Paper ID #29098

Exploratory Factor Analysis of Approaches to Teaching Inventory (ATI):

Use in an Evidence-Based Faculty Development Program for Promoting
Active Learning Pedagogical Strategies
Kristi Glassmeyer, Arizona State University
Kristi Glassmeyer is a PhD student in Educational Policy and Evaluation at Arizona State University.
Her research interests engage the intersection of organizational resources and processes with science and
engineering education for the purposes of policy implementation and educational change.
Lydia Ross, Arizona State University
Dr. Lydia Ross is a clinical assistant professor in the Mary Lou Fulton Teachers College. She also
serves as the executive director of the Association for Education Finance & Policy. She holds a PhD in
Educational Policy and Evaluation from Arizona State University. Her research focuses on equity and
access and in higher education, with a focus on STEM.
Dr. Eugene Judson, Arizona State University
Eugene Judson is an Associate Professor of for the Mary Lou Fulton Teachers College at Arizona State
University. He also serves as an Extension Services Consultant for the National Center for Women and
Information Technology (NCWIT). His past experiences include having been a middle school science
teacher, Director of Academic and Instructional Support for the Arizona Department of Education, a
research scientist for the Center for Research on Education in Science, Mathematics, Engineering and
Technology (CRESMET), and an evaluator for several NSF projects. His first research strand concentrates
on the relationship between educational policy and STEM education. His second research strand focuses
on studying STEM classroom interactions and subsequent effects on student understanding. He is a co-
developer of the Reformed Teaching Observation Protocol (RTOP) and his work has been cited more than
2200 times and he has been published in multiple peer-reviewed journals such as Science Education and
the Journal of Research in Science Teaching.
Prof. Stephen J Krause, Arizona State University

Stephen Krause is professor in the Materials Science Program in the Fulton School of Engineering at
Arizona State University. He teaches in the areas of introductory materials engineering, polymers and
composites, and capstone design. His research interests include faculty development and evaluating con-
ceptual knowledge and strategies to promote conceptual change. He has co-developed a Materials Concept
Inventory and a Chemistry Concept Inventory for assessing conceptual knowledge and change for mate-
rials science and chemistry classes. He is currently conducting research in two areas. One is studying
how strategies of engagement and feedback and internet tool use affect conceptual change and impact
on students’ attitude, achievement, and persistence. The other is on a large-scale NSF faculty develop-
ment program and its effect on change in faculty teaching beliefs, engagement strategies, and classroom
practice. Recent honors include coauthoring the ASEE Best Paper Award in the Journal of Engineering
Education in 2013 and the ASEE Mike Ashby Outstanding Materials Educator Award in 2018.

Dr. Lindy Hamilton Mayled, Arizona State University

Lindy Hamilton Mayled is the Director of Instructional Effectiveness for the Fulton Schools of Engineer-
ing at Arizona State University. She has a PhD in Psychology of Learning, Education, and Technology
and her research and areas of interest are in improving educational outcomes for STEM students through
the integration of active learning and technology-enabled frequent feedback.

American
c Society for Engineering Education, 2020
 Exploratory Factor Analysis of Approaches to Teaching
Inventory (ATI): Use in an Evidence-Based Faculty Development
Program for Promoting Active Learning Pedagogical Strategies
Abstract

While surveys/inventories can be very informative for researchers to better understand latent
constructs within social science research, critical analysis of these instruments is essential when
they are used outside of their initial contexts. This complete research paper reports on an
exploratory factor analysis of the Approaches to Teaching Inventory (ATI) as adapted for use in
measuring relational change of engineering faculty’s (N=65) instructional intent and teaching
strategies in their undergraduate engineering classes. Parallel analysis of data collected during
the JTFD professional development program, a National Science Foundation (NSF) funded
project, suggested an underlying structure of two or three factors. While the survey creators,
Trigwell and Prosser [1], claim a two-factor structure, each with two underlying subscales, in the
ATI, exploratory factor analyses global model fit suggested a three-factor model to be a better
fit. Interpretation of loading patterns and magnitudes indicated concerns with both two- and
three- factor models. Although the small sample size presents a limitation to the findings, critical
analysis of the ATI’s use in other disciplines should be considered.

