has been found that Higher Order Learning, Effective Teaching Practices, Quality of
Interactions, Supportive Environment, an interaction between Quality of Interactions and
Supportive Environment, class level, and participation in a research project work together as
significant predictors of sense of peer belonging. This means that given one knows a student
score for these six terms, their sense of peer belonging can be predicted, and this feeling has been
transformed into a mathematical equation. This model has been tested against several
assumptions and violations and has been found to be reliable and that 94.76% of the variation in
sense of belonging can be explained by these factors. These results are incredibly helpful for the
university to better understand the most significant factors that contribute and can be employed
to boost student sense of peer belonging.
Furthermore, just because all of the Engagement Indicators and High Impact Practices
were not included in the multiple regression model does not mean they do not contribute to sense
of peer belonging as concluded by post hoc analysis. Every EI and HIP besides Quantitative
Reasoning and study abroad have been found to have a significant difference in average sense of
belonging for their levels (low/high). While perhaps there should not be much emphasis put
towards Quantitative Reasoning and study abroad, the other variables should be considered when
the University of Wyoming is evaluating how to increase student sense of belonging. This report
also showed which EIs and HIPs students aren’t participating in as much, giving the institution a
better understanding of which of these variables students value more or have more access to.
In the future, using more annual NSSE surveys could provide further validation for these
results and give the university further insight in how sense of peer belonging among students can
Elliott 15
be increased. Given more data and time, other factors such as student retention could also be
evaluated to understand how EIs and HIPs contribute to a student returning to the institution.
Regardless, this dataset has shown that statistical analysis can provide an equation to predict
sense of peer belonging. Furthermore, there is now more insight into how and which
Engagement Indicators, High Impact Practices, and various demographics contribute to the
undergraduate experience and ultimately students’ sense of peer belonging at the University of
Wyoming.
Elliott 16
Appendix
Table 1
Table 2
Elliott 17
Figure 1
Table 3
Elliott 18
Table 4
Table 5
Table 6
Elliott 19
Figure 2
Variable N N* Mean SE Standard Minimum QI Median Q3 Maximum
Mean Deviation
Internship 796 3 2.9912 0.0309 0.8723 1.0000 3.0000 3.0000 4.0000 4.0000
Leadership 796 3 2.7010 0.0386 1.0902 1.0000 2.0000 3.0000 4.0000 4.0000
LearnCom 797 2 2.4291 0.0372 1.0510 1.0000 2.0000 2.0000 3.0000 4.0000
Abroad 795 4 2.3044 0.0338 0.9525 1.0000 2.0000 2.0000 3.0000 4.0000
Research 796 3 2.3392 0.0366 1.0330 1.0000 1.0000 2.0000 3.0000 4.0000
Capstone 796 3 2.6495 0.0343 0.9682 1.0000 2.0000 3.0000 3.0000 4.0000
Collaborative 799 0 10.952 0.0985 2.784 4.000 9.000 11.000 13.000 16.000
Learning
Elliott 20
Reflective & 799 0 19.340 0.150 4.241 7.000 16.000 19.000 22.000 28.000
Integrative
Learning
Student- 799 0 8.578 0.106 2.985 4.000 7.000 8.000 10.000 16.000
Faculty
Interactions
Higher-Order 799 0 11.310 0.0917 2.591 3.000 9.000 11.000 13.000 16.000
