You are on page 1of 10

The Role of Achievement Goal Orientations When

Studying Effect of Learning Analytics Visualizations


Sanam Shirazi Beheshitha, Marek Hatala Dragan Gašević, Srećko Joksimović
School of Interactive Arts and Technology Schools of Education and Informatics
Simon Fraser University University of Edinburgh
Surrey, Canada Edinburgh, UK
sshirazi,mhatala@sfu.ca dragan.gasevic,s.joksimovic@ed.ac.uk

ABSTRACT tion, sense-making and impact on students’ learning [35] existing


When designing learning analytics tools for use by learners we learning analytics visualizations and dashboards present trace
have an opportunity to provide tools that consider a particular data to students about their interaction with the learning environ-
learner’s situation and the learner herself. To afford actual impact ment such as use of resources [17, 24, 30], time spent on activities
on learning, such tools have to be informed by theories of educa- [2, 22], generated artifacts [2], or social interactions with others
tion. Particularly, educational research shows that individual dif- [7, 21, 27]. While some of the existing dashboards are targeted at
ferences play a significant role in explaining students’ learning providing general information that can facilitate awareness and
process. However, limited empirical research in learning analytics monitoring of learning activities , others go further and directly
has investigated the role of theoretical constructs, such as motiva- guide students to take actions to control their learning [5][8].
tional factors, that are underlying the observed differences be- In terms of evaluation, a line of existing research have focused on
tween individuals. In this work, we conducted a field experiment usability and students’ perceived usefulness [17, 30]. Some other
to examine the effect of three designed learning analytics visuali- research studies have also been conducted that indicate positive
zations on students’ participation in online discussions in authen- influence of learning analytics dashboards and visualizations on
tic course settings. Using hierarchical linear mixed models, our improving engagement [28], academic performance [2], test re-
results revealed that effects of visualizations on the quantity and sults and assessments [4, 22], and retention rates [2] of the overall
quality of messages posted by students with differences in population of students. A large number of studies that focus on
achievement goal orientations could either be positive or negative. assessing leaning impact have been carried out in limited lab set-
Our findings highlight the methodological importance of consid- tings [4, 22, 28]. There are also few studies that have been inves-
ering individual differences and pose important implications for tigated in course settings at large sale, such as [2].
future design and research of learning analytics visualizations.
As research on learning analytics dashboards and visualizations is
expanding, further empirical research is needed to understand the
Categories and Subject Descriptors varying impact of information selected to be presented through
K.3.1[Computers and Education] Distance Learning visualizations on different aspects of an individual student’s learn-
ing process and outcome. Research on educational psychology
General Terms shows that individuals differ in their readiness to profit from a
Human Factors, Measurement. particular treatment in a particular context [34]. This indicates the
possible varying effect of a treatment for individual students. In
Keywords our study we focus on theoretical constructs of so called aptitudes
that can shed light up on the observed differences between indi-
Learning Analytics, Visualizations, Dashboards, Achievement
viduals in learning context (e.g., motivational constructs, epistem-
Goal Orientation, Online Discussions
ic beliefs, approaches to learning, and attitudes) [36]. Our aim is
to investigate the effect of learning analytics visualizations learn-
1. INTRODUCTION ing behavior by taking into account individual differences.
Recent advancement in technology-enhanced learning offers a
In so doing, we conducted a field experiment to examine the ef-
powerful and yet challenging opportunity to observe learning
fects of different information presented through learning analytics
analytics from students’ perspective. Learning analytics tools,
visualizations on students’ learning behavior while controlling for
when put in the hands of students, can support their learning, par- their individual differences. In this work we focused on a motiva-
ticularly at higher education [23]. One way of presenting learning tional construct called achievement goal orientation [12].
analytics to students is through visualizations and dashboards
[35]. With the intent to offer opportunities for awareness, reflec- Achievement goal orientation (AGO) is a well-established moti-
vational construct describing "the purpose of engagement in an
Permission to make digital or hard copies of all or part of this work for
achievement behavior" [12]. In the early definitions, two main
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies goal orientations were identified: i) mastery goal, which was con-
bear this notice and the full citation on the first page. Copyrights for com- ceptualized in terms of development of task competence; and ii)
ponents of this work owned by others than ACM must be honored. Ab- performance goal, which was conceived as the illustration of per-
stracting with credit is permitted. To copy otherwise, or republish, to post formance competence [26]. In terms of valence, these achieve-
on servers or to redistribute to lists, requires prior specific permission ment goals were further distinguished by approaching success and
and/or a fee. Request permissions from Permissions@acm.org. avoiding failure, e.g., being able to accomplish a task or avoiding
LAK '16, April 25-29, 2016, Edinburgh, United Kingdom failing the test, respectively [11]. In recent AGO models, compe-
© 2016 ACM. ISBN 978-1-4503-4190-5/16/04…$15.00
DOI: http://dx.doi.org/10.1145/2883851.2883904

54
tence has been re-identified as the standard used in evaluation of
how well one is doing [13]. Task-based goals use absolute stand-
ards and define competence based on doing well or poorly relative
to the requirements of the task. Self-based and other-based goals
adopt intrapersonal and interpersonal standards, respectively, and
define competence in terms of doing well or poorly with respect
to how one has done before or can potentially do in the future, or
in comparison to others [13].
To discover possible association between individual differences
Figure 1: The design of the Class Average visualization
and information presented we designed three learning analytics
visualizations where each showed particular information about an
aspect of students’ participation in online discussions in a univer-
sity-level blended course. The visualizations were selected in a
way to potentially address students with particular goal orienta-
tions. We chose to focus on asynchronous online discussions, as
these are commonly exploited to support collaborative learning
[25] and can be seen as an environment in which students can
interact to build both collective and individual understanding
through conversation with their peers [20]. Critically, the level
and quality of students’ participation is largely influenced by stu-
dents’ agency [37]. Learning analytics in the form of reports and
visualizations have been suggested to be supportive of participa-
tion and productive engagement in online discussions for the pop-
ulation of students as a whole [41]. However, more attention to- Figure 2: The design of Top Contributors visualization
wards the impact of what is presented to students with differences
in achievement goals is warranted [40]. Our results not only sub-
stantiate this assumption, they also have significant implication
for broader learning analytics research.

2. METHOD
2.1 Study Design and Research Questions
To study the effects of different information presented through
visualization on the posting behavior of students with individual
differences, we conducted a field experiment in an authentic
blended course setting. Students participated in an online group
discussion activity on a topic related to the course content. Each
student was randomly assigned to an experimental condition in
which they had access to one of the three visualizations informing
them on how they are performing in the group discussion activity. Figure 3: The design of the Quality visualization
Students’ goal orientations were measured through a self-reported
instrument. view, and hence, they could judge other group members perfor-
We defined our research questions as follows: mance directly.

