Making Connections: The Social Networks of First-Year Students Enrolled in Learning Communities

The Social Networks of
College Students in
Learning Communities
Gale Stuart, Doctoral Candidate
UCLA Graduate School of Education & Information Studies
Social Research Methodology Division
Research Analyst
Texas A&M University-Corpus Christi
February 18, 2007
Learning Communities in
Higher Education
• Theoretical Rationale:
– Social learning
– Student involvement
– Peer interactions
– Small groups
– Connected curricula
2
Goals of Learning Communities
• Increase involvement
• Develop a sense of belonging
• Increase awareness of connections
between courses or disciplines
• Enhance critical thinking skills
3
Outcomes of Learning
Communities
•
•
•
•
•
Higher retention
Higher GPAs
Higher satisfaction with college
Higher intellectual skills functioning
Greater gains in social and personal
development
4
Focus of this study:
• Do the social relationships that students
may form in learning communities have
any impact on college outcomes such as
GPA, persistence, or satisfaction with the
college experience?
5
Method: Social Network
Analysis
• A technique that considers social relations,
from families up to nations. Social networks
have been found to play a critical role in
determining the way problems are solved,
how organizations are run, and the degree
to which individuals achieve their goals
• Attribute data versus Relational data
6
Applications of Social Network
Analysis:
•
•
•
•
•
Study the spread of HIV in a prison system
Understand terrorist networks
Identify key players in an organization
Improve the functioning of a project team
Expose financial flows to investigate criminal
behavior
• Map communities of expertise in medical
fields
• Study the adoption of contraceptive
techniques in third world countries
• Explore power relations between countries
7
Network Perspectives
• Ego-centric perspective
• Socio-centric perspective
8
Ego-centric network
A●
Ego ●
B
●
D●
●C
9
Types of Network Measures for
Ego-centric Networks
•
•
•
•
Number sent
Number received
Number reciprocated
Personal Network
Density
• Indegree centrality
• Outdegree centrality
• Betweenness centrality
15
18
23
20
17
21
25
14
6
9
10
22
19
11
3
8
24
12
13
5
2
7
1
4
16
10
Comparison of Ego-centric Measures
18
15
2
#sent
#received
#reciprocate
d
Density
7
4
8
7
14
22
1
4
7
0
.24
.43
0
29.1
7
29.1
7
12.5
Indegree
16.6
7
58.3
3
4.17
5.06
15.7
2
4.82
17
21
3
Outdegree
Betweennes
s
20
23
25
14
22
6
9
10
22
19
11
3
8
24
8
12
13
5
1
2
2
7
4
16
11
Socio-centric Networks
1
24
13
22
3
23
4
11
14
5
3
16
22
7
20
10
8
5
17
2
21
16
19
23
2
8
18
11
21
10
12
4
6
15
20
9
25
1
7
18
17
15
12
9
14
25
24
6
19
13
2
18
18
15
20
23
19
17
21
26
27
21
25
15
14
6
9
10
22
19
11
3
8
24
4
14
12
10
13
17
5
9
6
5
20
22
2
7
1
1
8
11
24
13
25
12
4
3
7
16
16
23
12
Types of Network Measures for
Socio-centric Networks
•
•
•
•
•
•
•
•
Number of links
Average number sent
Density
Percent reciprocated
Number of isolates
Average Path Length
Clustering Coefficient
Centralization
15
18
23
20
17
21
25
14
6
9
10
22
19
11
3
8
24
12
13
5
2
7
1
4
16
13
A Comparison of Friendship
Networks from Two Classes:
Friends Network 1
15
Friends Network 2
18
23
1
20
3
17
21
4
25
5
14
9
22
19
11
2
8
25
18
12
11
21
12
13
15
20
9
3
8
24
22
7
6
10
16
19
23
10
17
14
5
2
7
24
6
1
4
16
13
14
15
1
18
23
3
20
4
17
21
25
5
14
9
22
19
12
13
2
18
8
11
25
15
20
9
11
21
3
8
24
22
7
6
10
16
19
23
12
10
17
14
5
2
7
24
6
1
4
13
16
Network 1
Network 2
Number of Links
122
53
Average Number Sent
4.88
2.12
Density
20.33
8.83
Percent Reciprocated
54.43
47.22
2
6
Average Path Length
8.4
17.14
Clustering Coefficient
41.11
11.29
Centralization
41.30
22.10
Measure
Number of isolates
15
Site of Study
• Texas A&M University-Corpus Christi,
regional university in south Texas
• Enrollment approx. 8,500
• 38% Hispanic; 53% White
• 62% Female
• 65% Full-time
• Fall 2006 first-year class = 1,699
a
16
First Year Learning Community
Program Design
Triad B
Tetrad D
Tetrad G
Sociology
History
Music
Triad L
Political
Science
Tetrad R
Biology
Political
Science
Psychology
Chemistry
English
Composi
-tion
English
Composi
-tion
English
Composi
-tion
English
Composi
-tion
English
Composi
-tion
Freshman
Seminar
Freshman
Seminar
Freshman
Seminar
Freshman
Seminar
Freshman
Seminar
17
Fall 2006 Design
• 7 Triads/Tetrads, approximately 150
students each
• Approximately 6 Cohorts per
Triad/Tetrad, 25 students each meeting
in Freshman Seminar classes
• 52 total cohorts in Freshman Seminar
with a total of 1,243 first-year students
18
The Data
• On-line survey administered in
Freshman Seminar class in late
October 2006
• 70% Response rate
• Confidential not anonymous
• Background variables matched from
university student records
19
Items on the Instrument
• How many hours per week do you study?