Introduction

Active learning, or student-centered teaching, is a pedagogical practice where instructors

engage students directly in the learning process. This is done through in-class activities,
discussions, and fostering an interactive environment. Contrastingly, instructor-centered teaching
strategies involve transmission of information from the instructor to the students, often via
lecture. In teacher-centered classrooms, the teacher is the central focus, and students just listen to
the lecture to receive the information. Teacher-centered practices remain the dominant form of
teaching in higher education. However, there is a large body of literature supporting the efficacy
of active learning resulting in greater student comprehension and achievement [2, 3, 4]. This
research indicates students in active learning classes have improved understanding of class
content and are more likely to pass a class [5].

The context for this paper is a professional development program at Arizona State
University, an R1 institution in the southwestern United States. The JTFD program is part of the
NSF’s Improving Undergraduate Science Education initiative. The JTFD program aims to
increase faculty awareness and use of active learning strategies in the classroom. To that end,
faculty participated in a year-long professional development series. In the fall semester, faculty
attended 8 biweekly workshops, and in the spring semester they participated in six communities
of practice. As part of the program evaluation, surveys were given to understand shifts in faculty
beliefs and attitudes regarding active learning or student-centered teaching strategies. One survey
included in our evaluation strategy is the Approaches to Teaching Inventory (ATI), which is the
focus of this paper.

The Approaches to Teaching Inventory (ATI) was created based on a phenomenological

qualitative study of faculty teaching physical sciences to first year undergraduate students in
Australia [6]. Since its creation, the ATI has been adapted to various contexts in order to
relationally measure student- versus teacher-centered intent and strategies of use with regards to
faculty instruction [7]. Original analysis of the inventory utilized a principal components analysis
indicating the composite scores of the subscales including information transmission intent,
teacher-focused strategies, conceptual change intent, and student-focused strategies loaded onto
two components [8]. The inventory’s subscales of information transfer and teacher-focused
strategies were identified as loading onto a teacher-centered component while conceptual change
and student-focused strategies subscales were identified as loading onto a student-centered
component.

Considering potential correlations across the items of the ATI, we examined the ATI data
using an exploratory factor analysis. The purpose of this study was to examine the behavior of
the ATI items to provide insight into the use of its findings with regards to professional
development interventions. We examined the structure and behavior of the items of the ATI
based on the responses from engineering faculty participating in a NSF-funded professional
development program aimed at developing active-engagement and student-centered instructional
strategies. This study addresses the following research question:
1. What is the underlying factor structure of the Approaches to Teaching Inventory for
faculty of engineering disciplines at a large public university?

Methods

Participants

This study utilized self-reported responses on the ATI from engineering faculty from a
large, public university located in the southwestern United States. Individuals were participating
in a NSF-funded professional development program promoting more effective teaching and
learning practices. The JTFD professional development program began in the fall of 2016. Two
cohorts of faculty, representing different engineering disciplines each year, participated in a year-
long professional development program over two subsequent years. Numbers of participants as
well as their engineering disciplines are presented in Table 1.

Table 1

JTFD Participant Characteristics for Cohorts

Number of Faculty
Cohort Discipline
Participants
Cohort One (2015-2016) Disciplinary Leader Pairs 8
Civil 13
Construction 9
Cohort One (2016-2017)
Aerospace & Mechanical 13
Disciplinary Leader Pairs 6
Biomedical 18
Cohort Two (2017-2018) Chemical 7
Materials 8
The first cohort consisted of participants from the engineering disciplines of mechanical and
aerospace, civil, and construction engineering while the second cohort represented the
engineering disciplines of biomedical, chemical, and materials engineering.