RQ1: Is there an association between visualization type and the The Class Average visualization allows students to compare their
quantity of students’ posts when controlled for their self-reported posting performance with the average number of messages posted
achievement goal orientations? by the rest of the class (Figure 1). Comparison of the students
with the class average has been the most widely used approach
RQ2: Is there an association between visualization type and the when offering learning analytics dashboards and visualiza-
quality of students’ posts when controlled for their self-reported tions [6]. Students who have a stronger inclination towards per-
achievement goal orientations? formance orientation may find this visualization beneficial, with a
caveat that its effect on students’ participation and learning was
2.2 Learning Analytics Visualizations not always positive [6, 41]. We included this visualization mainly
The choice of learning analytics visualizations was guided by the because of its prevalence in deployed systems.
main goal of this study, i.e., to establish the association between
type of information visualized and its effect on students’ behavior. The Top Contributors visualization shows the count of posted
Secondly, we expected that the effect the visualizations will vary messages by the student in comparison to the top contributors in
with goals students are pursuing. The three visualizations selected the class. Top contributors are the top 5+ individuals in the class
aimed to potentially align with different types of motivations un- who have had the highest number of messages posted (Figure 2).
derlying students’ goals. Each visualization also considered which Not only students are able to see performance of the top contribu-
norm students will be evaluating themselves against, which varies tors, but the visualization increases their individual recognition by
for different goal orientations. When cumulative performance is showing their names and profile pictures. The norm shown by this
shown, we used class average values (up to 200 students), rather visualization is that of the best performing students and we expect
than the group average of 4-11 students. The reason is that our that it will positively motivate students with other-approach ten-
LMS system students always see all contributions in a single dency and somewhat those with self-approach tendency, while it

55
will be disturbing to students with avoidance valence for the same Table 1. Number of students assigned to each visualization
goals.
Condition C1SP C1SM C2SP C3SP
While both prior visualizations focused on count of messages, the
Quality visualization focuses the content of posted messages. It (Visualization) D1 D2 D3 D4 D5 D6 All
represents how many of the key concepts student has covered Class Average 25 11 8 N/A 7 7 58
within his/her posted messages and how well he/she has integrat-
Top Contributors 23 12 7 5 4 4 55
ed those with logically related ideas. Students can compare quality
of their message with that of the rest of the class, hence we expect Quality 13 17 5 11 5 5 56
that it will have positive effect on students pursuing mastery, both
those with task-approach or self-approach tendency. The key con-
cepts for each discussion topic were previously identified by the 2.5 Data Collection a Measurement
course instructor. The visualization (Figure 3) showed the quality The time stamped log data of students’ interaction with the visual-
for each key concept as a color-coded square. The color was de- ization was recorded. Also, the messages posted by each student
termined by computing the Latent Semantic Analysis (LSA), a and the group structures were captured within the Learning Man-
natural language processing technique for measuring the coher- agement System. The count of posted messages and count of vis-
ence of the text1, at the sentence level [14]. ualization views was computed for each student per discussion
across different courses.
More complex dashboards with several metrics may address dif-
ferent achievement goal orientations at the same time. For the The 3×2 AGQ instrument was used to investigate students’
purpose of this study our selected visualizations included only one Achievement Goal Orientations [13]. The instrument consists of
metric of student performance and the same metric was shown at 18 items, grouped into 6 scales corresponding to achievement
the class level. We were explicitly not concerned with providing a goals (task-approach, task-avoidance, self-approach, self-
comprehensive overview of students’ performance in a single avoidance, other-approach, and other-avoidance, whereby self and
cumulative view. task represent mastery goals and other represents performance
goals). The responses were recorded on a Likert-type scale, from
2.3 Online Group Discussion Activity 1 (not at all true of me) to 7 (very true of me). The total scores on
Design and facilitation of discussions in all participating courses every 3 items corresponding to a scale were used as the overall
followed guidelines suggested in collaborative learning literature measure on that AGO scale.
[29, 43]. The students were split into several groups of 4-11
members and were asked to participate over a period of 7-14 days. 2.6 Data Analysis
All of the groups in a particular discussion were given the same 2.6.1 Coh-Metrix Analyses
open-ended question related to the course content and were ex- Discourse analysis can be used to help identify effectiveness of
pected to engage in the discussion by exploring different aspects discussions and quality of argumentation in collaborative learning
of the question itself, proposing different ideas to address them, environments [33]. We used Coh-Metrix, a computational linguis-
selecting some ideas and finally deciding on one answer as a tics facility that provides various measures of text characteristics
group and justifying it with a clear rationale. Engagement in the (e.g., text coherence, linguistic complexity, characteristics of
discussion was mandatory and was considered a graded compo- words and readability scores), to analyze content of the messages
nent of the course (5% of final grade per discussion task). Mark- posted by students [18]. We adopted the five latent components
ing rubric was also provided which thoroughly explained the that in a recent study on a corpus of around 37,520 texts explained
marking criteria in terms of quantity and content of individual over 50% of the variability among texts [18]:
posts, as well as, tone and mechanics, collaboration between
group members, and quality of arguments in the final response. • Narrativity: the degree to which the text is a narrative and con-
veys a story. On the opposite end of the spectrum are exposito-
Each group had access to their private discussion space inside the ry texts.
Canvas Learning Management System used in the course. This
space was composed of the discussion activity description, link to • Deep Cohesion: the degree to which the ideas in the text are
visualization, and of the discussion thread itself (can be viewed at cohesively connected at a mental and conceptual level.
http://at.sfu.ca/gCXQNW (permalink)). Once students clicked on • Referential Cohesion: reflects the degree to which explicit
the visualization link, a new tab would open up and display the words and ideas in the text overlap with each other.
assigned visualization to the student.
• Syntactic Simplicity: reflects the degree to which sentences
2.4 Courses, Discussions and Participants have lower number of words and use more simple and familiar
The study was run in the Spring and Summer 2015 terms across structures rather than dense sentences and high frequency of
four different blended course offerings at the second and third embedded phrases.
levels in a multidisciplinary Design, Media Arts and Technology • Word Concreteness: the degree to which the text includes
program in a Canadian post-secondary institution. Table 1 shows words that are concrete and induce mental images in contrast to
the number of students (i.e., study participants) assigned to each abstract words.
visualization per discussion (D1- D6) across courses (C1SP,
In this study, the above discourse features were analyzed at the
C1SM, C2SP and C3SP).
level of each message using Co-Metrix. Then for each single stu-
dent we computed the average measures of all the messages per
discussion. Only the messages that included at least one of the key
1
concepts related to discussion topic (as identified by the instruc-
Coherence has been described as “the unifying element of good writing” tor) were included in this analysis. These are the messages gauged
and hence it can be used in a way to measure quality of text. to have traces of higher level of knowledge construction [19].
(http://www.elc.polyu.edu.hk/elsc/material/Writing/coherenc.htm)

