Content uploaded by Chris Hulleman
Author content
All content in this area was uploaded by Chris Hulleman on Aug 31, 2016
Content may be subject to copyright.
Content uploaded by Chris Hulleman
Author content
All content in this area was uploaded by Chris Hulleman on Jul 04, 2016
Content may be subject to copyright.
Journal of Educational Psychology
Making Connections: Replicating and Extending the Utility
Value Intervention in the Classroom
Chris S. Hulleman, Jeff J. Kosovich, Kenneth E. Barron, and David B. Daniel
Online First Publication, August 15, 2016. http://dx.doi.org/10.1037/edu0000146
CITATION
Hulleman, C. S., Kosovich, J. J., Barron, K. E., & Daniel, D. B. (2016, August 15). Making
Connections: Replicating and Extending the Utility Value Intervention in the Classroom. Journal
of Educational Psychology. Advance online publication. http://dx.doi.org/10.1037/edu0000146
Making Connections: Replicating and Extending the Utility Value
Intervention in the Classroom
Chris S. Hulleman and Jeff J. Kosovich
University of Virginia
Kenneth E. Barron and David B. Daniel
James Madison University
We replicated and extended prior research investigating a theoretically guided intervention based on
expectancy-value theory designed to enhance student learning outcomes (e.g., Hulleman & Harackie-
wicz, 2009). First, we replicated prior work by demonstrating that the utility value intervention, which
manipulated whether students made connections between the course material and their lives, increased
both interest and performance of low-performing students in a college general education course. Second,
we extended prior research by both measuring and manipulating one possible pathway of intervention
effects: the frequency with which students make connections between the material and their lives. In
Study 1, we measured connection frequency and found that making more connections was positively
related to expecting to do well in the course, valuing the course material, and continuing interest. In Study
2, we manipulated connection frequency by developing an enhanced utility value intervention designed
to increase the frequency with which students made connections. The results indicated that students
randomly assigned to either utility value intervention, compared with the control condition, subsequently
became more confident that they could learn the material, which led to increased course performance.
The utility value interventions were particularly effective for the lowest-performing students. Compared
with those in the control condition who showed a steady decline in performance across the semester,
low-performing male students randomly assigned to the utility value conditions increased their perfor-
mance across the semester. The difference between the utility value and control conditions for low-
performing male students was strongest on the final exam (d⫽.76).
Keywords: academic motivation, educational intervention, expectancy-value motivation, gender, utility
value
Supplemental materials: http://dx.doi.org/10.1037/edu0000146.supp
Optimizing student motivation and learning in the classroom is
a goal shared by most educators. However, there is no consensus
on the best methods. Rewarding students for classroom behavior or
performance, or threatening punishment, are strategies convention-
ally believed to increase motivation and engagement (Ash, 2008;
Kohn, 1999; Newby, 1991). Such strategies presume that learning
tasks are not inherently rewarding and, therefore, extrinsic reasons
for task engagement must be introduced. In contrast, tapping into
more intrinsic sources of motivation (Ames, 1992), such as fos-
tering individual interest in specific topics (Hidi & Renninger,
2006), self-determined motivation (Deci & Ryan, 1985), and self-
directed task involvement (Csikszentmihalyi, 1990), are strategies
more likely to be recommended by educational psychologists
(Boekaerts, 2002). By focusing on student perceptions and beliefs
about the value of the learning activity, contemporary models of
expectancy-value motivation highlight this more intrinsic source
of motivation (e.g., Brophy, 1999; Eccles et al., 1983). Although
the research generated by the expectancy-value framework has
been largely correlational (Wigfield & Cambria, 2010), recent
classroom studies reveal that interventions designed to enhance
perceptions of value can increase both interest and course perfor-
mance (e.g., Hulleman & Harackiewicz, 2009; Hulleman, Godes,
Hendricks, & Harackiewicz, 2010). The research presented herein
replicates and extends this prior work by further investigating a
theory-based intervention designed to enhance student motivation
and performance.
The Expectancy-Value Framework
Originally adapted from classic models of expectancy-value
motivation (e.g., Atkinson, 1957; Vroom, 1964), Eccles and her
colleagues (1983) proposed that motivation in educational contexts
Chris S. Hulleman, Center for Advanced Study of Teaching and Learn-
ing, University of Virginia; Jeff J. Kosovich, Curry School of Education,
University of Virginia; Kenneth E. Barron and David B. Daniel, Depart-
ment of Psychology, James Madison University.
This research was supported by the National Science Foundation,
through Grants DRL 1252463 and 1228661, to the first author, and by the
U.S. Department of Education, through Grant R305B090002, to the second
author. The opinions expressed are those of the authors and do not
represent views of the funding agencies. A previous version of this article
was presented at the annual conference of the Northeastern Educational
Research Association, October 2012.
Correspondence concerning this article should be addressed to Chris S.
Hulleman, Center for the Advanced Study of Teaching and Learning,
University of Virginia, 405 Emmet Street South, Charlottesville, VA
22903. E-mail: chris.hulleman@virginia.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Journal of Educational Psychology © 2016 American Psychological Association
2016, Vol. 108, No. 7, 000 0022-0663/16/$12.00 http://dx.doi.org/10.1037/edu0000146
1
is determined most proximally by an individual’s expectancy
beliefs and subjective task values. Expectancy beliefs are defined
as the belief that one can succeed at an activity, and have been
correlated with achievement outcomes and achievement choices,
such as continued persistence and course taking (for reviews see
Richardson, Abraham, & Bond, 2012; Robbins et al., 2004). Sub-
jective task values are defined as the perceived importance of a
task or activity, and four facets were originally proposed by Eccles
and colleagues (1983): intrinsic (enjoyment), utility (usefulness for
proximal or distal goals), attainment (importance for one’s sense of
self), and cost (psychological barriers to, and negative conse-
quences of, task engagement). A wealth of prior research has
demonstrated that task values are positively correlated with con-
tinued persistence and ongoing motivation in an activity (for
reviews see Wigfield & Cambria, 2010), except cost which is
negatively related (e.g., Conley, 2012; Flake, Barron, Hulleman,
McCoach, & Welsh, 2015). Consistent with more recent concep-
tualizations of the expectancy-value framework (e.g., Barron &
Hulleman, 2015), we consider cost to be a unique construct inde-
pendent of expectancy and value.
In particular, students’ perceptions of utility value have been
associated with achievement outcomes in longitudinal field studies
(e.g., Bong, 2001; Durik, Vida, & Eccles, 2006; Hulleman et al.,
2008). Originally defined as “the value a task acquires because it
is instrumental in reaching a variety of long- and short-range goals
(Eccles & Wigfield, 1995, p. 216),” measures of utility value have
captured the relationship between students’ current (e.g., classes,
hobbies) and future goals (e.g., college major, career). For exam-
ple, one of the original scales measuring utility value (1995)
included items that tapped students’ future plans (“How useful is
learning advanced high school math for what you want to do after
graduation?”) and current goals (“How useful is what you learn in
advanced high school math for your daily life outside school?”).
Recent measures of utility value have mirrored this connection to
both current and future goals (e.g., Hulleman et al., 2008). Fur-
thermore, because some goals are more personally important than
others, utility value has been conceptualized as having elements of
both intrinsic and extrinsic motivation (Hulleman et al., 2008;
Simons, Vansteenkiste, Lens, & Lacante, 2004).
Despite extensive correlational support, limited research has
tested the effectiveness of interventions based on expectancy-value
models. Our review of the current literature revealed only a hand-
ful of published papers investigating interventions based on the
expectancy-value framework in an educational context, all focused
on utility value (Acee & Weinstein, 2010; Brown, Smith, Thoman,
Allen, & Muragishi, 2015; Harackiewicz, Canning, Tibbetts, Prini-
ski, & Hyde, 2015; Harackiewicz, Rozek, Hulleman, & Hyde,
2012; Hulleman & Harackiewicz, 2009; Hulleman et al., 2010;
Johnson & Sinatra, 2013). To provide stronger claims about both
internal and external validity, three of the studies were conducted
as double-blind, randomized classroom experiments (Harackie-
wicz et al., 2015; Hulleman & Harackiewicz, 2009; Hulleman et
al., 2010, Study 2). Hulleman and colleagues evaluated a utility
value intervention that encouraged students to discover the rele-
vance of the material they were studying to their lives. Utility
value was manipulated through a writing prompt given to high
school science (Hulleman & Harackiewicz, 2009) and college
psychology (Hulleman et al., 2010) students as part of their reg-
ularly assigned coursework. Students were randomly assigned to
either write about the relevance and usefulness of the course
material in their own lives (relevance condition) or a summary of
the material they were currently studying (control condition). In
the high school sample, students completed writing assignments
every three to four weeks of a 20-week semester. Students aver-
aged about five essays throughout the semester. In the college
sample, students were given writing assignments in the 8th and
12th weeks of a 15-week semester. The key dependent variables in
both studies were end-of-semester interest in the course topic and
course grades. The researchers provided teachers with information
regarding whether students had completed the essays, but teachers
were blind to condition throughout the semester. Because students
wrote about course-related topics in both the relevance and control
conditions, knowledge activation was controlled (i.e., summariza-
tion; see Dunlosky, Rawson, Marsh, Nathan, & Willingham,
2013). The conditions thereby differed only in terms of the acti-
vation of utility value.
In these studies, the results indicated that the intervention was
more effective for students with lower perceived or actual com-
petence. In college psychology, students who performed more
poorly on initial course exams were more interested in the course
if they were in the relevance condition than the control condition.
In high school science, the interaction effect was replicated on both
science interest and grades for students who entered the course
with lower performance expectations. In fact, the effect on end-
of-semester GPA for students with low performance expectations
resulted in an increase in .80 GPA points. In both studies, the
intervention increased students’ perceptions of utility value, and
these increased perceptions led to improved performance and
interest. Furthermore, this pattern of intervention effectiveness was
also replicated with (a) undergraduate students who learned a
mental math technique in the laboratory (Hulleman et al., 2010,
Study 1), (b) first-generation college students enrolled in introduc-
tory science classes (Harackiewicz et al., 2015), and (c) high
school students whose parents received an intervention on how to
talk to their teenager about the value of math and science course-
work (Rozek, Hyde, Svoboda, Hulleman, & Harackiewicz, 2015).
Learning Why the Utility Value Intervention Works
Together, these initial studies demonstrated that interventions
designed to increase subjective task value subsequently increased
interest and performance. However, these studies also demon-
strated that the intervention effects were not the same for everyone
(for reviews see Durik, Hulleman, & Harackiewicz, 2015; Harack-
iewicz, Tibbetts, Canning, & Hyde, 2014). Given these interven-
tion effects, we sought to understand what might explain the
underlying mechanisms, as well as develop future interventions
that might work for all students regardless of their success expec-
tancies and performance history.
Although Hulleman and colleagues (2009, 2010; Harackiewicz
et al., 2015) have generally used relevance and utility value inter-
changeably in describing their intervention, there is a potentially
important distinction to be made. Whereas utility value refers to
usefulness to a proximal or distal goal, relevance simply refers to
the presence of a relationship between one topic or idea and
another topic or idea, which could include a goal but also includes
a broader set of relationships (Kosovich & Hulleman, 2016). For
example, math could be useful because it will help me in a future
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2HULLEMAN, KOSOVICH, BARRON, AND DANIEL
job (utility value), or it could relate to my life because store
cashiers need it even if I do not (relevance). Because the utility
value intervention is one type of relevance intervention, one pos-
sible mechanism for utility value intervention effects is that en-
couraging students to find a connection allows them to notice
relationships that they previously had not. Seeing such connections
may allow individuals to view new information from a different
perspective, and develop a more in-depth integration of their
knowledge (Bransford & Schwartz, 1999). In addition, simply
referencing the self when learning new material can lead to learn-
ing gains (e.g., Barney, 2007; for a review see Symons & Johnson,
1997). Consistent with this hypothesis is the finding that instruct-
ing individuals to find connections between learning situations can
increase the likelihood of adapting a skill from one situation to
another (i.e., cognitive transfer; Burke & Hutchins, 2007; Gentner,
Loewenstein, & Thompson, 2003; Gick & Holyoak, 1980). As
hypothesized in the early work of learning theorists (e.g.,
Thorndike & Woodworth, 1901), making a connection may en-
hance learning by instigating a set of processes that engenders a
different approach to studying that may increase learning. For
example, if a student finds a personal connection during a psy-
chology lecture, the student may be more interested in the assigned
reading and to discuss the material with friends. In general, the
student may be more motivated to actively process the material
during lecture and later when reading the book. Establishing rela-
tionships between new knowledge and old ideas may create a
richer cognitive architecture which the student can draw upon
when studying. As a result, students who make more connections
between course material and existing knowledge may be more
likely to find usefulness in the course, which may enhance moti-
vation.