Learning
Effective 799 0 14.284 0.108 3.042 5.000 12.000 14.000 16.000 20.000
Teaching
Practices
Quantitative 799 0 7.3166 0.0843 2.3839 3.0000 6.0000 7.0000 9.0000 12.0000
Reasoning
Discussions 799 0 11.676 0.107 3.019 4.000 10.000 12.000 14.000 16.000
with Diverse
Others
Learning 799 0 8.2653 0.0738 2.0870 3.0000 7.0000 8.0000 10.000 12.0000
Strategies
Quality of 799 0 26.318 0.198 5.584 5.000 23.000 27.000 30.000 45.000
Interactions
Supportive 799 0 21.747 0.179 5.072 6.000 18.000 22.000 25.000 32.000
Environment
Table 7
Elliott 21
Figure 3
Figure 4
Figure 5
Elliott 22
Figure 6
Figure 7
Figure 8
Elliott 23
Figure 9
Figure 10
Figure 11
Elliott 24
Figure 12
Figure 13
Figure 14
Elliott 25
Figure 15
Figure 16
Figure 17
Elliott 26
Figure 18
Figure 19
Elliott 27
Table 8
Model Summary
S R-Squared R-Squared (Adj) R-Squared
(Predicted)
1.71120 94.76% 94.72% 94.56%
Table 9
Elliott 28
Table 10
Table 10
Elliott 29
Table 11
First-Year/Sophomore Junior/Senior
When students are not/do not plan to When students are not/do not plan to
participate in research: participate in research:
PeerBelong = -0.64 + 0.0551*HO + PeerBelong = -0.85 + 0.0551*HO +
0.0940*ET + 1.2383*QI + 05413*SE – 0.0940*ET + 1.2383*QI + 05413*SE –
0.00654*QI*SE 0.00654*QI*SE
When students are/planning to When students are/planning to
participate in research: participate in research:
PeerBelong = -0.49 + 0.0551*HO + PeerBelong = -0.71 + 0.0551*HO +
0.0940*ET + 1.2383*QI + 05413*SE – 0.0940*ET + 1.2383*QI + 05413*SE –
0.00654*QI*SE 0.00654*QI*SE
Table 12
Elliott 30
--- When ET, QI, SE = high --- When ET, QI, SE = medium --- When ET, QI, SE = low
Figure 20
Elliott 31
--- When HO, QI, SE = high --- When HO, QI, SE = medium --- When HO, QI, SE = low
Figure 21
Elliott 32
--- When HO, ET, SE = high --- When HO, ET, SE = medium --- When HO, ET, SE = low
Figure 22
Elliott 33
--- When HO, ET, QI = high --- When HO, ET, QI = medium --- When HO, ET, QI = low
Figure 23
Elliott 34
--- When HO, ET, QI, SE = low
--- When HO, ET, QI, SE = medium
--- When HO, ET, QI, SE = high
Figure 24
Elliott 35
--- When HO, ET, QI, SE = low
--- When HO, ET, QI, SE = medium
--- When HO, ET, QI, SE = high
Figure 25
Elliott 36
Figure 26
Figure 27
Figure 28
Elliott 37
Figure 29
Figure 30
Figure 31
Elliott 38
Figure 32
Figure 33
Figure 34
Elliott 39
Figure 35
Figure 36
Figure 37
Elliott 40
Figure 38
Figure 39
Figure 40
Elliott 41
Figure 41
Table 13
Elliott 42
Table 14
Table 15
Elliott 43
Works Cited
Baumeister, Roy F., and Mark R. Leary. “The Need to Belong: Desire for Interpersonal
Attachments as a Fundamental Human Motivation.” Psychological Bulletin, vol. 117, no.
3, 1995, pp. 497–529., doi:10.1037/0033-2909.117.3.497.
Berger, Joseph B. "Students' Sense of Community in Residence Halls, Social Integration, and
First-Year Persistence." Journal of College Student Development, vol. 38, no. 5, 1997,
pp. 441. ProQuest, http://libproxy.uwyo.edu/login/?url=https://search-proquest-
com.libproxy.uwyo.edu/docview/195174527?accountid=14793.
“NSSE - National Survey of Student Engagement.” NSSE About NSSE,
nsse.indiana.edu/html/about.cfm.
Strayhorn, Terrell L. College Students Sense of Belonging: a Key to Educational Success for All
Students. Routledge, 2019.
Walton, Gregory M., and Geoffrey L. Cohen. “A Question of Belonging: Race, Social Fit, and
Achievement.” Journal of Personality and Social Psychology, vol. 92, no. 1, 2007, pp.
82–96., doi:10.1037/0022-3514.92.1.82.