56
2.6.2 Statistical Analysis Table 3. Inferential Statistic for Model fit assessment - RQ1
Due to the nested structure of the data and the crossed variables in χ2 Df R2 AIC
our analysis we identified hierarchical linear mixed models to be a
Null Model 0.70 251.21
suitable method [31]. The primary analyses for RQ1 focused on
Fixed Model 40.60** 20 0.91 250.61
association between the visualization type for those students who
used them to monitor their participation in discussion and the χ2 values show the differences between the model in the current
quantity of posted messages, after controlling for self-reported row and the model in the previous row.
AGOs. Hence, we identified student’s counts of posts as the de- Significance codes: *** p<0.001 , ** p<0.01 , *p<0.05
pendent variable. Table 4. Analysis of the fixed effects for the model - RQ1
The subsequent analysis was centered around RQ2 to find the Variable β SE 95% CI
association between the visualization type and the quality of post- Lower Upper
ed messages, measured through the discourse features, after con- Intercept (Class Average)** 0.478 0.174 0.130 0.826
trolling for the self-reported AGOs. Therefore, we identified five Viz (Top Contributors) 0.156 0.274 -0.705 0.392
dependent variables: Narrativity, Deep Cohesion, Referential Viz (Quality)* -0.511 0.228 -0.967 -0.054
Cohesion, Syntactic Simplicity and Word Concreteness. The in- TaskAp 0.002 0.185 -0.369 0.373
dependent variables in all models for both RQ1 and RQ2 were the TaskAv -0.067 0.346 -0.759 0.625
visualization type assigned to the student (i.e., Class Average, Top SelfAp 0.023 0.251 -0.480 0.525
Contributors, or Quality) and the covariates were the scores on six SelfAv 0.402 0.510 -0.619 1.423
AGO scales (i.e., task-approach, task-avoidance, self-approach, OtherAp*** -0.986 0.357 -1.700 -0.274
self-avoidance, other-approach, and other-avoidance). OtherAv* 0.707 0.480 -0.254 1.668
Six different linear mixed models were constructed, one for the
dependent variable in RQ1 (count of posts) and one for each of Viz (Top Contributors)*TaskAp -0.641 0.394 -1.428 0.146
the five dependent variables in RQ2 (Narrativity, Deep Cohesion, Viz (Top Contributors)*TaskAv -0.151 0.565 -1.281 0.980
Referential Cohesion, Syntactic Simplicity and Word Concrete- Viz (Top Contributors)* SelfAp. 1.076 0.628 -0.181 2.333
ness). The choice of the best fitting model for each dependent Viz (Top Contributors)* SelfAv -0.866 0.769 -2.404 0.671
variable was finalized after two steps of the model construction: Viz(Top Contributors)* OtherAp* 1.047 0.490 0.067 2.026
1) null model with student within a course as the only random Viz (Top Contributors)* OtherAv -0.724 0.694 -2.112 0.665
effect2 2) fixed model with the random effects introduced in the
null model and the interaction between visualization type and six Viz (Quality)*TaskAp -0.180 0.222 -0.623 0.263
AGO scale scores as the fixed effect. Viz (Quality)*TaskAv -0.016 0.391 -0.799 0.767
A comparison between the null random-effects only model and Viz (Quality)* SelfAp 0.024 0.370 -0.716 0.765
fixed-effects model allows us to determine whether the model that Viz (Quality)* SelfAv -0.206 0.589 -1.384 0.972
considers visualization types estimates quantity and quality of Viz (Quality)* OtherAp** 1.199 0.430 0.340 2.059
Viz (Quality)* OtherAv. -1.076 0.547 -2.169 0.018
posts when controlled for the self-reported AGO score better the
random effects model. Akaike Information Criterion (AIC) and Significance code: *** p<0.001 , ** p<0.01 , *p<0.05, . p<0.1
the likelihood ratio test were used to decide the best fitting model (marginal)
[15]. Primarily, the model with lower AIC was suggested to have All variables are scaled
a better fit. We used the likelihood ratio test to confirm AIC re-
sult. We also calculated an estimate of effect size (R2) for each 3.1 RQ1
model, which reveals the variance explained by the model [42]. According to the AIC and the likelihood ratio test the fixed model
that included the interaction between learning analytics visualiza-
3. RESULTS tion and AGO scales yielded a significantly better fit than the null
Since the students’ use of learning analytics visualizations was model (Table 3). The linear mixed-effect analysis uncovered a
voluntary, not all chose to engage with the visualizations. The significant interaction effect between the learning analytics visual-
subset of students who engaged with the visualization more than ization and other-approach scale scores, (F(2,79.11)=4.12,
once are considered the actual users of the visualization and the p<0.05) (Further details in Table 4).
focus of our analysis in RQ1 and RQ2 (Table 2).
Further investigation on interaction effect between learning ana-
Table 2. Count of visualization views for students who used lytics visualization and other-approach shows a marginal signifi-
visualizations cant difference in the count of posts between the users of the Class
Visualization N Median (25%,75%) Average visualization and the users of the Top Contributor visual-
ization (z=2.14, p<0.1) and significant difference between the
Class Average 38 7.00 (4.00, 9.00)
users of the Class Average visualization and the users of the Qual-
Top Contributors 22 6.50 (3.25, 15.50)
ity visualization (z=2.79, p<0.05). The other-approach scale is
Quality 38 5.00 (3.00, 10.00)
positively associated with counts of posts for the users of the Top
Contributors and Quality visualization, while the other-approach
scale is negatively associated with counts of posted messages for
2
In model construction, discussion groups were considered an additional the users of the Class Average visualizations.
levels in the nested structure of the random effect. Also, the total
activity count of students was accounted as another random effect. In all 3.2 RQ2
models our findings showed that considering either or both of these For all of the five Coh-Metrix principal components, fixed effect
variables did not yield a better model. models that included interaction between learning analytics visu-
alization and the six AGO scales resulted with better overall