What is the best way to investigate the frequency of connections
as a key pathway through which the utility value intervention
impacts outcomes? One approach is to measure the proposed
mechanism and conduct path analyses (e.g., Hulleman et al.,
2010). This approach is appealing because it is relatively simple
and falls within the range of most statistical packages (e.g., Tofighi
& MacKinnon, 2011). The limitation of this approach is that it is
correlational, and it does not account for other key variables that
may explain the effects of the intervention but have not been
measured. In contrast, the second approach, which is far less
common but more powerful, is to manipulate the mechanism (see
Baron & Kenny, 1986; Sigall & Mills, 1998). This approach
allows the researcher to randomly assign participants to different
levels of the variable to establish a cause-and-effect relationship.
The con to this approach, which is inherent to all intervention
studies, is that the effect of a manipulated variable may not be the
same as the effect of the measured variable (cf. Barron & Harack-
iewicz, 2001). Rather than choosing one method, both approaches
to enhancing learning outcomes will be investigated in this paper.
Interest as an Educational Outcome
Academic performance is a widely accepted educational out-
come and grades play a pivotal role in a student’s long-term
educational opportunities. A less-acknowledged but equally im-
portant outcome is interest (Hidi & Harackiewicz, 2000; Hulleman
et al., 2008). In a longitudinal study, Harackiewicz, Barron, Tauer,
and Elliot (2002) found that interest predicted course choice and
college major selection over six years, whereas prior performance
and college GPA did not. Interest can be thought of as two
different types (Hidi & Renninger, 2006). Situational interest is
the experience of engagement or attention during a task (Schraw &
Lehman, 2001). Individual interest is an enduring proclivity for the
task or behavior (Renninger & Wozniak, 1985). In the current
study, we focused on situational interest because it is a precursor
for the development of individual interest (Hidi & Renninger,
2006), which predicts long-term academic and career choices (e.g.,
Peters & Daly, 2013; Pike & Dunne, 2011), and is heavily influ-
enced by the learning context and therefore amenable to change
via short-term interventions (e.g., Durik & Harackiewicz, 2007;
Hulleman et al., 2010).
Current Studies
In Study 1, students’ perceptions of how often they connected
the material to their lives was measured and used to predict student
learning outcomes over the course of a semester. In Study 2,
connection frequency was manipulated through an experimental
intervention delivered as part of course embedded assignments.
Although we hypothesized that both measured and manipulated
connection frequencies will have similar effects on learning out-
comes, it is possible that measured and manipulated variables
capture different aspects of the same phenomena. Thus, it is crucial
to examine both types of effects when investigating the role of
connection frequency.
The utility value interventions utilized in the current studies
were based on the self-generated utility value interventions
used by Hulleman and colleagues (e.g., Hulleman et al., 2010;
Hulleman & Harackiewicz, 2009). In addition to replicating this
prior research, we extend it in five ways. First, we used an
online course management system to deliver the intervention,
instead of paper-and-pencil writing assignments used in prior
studies, which was seamlessly embedded within the course as a
regular course assignment. Second, we tested the mechanism of
the utility value intervention by measuring the self-reported
frequency with which students made connections between the
material and their lives throughout the semester (e.g., connec-
tion frequency). Third, we further examined the mechanism of
the utility value intervention by manipulating one hypothesized
mechanism— connection frequency. Fourth, we examined the
effects of the intervention on students’ expectancies and per-
ceived costs in the course. The theoretical model hypothesizes
that the utility value intervention effects are driven through
increased perceptions of value, particularly utility value. How-
ever, it is also possible that writing about the relevance of
course material could increase students’ expectancies that they
can learn the material and perform well in the course, or
decrease their perceived cost for learning. Fifth, because prior
research found differential intervention effects based on key
demographics, such as gender and initial performance, we also
examined whether the intervention was more effective for stu-
dents at-risk for poor performance. In the case of a college
general education course, students who initially perform poorly
in the course are most at-risk, as are male students (Voyer &
Voyer, 2014).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
3
UTILITY VALUE INTERVENTION IN THE CLASSROOM
Study Samples
The samples for both studies in this paper came from students
who were enrolled in two sections of a 15-week introductory
psychology course at a midsized university in the southeastern
United States. Both sections were taught by the same instructor. Of
the 589 students who were enrolled in the course, 501 students
(85%) completed the initial consent form and were eligible to
participate in our research. Of these students, 113 were randomly
selected to participate in Study 1 and 388 were randomly selected
for Study 2.
Study 1: A Longitudinal, Correlational Investigation
In Study 1, we explored a potential pathway for utility value
effects found in prior research. Specifically, we developed a new
measure that asked students to report on the frequency with which
they made connections between the material and their lives while
listening to lectures, studying for exams, and socializing with
friends. As a method for providing initial validity evidence for
both the idea and the measure, we examined whether students’
self-reports of connection frequency provided a pathway through
which utility value was related to student learning outcomes during
a semester-long undergraduate psychology course.
Method
Participants. Of 113 eligible introductory psychology stu-
dents, the final sample included 97 students who were over the age
of 18 and participated in the surveys. Students received extra credit
in the course for completing both surveys. The sample was 70%
female, 84% white (4% African American, 4% Asian), 86% non-
psychology majors, and 55% freshman (28% sophomore, 12%
junior, 3% senior). The mean age of participants was 18.7 years.
Self-reports of expectancy-value-cost motivation. Students
completed self-report surveys at three time points during the
semester: Time 1 measures were taken during the 2nd week, Time
2 measures were taken during the 8th week, and Time 3 measures
were taken during the 14th week of the semester. Measures of
expectancy, utility value, and cost were collected at Time 1 and 3,
and have been previously validated with students in middle school
(Kosovich, Hulleman, Barron, & Getty, 2015), high school (Hul-
leman & Harackiewicz, 2009), and college (Grays, 2013; Hulle-
man et al., 2008). Expectancy was measured using a 4-item scale
(e.g., “I expect to do well in this class,” ␣⫽92 and .93). Utility
Value was measured using a 6-item scale (e.g., “I can apply what
we’re learning in this class to the real world,” “The course material
is relevant to my future career plans,” ␣⫽.93 and .92). Cost was
measured using a 6-item scale (e.g., “Doing well in this class isn’t
worth all the things that I have to give up,” ␣⫽.81 and .87). All
self-report items used an 8-point Likert-type scale that ranged from
1(completely disagree)to8(completely agree; see Appendix A in
the Supplemental Online Material for complete list of items, and
Table 1 for descriptive information including reliabilities).
Connection frequency. To capture the number of connections
between students’ lives and the course material, a 3-item measure
of connection frequency was included at Time 2, just after the
second course exam (e.g., “When reading a chapter from the
textbook/During a regular class period or lecture/When studying
for quizzes and exams, how often do you connect the class material
to your life?”; ␣⫽.87). These items used a 6-point Likert-type
scale ranging from 1 (never)to6(all of the time).
Learning outcomes. Two major learning outcomes were col-
lected: academic performance and interest in the course. Students’
academic performance was measured using class exam scores.
There were 4 noncumulative exams, each covering four chapters in
the textbook, and administered during the 4th, 8th, 12th, and 16th
weeks of the course. There were 80 multiple-choice questions on
the first three exams, and 100 multiple-choice questions on the
fourth and final exam. Each question was worth one point. Stu-
dents completed the exams using Scantron answer sheets and the
exams were machine-scored. The grades were never curved and
the grading scale was: 90% to 100% A, 80% to 89% B, 70% to
Table 1
Descriptive Statistics for Major Variables in Study 1
Variable 12345 67891011
1. Time 1 interest
2. Time 1 expectancy .34
3. Time 1 utility value .79 .31
4. Time 1 cost ⫺.37 ⫺.20 ⫺.30
5. Initial exam .00 .23 .04 ⫺.12
6. Time 2 connections .33 .33 .25 ⫺.14 .14
7. Time 3 interest .75 .34 .62 ⫺.35 .11 .46
8. Time 3 expectancy .33 .60 .32 ⫺.39 .50 .34 .54
9. Time 3 utility value .61 .32 .69 ⫺.29 .06 .41 .77 .45
10. Time 3 cost ⫺.33 ⫺.13 ⫺.18 .66 ⫺.19 ⫺.30 ⫺.37 ⫺.50 ⫺.29
11. Final exam .01 .06 .13 .08 .60 .05 .03 .27 .02 .02
12. Female .19 ⫺.03 .18 ⫺.07 .10 ⫺.01 .16 .08 .21 ⫺.08 .20
Observed min. 3.00 4.50 2.33 1.83 31.00 2.00 3.33 4.00 2.00 1.67 35.00
Observed max. 8.00 8.00 8.00 6.00 75.00 6.00 8.00 8.00 8.00 8.00 98.00
Mean 6.00 6.50 6.09 3.60 63.16 3.88 6.14 6.48 6.01 3.96 81.85
SD 1.08 .86 1.11 .93 7.19 .91 1.19 .97 1.15 1.23 9.28
␣.92 .92 .93 .81 .83 .87 .92 .93 .92 .87 .87
Note. N ⫽97. Female is a dummy-coded variable: 0 ⫽male, 1 ⫽female. Correlations greater than |.20| are significant at p⬍.05. Correlations greater
than |.29| are significant at p⬍.01.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
4HULLEMAN, KOSOVICH, BARRON, AND DANIEL
79% C, 60% to 69% D, and below 60% F. Point biserial informa-
tion for each test revealed no bad questions. Further information
regarding the exams can be obtained from the authors upon re-
quest.
In addition to performance, we also collected a measure of
students’ interest in the course material using a 9-item scale (e.g.,
“I think the field of psychology is very interesting,” “I really enjoy
this class,” “I plan on taking more courses in psychology,” ␣⫽.92
and .92). This measure of interest has been used in prior research
(Harackiewicz, Durik, Barron, Linnenbrink-Garcia, & Tauer,
2008; Hulleman et al., 2010), and is designed to capture students’
emerging interest in psychology (Renninger & Hidi, 2011). Al-
though this measure of interest was collected at the same time as
the Time 3 measures of motivation, we used interest as an outcome
in our analyses because, conceptually, interest is one of our key
academic outcomes. Theoretical models of interest development
(e.g., Hidi & Renninger, 2006) and empirical research (Harackie-
wicz et al., 2008) demonstrate that perceptions of competence and
value are key antecedents of interest.
Procedure. The Time 1 survey was administered via an online
survey during the second week of the semester. Students had a
week to complete the survey, which included all Time 1 motiva-
tion items. During the 4th week of the semester, students took the
first exam. During the 8th week of the semester, students com-
pleted the connection-frequency items (Time 2). During the 14th
week of the semester, participants completed the Time 3 measures.
In the 16th week, participants completed the final exam. Students
earned course credit for completed surveys.
Results
Descriptive analyses. Our primary research question was
whether the newly developed measure of connection frequency
was related to self-reports of student motivation and course out-
comes. Our hypothesis was that greater endorsement of the con-
nection frequency items would be related to end-of-semester mo-
tivation, interest, and exam scores. Further, we examined whether
connection frequency during midsemester provided an indirect
pathway between initial motivation and final course outcomes.