• How many hours per week do you work?
(on/off campus)
• Pedagogical measures
• Social Support items
• Quality of Life items
• Attitudes toward Freshman Seminar items
• Sense of belongingness to the institution
item
• Satisfaction with college items
20
Three Network Items:
• Select up to 7 people from your Freshman
Seminar Class who:
– You consider to be friends
– You study with
– You would share a secret with
21
Dependent Variables
• Cumulative GPA in the Fall 2006 Semester
(from matched university records)
• Satisfaction with the College experience (from
survey items)
• Re-enrollment in the spring semester (not yet
available)
22
Preliminary Results – Predicting GPA
Ego-centric Network Measures
(n=571, r-square = .229)
Unstandardized
Coefficient
Standardized
Coefficient
p-value
Constant
1.163
0.000
SAT
0.001
0.233
0.000
Hispanic (compared with Whites)
-0.275
-0.157
0.000
Mother's education level
0.069
0.094
0.017
Number of hours studying
0.162
0.232
0.000
Number of hours socializing
-0.084
-0.181
0.000
Indegree Centrality (friends)
0.011
0.162
0.000
Number of hours working off
campus
-0.035
-0.113
0.003
Quality of Life Factor
0.049
0.083
0.033
Outdegree Centrality (friends)
-0.005
-0.086
0.03323
Preliminary Results – Predicting Mean GPA
Socio-centric Measures
(n=52, r-square = .318)
Unstandardized
Coefficient
Standardized
Coefficient
pvalue
Constant
1.832
0.000
High School Rank percent
0.011
0.423
0.001
Clustering Coefficient-Friends
nets
0.005
0.271
0.032
24
Once we control for High School Rank, the
clustering coefficient becomes important in
predicting average class GPA:
1
24
13
22
3
23
4
11
14
5
3
16
17
2
2
8
8
5
22
7
20
10
25
18
18
15
20
9
1
10
17
4
6
7
11
21
12
21
16
19
23
15
12
14
9
25
6
19
13
Mean GPA = 3.05
Mean GPA = 2.59
N= 24
N= 25
Clustering Coefficient = 34.63
Clustering Coefficient = 11.29 25
24
Preliminary Results – Predicting Mean Global
Satisfaction with the College Experience
Socio-centric Measures
n = 52, r-square = .429
Unstandardized
Coefficient
Constant
Q13-number of isolates
Standardized
Coefficient
1.763
p-value
0.233
- 0.055
- 0.396
0.001
Social Support Factor
0.35
0.353
0.003
No. hours socializing with
friends
0.214
0.236
0.036
26
Interpretation
• Once we control for social support and the
number of hours students spend
socializing with their friends, having at
least one person in their freshman
seminar class who they can trust is
strongly related to higher satisfaction with
their college experience.
27
Early Conclusions
• Aspects of the bonds that students make in their
Freshman Seminar classes do predict academic
achievement
• Analysis of satisfaction with the overall college
experience outcome indicates that having a
close bond with someone in their learning
community class has a positive influence
• Retention to the next term is an important
outcome that is not available for analysis at this
time
28
Research Implications of the
Method
• Social network analysis can be used to
investigate the relationships between
pedagogy and outcomes
• The importance of students’ relationships
with each other in the context of academic
success can be measured
• Can aid in early recognition of situations
that may require intervention
29
Thank you!
Contact Information:
Gale Stuart
Research Analyst
Texas A&M University-Corpus Christi
[email protected]
UCLA Doctoral Candidate
[email protected]
30