Measures

In order to measure participants’ shifts in attitude, motivation, and use in active learning
pedagogical practices, three surveys were administered including the ATI, Education Research
Awareness & Use, and Value, Expectancy, and Cost of Testing Educational Reforms Survey
(VECTERS). The focus of this study, the ATI, consists of 16 items measuring self-perception of
teaching. The scale of each measure is a 5-point Likert-type scale ranging from 0 = “strongly
disagree” to 4 = “strongly agree.” The inventory contains items identified in a phenomenological
study of university science teachers approaches to teaching [9]. The survey is hypothesized to
contain two scales representing two fundamentally different approaches to teaching. The
inventory consists of eight transmission/teacher-focused approach items (e.g., “I feel it is
important that this subject should be completely described in terms of specific objectives relating
to what students have to know for formal assessments”) and eight conceptual change/student-
focused approach items (e.g., “In my class/tutorial for this subject I try to develop a conversation
with students about the topics we are studying”). Additionally, these two approaches are each
thought to contain two subscales of intention and strategy including teacher focused strategies
(TFS), teacher focused intention (TFI), student focused strategy (SFS), and student focused
intention (SFI). Initial analysis of the inventory utilized a principal components analysis instead
of an exploratory factor analysis [10]. The use of principal components analysis accounting for
total variance (unique, common, and error) instead of an exploratory factor analysis has been
critiqued in the literature [11].

Procedure

The professional development program, targeting research-based active learning

strategies as well as providing support and best practices of how to implement these strategies
into the classroom, included 8 workshops during the fall semester and 6 communities of practice
sessions in the spring semester. All participants were requested to take the ATI at the beginning
(pre), middle (mid), and end (post) of the professional development program. This is one of
several instruments used to measure faculty attitudes and beliefs during this research project. The
survey, administered through Qualtrics, had 65 respondents for the pre survey, 45 respondents
for the mid survey, and 49 respondents for the post survey. This measurement model study
focused on the data collected from the pre survey because it contained the greatest number of
respondents across collection times. All respondents (N=65) completed the survey in its entirety.

Analysis

Since creation of the survey was not based on prevalent theoretical findings in the
literature [12], parallel analysis, which takes sample error into account, was used to estimate the
number of factors to be retained. Use of mean eigenvalues from 1000 samples recommended a
three-factor extraction while the 95th percentile eigenvalues supported a two-factor extraction.
We used the 16-item correlation matrix as well as calculations of the determinant, Kaiser-Meyer-
Olkin measure of sampling adequacy, and Bartlett’s test of sphericity to determine initial
factorability of the data. There was no missing data because participants completed the inventory
in its entirety. In order to determine whether the underlying assumptions of an exploratory factor
analysis would be met, measures of central tendency including means, standard deviations,
skewness, and kurtosis were calculated for the 16 items. The exploratory factor analysis was
performed with MPlus with weighted least square mean and variance adjusted (WLSMV) as an
estimator because it is a robust maximum likelihood estimator recommended for use with
ordered categorical data. Correlation between items in the inventory and factors underlying the
inventory and an oblique rotation of geomin was used [13]. Model fit was evaluated using Chi-
square, root mean square error of approximation (RMSEA), confirmatory factor index (CFI), and
standardized root mean square residual (SRMR). Factor loadings on the items for each model
were also examined for model interpretability.

Results

Descriptive Statistics

Given the Likert-scale used in the inventory, the underlying assumptions of normality
were analyzed. We used descriptive statistics and calculations for skewness and kurtosis for each
of the 16 items and found several items to be positively or negatively skewed with values greater
than 1 and less than -1 indicating a violation of normality (Table 2). Due to the violations of
normality of the ordered categorical data, WLSMV was used as an estimator in the analysis.