The measure of connection frequency was normally distributed
(skewness ⫽⫺.11; kurtosis ⫽.43), and the overall scale mean
was just above the scale midpoint of 3.5 (M⫽3.9 out of 6), which
reflects making connections between ‘sometimes’ and ‘often.’ And
95% of the scores fell between 2.33 and 5.33 (see Figure S1).
When inspecting the zero-order correlations (see Table 1), stu-
dents’ Time 2 self-reports of the frequency with which they
connected the course material with their lives was moderately and
positively correlated with Time 3 utility value (r⫽.41), expec-
tancy (r⫽.34), and interest (r⫽.46), and negatively correlated
with cost (r⫽⫺.30). Connection frequency was unrelated to final
exam scores (r⫽.05). Thus, the measure of connection frequency
is correlated in ways we would expect with other self-reported
motivation variables, and had a sufficiently normal distribution
and variance to warrant continuing to explore its role in the
development of motivation during the semester.
Regression analyses. While the pattern of correlations pro-
vided preliminary evidence that connection frequency may con-
tribute to learning outcomes, we ran a series of OLS regressions to
investigate the unique pattern of relationships between Time 2
connection frequency and Time 3 outcomes, controlling for Time
1 covariates (measures of motivation, interest, exam 1 score, and
gender). As shown in Table S1, connection frequency was a
unique and significant predictor of utility value (⫽.25, sr
2
⫽
.05), cost (⫽⫺.21, sr
2
⫽.04), and Time 3 interest (⫽.22,
sr
2
⫽.04). However, connection frequency was not a significant
predictor of Time 3 expectancy or final exam score. We next tested
whether Time 2 connection frequency served as an indirect path-
way () through which initial motivation could be related to
course outcomes. Using the method outlined by Tofighi and
MacKinnon (2011), we first regressed Time 2 connection fre-
quency on the Time 1 covariates and found that Time 1 expectancy
was the only significant predictor (⫽.24, sr
2
⫽.11). Next, we
utilized the prior regression equations to determine that Time 2
connection frequency served as an indirect pathway for Time 1
expectancy to contribute to changes in Time 3 utility value (⫽
.08, 95% CI [0.001, 0.20]), cost (⫽⫺.07, 95% CI
[⫺0.15, ⫺0.01]), and Time 3 interest (⫽.08, 95% CI [0.001,
0.19]). Finally, we examined whether connection frequency con-
tributed to learning outcomes through its relationships with Time
3 motivation. This path model is displayed in Figure 1, and
revealed that the relationship between connection frequency and
interest could be explained by the increases in Time 3 utility value
(⫽.15, 95% CI [0.043, 0.287]).
Study 1 Discussion
Study 1 demonstrated initial validity evidence for our measure
of the frequency with which students made connections between
the course material and their lives. The measure was normally
distributed, correlated with other self-reported motivation vari-
ables as expected, and contributed to students’ motivation and
learning outcomes. Consistent with this hypothesis, our measure of
connection frequency was uniquely related to increases in expec-
tancy, utility value, and interest, and decreases in cost, when
controlling for prior measures of those variables. In addition, we
found that connection frequency led to increased interest in psy-
chology by increasing students’ perceived utility value in the
course. Although connection frequency did not operate as an
indirect pathway between utility value and outcomes, as we had
hypothesized, it did provide an indirect pathway between initial
expectancy and outcomes. As a result, in Study 2 we examined
whether the utility value intervention operates through success
expectancies.
We also found in Study 1 that women outperformed men in the
course. This replicates an emerging finding that women earn
higher grades in school than men (e.g., Duckworth & Seligman,
2006; Voyer & Voyer, 2014). Our prior classroom-based interven-
tion research demonstrates that the utility value intervention works
better for low-performing students (Hulleman et al., 2010). Other
interventions designed to promote perceptions of utility value for
math and science found that the intervention differed depending
upon students’ gender and school performance (Rozek et al.,
2015). These three findings set up the need to examine whether the
intervention works differently for students based on demographics
associated with being at-risk for poor performance (e.g., gender,
initial performance levels), and experimental condition. We tested
this possibility in Study 2.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
5
UTILITY VALUE INTERVENTION IN THE CLASSROOM
Study 2: A Longitudinal, Experimental Investigation
In Study 2, we conducted a double-blind, randomized classroom
experiment that manipulated connection frequency by designing
an enhanced utility value intervention that encouraged students to
make more frequent connections between the course material and
their lives. The goal of this enhanced utility value intervention was
to increase the strength of the original utility value intervention by
adding a new element. Although we hypothesized that connection
frequency operated in the original utility value intervention, we
wondered whether spontaneous connections may be uncommon or
difficult to make (Bransford & Schwartz, 1999). We therefore
utilized a related line of research on implementation intentions
(e.g., Gollwitzer, 1999; Gollwitzer & Brandstatter, 1997), with the
goal of increasing the connections students make between the
material and their lives outside of the intervention.
When an individual forms an implementation intention, he or
she specifies the when, where, and how an intended behavior will
occur to promote goal attainment (Gollwitzer, 1999). The setting
of these intentions provides a salient anchor for when a specific
behavior should occur. In a study by Gollwitzer and Brandstatter
(1997), participants were randomly assigned to either adopt im-
plementation intentions for the completion of a self-reflection
essay assigned over winter break, or were simply given the goal of
turning in the essay. Students in the implementation intention
condition were far more likely to complete the essay on time, and
in less time, than a group of participants who were just given the
goal to complete the essay.
We integrated the implementation intentions framework into the
design of our enhanced utility value intervention. The enhanced
condition included an opportunity for students to set implementa-
tion intentions to make connections between their lives and the
course material on a routine basis during the semester (e.g., in
class, while studying, when socializing). By adopting implemen-
tation intentions to make connections between the course material
and their lives, we hypothesize that students will be more likely to
actively seek connections in the specific situations that they iden-
tify. An increase in connections should promote deeper processing
and engagement in learning, which in turn should enhance utility
value, interest, and course performance. Because spontaneous
connection-seeking does not always happen (Gentner et al., 2003),
setting implementation intentions may nudge individuals toward
this behavior.
Finally, we endeavored to make the interventions as easy to
implement as possible. Both utility value interventions were de-
signed so that they could be delivered via an online course man-
agement system used by the instructor and students. This allowed
us, as researchers, to keep the instructor blind to students’ exper-
imental condition, and enabled students to participate in the inter-
vention by using a familiar system. Although there was consider-
able set-up of the intervention required by the researchers,
including randomizing of groups, the delivery of such an interven-
tion to an entire class by a single instructor was done via an
assignment through the online course management system. This
approach is one solution for testing and scaling up psychological
interventions in classrooms (Harackiewicz & Borman, 2014;
Paunesku et al., 2015).
Method
Participants. Students in Study 2 were part of a separate
subsample of students enrolled in the same two sections of intro-
Figure 1. Path model of the relationships between connection frequency and learning outcomes in Study 1.
N⫽97. Values are standardized OLS regression coefficients that were statistically significant (p⬍.05). Only
significant paths are shown. Regression equations also controlled for gender (see Table S1 and text for details).
The direct effect of connection frequency on Time 3 Interest (⫽.22, p⬍.01) was reduced to nonsignificant
(⫽.10, p⫽.11) when the Time 3 motivation measures were included in the model.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
6HULLEMAN, KOSOVICH, BARRON, AND DANIEL
ductory psychology used in Study 1. Of the original 589 students
enrolled in the two sections, 388 were randomly selected to par-
ticipate in Study 2. The final sample included 357 students who
were over the age of 18, completed the final exam, and participated
in the interventions. Similar to Study 1, the Study 2 sample was
70% female, 84% white (6% African American, 4% Asian), 84%
nonpsychology majors, and 61% freshman (21% sophomore, 13%
junior, 4% senior). The mean age of participants was 18.6 years.
Measures. All self-report and performance measures were
identical to Study 1. Descriptive information on the scales, includ-
ing reliabilities, can be found in Table 2.
Procedure. The procedures were nearly identical to Study 1,
with two exceptions. First, the intervention prompts were delivered
after the first and second exams. Second, instead of being mea-
sured at Time 2 (after the second exam), connection frequency was
measured at the same time as the other motivation measures: Time
1 was the 2nd week of the semester and Time 3 was the 14th week.
This allowed us to examine whether the intervention worked by
increasing the frequency with which students connected the ma-
terial to their lives. As in Study 1, students received course credit
for completing the surveys and intervention prompts. Importantly,
neither the instructor nor teaching assistants knew the specific
content of the intervention, nor which students were assigned to
which condition.
Intervention #1. Following the first exam, students were re-
minded in class to participate in the first intervention assignment
and were given three days from the time that the links were
available to complete the activity. Students then received web links
to the intervention via email. The links were also posted in stu-
dents’ respective online group pages by the researchers. Upon
clicking the web-link, participants were randomly assigned to one
of three conditions: the control condition, the utility value condi-
tion, or the enhanced utility value condition. Thus, both the in-
structor and students were kept blinded to which students were in
which conditions.
In the control condition (n⫽119), participants received the
following prompt: “Below is a list of the units covered in GPSYC
101 so far. For each topic, summarize what you know in about 1
or 2 sentences. We are not asking you to elaborate on the material,
just to summarize the information that you can recall.” Underneath
the prompt were four text boxes labeled for each class unit (i.e.,
History, Careers, & Connections; Research; Biology & Behavior;
and Memory).
Both the utility value (n⫽116) and enhanced utility value (n⫽
122) conditions received the following prompt:
In the space below, we would like you to write 1 to 2 paragraphs about
how the material that you have been studying in GPSYC 101 relates
to your life. We are not asking you to summarize the material, just to
elaborate on its relevance to your life. So far, you have covered the
following units in your class: History, Careers, & Connections; Re-
search; Biology & Behavior; and Memory.
Below the prompt was a text-box for participants to type their short
essays. This prompt was adapted from Hulleman and colleagues
prior intervention prompts (Hulleman et al., 2010; Hulleman &
Harackiewicz, 2009).
In addition, enhanced utility value participants were then sent to
a new page featuring three additional prompts (see Appendix B in
the Supplemental Online Material). The first prompt asked partic-
ipants to identify the time and place where they might be able to
think about the relevance of class material to their own lives. The
second prompt asked participants to identify obstacles that might
prevent finding the relevance of class material. The third prompt
asked participants to identify strategies to overcome the obstacles
identified in the second prompt.
Intervention #2. Following the 2nd exam, students in the
control condition were given a pair of prompts in succession:
“Choose one of the specific topics from above. In 1 to 2 paragraphs
(75 to 125 words), summarize the details of the topic as best you
can.”
Table 2
Descriptive Statistics for Major Variables in Study 2
Variable 123456 789101112
1. Time 1 expectancy
2. Time 1 utility value .27
3. Time 1 cost ⫺.38 ⫺.36
4. Time 1 connections .30 .50 ⫺.26
5. Time 1 interest .27 .79 ⫺.38 .44
6. Initial exam .11 .01 ⫺.05 .00 .03
7. Time 3 expectancy .40 .20 ⫺.26 .16 .24 .34
8. Time 3 utility value .18 .67 ⫺.33 .30 .65 .06 .43
9. Time 3 cost ⫺.17 ⫺.33 .54 ⫺.15 ⫺.42 ⫺.27 ⫺.42 ⫺.44
10. Time 3 connections .19 .38 ⫺.27 .47 .39 .10 .34 .51 ⫺.34
11. Time 3 interest .07 .62 ⫺.32 .30 .78 .10 .33 .75 ⫺.50 .49
12. Final exam .09 .06 ⫺.10 .00 .08 .63 .40 .11 ⫺.31 .15 .13
13. Female .08 .33 ⫺.19 .17 .33 .07 .12 .28 ⫺.30 .21 .33 .12
Observed min. 4.75 2.17 1.67 1.00 1.44 35.00 1.50 1.00 1.83 1.50 1.67 39.00
Observed max. 8.00 8.00 7.00 6.00 8.00 76.00 8.00 8.00 8.00 6.00 6.00 100.00
Mean 6.46 6.12 3.52 3.77 6.11 61.62 6.34 6.47 5.97 3.99 3.99 79.48
SD .81 1.07 .87 .84 1.14 7.15 1.01 1.08 1.13 .88 .88 9.40
␣.90 .92 .80 .88 .93 — .93 .93 .86 .89 .93 —
Note. N ⫽357. Female is a dummy-coded variable: 0 ⫽male, 1 ⫽female. Correlations greater than |.10| are significant at p⬍.05. Correlations greater
than |.15| are significant at p⬍.01.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
7
UTILITY VALUE INTERVENTION IN THE CLASSROOM
Participants in the utility value and enhanced utility value con-
ditions were given a pair of prompts about relevance: (a) “Choose
a topic from above that is personally useful and meaningful to you.
