Presented by Hillary E. Merzdorf, Nathan M. Hicks and Kerrie A. Douglas of Purdue University at The Open University, Milton Keynes, UK on 15 June 2017. This presentation formed part of the FutureLearn Academic Network section (FLAN Day) of the 38th Computers and Learning Research Group (CALRG) conference. For full details, see http://cloudworks.ac.uk/cloudscape/view/3004
call girls in Kamla Market (DELHI) 🔝 >༒9953330565🔝 genuine Escort Service 🔝✔️✔️
Psychometric Analyses of the Expectancy-Value-Cost Scale in Advanced Nanotechnology MOOCs
1. +
Psychometric Analyses of the
Expectancy-Value-Cost Scale in
Advanced Nanotechnology
MOOCs
MOOC
Evaluation
Research Group
Hillary E. Merzdorf
Nathan M. Hicks
Kerrie A. Douglas
6/16/2017
2. +
The Expectancy-Value-Cost model of
motivation may provide good
information about learners in advanced
STEM MOOCs, but may require revision
for these populations
How do we know this?
6/16/2017
2
4. +
How to model motivation?
Extrinsic-Intrinsic Model, derived from Self-
Determination Theory1,2
Not helpful with classifying users
Did not seem to align with MOOC-related factors
Nearly all learners had high intrinsic motivation
Extrinsic motivators may work differently in MOOCs than
more traditional learning environments
6/16/2017
4
1. Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behavior.
New York and London: Plenum.
2. Douglas, K. A., Mihalec-Adkins, B. P., Hicks, N. M., Diefes-Dux, H. A., Bermel, P., & Madhavan,
K. (2016). Learners in advanced nanotechnology MOOCs: Understanding their intentions and
motivation. In American Society for Engineering Education’s 123rd Annual Conference &
Exposition, New Orleans, LA.
5. +
How to model motivation?
Alternative Model: Expectancy-Value-Cost model3
Expectancy = one’s belief they can accomplish a task
currently (ability beliefs) or in the future (expectancy beliefs)
Value = how much one wants to do the task based on
whether its perceived enjoyment (interest), usefulness
(utility), and ability to affirm one’s identity (attainment)
Cost = what one believes they will have to give up or expend
in order to accomplish the task, including time, effort, or self-
image
6/16/2017
5
3. Barron, K. E., & Hulleman, C. S. (2015). Expectancy-Value-Cost Model of Motivation. In J. D.
Wright (Ed.), International encyclopedia of the social and behavioral sciences, 2nd edition, Vol.
8 (pp. 503–509). Oxford: Elsevier.
7. +
Why is EVC relevant in a MOOC?
EVC highlights factors that are more relevant in a learning
environment than the intrinsic-extrinsic model (which is more
relevant in work-related settings)
MOOCs are voluntary, free, and low-risk, but require time and
effort
Learners are free to enroll regardless of background or abilities or
familiarity with the subject
Learners come from many backgrounds and likely have varying
levels of competing obligations
But…
6/16/2017
7
https://ehtrust.org/wp-content/uploads/2014/07/kids-reading-sm.jpg
8. +
Context of study
Pre-course surveys from two courses:
Nanophotonic Modeling
Principles of Biosensors
Discover the underlying traits with factor analysis
Investigate item functioning with item response theory (IRT)
6/16/2017
8
9. +
Demographics
6/16/2017
9
Category n Percentage
Course
Nanophotonic
Modeling
365 56.8
Principles of
Electronic
Biosensors
278 43.2
Gender
Male 302 47.0
Female 67 10.4
Transgender 0 0.0
Prefer not to answer 7 1.1
Non-respondent 267 41.5
Category n Percentage
Education
Less than a four-year degree 75 11.7
Four-year degree 110 17.1
Master’s degree 133 20.7
Doctoral or Professional
degree
68 10.6
Non-respondent 257 40.0
Academic Status
Part-time student (either in-
person or online)
219 34.1
Full-time student (either in-
person or online)
161 25.0
Non-respondent 263 40.9
10. +
Demographics
6/16/2017
10
Category n Percentage
Age
24 or under 131 20.4
25-34 150 23.3
35 or older 101 15.7
Non-respondent 261 40.6
Category n Percentage
Employment Status
Employed part-time,
unemployed, or retired
185 28.8
Employed full-time 196 30.5
Non-respondent 262 40.7
12. +
Exploratory Factor Analysis
6/16/2017
12
90% CI for
RMSEA
Model 𝜒2 𝑑𝑓 CFI TLI RMSEA LL UL
1-factor 1391.944* 27 .833 .778 .352 .337 .368
2-factor 539.212* 19 .936 880 .259 .241 .278
3-factor 11.040 12 1.000 1.000 .000 .000 .047
4-factor 2.308 6 1.000 1.003 .000 .000 .030
Item
Factor 1
loadings
Factor 2
loadings
Factor 3
loadings
(1) I know I can learn the material in this course. .879 .514 -.292
(2) I believe that I can be successful in this course. .944 .524 -.350
(3) I am confident that I can understand the
material in this course. .880 .549 -.263
(4) I think this course is or will be important. .545 .940 -.261
(5) I value this course. .551 .928 -.269
(6) I think this course is or will be useful. .496 .919 -.167
(7) Because of other things that I do, I do not
expect to have time to put into this course.