Table 2

Means, and Standard Deviations for Approaches to Teaching Inventory Items for JTFD
Participants (N=65)
Item M SD Skewness Kurtosis
I feel it is important that this subject should be
completely described in terms of specific
1 2.92 1.035 -0.626 -0.738
objectives relating to what students have to
know for formal assessments.
I feel it is important to present a lot of facts in
2 classes so that students know what they have to 2.20 1.003 0.349 -0.943
learn for this subject/course.
In this subject/course I concentrate on covering
3 the information that might be available from a 2.11 1.161 0.031 -1.032
good textbook.
I think an important reason for running
4 teaching sessions in this subject/course is to 1.69 1.045 0.147 -1.064
give students a good set of notes.
In my class/tutorial for this subject I try to
5 develop a conversation with students about the 3.17 0.821 -0.851 0.379
topics we are studying.
 In this subject/course, I only provide the
6 students with the information they will need to 0.77 0.825 1.32 2.762
pass the formal assessments.
I encourage students to restructure their
existing knowledge in terms of the new way of
7 3.06 0.916 -0.753 -0.187
thinking about the subject that they will
develop.
I structure this subject/course to help students
8 2.34 1.035 -0.642 -0.218
to pass the formal assessment items.
I feel a lot of teaching time in this
9 subject/course should be used to question 2.17 0.945 0.108 -0.104
students’ ideas.
I feel that it is better for students in this
10 subject/course to generate their own notes 2.98 1.082 -1.037 0.428
rather than always copy mine.
I feel that I should know the answers to any
11 questions that students may put to me during 2.05 1.230 -0.194 -0.959
this subject/course.
In teaching sessions for this subject/course, I
12 use difficult or undefined examples to provoke 2.05 1.165 -0.092 -0.989
debate.
I set aside some teaching time so that the
students can discuss, among themselves, the
13 2.42 1.298 -0.428 -1.005
difficulties that they encounter in studying this
subject.
I design my teaching in this subject with the
assumption that most of the students have very
14 2.18 1.171 -0.311 -0.857
little useful knowledge of the topics to be
covered.
I feel that the assessment in this subject should
be an opportunity for students to reveal their
15 2.86 0.704 -0.352 0.277
changed conceptual understanding of the
subject.
I make available opportunities for students in
16 this subject/course to discuss their changing 2.60 1.058 -0.353 -0.429
understanding of the subject.

Initial factorability indicated the existence of a linear combination of at least one

underlying factor within the data (Table 3) [14, 15].
 Table 3

Measures for Assessing the Correlation Matrix

Measure Value
Determinant 0.01
KMO 0.627
Bartlett's Test of Sphericity p < .001

First, the determinant of the matrix was not equal to zero indicating the data matrix can
be explained by a linear combination. Second, measurement of shared variance in the items,
measured using the Kaiser-Meyer-Olkin Test of Sampling Adequacy (KMO), was 0.627
indicating a mediocre degree of common variance among items. Furthermore, Bartlett’s test
produced a significant test result, χ2(120, N = 65) = 264.57, p < .001, supporting the rejection of
the null hypothesis that the observed correlation matrix of the data is equal to the identity matrix.
Lastly, the correlation matrix of the 16 items was analyzed (Table 4). Multiple correlations
exceeded .30 supporting the existence of enough commonality between the items to justify the
existence of a common factor [16].
Table 4