In 1 to 2 paragraphs (75 to 125 words), describe how learning
about this topic is useful to your life right now” and, (b)
Choose a topic from above that is personally useful and meaningful to
you (it may be the same topic as before). In 1 to 2 paragraphs (75 to
125 words), describe how learning about this topic will be beneficial
to you in the future (e.g., education, career, daily life).
Enhanced utility value participants were also given several items
prompting reflection on their implementation intentions (see Ap-
pendix B in the Supplemental Online Materials). The reflection
items asked students to recall what implementation intentions they
had discussed in the previous intervention and to reflect on ways
they could improve their strategies.
Results
Analytic plan. Our primary research question involved exam-
ining whether the intervention conditions would promote learning
outcomes compared with the control condition (see Primary anal-
yses). Our main hypothesis was that the utility value interventions
would promote interest and achievement at the end of the semester
compared with the control condition. We also hypothesized that
the enhanced utility value condition would have an additional
effect above and beyond the utility value condition. This first set
of questions led us to conduct intent-to-treat, OLS regression
analyses on the outcomes (interest, exam scores) as a function of
the experimental conditions using hierarchical multiple regression.
Second, we tested whether the intervention was more effective for
students most at-risk for poor performance (see At-risk student
analyses). This involved adding interaction terms between initial
exam scores, gender, and the experimental conditions to the OLS
regression models used in the primary analyses. Third, we tested
whether the effects of the utility value interventions could be
explained, at least partially, by increased motivation (i.e., expec-
tancy, utility value, cost) and connection frequency. This would
demonstrate that both motivation and connection frequency were
indirect pathways through which the intervention impacted out-
comes. This question was examined using path modeling and
indirect effects analyses (see Indirect effects analyses). Finally, we
conducted a fidelity analysis to examine whether students re-
sponded to the utility value writing prompts as expected (see
Intervention fidelity analyses).
Descriptive analyses. A comparison of the unadjusted raw
means in Table S2 reveals no significant difference on the Time 1
covariates (all Fs⬍2.7, all ps⬎.10), indicating balanced ran-
domization across the three conditions. Second, students success-
fully responded to the intervention prompts. There were no differ-
ences in the number of words written in the control (M⫽178.0,
SD ⫽57.1) or utility value conditions (M⫽182.8, SD ⫽57.7;
d⫽.06; p⫽.57), indicating students committed a similar level of
effort and thinking in each condition. Example essays can be found
in the Supplemental Online Materials (see Appendix C). In addi-
tion, the connection frequency variable was again normally dis-
tributed at both Time 1 (skewness ⫽.19, kurtosis ⫽.08) and Time
3 (skewness ⫽.06, kurtosis ⫽⫺.28), with a mean near the
midpoint of the scale at both Time 1 (M⫽3.77 of 6, SD ⫽.83)
and Time 3 (M⫽3.99, SD ⫽.88). However, as expected, there
were small, raw mean differences in favor of the utility value
intervention conditions compared with the control condition on
Time 3 measures of interest (d⫽.24), expectancy (d⫽.23), utility
value (d⫽.24), and final exam scores (d⫽.23), but not on cost
or connection frequencies.
The pattern of correlations (see Table 2) was similar to Study 1
and revealed positive relationships between Time 1 connection
frequency and Time 3 expectancy (r⫽.16) and utility value (r⫽
.30), and a negative relationship with cost (r⫽⫺.15). Time 3
connection frequency was positively related to interest (r⫽.49)
and final exam scores (r⫽.15).
Primary analyses. Did the utility value intervention condi-
tions enhance academic outcomes compared with the control
group? To answer this question, we used hierarchical, OLS regres-
sion to examine the effects of the interventions on interest in
psychology and final exam scores. Regression allows us to exam-
ine unique relationships among the predictors of interest. We first
examined intervention differences using an intent-to-treat model
(Model 1) that included only the contrast codes for the experimen-
tal conditions. The utility value code compared whether both
utility conditions were better than the control (control ⫽⫺2,
utility ⫽⫹1, enhanced utility ⫽⫹1), and the enhanced utility
value code compared whether the enhanced utility value condition
was better than the utility value condition (utility value ⫽⫺1,
enhanced utility value ⫽⫹1). We did not include the covariates in
Model 1 because there were no differences between experimental
groups on the covariates, which meant that we could avoid the
additional assumptions required when including covariates
(Rosenbaum et al., 2002). However, adding the Time 1 motivation
covariates (expectancy, value, cost) and connection frequency to
the regression models predicting final exam scores and Time 3
interest did not alter the pattern of effects or statistical significance
on either final exam scores or interest (see Model 3 in Tables S3
and S4).
Second, to examine whether the intervention was more effective
for students at-risk of poor performance, we tested interactions
between the intervention contrast codes and initial exam scores
and gender in Model 2. Based on prior research (e.g., Hulleman &
Harackiewicz, 2009; Hulleman et al., 2010) and Study 1, we
identified male students who performed poorly on the first exam as
most at-risk. Therefore, we added initial exam scores (mean-
centered) as a continuous variable, gender (0 ⫽male, 1 ⫽female),
four two-way interaction terms (utility contrast by gender, utility
contrast by initial exams, enhanced contrast by gender, enhanced
contrast by initial exams), and two three-way interaction terms to
our model (utility contrast by gender by initial exams, enhanced
contrast by gender by initial exams). This became Model 2.
Using the methods outlined by Aiken and West (1991), we
probed significant interactions in three ways. First, we calculated
predicted values at one standard deviation above and below the
continuous moderator (in this case initial exam scores) by using
the regression equation that contains the continuous variable. Sec-
ond, we calculated simple slopes at one standard deviation above
and below the continuous moderator to test for significant differ-
ences. Third, we calculated standardized mean differences be-
tween conditions at one standard deviation above and below the
continuous predictor by estimating a regression equation with
standardized predictors and standardized outcomes. Finally, to aid
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8HULLEMAN, KOSOVICH, BARRON, AND DANIEL
interpretation of predicted values, we standardized the dependent
variable when calculating predicted values so that they can be
interpreted on a standardized metric.
Intent-to-treat analyses on learning outcomes. The analyses
of Model 1 (see Tables S3 and S4) revealed a significant effect of
the utility contrast on both final exam scores (⫽.12, sr
2
⫽.01,
p⫽.03) and interest (⫽.11, sr
2
⫽.01, p⫽.04). Students
randomly assigned to either utility condition performed better on
the final exam (M⫽80.3) and were more interested in psychology
at the end of the course (M⫽6.14) compared with students in the
control condition (M
Exam
⫽77.96, d⫽.25; M
Interest
⫽5.85, d⫽
.24). The difference between the utility and enhanced utility con-
dition was not significant for either performance or interest.
At-risk student analyses. The analyses of Model 2 (see Ta-
bles S3 and S4) revealed a significant utility value contrast by
initial exam interaction on both final exam scores (⫽⫺.22,
sr
2
⫽.02, p⫽.01) and interest (⫽1.11, sr
2
⫽.01, p⫽.03). As
presented in Figure 2, when compared with students in the control
condition, low-performing students in the utility value conditions
performed better on the final exam (d⫽.82) and were more
interested in the course material at the end of the semester (d⫽
.13). Initial exam scores were also a significant predictor of final
exam scores (⫽.60, sr
2
⫽.10). Female students also reported
more interest in psychology at the end of the semester than male
students (⫽.32, sr
2
⫽.10). Importantly, the enhanced utility
value contrast was nonsignificant, indicating that there was no
additional benefit of the enhanced utility value condition above
and beyond the regular utility value condition (see Tables S3 and
S4 for complete regression results).
The two-way interaction between the utility value contrast and
initial performance on final exam performance was qualified by a
significant three-way interaction between gender, initial perfor-
mance, and the utility value contrast (⫽.19, sr
2
⫽.01). As
presented in Figure 3, the benefits of the utility value intervention
appeared for low-performing male students who increased their
exam performance by over three-quarters of a standard deviation
in the utility value conditions compared with the control condition
(d⫽.76). By the final exam, male students in the utility value
conditions were performing as well as female students in the
control conditions, which was equivalent to going fromaCtoaB
-1.5
-1
-0.5
0
0.5
1
1.5
Low Exam High Exam
Final Exam Score
Inial Exam Scores
Control Ulity
-0.2
-0.1
0.0
0.1
0.2
Low Exam High Exam
Psychology Interest
Inial Exam Scores
Control Ulity
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Low Exam High Exam
Time 3 Expectancy
Inial Exam Scores
Control Ulity
Figure 2. Interaction between the utility value interventions and initial exam scores on final exam scores, Time
3 interest in psychology, and Time 3 success expectancies in Study 2. N⫽357. Predicted values for Low and
High Exam were computed based on estimates for one standard deviation below (Low Exam) and above the
mean (High Exam) on Initial Exam Scores (Aiken & West, 1991). We calculated standardized mean differences
between the utility value and control conditions by using predicted values from a regression equation in which
the outcome variables were standardized. Doing so results in predicted values that are in standardized units of
both the predictor and the criterion (Cohen, Cohen, West, & Aiken, 2002). For example, in the upper left panel
of this Figure, students with low expectancies in the control group had a predicted value of ⫺1.22 on final exam
scores, whereas low expectancy students in the combined utility value conditions had a predicted value of ⫺0.41
on final exam scores. These values produce an adjusted, standardized mean difference of d⫽0.82. See Model
2 in Tables S3 and S4 for complete details.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
9
UTILITY VALUE INTERVENTION IN THE CLASSROOM
(see Figure 4). In other words, the students who traditionally
perform most poorly in general education courses, males who
initially struggle in the course, benefitted the most from the inter-
vention. Importantly, the inclusion of the motivation covariates did
not change the results (see Model 3 in Tables S3 and S4). The
utility value conditions did not significantly affect high-
performing students’ exam scores, regardless of whether they were
males or females, or low-performing females.
Indirect effects analyses. We also examined whether changes
in connection frequency and expectancy-value-cost motivation
were induced by the utility value interventions, and whether these
changes contributed to further motivation and learning in the
course. We also hypothesized that expectancy and perceptions of
utility value might contribute to learning outcomes. We used path
modeling within a multiple regression framework for two reasons.
First, it matches the regression framework we used to analyze the
intent-to-treat effects. Second, this was the same technique used in
Study 1, with two exceptions. First, because we measured connec-
tion frequency at two time points, we first examined whether the
interventions increased students’ reports of connection frequency
by regressing Time 3 connection frequency on the contrast codes,
gender, initial exam scores, initial interest, and initial motivation
(see Table S5 for regression results). Second, to test whether the
motivation measures were pathways for the intervention effects,
we included Time 3 measures of expectancy, value, cost, and
connection frequency in the regression models predicting interest
and final exam scores (see Model 4 in Tables S3 and S4).