-.168 -.183 .839
(8) I think I will be unable to put in the time
needed to do well in this course.
-.270 -.203 .908
(9) I think I may have to give up too much to do
well in this course. -.173 -.067 .693
17. +
What does it all mean?
The instrument does measure expectancy, value, and cost in
this population
However, slight interpretation variability across sub-
populations
Advanced STEM MOOC learners have strong confidence in
their abilities, and value the courses highly, but experience
varying influence of cost
In order for this instrument to provide meaningful information,
items need to be revised to better differentiate levels of
expectancy and value, given the population
6/16/2017
17
18. +
Recommendations and implications
Given that most participants agree to expectancy and value,
questions should be more nuanced to better explore varying
sources of agreement
Measuring motivation with EVC may have direct relationship to
learning behaviors
If we can modify instrument to differentiate each dimension, it
can help us
Personalize learning experiences
Give learners greater autonomy
Deliver more targeted interventions
6/16/207
18
19. +
Thank you
Questions?
19
National Science Foundation
PRIME #1544259 and EEC #1227110
Any opinions, findings, and conclusions or recommendations expressed in this material
are those of the authors and do not necessarily reflect the views of the National Science
Foundation.
Hillary E. Merzdorf (hmerzdor@purdue.edu)
Nathan M. Hicks (hicks80@purdue.edu)
Kerrie A. Douglas (douglask@purdue.edu)
School of Engineering Education,
Purdue University
West Lafayette, Indiana
Editor's Notes
The design of a course must be developed based upon the needs of the stakeholders. One aspect of the learners’ needs stems from how the learners are motivated. Further, the motivational characteristics of the learners affect the experiences they have in the course, influencing their behaviors and outcomes.
Expectancy consists of ability beliefs (what they feel capable of doing now) and expectancy beliefs (what they think they’ll be capable of doing in the future). Value consists of intrinsic value (whether or not they inherently enjoy it), utility value (achieving short-term and long-term goals), and attainment value (affirms important aspect of their identity). Costs are negative aspects of engaging in an activity (effort, time, loss of other valued activities, and negative psychological states that could result from participation and particularly failure).
Previous instrument development for the EVC scale was done in K-12 populations in live learning environments, so it is necessary to make sure that the instrument and items function properly in different populations and contexts.
Brief explanations of what each of these techniques investigates:
Exploratory factor analysis allows us to explore different numbers of possible underlying traits that influence the way people respond to questions
Confirmatory factor analysis allows us to determine how well a specifically chosen factor structure fits the data (that is, which items we believe will be dictated by learners’ values of those underlying traits)
Measurement invariance allows us to determine the extent to which the factor structure is consistent for different sub-populations
Item Response Theory allows us to model learners’ probabilities of responding certain ways on items based on an estimation of their underlying trait values
Don’t worry too much about what all these numbers mean, but they basically tell us that the 3 and 4 factor models fit and the 1 and 2 factor models do not. However, the 4-factor model does not allow for reasonable substantive interpretation whereas the 3-factor model aligns with the proposed factor structure.
The Item loadings indicate that the three questions that should correspond to each factor do, in fact, load most heavily together.
Highest probability of selecting Strongly Agree
Items measure high trait level
Remaining four categories are not very functional
Collapsed response categories (lowest category is Disagree only)
These were the most discriminating items (that is, they did the best job of differentiating learners at certain levels), but did so at such low values of the underlying trait as to be rendered nearly useless for actually differentiating learners.
Highest probability of selecting Strongly Disagree or higher OR Strongly
Even but low probability for middle response categories
Response scale is more precise because small increases or decreases in theta shift categories
Results from current items are nearly dichotomous (either you agree or don’t agree, and most agree). Thus, rather than seeking whether they agree with basic questions of expectancy and value, we need to assume those are present and look for more nuanced discrepancies.
A stronger instrument will give us a better understanding of the motivations of our learners, which will allow us to structure courses to better meet the needs of the learners