Correlations for Approaches to Teaching Inventory for Engineering Faculty Participants (N = 65)
Item 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
1 1.000
2 0.346 1.000
3 0.254 0.424 1.000
4 0.166 0.283 0.543 1.000
5 0.163 -0.004 -0.265 -0.121 1.000
6 0.144 0.132 0.124 0.333 -0.034 1.000
7 -0.110 -0.048 -0.197 -0.094 0.297 -0.043 1.000
8 0.185 0.220 0.138 0.329 -0.105 0.349 -0.039 1.000
9 0.077 0.079 -0.159 -0.057 0.426 0.071 0.204 0.100 1.000
10 -0.127 -0.040 -0.210 -0.115 0.548 0.048 0.332 0.144 0.385 1.000
11 0.211 0.246 0.412 0.145 -0.132 0.041 -0.210 0.012 -0.208 -0.035 1.000
12 -0.075 0.005 -0.050 0.063 0.171 0.109 0.070 0.194 0.220 0.348 0.053 1.000
13 -0.011 -0.005 -0.248 0.050 0.285 0.208 0.241 0.126 0.184 0.327 -0.042 0.421 1.000
14 0.063 0.061 0.123 0.009 -0.196 0.287 -0.113 0.167 -0.085 0.027 0.211 -0.075 0.237 1.000
15 0.007 -0.027 -0.058 -0.080 0.311 0.052 0.255 -0.085 0.271 0.120 0.007 0.274 0.303 0.107 1.000
16 0.014 -0.071 -0.295 -0.212 0.367 0.054 0.348 0.097 0.350 0.267 -0.106 0.408 0.510 0.061 0.407 1.000
Exploratory Factor Analyses

Global fit indices supported a three-factor model as a being a better fit to the data. Chi-
square model fit statistics, χ2(62, N = 65) = 95.61, p = 0.055, (Table 5) indicated the three-factor
model does not significantly deviate from the observed data. The RMSEA, being greater than
.05, indicates adequate fit for both a two- or three-factor model. However, the comparative fit
index, .934, and the standardized root mean square residual, 0.08, support a three-factor model as
being a better fit with the observed data as opposed to the two-factor model.

Table 5

Summary of Model Fit

Model Chi-Square df RMSEA CFI SRMR
2-factor 125.13* 89 0.079 0.884 0.1
3-factor 95.61 62 0.065 0.934 0.08
Note. df = degrees of freedom; RMSEA = Root mean square error of approximation; CFI =
Confirmatory factor index; SRMR = Standardized root mean square residual; *p < .01

The rotated loadings for both the two- and three-factor models indicated the existence of
an underlying factor for the items after having controlled for the effects on other factors. The
factor structure coefficients, indicating the correlation between the factor and the items, were
compared for the two- and three-factor extractions (Table 6). The rotated and structure matrices
have identical patterns so the rotated loadings are presented for clarity purposes.
 Table 6

Rotated Loadings of Two- and Three-Factor Extractions of 16 Items in ATI with Geomin
Rotation
Extraction Two-Factor Three-Factor
Item 1 2 1 2 3
1 0.47* 0.07 0.51* 0.16 0.03
2 0.56* 0.01 0.67* 0.17 -0.05
3 0.74* -0.34* 0.83* -0.20 -0.01
4 0.70* -0.07 0.70* -0.02 0.17
5 -0.13 0.66* 0.01 0.85* -0.30
6 0.46* 0.17 0.32* -0.01 0.55*
7 -0.20 0.47* -0.19 0.47* -0.05
8 0.53* 0.23 0.41* 0.10 0.47*
9 0.00 0.58* 0.05 0.63* -0.07
10 0.00 0.66* 0.00 0.66* 0.03
11 0.34* -0.14 0.36* -0.10 0.05
12 0.20* 0.56* 0.13 0.48* 0.27
13 0.15 0.71* -0.01 0.52* 0.52*
14 0.28* 0.05 0.14 -0.15 0.51*
15 0.03 0.55* 0.00 0.50* 0.15
16 -0.07 0.76* -0.17 0.63* 0.29
Note. Factor loadings and correlations > .40 are in boldface. * p < .05