Regressing Time 3 connection frequency on the Time 1 cova-
riates and contrast codes for conditions revealed that initial interest
in psychology (⫽.17, sr
2
⫽.01), Time 1 connection frequency
(⫽.35, sr
2
⫽.09), and initial exam scores (⫽.10, sr
2
⫽.01)
were the only significant predictors. In terms of motivation, al-
though the utility value interventions did not impact utility value or
cost, they did impact students’ success expectancies. When pre-
dicting Time 3 expectancies, there was a significant interaction
between the utility value contrast and initial exam scores
(⫽⫺.11, sr
2
⫽.01). As presented in Figure 2, the utility value
conditions increased low-performing students’ expectancies com-
pared with the control condition (d⫽.20), whereas there was no
effect for high-performing students (d⫽⫺.04). Time 3 expec-
tancy in turn was a significant predictor of final exam scores (⫽
.21, p⬍.01, sr
2
⫽.02). As presented in the top panel of Figure 5,
low-performing male students in the utility conditions had higher
exam scores than their counterparts in the control condition (d⫽
.76), which was partially explained by an increase in expectancy
for students who performed poorly on the first exam (⫽.15,
95% CI [0.019, 0.322]). Importantly, these interaction effects were
unaffected by the inclusion of expectancy in the regression model
predicting final exam scores.
Although perceived utility value, cost, and connection fre-
quency could not provide indirect pathways for the intervention
effect (because they were not predicted by the intervention), the
motivation variables were significant predictors of outcomes. Be-
cause we controlled for Time 1 measures of motivation and inter-
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Low Exam High Exam
serocS maxE laniF dezidradnatS
Females
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Low Exam High Exam
Males
Control
Ulity
Figure 3. Three-way interaction between gender, initial exam scores, and utility conditions on final exam
scores in Study 2.N⫽357. Predicted values for Low and High Exam were computed based on estimates for
one standard deviation below (Low Exam) and above the mean (High Exam) on First Exam scores (Aiken &
West, 1991). See Model 2 in Tables S3 and S4 for complete details.
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
1234
serocS maxE dezidradnatS
Exam
Male Control Female Control Male Utility Female Utility
Figure 4. Unadjusted final exam scores by experimental condition and
gender in Study 2. N⫽357.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
10 HULLEMAN, KOSOVICH, BARRON, AND DANIEL
est, the Time 3 measures could be considered residual (or change)
scores. As presented in the bottom panel of Figure 5, increases in
perceived cost during the semester were associated with declines
in interest during the semester (⫽⫺.10, sr
2
⫽.01), whereas
increases in perceptions of utility value (⫽.37, sr
2
⫽.05) and
connection frequency (⫽.11, sr
2
⫽.01) were associated with
increases in interest.
Intervention fidelity analyses. Although connection fre-
quency was predictive of later interest, we did not increase con-
nection frequency through our enhanced utility value manipula-
tion. Because there were no differences between the utility and the
enhanced utility conditions, this meant that the additional elements
to the enhanced utility condition had no effect on outcomes above
and beyond the utility writing. To further understand the effects of
the intervention, we analyzed the extent to which participants
responded to the intervention prompts as intended (i.e., interven-
tion fidelity; see O’Donnell, 2008). To capture intervention fidel-
ity, we identified three core components of the intervention
prompts (Nelson et al., 2012): (a) the degree to which individuals
provided satisfactory responses to their requisite prompts in all
three conditions (indicating general compliance to the prompts
across conditions); (b) the degree to which essays contained per-
sonalized connections between the material and their lives (indi-
cating responsiveness to the utility value prompts); and (c) the
degree to which individuals specified implementation intentions
(indicating responsiveness to the enhanced utility value prompts).
For the purposes of assessing intervention fidelity to the essay
prompts, independent raters were trained on a brief rubric that
contained three elements: writing quality (expected to be equal
across conditions), personalization of connections (expected to be
higher in the utility value conditions than in the control), and
implementation intentions (expected to be higher in the enhanced
utility condition compared with the control and utility condition).
Writing quality. To assess common elements of writing qual-
ity across all conditions, raters coded essays on a four-point rating
scale that included 0 (Less than a sentence), 1 (Typed incoherent
thoughts or thoughts unrelated to the topic),2(A series of clear,
unrelated sentences addressing the same topic),3(Groups of
Figure 5. Path model of intervention effects on final exam scores (top) and interest (bottom) in Study 2. N⫽
357. Values are standardized OLS regression coefficients that were statistically significant (p⬍.05). Ovals with
“(Residual)” in them are residual values having controlled for Time 1 measures. Other control variables
included: Time 1 motivation (expectancy, utility value, cost, interest), gender, and initial exam scores. See text
for details. The two-way interaction between the utility intervention and initial exam scores on interest was
significant for students with low exam scores (⫽.19), and reduced to nonsignificant (⫽.06) when the
motivation measures were included in the model. Predicted values for Low and High Exam were computed based
on estimates for one standard deviation below (Low Exam) and above the mean (High Exam) on First Exam
scores (Aiken & West, 1991).
ⴱ
The significant three-way interaction between the utility intervention, initial exam
scores, and gender on final exam scores (⫽.19, sr
2
⫽.01) revealed that low-performing males benefitted the
most from the intervention (d⫽.76).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
11
UTILITY VALUE INTERVENTION IN THE CLASSROOM
clear, related sentences addressing the same topic). Raters were
allowed to use half points. Rater reliability was assessed using
adjacent percent agreement, or the degree to which independent
ratings were less than one point away on a rating scale (e.g., a
rating of 2 and 2.5 would be considered agreement). In the case of
disagreements, scores were averaged to compute the final rating. A
score of 2 or higher on the writing quality scale was considered to
be adequate fidelity. For writing quality, raters demonstrated 83%
adjacent agreement across all interventions. The results indicated
that the control group produced slightly lower quality essays during
the first intervention (M
control
⫽2.21, SD ⫽0.62; M
Utility
⫽2.56, SD ⫽
0.49; M
Enhancned
⫽2.57, SD ⫽0.47), whereas the utility value
condition was slightly lower during the second intervention (M
control
⫽
2.92, SD ⫽0.35; M
Utility
⫽2.66, SD ⫽0.64; M
Enhancned
⫽2.85,
SD ⫽0.39). Despite the minor differences in writing quality, we
note that all three groups received higher average ratings for the
second set of essays, and all three group averages were between
the highest points on the rating rubric. On average, 95% of essays
were rated as having adequate fidelity, and only the control con-
dition during first intervention (85%) was below 94%. These
results suggested that students tended to write acceptable essays
that addressed the prompted topics.
Personalized connections. To assess the degree to which
personalization was present in both the utility and control essays,
raters coded essays on a four-point rating scale that included 0
(Essay is not focused on the self or a significant other),1(Essay
implies or suggests personal importance, but does not say how), 2
(Essay references personal relevance rather than general),3(Es-
say references personal relevance and provides a strong example
of why). Essays were scored in the same manner as for writing
quality. Raters demonstrated 92% adjacent agreement across all
interventions. The results indicated that the control group dis-
played substantially lower personalization in their essays during
the first intervention (M
control
⫽0.04, SD ⫽0.34; M
Utility
⫽2.24,
SD ⫽0.67; M
Enhancned
⫽2.21, SD ⫽0.69; d⫽3.7 between the
control and combined utility value conditions), as well as during
the second intervention (M
control
⫽0.02, SD ⫽0.15; M
Utility
⫽
2.34, SD ⫽0.74; M
Enhancned
⫽2.50, SD ⫽0.55; d⫽3.8 between
the control and combined utility value conditions). On average,
essays were rated as having adequate fidelity 89% of the time in
the utility groups. The control essays were rated as being person-
alized 0.8% of the time, as would be expected. These differences
suggest that the hypothesized driving feature of the intervention
was present in the utility conditions, but not the control condition.
Implementation intentions. The enhanced utility value con-
dition was created by including elements of implementation inten-
tion interventions. To that end, students were asked to respond to
prompts about specific times to think about connections, about
obstacles that might prevent them from thinking about connec-
tions, and about solutions to overcoming those obstacles. To
measure fidelity of student responses in these cases, we used word
counts for each individual prompt. Raters were asked to flag
sentences that used language that could be used to answer any one
of the three implementation intention prompts (i.e., times, obsta-
cles, solutions; Rater Agreement ⫽100%). This approach was
used because or initial analyses indicated that the presence (vs.
absence) of implementation intentions was highly correlated with
elaboration. Neither the control group nor the utility value group
were coded as including implementation intentions (all mean word
counts ⬍1.55 words, median and mode word counts were 0 in both
conditions at both time points). In contrast, the enhanced utility
value prompts induced substantially more writing about imple-
mentation intentions during the first intervention (M
Times
⫽33.19,
SD ⫽25.27; M
Obstacles
⫽31.26, SD ⫽23.61; M
Solutions
⫽37.53,
SD ⫽18.68), and slightly more during the second intervention
(M
Times
⫽10.16, SD ⫽9.01; M
Obstacles
⫽9.62, SD ⫽8.01;
M
Solutions
⫽10.37, SD ⫽7.97). To have adequate fidelity, students
need to have indicated at least one time, one obstacle, and one
solution. On average, control and utility essays were rated as
having adequate fidelity 1% of the time, whereas essays in the
enhanced utility condition were rated as having adequate fidelity
50% of the time.
Fidelity results summary. These results suggested that stu-
dents generally demonstrated reasonable fidelity to the implemen-
tation intention prompts, although fidelity to the implementation
intentions aspect of the enhanced utility condition was not quite as
strong as fidelity to the personalization aspect of the regular utility
condition. See Tables S6 and S7 for more details.
Study 2 Discussion
The results of Study 2 partially replicated prior research dem-
onstrating that a theoretically guided intervention based on the
expectancy-value framework could enhance student learning out-
comes (Hulleman & Harackiewicz, 2009; Hulleman et al., 2010).
As found in prior research, the utility value intervention worked
best for students who were at-risk for poor overall course perfor-
mance. On both interest in psychology and performance, the
interaction between initial exam performance and the utility value
intervention revealed positive effects for low performers and null
effects for high performers. Furthermore, the three-way interaction
on performance revealed that male students who had performed
poorly on the first exam especially benefitted from the utility value
intervention. We also replicated the effects of Study, which dem-
onstrated that a new measure of connection frequency was a
pathway through which students developed interest in psychology
over the course of the semester. This directly supports our hypoth-
esis that an important aspect of finding value in a topic, and
eventually developing interest, is for students to make connections
between the course content and their lives.
In addition to the utility value intervention used in prior re-
search, we developed an additional intervention intended to in-
crease connection frequency. The enhanced utility value interven-
tion encouraged students to make more connections between the
psychology they were learning and their lives during their daily
routines. Unfortunately, our analyses did not reveal an additional
benefit of the enhanced condition above and beyond the utility
value condition. However, our investigation led to a surprising
finding: instead of further bolstering students’ utility value, as
found in prior research (e.g., Hulleman et al., 2010), the utility
value conditions increased students’ expectancies for success. This
was surprising because our theory and prior research had focused
on perceptions of utility value as the primary mechanism of
intervention effects. We consider these surprising findings in turn
below.
What happened in the enhanced utility value condition?
Our qualitative examination of the written responses to our inter-
vention prompt indicated that students had, for the most part,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
12 HULLEMAN, KOSOVICH, BARRON, AND DANIEL
engaged in the intervention how we had intended. Students were
prompted to make an implementation intention about when they
would make connections, identify obstacles to making connec-
tions, and identify strategies to overcome those obstacles. These
aspects of creating an intention to perform a specific behavior are
aligned with the research literature in this area (e.g., Gollwitzer,
1999). So, what are we to conclude about the enhanced interven-
tion? First, because this was our first attempt at creating this type
of intervention, it is possible that our manipulation did not ade-
quately activate behavioral commitment. The lack of effects on
self-reported connection frequency seem to support this concern.
One implication is to revise the intervention by including more
aspects intended to activate behavioral commitment. However,
instead of being an implementation issue, it is possible that our
focus is on the wrong variable. Although our connection frequency
measure may correlate with positive outcomes, it could be that
when students are prompted to make connections on their own,
that the quality of that connection also matters. Unfortunately, both
our enhanced intervention prompts and connection frequency mea-
sures were solely focused on frequency and not quality. A second
implication is to develop a different intervention that encourages
students to make more high quality connections, similar to the
types of connections that they are making when instructed to write
about relevance, and to develop a measure that captures both
quantity and quality of connections.