57
goodness of fit measures (AIC, likelihood ratio test and R2) than analytics visualization and other-avoidance (F(2,81.14)=3.65,
null models (Table 5). As an example of analysis performed, Ta- p<0.05). Also, the interaction between learning analytics visuali-
ble 6 shows analysis of the fixed models for Deep Cohesion, simi- zation and task-avoidance was marginally significant,
lar tables for remaining components can be viewed at F(2,81.62)=2.94, p<0.1 (Further details in Table 6).
http://at.sfu.ca/shKRxa (permalink). In the context of online dis- Further investigation on the interaction effect between learning
cussions we believe Deep Cohesion should be given higher analytics visualization and task-approach shows a significant
weight given the importance of text cohesion for knowledge con- difference in the deep cohesion scores between the users of the
struction as emphasized by cognitive scientists [10]. Class Average visualization compared to the deep cohesions
scores of the users of the Top Contributors visualization (z=4.33,
3.3 Narrativity p<0.001), and between deep cohesions scores of the users of the
The linear mixed model for narrativity further revealed significant
Top Contributors and Quality visualizations (z=-3.99, p<0.001).
interaction effect between learning analytics visualization and
The positive association between task-approach scales and deep
task-approach (F(2,81.52)=9.27, p<0.001), learning analytics
cohesion was largest for the Top Contributors, while much small-
visualization and task-avoidance (F(2,81.02)=5.26, p<0.01),
er positive association was found for the Quality visualization
learning analytics visualization and self-approach
followed by the negative association for the users of the Class
(F(2,80.66)=3.64, p<0.05), and learning analytics visualization
Average visualization.
and self-avoidance (F(2,81.36)=4.08, p<0.05). Also, the interac-
tion between learning analytics visualization and other-avoidance Further exploration on the interaction effect between learning
is marginally significant, F(2,80.62)=2.99, p<0.1. analytics visualization and self-avoidance exhibited a significant
difference in the scores of deep cohesion between the Class Aver-
Further investigation on interaction effect between learning ana-
age visualization users and Quality visualization users (z=2.47,
lytics visualization and task-approach shows a significant differ-
p<0.05), and marginally significant difference between Top Con-
ence between the scores of narrativity of the users of the Top Con-
tributor visualization users and Quality visualization users
tributors visualization compared to the scores of narrativity of the
(z=2.21, p<0.1). Self-avoidance scale scores were positively asso-
users of the Quality visualization (z=-3.22, p<0.01) and between
ciated with deep cohesion scores for the users of the Quality visu-
the scores of narrativity of the Class Average visualization and
alization, whereas this self-avoidance scale scores were negatively
those of the users of Top Contributors (z=4.31, p<0.001). The
associated with the deep cohesion scores for the messages posted
positive association between the task-approach scale and narrativ-
by the Top Contributors and Class Average visualization users.
ity scores was largest for Top Contributors, followed by the posi-
tive association for the users of the Quality visualization, while a Further investigation on interaction effect between learning ana-
negative association was found for the users of the Class Average lytics visualization and other-avoidance showed a significant
visualization. difference in the scores of deep cohesion between the users of
Class Average visualization with those of Quality visualization
Probing the interaction effect between learning analytics visuali-
(z=-2.58, p<0.05). The association between other-avoidance scale
zation and task-avoidance shows a significant difference in narra-
scores and deep cohesion scores was negative for the users of the
tivity scores between the users of Top Contributors compared to
Quality visualization, while the association was positive for the
the users of Quality (z=-3.00, p<0.01). The effect of task-
users of both Top Contributors and Class Average visualizations.
avoidance was negative on narrativity for the users of the Quality
visualization, while this effect was positive on the narrativity 3.5 Syntactic Simplicity
scores of the users of the other two visualizations. Analysis for syntactic simplicity principal component revealed
Further exploration on the interaction effect between learning significant interaction effect between learning analytics visualiza-
analytics visualization and self-approach exhibited a significant tion and self-avoidance (F(2,80.99)=3.46, p<0.05).
difference in the scores of narrativity between the users of Class Further exploration on the interaction effect between learning
Average visualization and those of the Quality visualization (z=- analytics visualization and self-avoidance exhibited a significant
2.32, p<0.05). Self-approach scale scores were positively associ- difference in the scores of deep cohesion between the Top Con-
ated with narrativity scores for the users of the Class Average tributors visualization users and the Quality visualization users
visualization, whereas this self-avoidance scale scores were nega- (z=2.56, p<0.05). Self-avoidance scale scores were positively
tively associated with the narrativity scores for the messages post- associated with syntactic simplicity scores for the users of the
ed by the users of Top Contributors and Quality visualization. Quality visualization, whereas this self-avoidance scale scores
Finally, exploring the interaction effect between learning analytics were negatively associated with the syntactic simplicity for the
visualization and self-avoidance goal-orientation exhibits signifi- messages posted by the Top Contributors and Class Average visu-
cant difference in the scores of narrativity between the users of the alization users.
Top Contributors and Quality visualizations (z=2.61, p<0.05).
Self-avoidance scale scores were positively associated with narra- 3.6 Referential Cohesion
tivity scores for the users of the Quality visualization, whereas Analysis of mixed models for referential cohesion revealed a sig-
these self-avoidance scores were negatively associated with both nificant interaction effect between learning analytics visualization
narrativity scores of the users of both Top Contributors and Class and task-approach scales (F(2,78.05)=7.44, p<0.01), learning
Average visualizations. analytics visualization and self-avoidance (F(2,75.33)=3.93,
p<0.05), and learning analytics visualization and other-approach
3.4 Deep Cohesion (F(2,73.33)=3.61,p<0.05).
The deep cohesion model revealed significant interaction effects Further investigation of the interaction effect between learning
between learning analytics visualization and task-approach analytics visualization and task-approach showed a significant
(F(2,82.38)=10.02, p<0.001), learning analytics visualization and difference in the scores of referential cohesion between the users
self-avoidance scales (F(2,82.28)=4.36, p<0.05), and learning of the Top Contributor visualization and the users of the Quality