The two-factor model rotated loadings aligned well with the two-factor model upon
which Trigwell and Prosser [17] developed the ATI with items identified as information transfer
and teacher-focused (e.g., items 1, 2, 3, 4, 6, 8, 11, and 14) predominantly combining to made up
factor one. Complementarily, items identified as conceptual change and student-focused (e.g.,
items 5, 7, 9, 10, 12, 13, 15, 16) largely combined for factor two. Two items, three and twelve,
load significantly on both factors while still loading more dominantly on their intended factor in
the two-factor model. Item three states, “In this subject/course I concentrate on covering the
information that might be available from a good textbook” and loads more heavily on
information transfer/teacher-focused (factor one) while item 12 states, “In teaching sessions for
this subject/course, I use difficult or undefined examples to provoke debate” and loads more
heavily on conceptual change/student-focused (factor two). Together these two items represent
components typical of rigorous higher education classes. Additionally, item 14 loads
significantly onto the information transfer/teacher-focused factor in the two-factor extraction, it
does not exceed the traditional cutoff value of .30. Item 14 states “I design my teaching in this
subject with the assumption that most of the students have very little useful knowledge of the
topics to be covered.”

In addition to being a better fit, the presence of the third factor splits the loading of some
items onto multiple factors suggesting teaching to be more complex than just intention and
strategy (Table 7).
Table 7

Rotated Factor Loadings (Pattern Matrix) for 16-item ATI based on Exploratory Factor
Analysis with Geomin Rotation (N=65)

2-
Item Preparation Instruction Assessment
Factor
1. I feel it is important that this subject
should be completely described in terms of
ITTF 0.51* 0.16 0.03
specific objectives relating to what students
have to know for formal assessments.
2. I feel it is important to present a lot of
facts in classes so that students know what ITTF 0.67* 0.17 -0.05
they have to learn for this subject/course.
3. In this subject/course I concentrate on
covering the information that might be ITTF 0.83* -0.20 -0.01
available from a good textbook.
4. I think an important reason for running
teaching sessions in this subject/course is to ITTF 0.70* -0.02 0.17
give students a good set of notes.
5. In my class/tutorial for this subject I try to
develop a conversation with students about CCSF 0.01 0.85* -0.30
the topics we are studying.
6. In this subject/course, I only provide the
students with the information they will need ITTF 0.32* -0.01 0.55*
to pass the formal assessments.
7. I encourage students to restructure their
existing knowledge in terms of the new way
CCSF -0.19 0.47* -0.05
of thinking about the subject that they will
develop.
8. I structure this subject/course to help
ITTF 0.41* 0.10 0.47*
students to pass the formal assessment items.
9. I feel a lot of teaching time in this
subject/course should be used to question CCSF 0.05 0.63* -0.07
students’ ideas.
10. I feel that it is better for students in this
subject/course to generate their own notes CCSF 0.00 0.66* 0.03
rather than always copy mine.
11. I feel that I should know the answers to
any questions that students may put to me ITTF 0.36* -0.10 0.05
during this subject/course.
12. In teaching sessions for this
subject/course, I use difficult or undefined CCSF 0.13 0.48* 0.27
examples to provoke debate.
 13. I set aside some teaching time so that the
students can discuss, among themselves, the
CCSF -0.01 0.52* 0.52*
difficulties that they encounter in studying
this subject.
14. I design my teaching in this subject with
the assumption that most of the students have
ITTF 0.14 -0.15 0.51*
very little useful knowledge of the topics to
be covered.
15. I feel that the assessment in this subject
should be an opportunity for students to
CCSF 0.00 0.50* 0.15
reveal their changed conceptual
understanding of the subject.
16. I make available opportunities for
students in this subject/course to discuss their CCSF -0.17 0.63* 0.29
changing understanding of the subject.
Note. Rotated loadings > .4 are bolded. ITTF = information transfer/teacher-focused; CCSF =
conceptual change/student-focused, * p < .05