Why did the intervention boost success expectancies and not
utility value? One obvious reason why we found effects on
success expectancies not found in prior research is that success
expectancies had not been previously hypothesized as a pathway
of intervention effects, and thus had not been measured or ana-
lyzed in this way. However, mediation by success expectancies
seems quite plausible. Several studies of the utility value interven-
tion in both classroom-based field experiments and laboratories
have found that the version of the utility value intervention used in
this study, where students write about the connections they see as
relevant to them, is most effective in boosting performance and
interest for students with poor performance histories or low ex-
pectations (Durik et al., 2015). Thus, even though utility value was
found to be a mediator in previous studies, it is quite plausible that
the utility value intervention was working as a proxy for success
expectancies. Conventional wisdom suggests that people like what
they are good at and do better at what they like. This adage is
supported by the finding that expectancies and values, when mea-
sured via self-reports, are positively correlated (Robbins et al.,
2004). We found this to be true in both of our studies, with
expectancy and utility value being moderately correlated (rs from
.27 to .45). If this adage is true, it may be that the utility value
interventions have been affecting expectancies and utility value,
ultimately increasing both interest and performance. Yet prior
investigations did not examine whether expectancies helped ex-
plain the intervention effects.
In addition to the surprising expectancy effect, it was also
surprising that we did not find mediation of the intervention effect
attributable to utility value in Study 2. Prior laboratory and field
studies of the utility value intervention by Hulleman and col-
leagues (Hulleman et al., 2010; Hulleman & Harackiewicz, 2009)
revealed that the effects of the intervention on outcomes could be
explained, at least partially, by increases in perceptions of utility
value. In our case there are at least three possibilities why our
study might be different. First, this study was the first to manip-
ulate utility value via an online assignment, so it may be that this
difference changes how the intervention affects the mediating
motivational mechanisms. In prior classroom studies, students
either hand-wrote their responses to the intervention prompt in
notebooks (2009), or had several weeks to hand-write or type a two
to three page paper (2010, Study 2). In fact, there is some research
to suggest that the physical act of writing activates different areas
of the brain than typing the same text (Mueller & Oppenheimer,
2014). Further work is needed to examine this possible explana-
tion.
Second, contextual differences in course instructor, instructional
practices, or student characteristics could be responsible. The
small sample of contexts (this is only the third published test of
this intervention in the field) makes the influence of such contex-
tual differences difficult to discern. However, there are important
contextual aspects of this study worth noting. The instructor teach-
ing the course is known for being highly motivating, and has
received national teaching awards from The Princeton Review
(2012) and the American Psychological Association (2012). Per-
haps even more importantly for these particular findings, the
student body at his university voted him as the best professor
during 2013 (Jacobs, 2013). Evidence of his teaching prowess is
also apparent in our survey data. Students reported making more
frequent connections to psychology at the end of the semester than
at the beginning. In fact, the change in connection frequency from
the fourth to the thirteenth week was over a quarter of a standard
deviation in the control conditions (d⫽.25) and combined utility
value conditions (d⫽0.27). As a point of reference, this change
was much larger than for either expectancy or value, which both
slightly decreased over the semester (both ds⫽⫺.13). These
contextual differences could change the way the intervention af-
fects motivational dynamics. For example, with such a strong base
for value and making connections being provided by the context,
the measure of utility value may have reached a ceiling which
could not be adjusted by the intervention. In this strong value
context, the utility value intervention may have emboldened stu-
dents to believe they could succeed because they trusted the
teacher to make the content not only interesting, but also learnable.
As demonstrated in K through 12 settings, trust in school is an
important predictor of increased achievement scores (Bill & Me-
linda Gates Foundation, 2010; Bryk & Schneider, 2002), and
future research could investigate trust in relation to the utility value
intervention.
Third, it can be difficult to capture mediation for many reasons.
There are examples of other interventions which are based on
social psychological theory (e.g., growth mindsets, belonging un-
certainty, values affirmation) that have effects on outcomes but do
not always capture effects on measured variables. Self-reports are
prone to biases, such as social desirability and reference bias,
which are sensitive to context (Duckworth & Yeager, 2015). This
is particularly true for utility value as students’ responses are likely
affected by the specific unit or topic that students are studying,
which varies from week to week.
General Discussion
The results of Study 2 add to the growing body of literature that
social psychological interventions in general can promote student
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
13
UTILITY VALUE INTERVENTION IN THE CLASSROOM
learning outcomes, and that utility value interventions in particular
can be beneficial. In addition to replicating prior research on utility
value, we extended this growing body of work in several ways. To
further understand the mechanism of the utility value intervention
effect, we both measured (Studies 1 and 2) and manipulated (Study
2) one process of the utility intervention effect. This method can
provide additional validity evidence in support of the mediating
mechanism (Baron & Kenny, 1986; Sigall & Mills, 1998). Spe-
cifically, we hypothesized that the frequency with which students
made connections between the material and their lives throughout
the semester might be one way that the utility value intervention
increases motivation and performance. By continually making
connections, students might be energizing their study behavior and
integrating their knowledge in deeper ways (Bransford &
Schwartz, 1999). Although we were unable to successfully manip-
ulate this mechanism in Study 2, we found that making more
connections between the material and students’ lives was posi-
tively related to expecting to doing well in the course, and valuing
the course material. In addition, students who made more connec-
tions perceived fewer costs for learning the material. Because this
was an exploratory investigation of this measure, future research is
needed to further validate the measure and understand its relation-
ship with motivational processes and learning outcomes.
Implications for Theory and Research
At a theoretical level, the current research lends additional
support for understanding motivation and achievement in educa-
tional contexts using an expectancy-value framework. In particu-
lar, the role that expectancy and utility value both play in deter-
mining key academic outcomes within the context of interventions
was further elucidated. Although prior theoretical work allowed
for expectancies and values to be positively related, it was only
through experimental research that we learned that a utility value
intervention can actually increase expectations for success. We
also uncovered an important proximal process, or mechanism,
through which the utility value intervention has its effects: con-
nection frequency. In both studies, students who reported more
frequently seeing connections between the course material and
their lives reported more interest in the material at the end of the
semester. Further, in Study 1, this link between connection fre-
quency and outcomes was explained by a concomitant increase in
perceptions of utility value. Although this finding is correlational,
it corroborates our inclination that the frequency of connections is
an important aspect of finding value, and developing interest, in
academic content. As in past research, encouraging students to
make a connection at a single point in time through a utility value
intervention boosted utility value and learning outcomes. In ex-
tending prior work, we demonstrated that making frequent con-
nections between the material and students’ lives also boosts utility
value and learning outcomes.
Our findings are also consistent with models of interest devel-
opment (Hidi & Renninger, 2006; Renninger & Hidi, 2011), which
posit that perceiving value in a particular domain or activity is a
crucial aspect of developing an enduring interest. In addition to
providing support for the role of value in interest development
through an intervention study, we also found a new pathway for
interest development: connection frequency. Although not explic-
itly outlined in their four-phase model of interest, connection
frequency is likely related to two important factors in this model:
perceived knowledge and value. By making more connections,
students are building additional knowledge about how learning
content relates to their lives, and also creating a foundation for
perceiving personal value in a topic. Future work will need to
further develop our understanding of connection frequency and
quality in interest development. Importantly, this link to interest
development, and in particular understanding which factors are
amenable to manipulation within the classroom context, are espe-
cially important for educational practice.
Implications for Practice
On the surface, our effort to further explore the mechanisms of
the utility value intervention is a theoretical question. Why should
practitioners care why an intervention works so long as it works?
However, our work uncovered a theoretically surprising finding
related to this question: The utility value intervention, in this
context, increased low performing students’ outcomes by virtue of
enhancing success expectancies. One interpretation of this finding
is that an intervention designed to enhance value actually enhances
students’ expectancies. For practitioners, solving their local chal-
lenges involves aligning the sources of the problem (i.e., students
with low expectancies) with an intervention that targets that
source. This means looking beyond the surface features of the
intervention to understand the mechanics of how the intervention
works (Nelson, Cordray, Hulleman, Darrow, & Sommer, 2012).
In addition, this research is an example of designing an inter-
vention in an online environment to facilitate achieving both
research and practice goals. On the research end, the online deliv-
ery enabled us to randomize students behind the scenes so that
both students and the instructor were blind to differences in inter-
vention activities across students. Data entry and cleaning were
minimized, and the data were immediately available to us. We
were able to inform the instructor which students had completed
the interventions and surveys so that he could assign course credit,
and students had a common means of accessing the ‘assignment’
regardless of condition. On the practice end, the online environ-
ment minimized the impact on instructional time (students com-
pleted the intervention and surveys on their own time outside of
class). An online environment is also a cost-effective means of
scaling psychological interventions, as most undergraduate insti-
tutions use some version of course management software within
which the intervention and surveys can be imbedded.
At-Risk Students and Academic Achievement
Students disengage from school, and eventually drop out, as a
result of a number of factors, including lack of academic prepa-
ration, poor knowledge of effective learning strategies, and low
motivation (e.g., Allensworth & Easton, 2007). Poor performance
in introductory and general education courses, especially those
taken early in students’ academic career (e.g., middle school
prealgebra or college-level general education courses), can have an
especially salient impact on their academic trajectories (e.g., Casil-
las et al., 2012). Interestingly, one group that has consistently
underperformed in school has been males. A recent meta-analyses
demonstrated that females outperform males in school, and this
performance gap cuts across grade-levels, subject areas, and pub-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
14 HULLEMAN, KOSOVICH, BARRON, AND DANIEL
lication year (Voyer & Voyer, 2014), indicating that this is not a
recent phenomenon. A separate review indicated that gender dif-
ferences in intelligence, personality, and motivation partially ex-
plained this performance gap (Spinath, Eckert, & Steinmayr,
2014). In particular, girls are more self-disciplined than boys,
which leads to increased learning and academic achievement
(Duckworth & Seligman, 2006).
The results from Study 2 are consistent with this emerging work,
and provide additional evidence that the utility value intervention
helps students at risk for underperformance. In both of our studies,
male students performed more poorly than their female counter-
parts. In Study 2, the utility value intervention reduced this gap by
over 75%. These results also align with other social psychological
interventions that boost academic achievement of at-risk student
groups (e.g., Aronson, Fried, & Good, 2002; Cohen, Garcia, Apfel,
& Master, 2006; Walton & Cohen, 2011). For example, the utility
value intervention has boosted the performance of students who
initially doubted their ability to succeed in high school science
(Hulleman & Harackiewicz, 2009) and undergraduate science
(Hulleman, An, Hendricks, & Harackiewicz, 2007), first-
generation under-represented minority students in college biology
(Harackiewicz et al., 2015), and underperforming undergraduate
psychology students (Hulleman et al., 2010). Further research will
need to determine whether the gender effect was simply a proxy
for identifying low-performing students, or whether it identified an
important difference between male and female students in partic-
ular.
Limitations and Future Directions
There are some important limitations to this study, including
those that are common to field experiments, such as studying
intervention effects within a single context, which necessarily
constrains generalizations about effectiveness (cf. Shadish, Cook,
& Campbell, 2002; Shavelson, Phillips, Towne, & Feuer, 2003). In
our two studies, four limitations stand out as particularly important
for future research. First, our measure of connection frequency,
which was newly developed for this study and central to our
research questions about the pathways of the utility value inter-
vention effects, needs further validation. How does it correlate
with other measures of motivation, and how sensitive is it to
differences in teaching style or learning content? Furthermore, the
measure does not capture the quality of connections that students
make. It is highly likely that students who make high-quality
connections will benefit more than students who make low-quality
connections to their lives. There are at least two possible reasons
for this. From a motivation perspective, making higher quality
connections could deepen students’ desire to digest the material
and engage in learning. From a neuroscience perspective, research
shows that new experiences can be associated with existing mem-
ories when these experiences are strongly activated (e.g., Mc-
Gaugh, 2000). When two neural pathways are activated in tandem,
the intensity in activation can trigger a process known as long term
potentiation (Purves et al., 2001). Long-term potentiation leads to
new synaptic connections between the two pathways, resulting in
the two pathways being activated together in future experiences of
either. This would mean that making a connection between course
content and a common daily experience (e.g., working at a job)
could then lead to the activation of that content whenever the daily
experience reoccurs. Regardless of the underlying reason, the
importance of quantity and quality of connections could be exam-
ined in future research. For example, student essays in response to
the utility value intervention prompts could be coded for quality of
connections, and then linked to outcomes (e.g., Hulleman &
Cordray, 2009). Alternatively, an accompanying student self-
report measure of connection quality could be assessed so that the
independent effects of connection quantity and quality could be
examined.