58
visualization (z=-3.066, p<0.01), and between the Class Average Table 5. Inferential Statistic for Model fit assessment - RQ2
visualizations users and Top Contributor users (z=3.86 ,p<0.001).
Narrativity
The positive association between task-approach scales and refer-
ential cohesion was largest for the Top Contributors, while much χ2 df R2 AIC
smaller positive association was found for the Quality visualiza- Null Model 0.51 251.70
tion followed by the negative association for the users of the Class Fixed Model 74.42*** 20 0.68 217.28
Average visualization. Deep Cohesion
Probing the interaction effect between learning analytics visuali- χ2 df R2 AIC
zation and self-avoidance shows a significant difference in refer- Null Model 0.36 246.40
ential cohesion scores between the users of Top Contributors Fixed Model 56.64*** 20 0.44 229.76
compared to the users of Quality (z=2.77, p<0.05) and marginally Syntactic Simplicity
significant difference between the users of Class Average visuali- χ2 df R2 AIC
zation compared to the users of Top Contributors visualization Null Model 0.14 248.32
(z=-2.22, 0<0.1). Self-avoidance scale scores were positively as- Fixed Model 44.10** 20 0.28 244.22
sociated with referential scores for the users of the Quality visual- Referential Cohesion
ization and Class Average, whereas this self-avoidance scale χ2 df R2 AIC
scores were negatively associated with the referential cohesions Null Model 0.69 245.88
for the messages posted by the users of Top Contributors visuali- Fixed Model 57.47*** 20 0.77 228.42
zation.
Word Concreteness
Further exploration on the interaction effect between learning χ2 df R2 AIC
analytics visualization and task-approach exhibited a significant Null Model 0.44 226.46
difference in the scores of referential cohesion between the users Fixed Model 52.99*** 20 0.68 213.47
of Top Contributors visualization and those of Quality visualiza-
χ2 values show the differences between the model in the current
tion (z=2.68, p<0.5). The other-approach scale scores were posi-
row and the model in the previous row.
tively associated with referential cohesion scores for the users of
Significance codes: *** p<0.001 , ** p<0.01 , *p<0.05
the Quality visualization, whereas this self-avoidance scale scores
were negatively associated with the deep cohesion scores for the Table 6. Analysis of the fixed effects for the model - RQ2
messages posted by the Top Contributors and Class Average visu- Deep Cohesion
alization users.
Variable β SE 95% CI
3.7 Word Concreteness Lower Upper
Further analysis of the models for word concreteness uncovered a Intercept (Class Average) 0.252 0.179 -0.105 0.609
significant interaction between learning analytics visualization Viz (Top Contributors) 0.195 0.235 -0.276 0.665
and task-approach (F(2,80.24)=4.41, p<0.05), learning analytics Viz (Quality) -0.165 0.198 -0.561 0.231
visualization and task-avoidance (F(2,80.17)=4.00, p<0.05), TaskAp -0.019 0.160 0.339 0.301
learning analytics visualization and other-approach TaskAv* 0.681 0.308 0.065 1.296
(F(2,80.57)=3.68, p<0.05), and learning analytics visualization SelfAp 0.009 0.212 -0.416 0.433
and other-avoidance scales (F(2,80.06)=4.35, p<0.05). SelfAv. -0.867 0.468 -1.803 0.070
OtherAp* -0.771 0.322 -1.414 -0.128
Further investigation of the interaction effect between learning OtherAv* 1.107 0.446 0.214 1.999
analytics visualization and task-approach showed a significant
difference in the word concreteness scores between users of the Viz(TopContributors)*TaskAp*** 1.523 0.351 0.820 2.225
Top Contributor visualization and the Quality visualization (z=- Viz (Top Contributors)*TaskAv -0.767 0.499 -1.764 0.231
2.59, p<0.05), as well as, users of the Top Contributors and Class Viz (Top Contributors)* SelfAp -0.740 0.690 -2.119 0.640
Average visualization (z=2.90, p<001). The positive association Viz (Top Contributors)* SelfAv -0.074 0.741 -1.554 1.406
between the task-approach scale and word concreteness scores Viz (Top Contributors)* OtherAp 0.162 0.433 -0.705 1.028
was largest for Top Contributors, followed by the users of the Viz (Top Contributors)* OtherAv
Quality visualization and the Class Average visualization. -0.604 0.608 -1.820 0.612
Probing the interaction effect between learning analytics visuali- Viz (Quality)*TaskAp 0.029 0.258 -0.488 0.545
zation and task-avoidance showed a significant difference in the Viz (Quality)*TaskAv* -0.886 0.370 -1.626 -0.146
word concreteness scores between the users of the Top Contribu- Viz (Quality)* SelfAp -0.167 0.349 -0.864 0.530
tors visualization and the Quality visualization (z=2.63, p<0.05). Viz (Quality)* SelfAv* 1.375 0.557 0.262 2.489
Further analysis showed a positive effect on word concreteness Viz (Quality)* OtherAp 0.508 0.401 -0.293 1.310
scores for the Quality visualization users, while this effect was Viz (Quality)* OtherAv** -1.333 0.516 -2.365 -0.300
negative on the word concreteness scores for the users of the other Significance code: *** p<0.001 , ** p<0.01 , *p<0.05, . p<0.1
two visualizations. (marginal)
Further investigation of the interaction effect between learning All variables are scaled
analytics visualization and other-approach showed a significant
on the word concreteness scores for the users of the Top Contribu-
difference in the word concreteness scores between users of the
tors visualization.
Class Average visualization and the users of the Top Contributors
visualization (z=-2.69, p<0.05). Further analysis showed a posi- Finally, the interaction effect between learning analytics visualiza-
tive effect on word concreteness scores for the users of the Class tion and other-avoidance showed a significant difference in the
Average and Quality visualization, while this effect was negative word concreteness scores between the users of the Class Average