While the conceptual change/student-centered items from the two-factor model loaded
predominantly on a single factor in the three-factor model, the items aligned with information
transfer/teacher-focused in the two-factor model load onto two factors rather than a single factor.
Looking closer at the wording of the items as well as their alignment suggest the factors may be
more representative of an instructor’s preparation for class, instructional strategies, and
assessment strategies. Seven of the eight conceptual change/student-centered items (e.g., five,
seven, nine, ten, 12, 15, and 16) from the two-factor model load onto factor two in the three-
factor model. The items overlap in that they would take place during the class. For example,
items five, nine, and 12 engage discussions, questions, or problem-solving which would typically
take place during a class. Items one, two, three, four, and eleven with loadings of .51, .67, .83,
.70, and .36, respectively, load on the first factor. Wording of these items represent
characteristics representative of instructor’s preparation or structuring of the course. For
instance, item one references the use of objectives to communicate the learning goals of the
course. Items six, 13, and 14 with loadings of .55, .52, and .51, respectively, are more aligned
with the third factor. These items are either related to assessment or explicitly have the word
assessment within them. Although most items load onto a single factor, two items load similarly
on two factors. For example, item 13 loads onto the instruction and assessment factors similarly.
Item 13 states “I set aside some teaching time so that the students can discuss, among
themselves, the difficulties that they encounter in studying this subject.” This cross loading could
be due to the structure or word selection of the item having resulted in different interpretations
and responses from participants. Additionally, item eight states, “I structure this subject/course to
help students to pass the formal assessment items” includes terminology overlapping both
preparation as well as assessment of the course. The better fit of the three-factor model with this
data portrays a picture more aligned with teachers’ pedagogical content knowledge rather than
the teacher-focused/student-focused two-factor model used to develop the inventory.
Discussion

The better fit of the three-factor model with this data suggests that although we would
like to inventory higher education instructors’ instructional approaches and shift them to include
more active strategies, we must not disregard the pedagogical content knowledge underlying the
behaviors of all teachers. According to the three-factor model, the model has more items
representing the strategies taking place during instructional time. However, these items don’t
necessarily present the spectrum of approaches teachers (i.e., more direct instruction to more
active engagement) use during instructional time to facilitate learning. Although, this may not be
the ideal instrument to measure relational change between more teacher-focused and student-
centered approaches, it can inform us about instructors’ perceptions of teaching as well as their
beliefs and intentions regarding strategies they employ. The use of the ATI in research is
indicative of the need for an instrument providing insight into the spectrum of teaching
approaches as we look to shift instructional approaches in higher education courses.

While there were no missing data, the small sample size (N = 65) and ordered categorical
nature of the items present limitations for the findings of this study and its ability to be
generalized. The use of WLSMV as an estimator in the Mplus code should have accounted for
the categorical nature of the items but comparison with models using MLM can be performed.
Unfortunately, there is no way to increase the sample size which provides a challenge to the
interpretation of the findings. As such, future researchers should consider examining the ATI
with a new lens of three factors, rather than two. These findings also speak to the complicated
nature of teaching and measuring pedagogical practices. While it is important to try to measure
the spectrum of teaching being done in higher education courses or the change that may or may
not take place over the course of professional development opportunities, it is important to note
the limitations of the instruments we use and complement them with other measurements that
can give us a more complete picture of the impact any intervention may have. In this particular
research project, ATI was one of several instruments used to measure faculty attitudes and
beliefs regarding instruction. As such, findings from the inventory were complemented by other
instruments to shed light onto changes in faculty beliefs. These findings also speak to the
complicated nature of teaching and measuring pedagogical practices. As such, the complexities
around teaching need to be accounted for and further addressed in planning of and evaluation of
professional development programs aimed at shifting pedagogical practices.

Acknowledgments
The authors gratefully acknowledge support of this work by the National Science Foundation
under Grant No. 1524527.
 References

[1] Trigwell, K., & Prosser, M. (1996). Congruence between intention and strategy in university
science teachers’ approaches to teaching. Higher Education, 32(1), 77–87. Retrieved from
https://www.jstor.org/stable/3447897
[2] Felder, R. M., & Brent. R. (2016). Teaching & learning STEM: A practical guide. San
Francisco, CA: Jossey-Bass.
[3] Felder, R. M., & Brent, R. (1996). Navigating the bumpy road to student-centered
instruction. College teaching, 44(2), 43-47.
[4] Jungst, S., Likclider, L. L., & Wiersema, J. (2003). Providing support for faculty who wish to
shift to a learning-centered paradigm in their higher education classrooms. The Journal of
Scholarship of Teaching and Learning, 3(3), 69-81.
[5] Freeman, S., Eddy, S. L., McDonough, M., Smith, M. K., Okoroafor, N., Jordt, H., &
Wenderoth, M. P. (2014). Active learning increases student performance in science,
engineering, and mathematics. PNAS, 11(23), 8410-8415.
[6] Trigwell, K., & Prosser, M. (1996). Congruence between intention and strategy in university
science teachers’ approaches to teaching. Higher Education, 32(1), 77–87. Retrieved from
https://www.jstor.org/stable/3447897
[7] Ross, L., Judson, E., Krause, S. J., Ankeny, C. J., Culbertson, R. J., & Hjelmstad, K. D.
(2017, June). Relationships between engineering faculty beliefs & classroom practices. Paper
presented at the annual meeting of the American Society for Engineering Education (ASEE),
Columbus, OH.
[8] Trigwell, K., & Prosser, M. (1996). Congruence between intention and strategy in university
science teachers’ approaches to teaching. Higher Education, 32(1), 77–87. Retrieved from
https://www.jstor.org/stable/3447897
[9] Trigwell, K., & Prosser, M. (1996). Congruence between intention and strategy in university
science teachers’ approaches to teaching. Higher Education, 32(1), 77–87. Retrieved from
https://www.jstor.org/stable/3447897
[10] Trigwell, K., & Prosser, M. (1996). Congruence between intention and strategy in
university science teachers’ approaches to teaching. Higher Education, 32(1), 77–87.
Retrieved from https://www.jstor.org/stable/3447897
[11] Meyer, J. H. F., & Eley, M. G. (2006). The Approaches to Teaching Inventory: A critique
of its development and applicability. British Journal of Educational Psychology, 76, 633–
649. https://doi.org/10.1348/000709905X49908
[12] Meyer, J. H. F., & Eley, M. G. (2006). The Approaches to Teaching Inventory: A critique
of its development and applicability. British Journal of Educational Psychology, 76, 633–
649. https://doi.org/10.1348/000709905X49908
[13] Beavers, A. S., Lounsbury, J. W., Richards, J. K., Huck, S. W., Skolits, G. J., & Esquivel,
S. L. (2013). Practical considerations for using exploratory factor analysis in educational
research. Practical Assessment, Research & Evaluation, 18(6), 1–13. Retrieved from
http://pareonline.net/getvn.asp?v=18&n=6
[14] Beavers, A. S., Lounsbury, J. W., Richards, J. K., Huck, S. W., Skolits, G. J., & Esquivel,
S. L. (2013). Practical considerations for using exploratory factor analysis in educational
research. Practical Assessment, Research & Evaluation, 18(6), 1–13. Retrieved from
http://pareonline.net/getvn.asp?v=18&n=6
[15] Pett, M. A., Lackey, N. R., & Sullivan, J. J. (2003). Making sense of factor analysis.
 Thousand Oaks, CA: Sage Publications, Inc.
[16] Beavers, A. S., Lounsbury, J. W., Richards, J. K., Huck, S. W., Skolits, G. J., & Esquivel,
S. L. (2013). Practical considerations for using exploratory factor analysis in educational
research. Practical Assessment, Research & Evaluation, 18(6), 1–13. Retrieved from
http://pareonline.net/getvn.asp?v=18&n=6
[17] Trigwell, K., & Prosser, M. (1996). Congruence between intention and strategy in
university science teachers’ approaches to teaching. Higher Education, 32(1), 77–87.
Retrieved from https://www.jstor.org/stable/3447897