Académique Documents
Professionnel Documents
Culture Documents
Methodology
A Twitter Archiving Google Spreadsheet (TAGS; https://tags.hawksey.info/) was used to capture
#Thoughtvectors activity from the Twitter Application Programming Interface (API). As shown in Figure
1, the spreadsheet was monitored and maintained from the first use of the #Thoughtvectors hashtag
until one month after the formal course (UNIV 200) had ended.
Using public data available on Twitter and the course site for UNIV 200: Inquiry and the Craft of
Argument (Thoughtvectors.net), the Thoughtvectors community participants were placed into the four
subgroups defined in Figure 2. A comprehensive quantitative content analysis of #Thoughtvectors tweets
was performed to extract information related to mentions, retweets, and links. This additional
information was added to the spreadsheet to enhance the analysis. Descriptive statistics were generated
through Excel (Microsoft, 2013). Social network analysis was performed through the open source Excel
template, NodeXL, available through the Social Media Research Foundation and CodePlex
(http://nodexl.codeplex.com).
Students
Instructors
Thoughtvectors
Community
Open Participants
Others
,
Figure 2. The Thoughtvectors Community
Results
Research Question 1: What types of microblogging patterns can be studied through publicly available
Twitter data?
Table 1 outlines the descriptive statistics and social network analysis metrics made available by the
#Thoughtvectors TAGS data. While these numbers represent the entire time period collected (February
to August, 2014), the tweet timestamps automatically collected by TAGS allows for the application of
dynamic filters and isolation of specific time periods. Figure 3 provides sociograms of the whole
community and student components of the #Thoughtvectors network. Node colors represent subgroups,
while node size correlates to out-degree centrality..
Table 1. Descriptive Statistics and Social Network Analysis Metrics.
STUDENTS
INSTRUCTORS
OPEN
PARTICIPANTS
OTHERS
WHOLE
COMMUNITY
TOTAL TWEETERS
64
28
480
580
TOTAL PARTICIPANTS
68
28
671
775
UNIQUE TWEETS
453
1212
1255
997
3917
MENTIONS*
288
>1090
>868
>957
>3203
LINKS
135
691
762
619
2207
RETWEETS
RANGE OF ACTIVITY (# OF
TWEETS)
MEAN TWEETS/TWEETER
107
534
180
731
1552
1-93
28-775
1-344
1-69
0-775
203
46
MEDIAN TWEETS/TWEETER
94
18
0-31
0-198
1-75
0-32
0-198
53
13
0-17
0-176
1-83
0-38
0176
44
14
2.16
0-12123
2983-296355 0-88356
0-31248
0-296355
804
56680
263
1286
DESCRIPTIVE STATISTICS
OUT-DEGREE CENTRALITY
RANGE
MEAN OUT-DEGREE
CENTRALITY
BETWEENESS CENTRALITY
RANGE
MEAN BETWEENNESS
CENTRALITY
11239
*Mention data are estimated because Tweets mentioning more than four individuals were clustered in the analysis. In
general, few Tweets mentioned more than four people.
Figure 3. Sociograms of whole community and student #Thoughtvectors Twitter activity from February to August 2014.
Who's Tweeting?
12%
25%
31%
32%
Students
70%
60%
Students
% Tweets
40%
Instructors
30%
20%
Open
Participants
10%
Others
0%
0
Open Participants
Others
50%
Instructors
4+
# Mentions
Figure 5. Mentioning behaviors of #Thoughtvectors subgroups.
Who are Thoughtvectors participants mentioning? Do the subgroups have unique patterns in who they
choose to mention? Are Thoughtvectors participants retweeting, and if so, who are they retweeting?
33%
Students
Instructors
Open Participants
43%
Others
No
Yes
38%
39%
No
27%
44%
56%
70%
76%
57%
62%
61%
60%
86%
73%
56%
44%
30%
24%
40%
14%
Instructors
Open Participants
Others
12%
37%
26%
35%
30%
34%
22%
31%
27%
39%
26%
28%
18%
42%
STUDENTS
33%
17%
12%
13%
INSTRUCTORS
OPEN PARTICIPANTS
4%
OTHERS
12%
WHOLE COMMUNITY
Research Question 3: What aspects of these data might be useful in characterizing individual
performance within Twitter-based learning communities?