59
visualization and the users of both Top Contributors (z=2.67, to students oriented towards performance goals (i.e. other-based
p<0.05) and Quality visualizations (z=2.64, p<0.05). The associa- and self-based goals in 3×2 AGO model), it can encourage them
tion between other-avoidance scale scores and word concreteness to engage more productively in the discussion activity [40].
scores was negative for the users of the Class Average visualiza-
tion, while the association was positive for the Top Contributors 4.1.2 Different Visualizations and Students’ Quality
and Quality visualizations. of Posts Considering their AGOs
Our results showed that after controlling for achievement goals,
4. DISCUSSION AND CONCLUSIONS some learning analytics visualizations had positive and some had
The overall goal of this study was to investigate the effect of dif- negative effects on students’ quality of posts observed through
ferent information presented through learning analytics visualiza- discourse features (i.e., Narrativity, Deep Cohesion, Referential
tions on the posting behavior of students with different self- Cohesion, and Word Concreteness). For each achievement goal,
reported achievement goal orientations in online group discussion summary of significant associations are reported in Table 7.
activities.
In Table 7, positive associations show that higher scores on a
4.1 Interpretation of the results specific AGO scale are associated with higher scores on a specific
discourse component when using the visualization, whereas nega-
4.1.1 Different Visualizations and Students’ Quantity tive associations indicate that higher scores on an AGO scale are
of Posts Considering their AGOs associated with lower scores on discourse features for a particular
Our analysis showed that after controlling for achievement goals, visualization. Table 7 uncovers non-homogenous findings across
some learning analytics visualizations had positive and some had different goal orientations and different visualizations.
negative effects on students’ quantity of posts. Out of the five visible features presented in Table 7, the most
For students who used Top Contributors and Quality visualiza- highlighted and frequent discourse component is deep cohesion.
tion, higher scores on other-approach scale were significantly For long, the importance of cohesion in text and oral communica-
associated with higher numbers of posts, whereas for those who tion has been emphasized by cognitive scientists who aimed at
used Class Average, the association with count of posts was nega- understanding how human mind constructs meaning from dis-
tive. course [10]. In fact, measuring cohesion was the main driver for
the development of Coh-Metrix which later expanded to other
The positive effect of Top Contributors visualization on the quan-
tity of posts is in alignment with prior research showing that stu- discourse features. There are almost consistent findings in the
collaborative learning literature showing positive outcomes of
dents with other-approach goals assess their competence level in
deep cohesion. Higher levels of deep cohesion show deeper inte-
terms of normative standards and aim at outperforming their peers
gration of the ideas with background knowledge and fever con-
[13]. In this case, the students who used the Top Contributors
ceptual gaps, as well as, better individual and group performance
visualization may have interpreted the norm based on the contri-
[9].
bution level of those who had the highest number of postings in
the class. Another interpretation is that they may have strived to Our non-homogenous results across different visualizations show
gain visibility by the rest of the class, which means being listed as that using a particular visualization followed a positive association
top contributors themselves. Hence, this positive association be- between a certain goal and a discourse feature, while another vis-
tween the other-approach scale scores and numbers of posts for ualization may have followed a negative association for the same
users of this visualization is not surprising. The Quality visualiza- achievement goal and the same discourse component. For in-
tion may have motivated students with orientation towards other- stance, those with higher tendency towards self-avoidance goals
approach goal to outperform the rest of the class in terms of the constructed messages with higher deep cohesion when using the
depth and breadth of the key concepts covered in their messages. Quality visualization but lower deep cohesion when using the
In order to reach that goal, this visualization may have indirectly Class Average or Top Contributors visualizations. As discussed
encouraged them to contribute more. previously, high deep cohesion is associated with positive out-
comes and thus, it is highly desirable [9]. Students with avoidance
For the Class Average visualization the students’ judgment of
goals often suffer from the lack of task focus and hence, are more
how their peers were doing may have been influenced by the dis-
likely to experience low deep cohesion. It seems that the Quality
played average performance of the entire class. Research shows
that students who adopt normative standards, through other- visualization may have played a positive role in directing the stu-
dents with high self-avoidance goals towards overcoming task
approach, usually rely on the instructor’s criteria, as they believe
disrupting thoughts and integrating more cohesive messages,
this can best lead to outperforming their peers if no other visible
while the other two visualizations may have played a negative
norm exists [32]. In light of this, real-time updates presented in
role. A possible explanation is that presentation of information in
the visualization may push the instructor’s clearly expressed crite-
the Quality visualization was more focused on improvements of
ria behind the analytics metrics. If the class average is below
self over time (key concepts covered), which can increase feeling
teacher’s expectation at any given time, students with other-
of self-efficacy and self-confidence, and hence, improve the task
approach tendency may follow that as their normative standards
focus [32].
for their goal.
Similarly, our non-homogenous findings across different
Previous research shows that normative goal-standards can range
achievement goals indicate that using a particular visualization
from modest to extreme [32]. It might be that learning analytics
visualization can be an influencing factor in determining the end followed a positive association between one achievement goal and
a discourse feature, whereas the same visualization may have
points of this range. The Top Contributors and Quality visualiza-
followed negative association for the same discourse feature and
tions encourages setting a higher standard to outperform in com-
another achievement goal. For instance, despite positive outcomes
parison to the class average, which means it is more challenging
of Quality visualization for students oriented towards self-
to achieve and requires more effort. This is in accordance with the
avoidance goals, the role this visualization played on construction
idea that if desirable participation behaviors are explicitly exposed

60
of deep cohesive messages appeared to be negative on the indi- Table 7. Summary of Mixed Model Analysis for Interaction
viduals with higher tendency towards task-avoidance. It seems between Learning Analytics Visualization and AGO Scale on
that for students with task-avoidance strivings, seeing the con- Quality of Posts
cepts they have not covered increased their stress of doing poorly
in the discussions. Assoc.
AGO Scale Visualization Dependent Variable Direction
A highlighted aspect of the summary table is the presence of Task- Class Average Narrativity -
negative valence goals. In the literature, avoidance goals – regard- Approach Deep Cohesion -
less of the competence definition – have mostly been associated
Referential Cohesion -
with negative outcomes because of their tendency to avoid failure.
Low cognitive engagement, low self-efficacy, high anxiety and Word Concreteness +
feeling of shame, confusion, disorganized study habits, task- Top Narrativity +
disrupting thoughts, help-avoidance, poor performance and inter- Contributors Deep Cohesion +
est are among destructive outcomes of mastery-avoidance (task- Referential Cohesion +
avoidance) and performance-avoidance goals (other-avoidance Word Concreteness +
and self-avoidance) [32]. Therefore, providing feedback to help
Quality Narrativity +
reduce some of the negative aspects of these avoidance goals is
desirable in addition to the prevision of the information shown in Deep Cohesion +
the learning analytics visualizations. Referential Cohesion +
The most visible achievement goal with positive valance in Ta- Word Concreteness +
ble 7 is task-approach. Prior research shows that students with a Task Class Average Narrativity +
high task-approach tendency in a particular context compared to Avoidance Deep Cohesion +
others, find the topic interesting, have positive feelings about the Word Concreteness -
task and perceive it as valuable, use deep learning strategies and Top Narrativity +
appreciate both cooperativeness and help seeking [32]. Therefore, Contributors Deep Cohesion +
it is not surprising that their deep approach to learning can help
them mentally connect ideas and construct messages that show Word Concreteness -
stronger signs of deep cohesion [1]. Our findings indicate all the Quality Narrativity -
Quality and Top Contributors visualization had a positive effect Deep Cohesion -
on deep cohesion when controlled for task-approach scores. This Word Concreteness +
finding is not surprising for the Quality visualization, as it directly Self- Class Average Narrativity +
promotes coherent discussion of some key concepts and logical Approach Top Narrativity -
integration with related ideas. As for Top Contributors, quality
may indirectly be promoted through externalization of high stand- Contributors
ards on the contribution level. Therefore, it may encourage deeper Quality Narrativity -
investigation into the topic of discussion. Self- Class Average Narrativity -
Avoidance Deep Cohesion -
4.2 Implications for Theory and Practice Syntactic Simplicity -
The findings present some methodological, theoretical, and prac- Referential Cohesion +
tical implications. On a methodological side, the study shows the
importance of assessing learning analytics visualizations in au- Top Narrativity -
thentic course settings to evaluate the actual effect of the present- Contributors Deep Cohesion -
ed information on students’ behaviors and outcomes. Combining Syntactic Simplicity -
traditionally collected data through self-reported surveys, such as Referential Cohesion -
individual achievement goal orientation, with fine grained data Quality Narrativity +
such as interaction logs and generated artifacts. In this study, the Deep Cohesion +
effect of different learning analytics visualizations on students’
Syntactic Simplicity +
behavior was uncovered only when looking at the fine-grained
data and after controlling for students’ achievement goals, as mo- Referential Cohesion +
tivational constructs. In addition, analysis of discourse patterns Other- Class Average Referential Cohesion -
provided in-depth insight into the quality of students’ contribu- Approach Word Concreteness +
tions that complemented traditional metrics that rely on quantity Top Referential Cohesion -
of contributions. Contributors Word Concreteness -
The study poses some important theoretical and practical implica- Quality Referential Cohesion +
tions for the further research and use of learning dashboards and Word Concreteness +
tools by encouraging adoption of effective instructional practices
to support their use. From the instructional design point of view, Other- Class Average Deep Cohesion +
our findings show the potentials of learning analytics visualiza- Avoidance Word Concreteness -
tions as a feedback mechanism for students in online learning Top Deep Cohesion +
environments. In our study, the instructional design of the discus- Contributors Word Concreteness +
sion activity followed guidelines based on theories and practices Quality Deep Cohesion -
for effective and productive discussions. We are continuing to Word Concreteness +
investigate both the effect of pedagogical framing of learning
analytics visualization and the effect of connection of information

61
presented to the learning activities on students’ learning out- the tool, while considering the platform and affordances it pro-
comes. Also, our results confirm the findings of the limited re- vides.
search in this area that reveals learning analytics in the form of Secondly, in this study, we considered achievement goal orienta-
dashboards or reports can lead to the change of activities in online tion, a theoretical construct that could reveal individual differ-
discussions that are sometimes intentional and goal-oriented and ences with respect to motivational factors in educational context.
sometimes unconscious [41]. However, other aptitude constructs that illuminate on students'
Our research has implications for direction of empirical studies preferred approaches to learning [3] can also help understand how
around learning analytics visualizations and subsequently their particular students interact with learning analytics visualizations
designs. The findings of our field study reveal that the effect of a and how those visualizations affect their learning behaviors. Addi-
particular learning analytics visualization on students’ behavior tionally, since we are dealing with visual information and writing,
differs when students are inclined to different achievement goals. further linking motivational disposition to other individuals traits,
This can motivate further empirical studies to investigate the con- such as attention and perception, processing and evaluation, and
nection between other theoretical constructs that underlie individ- in case of the discussion argumentative writing, is needed to build
ual differences and effectiveness of learning analytics dashboards. fuller understanding how visualizations influence individuals. In
Such studies can help move towards developing a body of our current set of studies we are also probing for an individual’s
knowledge that could guide design and application of learning numeracy, graph literacy and other related cognitive characteris-
analytics tools that are theoretically informed. tics.
For instance, our results showed that the use of a particular learn- Our findings open several other directions for future research.
ing analytics visualization can be associated with positive changes First, in learning analytics of discussion activities, the listening
on students’ learning behavior with tendency towards a certain behaviors, i.e. reading other students’ post, is critical for effective
goal, even for avoidance goals. We know that avoidance goals discussion [39, 40]. As listening behavior constitutes over 75% of
have been mostly associated with negatives outcomes. Hence, our discussion activities, understanding the effect of learning analytics
findings encourages further examination of the role that personal- visualizations on listening behavior for would complement our
ized interventions can play in encouraging positive changes that research. Next, instructional scaffolds can produce desirable ef-
may lead to improved learning processes and outcomes. If the fects on development of critical thinking [16]. From this arises the
feedback provided through these visualizations alleviates negative question how scaffolding, or the lack thereof, fosters the positive
outcomes associated with pursuit of avoidance goals, such as anx- association between engaging with the visualizations and posting
iety and low self-efficacy, it may even have the potential to direct behaviors. Following a suggested framework [38], in a follow up
students towards pursuit of approach goals which according to study, we will investigate the effect of reflection and goal setting
research have been associated with more positive learning out- by embedding a space around the visualization for students to set
comes [32]. their goals and write a reflection journal and have it appear every
On the opposite side, our results show that each of the three visu- time they view the visualizations.
alizations can be negatively associated with learning behavior of This work focused on learning analytics for discussions. Investi-
students with certain individual difference. For example, showing gating the association between individual characteristics and dif-
the Top Contributors visualization to students with tendency to- ferent ways of visualizing other learning activities is needed to
wards self-avoidance was negatively associated with four dis- generalize our findings.
course features in their postings. This was discovered only after Acknowledgement. This research was supported by the Social
carefully analyzing the interaction effect of visualizations with Sciences and Humanities Research Council of Canada.
goal orientations. Other examples can be extracted from Table 7.
Such insight can encourage adoption of more stringent require- 5. REFERENCES
ments for empirical evaluation of learning analytics visualizations [1] Akyol, Z. and Garrison, D.R. 2011. Understanding cognitive
before deploying them to wide-scale use. presence in an online and blended community of inquiry: As-
The choice of visualizations in this research was guided by their sessing outcomes and processes for deep approaches to learn-
ability to engage different goals individual students may pursue. ing. British J. of Educational Technology. 42, 2, 233 – 250.
We did not tap into knowledge in information visualization field. [2] Arnold, K. and Pistilli, M. 2012. Course signals at Purdue:
As our results show, both what is being presented and how, very using learning analytics to increase student success. The 2nd
likely have different effect on individual students, with some vis- Int. Conf. on Learning Analytics and Knowledge, 267 – 270.
ualizations being more effective than others. A systematic study is [3] Biggs, J.B. 1987. Student Approaches to Learning and Study-
needed to understand the effect of different learning analytics ing. Research Monograph. ERIC.
visualization designs by controlling for certain individual differ-
[4] Brusilovsky, P., Hsiao, I.-H. and Folajimi, Y. 2011. Quiz-
ences, eventually leading to clear guidelines how to provide per-
Map: open social student modeling and adaptive navigation
sonalized learning analytics [4].
support with TreeMaps. Towards Ubiq. Learning. 71–82.
4.3 Limitations and Future Work [5] Bull, S. and Kay, J. 2008. Metacognition and open learner
The current work has several limitations that require further re- models. The 3rd Workshop on Meta-Cognition and Self-
search to complement our results. First, our learning analytics Regulated Learning in Educational Technologies, at
visualizations were integrated into the learning management sys- ITS2008, 7–20.
tem by providing a link that required additional effort and motiva- [6] Corrin, L. and de Barba, P. 2014. Exploring students’ inter-
tion on students’ part to click and be directed to the visualization. pretation of feedback delivered through learning analytics
This may have affected how many students and how often viewed dashboards. Proc. of the ascilite 2014 conference, 201-205.
the visualizations. Future work should explore other integration [7] Dawson, S., Bakharia, A. and Heathcote, E. 2010. SNAPP:
options and their influence on the adoption and engagement with Realising the affordances of real-time SNA within networked

62
learning environments. Proc. of the 7th Int. Conf. on Net- [25] Luppicini, R. 2007. Review of computer mediated communi-
worked Learning, 125–133. cation research for education. Instructional Science. 35, 2,
[8] Dimitrova, V. 2003. STyLE-OLM: Interactive open learner 141–185.
modelling. Int. Journal of Artificial Intelligence in Education [26] Maehr, M.L. 1989. Thoughts about motivation. Research on
(IJAIED). 13, 35–78. motivation in education: Goals and cognitions. 3, 1, 299–
[9] Dowell, N.M., Cade, W.L., Tausczik, Y., Pennebaker, J. and 315.
Graesser, A.C. 2014. What works: Creating adaptive and in- [27] Mazza, R. and Milani, C. 2004. Gismo: a graphical interac-
telligent systems for collaborative learning support. Intelli- tive student monitoring tool for course management systems.
gent Tutoring Systems, 124–133. Int. Conf. on Technology-Enhanced Learning, 1–8.
[10] Dowell, N.M., Graesser, A.C. and Cai, Z. Language and [28] Nakahara, J., Hisamatsu, S., Yaegashi, K. and Yamauchi, Y.
Discourse Analysis with Coh-Metrix: Applications from Ed- 2005. iTree: Does the mobile phone encourage learners to be
ucational Material to Learning Environments at Scale, Jour- more involved in collaborative learning? Proc. of the Conf.
nal of Learning Analytics, in press. on Computer Support for Collaborative Learning: Learning
[11] Elliot, A.J. 1999. Approach and avoidance motivation and 2005: the next 10 years! , 470–478.
achievement goals. Educ. Psychologist. 34, 3, 169-189. [29] Rovai, A.P. 2007. Facilitating online discussions effectively.
[12] Elliot, A.J., Elliot, A.J. and Dweck, C.S. 2005. A conceptual The Internet and Higher Education. 10, 1, 77–88.
history of the achievement goal construct. Handbook of [30] Santos, J.L., Verbert, K., Govaerts, S. and Duval, E. 2013.
Competence and Motivation. 16, 52–72. Addressing learner issues with StepUp!: an Evaluation. The
[13] Elliot, A.J., Murayama, K. and Pekrun, R. 2011. A 3×2 3rd Int. Conf. on Learning Analytics and Knowledge, 14–22.
achievement goal model. Journal of Educational Psychology. [31] Schielzeth, H. and Nakagawa, S. 2013. Nested by design:
103, 3, 632. model fitting and interpretation in a mixed model era. Meth-
[14] Foltz, P.W., Kintsch, W. and Landauer, T.K. 1998. The ods in Ecology and Evolution. 4, 1, 14–24.
measurement of textual coherence with latent semantic anal- [32] Senko, C., Hulleman, C.S. and Harackiewicz, J.M. 2011.
ysis. Discourse Processes. 25, 2-3, 285–307. Achievement goal theory at the crossroads: Old controver-
[15] Friedman, J., Hastie, T. and Tibshirani, R. 2001. The ele- sies, current challenges, and new directions. Educational
ments of statistical learning. Springer series in statistics Psychologist. 46, 1, 26–47.
Springer, Berlin. [33] Shum, S.B. and Ferguson, R. 2012. Social learning analytics.
[16] Gašević, D., Adesope, O., Joksimović, S. and Kovanović, V. Journal of Educational Technology & Society. 15, 3, 3–26.
2015. Externally-facilitated regulation scaffolding and role [34] Snow, R.E. 1991. Aptitude-treatment interaction as a frame-
assignment to develop cognitive presence in asynchronous work for research on individual differences in psychotherapy.
online discussions. The Internet and Higher Education. 24, J. of Consulting and Clinical Psychology. 59, 2, 205-216.
53–65. [35] Verbert, K., Duval, E., Klerkx, J., Govaerts, S. and Santos,
[17] Govaerts, S., Verbert, K., Duval, E. and Pardo, A. 2012. The J.L. 2013. Learning analytics dashboard applications. Ameri-
student activity meter for awareness and self-reflection. can Behavioral Scientist. 57, 10, 1500-1509.
CHI’12 Extended Abstracts on Human Factors in Computing [36] Winne, P.H. 2010. Improving measurements of self-
Systems, 869–884. regulated learning. Educ. Psychologist. 45, 4, 267–276.
[18] Graesser, A.C., McNamara, D.S. and Kulikowich, J.M. 2011. [37] Winne, P.H. and Hadwin, A.F. 1998. Studying as self-
Coh-Metrix providing multilevel analyses of text characteris- regulated learning. Metacognition in Educational Theory and
tics. Educational Researcher. 40, 5, 223–234. Practice. 93, 27–30.
[19] Joksimovic, S., Gasevic, D., Kovanovic, V., Adesope, O. and [38] Wise, A.F. 2014. Designing pedagogical interventions to
Hatala, M. 2014. Psychological characteristics in cognitive support student use of learning analytics. Proc. of the 4th Int.
presence of communities of inquiry: A linguistic analysis of Conf. on Learning Analytics and Knowledge, 203–211.
online discussions. The Internet and Higher Education. 22, [39] Wise, A.F., Hausknecht, S.N. and Zhao, Y. 2014. Attending
1–10. to others’ posts in asynchronous discussions: Learners’
[20] Kanuka, H. and Anderson, T. 2007. Online social inter- online “listening” and its relationship to speaking. Int. Jour-
change, discord, and knowledge construction. International nal of Computer-Supported Collaborative Learning. 9, 2,
Journal of E-Learning & Distance Education. 13, 1, 57–74. 185–209.
[21] Kay, J., Maisonneuve, N., Yacef, K. and Reimann, P. 2006. [40] Wise, A.F., Speer, J., Marbouti, F. and Hsiao, Y.-T. 2013.
The big five and visualisations of team work activity. Proc. Broadening the notion of participation in online discussions:
of the 8th Int. Conf. on Intel. Tutoring Systems, 197–206. examining patterns in learners’ online listening behaviors.
[22] Kerly, A., Ellis, R. and Bull, S. 2008. CALMsystem: a con- Instructional Science. 41, 2, 323–343.
versational agent for learner modelling. Knowledge-Based [41] Wise, A., Zhao, Y. and Hausknecht, S. 2014. Learning ana-
Systems. 21, 3, 238–246. lytics for online discussions: Embedded and extracted ap-
[23] Kruse, A. and Pongsajapan, R. 2012. Student-centered learn- proaches. Journal of Learning Analytics. 1, 2, 48–71.
ing analytics. CNDLS Thought Papers, 1–9. [42] Xu, R. 2003. Measuring explained variation in linear mixed
[24] Leony, D., Pardo, A., de la Fuente Valentín, L., de Castro, effects models. Statistics in Medicine. 22, 22, 3527–3541.
D.S. and Kloos, C.D. 2012. GLASS: a learning analytics vis- [43] Yuan, J. and Kim, C. 2014. Guidelines for facilitating the
ualization tool. Proc. of the 2nd Int. Conf. on Learning Ana- development of learning communities in online courses.
lytics and Knowledge, 162–163. Journal of Computer-Assisted Learning. 30, 3, 220–232.

63

You might also like