The final three limitations are related to the possible reasons
why the intervention effect sizes in our experimental study were
not enhanced as we had hoped. Second, because of the potential
influence of classroom contextual factors (e.g., instructor, instruc-
tional practices, peer norms), future research needs to be con-
ducted within a wide variety of classrooms and instructional styles.
Ideally, intervention effects would be examined in enough class-
rooms to examine between classroom differences in the effective-
ness of the utility value intervention. Not every intervention will
replicate in every context, whether due to implementation chal-
lenges or other issues (e.g., Dee, 2015). Thus, the need for inde-
pendent replication work is necessary for the utility intervention
just as it is for any other intervention in education contexts.
Third, this was the first time we used an online medium to
deliver the intervention, and we adapted the intervention so as to
reduce the time burden on students. As a result, intervention
dosage, and quite possibly intervention strength, was reduced
compared with prior studies of the utility value intervention. In this
study, students were asked to write less text and to do it less
frequently than prior versions of the utility value intervention. For
example, in the previous college psychology study published by
Hulleman et al. (2010), students completed two take-home essays
of 1 to 2 pages, compared with two online essays of 2 to 3
paragraphs. Thus, both of these factors (highly motivating class-
room context and intervention strength) may have conspired to
mute the salience of the intervention on students’ perceptions of
utility value. However, despite these implementation changes, the
overall effect sizes on learning outcomes in this study (ds from .23
to .24) were similar to other value interventions (Lazowski &
Hulleman, 2015). Future research could systematically vary these
implementation factors to identify necessary levels of dosage and
frequency required to obtain utility value intervention effects.
Fourth, the control condition used in Study 2, as well as in the
other published randomized field experiments of the utility value
intervention (Harackiewicz et al., 2015; Hulleman & Harackie-
wicz, 2009; Hulleman et al., 2010), was not a do-nothing control
group. Rather, the control condition consisted of asking students to
summarize the material they had been learning recently in the
course. In the research literature on the cognitive psychology of
learning, this control condition is known as summarization and has
been found to enhance learning (Dunlosky et al., 2013). Essen-
tially, these studies (including ours) tested a motivation interven-
tion in comparison to a cognitive intervention, and found that the
motivation intervention produced better outcomes. This means that
the effect size for the utility value intervention, both in the research
presented here as well as prior published work, likely has been
underestimated because the comparison group contained a cogni-
tive intervention. To obtain a more pure effect size, future research
could use a more inert comparison group.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
15
UTILITY VALUE INTERVENTION IN THE CLASSROOM
Conclusion
The methods in this study demonstrate the value of experimental
tests of psychological theories. Without such intervention studies,
we would know very little about what happens in classrooms when
we try to enhance student motivation (cf. Shavelson et al., 2003).
Our combined longitudinal and experimental approach provides
initial validity evidence for the role of connection frequency and
motivation in explaining utility value intervention effects. More
generally, this research contributes additional validity evidence to
the growing body of research related to the impact of social-
psychological interventions on educational outcomes (Lazowski &
Hulleman, 2015; Yeager & Walton, 2011). When such theoreti-
cally guided interventions are thoughtfully implemented within
academic contexts, surprisingly strong and consistent effects have
been found on interest in academic topics, course performance,
and persistence (Wilson, 2006). Importantly, this research demon-
strates that these effects are not “magic,” but rather rely on targeted
psychological mechanisms that can gain influence over time (Co-
hen, Garcia, Purdie-Vaughns, Apfel, & Brzustoski, 2009; Garcia &
Cohen, 2011). The hopeful message is that, by engineering the
psychological situation, educational practitioners can significantly
impact student learning and development.
References
Acee, T. W., & Weinstein, C. E. (2010). Effects of a value-reappraisal
intervention on statistics students’ motivation and performance. Journal
of Experimental Education, 78, 487–512. http://dx.doi.org/10.1080/
00220970903352753
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Newbury Park, CA: Sage.
Allensworth, E. M., & Easton, J. Q. (2007). What matters for staying on
track and graduating in Chicago Public High Schools. Chicago, IL:
Consortium on Chicago School Research. Retrieved December 17, 2007.
American Psychological Association. (2012). Psychology’s top honors:
Div. 2 (Society for Teaching of Psychology). Robert S. Daniel Teaching
Excellence Award (four-year college): David B. Daniel, PhD. Retrieved
from http://www.apa.org/monitor/2012/09/top-honors.aspx
Ames, C. (1992). Classrooms: Goals, structures, and student motivation.
Journal of Educational Psychology, 84, 261–271. http://dx.doi.org/10
.1037/0022-0663.84.3.261
Aronson, J., Fried, C. B., & Good, C. (2002). Reducing the effects of
stereotype threat on African-American college students by shaping the-
ories of intelligence. Journal of Experimental Social Psychology, 38,
113–125. http://dx.doi.org/10.1006/jesp.2001.1491
Ash, K. (2008). Promises of money meant to heighten student motivation.
Education Week. Retrieved on February, 14, 2008.
Atkinson, J. W. (1957). Motivational determinants of risk-taking behavior.
Psychological Review, 64, 359 –372. http://dx.doi.org/10.1037/
h0043445
Barney, S. T. (2007). Capitalizing on the self-references effect in general
psychology: A preliminary study. Journal of Constructivist Psychology,
20, 87–97. http://dx.doi.org/10.1080/10720530600992915
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
51, 1173–1182. http://dx.doi.org/10.1037/0022-3514.51.6.1173
Barron, K. E., & Harackiewicz, J. M. (2001). Achievement goals and
optimal motivation: Testing multiple goal models. Journal of Personal-
ity and Social Psychology, 80, 706 –722. http://dx.doi.org/10.1037/0022-
3514.80.5.706
Barron, K. E., & Hulleman, C. S. (2015). Expectancy-value-cost model of moti-
vation. In J. D . Wright (Ed.), International encyclopedia of the social &
behavioral sciences (2nd ed., Vol. 8, pp. 503–509). Oxford, UK:
Elsevier Ltd. http://dx.doi.org/10.1016/B978-0-08-097086-8.26099-6
Bill & Melinda Gates Foundation. (2010). Learning about teaching: Initial
findings from the Measures of Effective Teaching Project. Research
paper retrieved from http://www.metproject.org/downloads/
Preliminary_Findings-Research_Paper.pdf
Boekaerts, M. (2002). Motivation to learn. Educational Practice Series #10.
International Bureau of Education. Bellegarde, France: SADAG. Re-
trieved from http://www.ibe.unesco.org/fileadmin/user_upload/archive/
publications/EducationalPracticesSeriesPdf/prac10e.pdf
Bong, M. (2001). Role of self-efficacy and task-value in predicting college
students’ course performance and future enrollment intentions. Contem-
porary Educational Psychology, 26, 553–570. http://dx.doi.org/10.1006/
ceps.2000.1048
Bransford, J. D., & Schwartz, D. L. (1999). Rethinking transfer: A simple
proposal with multiple implications. Review of Research in Education,
24, 61–100.
Brophy, J. (1999). Toward a model of the value aspects of motivation in
education: Developing appreciation for particular learning domains and
activities. Educational Psychologist, 34, 75– 85. http://dx.doi.org/10
.1207/s15326985ep3402_1
Brown, E. R., Smith, J. L., Thoman, D. B., Allen, J. M., & Muragishi, G.
(2015). From bench to bedside: A communal utility value intervention to
enhance students’ biomedical science motivation. Journal of Educa-
tional Psychology, 107, 1116 –1135. http://dx.doi.org/10.1037/
edu0000033
Bryk, A., & Schneider, B. (2002). Trust in schools: A core resource for
improvement. New York, NY: Russell Sage Foundation.
Burke, L. A., & Hutchins, H. M. (2007). Training transfer: An integrative
literature review. Human Resource Development Review, 6, 263–296.
http://dx.doi.org/10.1177/1534484307303035
Casillas, A., Robbins, S., Allen, J., Kuo, Y. L., Hanson, M. A., &
Schmeiser, C. (2012). Predicting early academic failure in high school
from prior academic achievement, psychosocial characteristics, and be-
havior. Journal of Educational Psychology, 104, 407– 420. http://dx.doi
.org/10.1037/a0027180
Cohen, G. L., Garcia, J., Apfel, N., & Master, A. (2006). Reducing the
racial achievement gap: A social-psychological intervention. Science,
313, 1307–1310. http://dx.doi.org/10.1126/science.1128317
Cohen, G. L., Garcia, J., Purdie-Vaughns, V., Apfel, N., & Brzustoski, P.
(2009). Recursive processes in self-affirmation: Intervening to close the
minority achievement gap. Science, 324, 400 – 403. http://dx.doi.org/10
.1126/science.1170769
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2002). Applied multiple
regression/correlation analyses for the behavioral sciences (3rd ed.).
London, UK: Routledge.
Conley, A. M. (2012). Patterns of motivation beliefs: Combining achieve-
ment goal and expectancy-value perspectives. Journal of Educational
Psychology, 104, 32– 47. http://dx.doi.org/10.1037/a0026042
Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience.
New York, NY: Harper & Row.
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-
determination in human behavior. New York, NY: Plenum Press. http://
dx.doi.org/10.1007/978-1-4899-2271-7
Dee, T. S. (2015). Social identity and achievement gaps: Evidence from an
affirmation intervention. Journal of Research on Educational Effective-
ness, 8, 149 –168. http://dx.doi.org/10.1080/19345747.2014.906009
Duckworth, A. L., & Seligman, M. E. (2006). Self-discipline gives girls the
edge: Gender in self-discipline, grades, and achievement test scores.
Journal of Educational Psychology, 98, 198 –208. http://dx.doi.org/10
.1037/0022-0663.98.1.198
Duckworth, A. L., & Yeager, D. S. (2015). Measurement matters: Assess-
ing personal qualities other than cognitive ability for educational pur-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
16 HULLEMAN, KOSOVICH, BARRON, AND DANIEL
poses. Educational Researcher, 44, 237–251. http://dx.doi.org/10.3102/
0013189X15584327
Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham,
D. T. (2013). Improving students’ learning with effective learning tech-
niques: Promising directions from cognitive and educational psychol-
ogy. Psychological Science in the Public Interest, 14, 4 –58. http://dx
.doi.org/10.1177/1529100612453266
Durik, A. M., & Harackiewicz, J. M. (2007). Different strokes for different
folks: How individual interest moderates the effects of situational factors
on task interest. Journal of Educational Psychology, 99, 597– 610.
http://dx.doi.org/10.1037/0022-0663.99.3.597
Durik, A. M., Hulleman, C. S., & Harackiewicz, J. M. (2015). One size fits
some: Instructional enhancements to promote interest don’t work the
same for everyone. In K. A. Renninger, M. Nieswandt, & S. Hidi (Eds.),
Interest in mathematics and science learning (pp. 49 – 62). Washington,
DC: American Educational Research Association.
Durik, A. M., Vida, M., & Eccles, J. S. (2006). Task values and ability
beliefs as predictors of high school literacy choices: A developmental
analysis. Journal of Educational Psychology, 98, 382–393. http://dx.doi
.org/10.1037/0022-0663.98.2.382
Eccles, J. S., Adler, T. F., Futterman, R., Goff, S. B., Kaczala, C. M.,
Meece, J. L., & Midgley, C. (1983). Expectancies, values, and academic
behaviors. In J. T. Spence (Ed.), Achievement and achievement motiva-
tion (pp. 74 –146). San Francisco, CA: Freeman.
Eccles, J. S., & Wigfield, A. (1995). In the mind of the actor: The structure
of adolescents’ achievement task values and expectancy-related beliefs.