Individuals in the #Thoughtvectors twitter community engaged at different levels. Since each subgroup
and individuals within each subgroup had different motivations for participation, this finding was
expected. Moreover, some instructors required students to engage in Twitter-related activity while others
did not. Nevertheless, the available data and data visualizations allowed for differentiation between
students and their activity in ways that may be meaningful for assessments in more formalized
environments. Figure 7 demonstrates how descriptive statistics and social network metrics might be
compiled into a dashboard to document varying levels of student participation in a connected learning
environment.
Student A
#Tweets: 85
#Mentions: 66
#Links: 11
#Retweets: 64
In-Degree Centrality: 5
Out-Degree Centrality: 17
Betweeness Centrality: 3110
Student B
#Tweets: 24
#Mentions: 5
#Links: 12
#Retweets: 6
In-Degree Centrality: 5
Out-Degree Centrality: 9
Betweeness Centrality: 1108
Student C
#Tweets: 0
#Mentions: 0
#Links:0
#Retweets: 0
In-Degree Centrality: 1
Out-Degree Centrality: 0
Betweeness Centrality: 0
Conclusions
This study was an initial step towards understanding how we might document connected learning
in formal higher education settings. It is not our intention to suggest that learning can be documented
wholly through basic descriptive statistics, but rather that these data offer a potentially interesting and
scalable avenue for further research.
Conclusion 1. It is possible to use TAGS data to develop dashboards for the assessment of student
microblogging behaviors.
The data collected from TAGS are appropriate for descriptive statistics and data visualizations consistent
with formative and summative assessments of group and individual student performance.
Conclusion 2. It is better to tailor data archiving spreadsheets to the pedagogical objectives of the
microblogging activity.
In this case researchers hoped to capture the incidence of student connection-making because of the
pedagogical importance placed on making connections across space, time, and disciplines in connected
learning environments. Information regarding mentions, links, and retweets was available through
TAGS, but it had to be mined from text cells in a time-consuming content analysis. Such an analysis
would not be feasible in an authentic, real-time educational context of similar size. Ideally archiving
spreadsheets could be designed to mine much of this information automatically.
Conclusion 3. If assessment strategies around connection-making are to be implemented, instructors
must send clear messages to their students regarding the usefulness of microblogging as a connectionmaking tool.
Individuals in the #Thoughtvectors community tweeted in very different ways. Heterogeneity in student
use might have been related to the lack of consistent and formalized pedagogical messaging around the
use of Twitter for class-related activity. If microblogging is to be seen as a powerful learning tool that
affords (1) connection of ideas across space, time, and disciplines; (2) strategic navigation through
distributed discourse; and (3) signal amplification; and if students are to practice using it as such, explicit
pedagogical messaging around these goals should be in place before the students ability to perform
such tasks are assessed.
Limitations. The current study is limited by the fact that the research was performed only with publicly
available data; identifications of individuals within each subgroup could not be confirmed with official
registration lists, nor could any of the metrics be compared with more traditional assessments such as
final grades. However, given that the primary purpose of the study was exploratory and focused on how
best to design data collection spreadsheets for future research, these limitations were considered
acceptable.
Future Research. Next steps in the research agenda include
Gaining a better understanding of how students and faculty use social media tools in connected
learning environments.
Building evidence to suggest certain digital practices or patterns of practice (e.g. mentioning,
linking) promote the sort of digital engagement, digital literacies, and connection-making valued
in connected learning environments.
Developing archiving spreadsheets specific to the purpose of documenting student learning in
connected learning environments.
Developing meaningful, scalable, and flexible assessment systems for connected learning that
may incorporate the data from digital artifact archiving spreadsheets similar to or based on
TAGS.