Personality and Social Psychology Bulletin, 21, 215–225.
Flake, J., Barron, K. E., Hulleman, C. S., McCoach, D. B., & Welsh, M. E.
(2015). Understanding cost: The Forgotten component of expectancy-
value theory. Contemporary Educational Psychology, 41, 232–244.
Garcia, J., & Cohen, G. L. (2011). A social psychological approach to
educational intervention. In E. Shafir (Ed.), Behavioral foundations of
policy. Princeton, NJ: Princeton University Press.
Gentner, D., Loewenstein, J., & Thompson, L. (2003). Learning and
transfer: A general role for analogical encoding. Journal of Educational
Psychology, 95, 393– 405.
Gick, M. L., & Holyoak, K. J. (1980). Analogical problem solving.
Cognitive Psychology, 12, 306 –355. http://dx.doi.org/10.1016/0010-
0285(80)90013-4
Gollwitzer, P. M. (1999). Implementation intentions: Strong effects of
simple plans. American Psychologist, 54, 493–503. http://dx.doi.org/10
.1037/0003-066X.54.7.493
Gollwitzer, P. M., & Brandstätter, V. (1997). Implementation intentions
and effective goal pursuit. Journal of Personality and Social Psychol-
ogy, 73, 186 –199. http://dx.doi.org/10.1037/0022-3514.73.1.186
Grays, M. P. (2013). Measuring motivation for coursework across the
academic career: A Longitudinal invariance study. [Unpublished doc-
toral dissertation.] James Madison University, Harrisonburg, VA.
Harackiewicz, J. M., Barron, K. E., Tauer, J. M., & Elliot, A. J. (2002).
Predicting success in college: A longitudinal study of achievement goals
and ability measures as predictors of interest and performance from
freshman year through graduation. Journal of Educational Psychology,
94, 562–575. http://dx.doi.org/10.1037/0022-0663.94.3.562
Harackiewicz, J. M., & Borman, G. (2014, April). Scaling up social
psychological interventions to address achievement gaps in education.
Symposium at the annual conference of the American Educational
Research Association. San Francisco, CA.
Harackiewicz, J. M., Canning, E. A., Tibbetts, Y., Giffen, C. J., Blair, S. S., Rouse,
D. I., & Hyde, J. S. (2014). Closing the social class achievement gap for
first-generation students in undergraduate biology. Journal of Educational Psy-
chology, 106, 375–389. http://dx.doi.org/10.1037/a0034679
Harackiewicz, J. M., Canning, E. A., Tibbetts, Y., Priniski, S. J., & Hyde, J. S.
(2015). Closing achievement gaps with a utility-value intervention: Disentan-
gling race and social class. Journal of Personality and Social Psychology.
Advance online publication. http://dx.doi.org/10.1037/pspp0000075
Harackiewicz, J. M., Durik, A. M., Barron, K. E., Linnenbrink-Garcia, L.,
& Tauer, J. M. (2008). The role of achievement goals in the development
of interest: Reciprocal relations between achievement goals, interest, and
performance. Journal of Educational Psychology, 100, 105–122. http://
dx.doi.org/10.1037/0022-0663.100.1.105
Harackiewicz, J. M., Rozek, C. R., Hulleman, C. S., & Hyde, J. S. (2012).
Helping parents to motivate adolescents in mathematics and science: An
experimental test of a utility-value intervention. Psychological Science,
23, 899 –906. http://dx.doi.org/10.1177/0956797611435530
Harackiewicz, J. M., Tibbetts, Y., Canning, E., & Hyde, J. S. (2014).
Harnessing values to promote motivation in education. In S. A. Kara-
benick & T. C. Urdan (Eds.), Advances in motivation and achievement
(Vol. 18, pp. 71–105). Bingley, UK: Emerald Publishing. http://dx.doi
.org/10.1108/S0749-742320140000018002
Hidi, S., & Harackiewicz, J. M. (2000). Motivating the academically
unmotivated: A critical issue for the 21st century. Review of Educational
Research, 70, 151–179. http://dx.doi.org/10.3102/00346543070002151
Hidi, S., & Renninger, K. A. (2006). The four-phase model of interest
development. Educational Psychologist, 41, 111–127. http://dx.doi.org/
10.1207/s15326985ep4102_4
Hulleman, C. S., An, B., Hendricks, B., & Harackiewicz, J. M. (2007,
June). Interest development, achievement, and continuing motivation:
The pivotal role of utility value. Poster presented at the Institute of
Education Sciences Research Conference, Washington, DC.
Hulleman, C. S., & Cordray, D. S. (2009). Moving from the lab to the field:
The Role of fidelity and achieved relative intervention strength. Journal
of Research on Educational Effectiveness, 2, 88 –110. http://dx.doi.org/
10.1080/19345740802539325
Hulleman, C. S., Durik, A. M., Schweigert, S., & Harackiewicz, J. M.
(2008). Task values, achievement goals, and interest: An integrative
analysis. Journal of Educational Psychology, 100, 398 – 416. http://dx
.doi.org/10.1037/0022-0663.100.2.398
Hulleman, C. S., Godes, O., Hendricks, B. L., & Harackiewicz, J. M.
(2010). Enhancing interest and performance with a utility value inter-
vention. Journal of Educational Psychology, 102, 880 – 895. http://dx
.doi.org/10.1037/a0019506
Hulleman, C. S., & Harackiewicz, J. M. (2009). Promoting interest and
performance in high school science classes. Science, 326, 1410 –1412.
http://dx.doi.org/10.1126/science.1177067
Jacobs, P. (2013, October 1). America’s hottest professor is more than just
a pretty face—Here’s why students are crazy about his class. Business
Insider. Retrieved from http://www.businessinsider.com/interview-with-
americas-hottest-professor-david-daniel-2013-9
Johnson, M. L., & Sinatra, G. M. (2013). Use of task-value instructional
inductions for facilitating engagement and conceptual change. Contem-
porary Educational Psychology, 38, 51– 63. http://dx.doi.org/10.1016/j
.cedpsych.2012.09.003
Kohn, A. (1999). Punished by rewards. Boston, Massachusetts: Houghton
Mifflin.
Kosovich, J. J., & Hulleman, C. S. (2016). A utility value framework:
Task-goal relevance in achievement motivation. Manuscript under re-
view.
Kosovich, J. J., Hulleman, C. S., Barron, K. E., & Getty, S. (2015). A
practical measure of student motivation: Establishing validity evidence
for the expectancy-value-cost scale in middle school. The Journal of
Early Adolescence, 35, 790 – 816.
Lazowski, R. A., & Hulleman, C. S. (2015). Motivation interventions in
education: A meta-analytic review. Review of Educational Research, 86,
602– 640.
McGaugh, J. L. (2000). Memory—A century of consolidation. Science,
287, 248 –251. http://dx.doi.org/10.1126/science.287.5451.248
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
17
UTILITY VALUE INTERVENTION IN THE CLASSROOM
Mueller, P. A., & Oppenheimer, D. M. (2014). The pen is mightier than the
keyboard: Advantages of longhand over laptop note taking. Psychological
Science, 25, 1159 –1168. http://dx.doi.org/10.1177/0956797614524581
Nelson, M. C., Cordray, D. S., Hulleman, C. S., Darrow, C. L., & Sommer,
E. C. (2012). A procedure for assessing intervention fidelity in experi-
ments testing educational and behavioral interventions. The Journal of
Behavioral Health Services & Research, 39, 374 –396. http://dx.doi.org/
10.1007/s11414-012-9295-x
Newby, T. J. (1991). Classroom motivation: Strategies of first-year teach-
ers. Journal of Educational Psychology, 83, 195–200. http://dx.doi.org/
10.1037/0022-0663.83.2.195
O’Donnell, C. L. (2008). Defining, conceptualizing, and measuring fidelity
of implementation and its relationship to outcomes in K–12 curriculum
intervention research. Review of Educational Research, 78, 33– 84.
Paunesku, D., Walton, G. M., Romero, C., Smith, E. N., Yeager, D. S., &
Dweck, C. S. (2015). Mind-set interventions are a scalable treatment for
academic underachievement. Psychological Science, 26, 784 –793.
http://dx.doi.org/10.1177/0956797615571017
Peters, D. L., & Daly, S. R. (2013). Returning to graduate school: Expec-
tations of success, values of the degree, and managing the costs. The
Journal of Engineering Education, 102, 244 –268. http://dx.doi.org/10
.1002/jee.20012
Pike, A. G., & Dunne, M. (2011). Student reflections on choosing to study
science post-16. Cultural Studies of Science Education, 6, 485–500.
http://dx.doi.org/10.1007/s11422-010-9273-7
Purves, D., Augustine, G. J., Fitzpatrick, D., Katz, L. C., LaMantia, A. S.,
McNamara, J. O.,...Williams, S. M. (2001). Plasticity of mature
synapses and circuits. In D. Purves, G. J. Augustine, D. Fitzpatrick
(Eds.), Neuroscience (2nd ed.). Sunderland, MA: Sinauer Associates.
Retrieved from http://www.ncbi.nlm.nih.gov/books/NBK10878/
Renninger, K. A., & Hidi, S. (2011). Revisiting the conceptualization,
measurement, and generation of interest. Educational Psychologist, 46,
168 –184.
Renninger, K., & Wozniak, R. H. (1985). Effect of interest on attentional
shift, recognition, and recall in young children. Developmental Psychol-
ogy, 21, 624 – 632. http://dx.doi.org/10.1037/0012-1649.21.4.624
Richardson, M., Abraham, C., & Bond, R. (2012). Psychological correlates
of university students’ academic performance: A systematic review and
meta-analysis. Psychological Bulletin, 138, 353–387. http://dx.doi.org/
10.1037/a0026838
Robbins, S. B., Lauver, K., Le, H., Davis, D., Langley, R., & Carlstrom, A.
(2004). Do psychosocial and study skill factors predict college out-
comes? A meta-analysis. Psychological Bulletin, 130, 261–288. http://
dx.doi.org/10.1037/0033-2909.130.2.261
Rosenbaum, P. R., Angrist, J., Imbens, G., Hill, J., Robins, J. M., &
Rosenbaum, P. R. (2002). Covariance adjustment in randomized exper-
iments and observational studies. Statistical Science, 17, 286 –327.
http://dx.doi.org/10.1214/ss/1042727942
Rozek, C. S., Hyde, J. S., Svoboda, R. C., Hulleman, C. S., & Harackie-
wicz, J. M. (2015). Gender differences in the effects of a utility-value
intervention to help parents motivate adolescents in mathematics and
science. Journal of Educational Psychology, 107, 195–206. http://dx.doi
.org/10.1037/a0036981
Schraw, G., & Lehman, S. (2001). Situational interest: A review of the
literature and directions for future research. Educational Psychology
Review, 13, 23–52. http://dx.doi.org/10.1023/A:1009004801455
Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and
quasi-experimental designs for generalized causal inference. Boston,
MA: Houghton Mifflin Company.
Shavelson, R. J., Phillips, D. C., Towne, L., & Feuer, M. J. (2003). On the
science of education design studies. Educational Researcher, 32, 25–28.
http://dx.doi.org/10.3102/0013189X032001025
Sigall, H., & Mills, J. (1998). Measures of independent variables and
mediators are useful in social psychology experiments: But are they
necessary? Personality and Social Psychology Review, 2, 218 –226.
http://dx.doi.org/10.1207/s15327957pspr0203_5
Simons, J., Vansteenkiste, M., Lens, W., & Lacante, M. (2004). Placing
motivation and future time perspective theory in a temporal perspective.
Educational Psychology Review, 16, 121–139. http://dx.doi.org/10
.1023/B:EDPR.0000026609.94841.2f
Spinath, B., Eckert, C., & Steinmayr, R. (2014). Gender differences in
school success: What are the roles of students’ intelligence, personality,
and motivation? Educational Research, 56, 230 –243. http://dx.doi.org/
10.1080/00131881.2014.898917
Symons, C. S., & Johnson, B. T. (1997). The self-reference effect in
memory: