Professional Documents
Culture Documents
Objectives
◆ 505
506–––◆–––Issues Regarding Methods and Methodology
world) and quantum mechanics (Howard, model. Q methodology fits well into this
2005; Ramlo, 2006; Stephenson, 1982b, idea of mixed methods, although what
1988). Stephenson expanded on the work exactly that model consists of is still in
of Cyril Burt (the creator of Q factor debate (Creswell, 2010; Tashakkori &
analysis) and Burt’s and his mentor, Charles Teddlie, 2009).
Spearman, to develop Q methodology. It is important to realize that Q method-
Stephenson was able to blend the con- ology is not simply a statistical technique
cepts of factor analysis with concepts but, instead, a complete methodology
from quantum mechanics (Ramlo, 2006; (McKeown & Thomas, 1988; Stephenson,
Stephenson, 1982a, 1982b, 1987, 1988) in 1953) where the focus is on measuring sub-
order to address his desire to objectively jectivity (Brown, 1980, 2008; McKeown &
measure subjectivity (Stephenson, Brown, Thomas, 1988; Stephenson, 1953). Thus, Q
& Brenner, 1972). methodology is a set of procedures, theory,
Like Cyril Burt, Stephenson learned and philosophy that focuses on the study of
about factor analysis from its creator, subjectivity, where subjectivity is typically
Charles Spearman (Brown, 1998). As a associated with qualitative research and
PhD in both psychology and physics, objectivity is usually associated with quanti-
Stephenson was able to blend the concepts tative research (Brown, 2008; Stenner &
of factor analysis with concepts from quan- Stainton-Rogers, 2004). Similarly, Q method-
tum mechanics, a field of physics that stud- ology fits into the qualitative framework of
ies particles at the subatomic level where naturalistic contextualization of research
one can never measure the exact location of (Stenner & Stainton-Rogers, 2004). Com-
a particle but instead only can attempt to pared to typical qualitative research,
predict its “behavior.” Although connected though, Q methodology maintains the rela-
mathematically, quantum mechanics was tionship among themes within the data as it
less readily accepted as Newtonian laws of minimizes the impact of the researcher’s
motion. Even the great Einstein took issue frame of reference (Stainton-Rogers, 1995).
with some of the ideas of quantum mechan- It does this through complex statistical analy-
ics (Howard, 2005; Sauer, 2008), not unlike sis including correlation and factor analysis
some key social science researchers who (Brown, 1980; Stephenson, 1953).
seemingly disregarded Q methodology, at This sophisticated use of statistics has
least in part, due to its subjectivity and, been what has led to the designation of
potentially, its mixed methods framework Q methodology as quantitative (Brown,
since, as Stenner and Stainton-Rogers 2008). Yet within Q methodology aspects
(2004) indicate, such a mixed methods of these statistical analyses, especially
hybrid ought to be discomforting. This dis- within the factor analysis, there also exist
comfort emerges from the reorganization of both qualitative and quantitative aspects
distinct ideas which come together, cross (Brown, 2008; Stenner & Stainton-Rogers,
boundaries, and form something new. 2004). Even the issue within mixed meth-
A similar discussion between two com- ods studies of how to collect two different
peting conceptions of research, such as the strands of data (Creswell, 2010) is not a
debate between Newtonian and quantum problem within Q methodology. These
physics, is found within Mixed Methods mixed methodological aspects of Q will be
Research: Exploring the Interactive discussed later in this chapter.
Continuum (Ridenour & Newman, 2008), Certainly, Q methodology was not origi-
in which Ridenour and Newman attempt to nally identified as a mixed method since,
connect quantitative and qualitative con- as Creswell (2010) states, mixed method
ceptualizations. As they explain, mixed research began around 1988. Perhaps this is
methods research does not consist of a why, since its inception 75 years ago when
dichotomy of quantitative and qualitative, Stephenson first published an article describ-
but instead represents a third research ing Q methodology in Nature (Stephenson,
508–––◆–––Issues Regarding Methods and Methodology
1935), Q methodology has held a controver- views of any topic. Thus, Q methodology
sial position in social science research that has allows researchers to investigate research
led to its relatively small following (Brown, questions that involve determining the var-
1998). It is only recently that Q methodology ious views within a group about a specific
has become more widely accepted in journals topic, as well as using those views to inves-
in a variety of disciplines (A. Wolf and tigate how they affect some other aspect of
S. Ramlo, personal communication, April 29, the study.
2009), possibly due to the greater acceptance Q methodology is a measure of subjec-
of mixed methodology research. tivity that represents an individual’s
Q factor analysis, which is different from feelings, opinions, perspectives, or prefer-
Q methodology, also fits into the conception ences (Brown, 1980, 2008; McKeown &
of mixed methods research. Although Q fac- Thomas, 1988; Siegesmund, 2008; Stenner
tor analysis also groups people, it does not & Stainton-Rogers, 2004; Stephenson,
include the sorting of items into a grid distrib- 1953). This is consistent with qualitative
ution as a means of measuring subjectivity researchers’ focus on the investigation of
the way Q methodology does. In other words, subjectivity (Siegesmund, 2008). Certainly,
the participants’ sorting of the statements in Q methodology represents a unique way
Q methodology is what determines the factors to measure subjectivity (McKeown &
and groupings, which makes these cate- Thomas, 1988; Stephenson, 1953) because
gories of operant subjectivity (Brown, 1998). it allows participants to provide their per-
Instead, Cattel (1978) described Q factor spectives by sorting items, typically state-
analysis as a means of determining dimen- ments related to the topic, into a sorting
sions or profiles of people. Within Q factor grid determined by the researcher. These Q
analysis the factors or groupings are based on sorts are then analyzed via factor analysis,
various characteristics or data collected, but which allows those of similar views to be
they are not categories of operant subjectivity grouped into factors. Thus, within Q,
such as those in studies involving Q method- people, not items, are grouped and, there-
ology. Even though the techniques are differ- fore, researchers must have a sufficient
ent, sometimes the term “Q factor analysis” is number of items to determine differences
used incorrectly to refer to studies that actu- among the participants, not a sufficient
ally involve Q methodology. Adding to the number of participant to determine differ-
confusion, the factor analysis of the Q sorts in ences among the items as is typically
Q methodology is often referred to as Q fac- required within R factor analysis (Brown,
tor analysis (McKeown & Thomas, 1988; 1980; McKeown & Thomas, 1988;
Stephenson, 1953). Stephenson, 1953).
Differences between Q methodology
and Q factor analysis are discussed in detail
by their respective creators elsewhere PERFORMING A
(Stephenson & Burt, 1939). Thus, both of Q METHODOLOGY STUDY
these preexisting yet different research
approaches fit the new age of social science Q studies commence with the develop-
research that is mixed methods. Specifically, ment of a collection of items, typically
these two methods, Q factor analysis and Q statements or pictures, which is qualitative
methodology, use sophisticated statistical in design.1 This collection of items is called
techniques to reduce large amounts of data. “the concourse within Q methodology,”
The data collection is typically based on and represents the communications about
qualitative research, as is the naming of the the topic. These items are typically taken
factors, or groupings of people. from interviews, focus groups, and other
Stephenson presented the idea of Q metho - sources of dialogue such as newspapers
dology as a way of investigating people’s (McKeown & Thomas, 1988). Often the
Using Q Methodology and Q Factor Analysis–––◆–––509
concourse is extracted via a theme analysis placed into the grid shown in Figure 20.1
(Strauss & Corbin, 1990). The Q sample, (Ramlo, 2008a, 2008c). Participants may
which includes the items for participants to be interviewed during the sorting process or
sort, is then selected from the concourse may be asked to make written comments
(Brown, 1980; McKeown & Thomas, regarding their sorting selections in order to
1988; Stephenson, 1953). Typically, the better inform the researcher’s interpretation
Q sample consists of 30 to 60 items selected of the results (Brown, 1980; McKeown &
as representative of the concourse (Brown, Thomas, 1988).
2008). It is the Q sample that is sorted by The analyses of Q sorts involve correla-
those participating in a Q study (Brown, tion, factor analysis, and the calculation of
1980, 2008; McKeown & Thomas, 1988; factor scores (Brown, 1980, 1986, 2008;
Stephenson, 1953). Stephenson, 1953). Although sophisticated
The sorting process is inherently subjec- mathematically, parts of this factor analysis
tive because participants judge each Q sam- process are qualitative. The first step of the
ple item relative to the others while placing factor analysis process in Q methodology is
them into a distribution based on a condi- to select the factor extraction method.
tion of instruction, both of which are pro- Typical software used specifically for Q
vided by the researcher (Brown, 1980; methodology studies, such as the PQ
McKeown & Thomas, 1988; Stephenson, Method (Brown, 2008; Schmolck, 2002),
1953). An example would be to have offers two choices for factor extraction:
students sort 34 items related to their views principal components and Centroid.
of learning in a first-semester college Principal components analysis is a common
physics course. Each item is presented on an extraction method that is frequently used in
individual strip of paper, which is then R factor analysis where there are 1s in the
2 4 5 5 6 5 5 4 2
Least Most
like my like my
view neutral view
−4 −3 −2 −1 0 1 2 3 4
510–––◆–––Issues Regarding Methods and Methodology
Qsort ID 1 2 3 4
NOTES: The Q sort ID in this table contains demographic information; the first letter represents the students’ major
(C = construction, M = mechanical engineering technology, E = Electronic engineering technology, S = Surveying &
Mapping), the second letter represents the students’ undergraduate level (F = freshman, S =sophomore, J = junior),
the third letter represents the grade received by the student. The first numerical part of the ID represents self-reported
age, and the second set of numbers represents the students’ score on the FMCE at posttest.
This example begins with the desire of a those of their students in these select
small group of faculty to investigate courses. Using Q methodology allowed
students’ views of learning and knowledge, them to avoid the lengthy interview process
also known as personal epistemology typical of qualitative epistemology studies
(Chan & Elliott, 2004; Hofer & Pintrich, (Duell & Schommer-Aikins, 2001; Elby &
2002), in a variety of classes at a large, Hammer, 2001; Schraw, Bendixen, &
Midwestern public university. This faculty Dunkle, 2002). In addition, Q allowed
group also sought to compare instructors’ them to determine the various epistemo-
beliefs about learning and knowledge to logical views of students, unlike studies
512–––◆–––Issues Regarding Methods and Methodology
that have used Likert-scale surveys (Adams, However, the physics course investigator
Perkins, Dubson, Finkelstein, & Wieman, wanted to delve more into how students’
2005; Halloun & Hestenes, 1998; Perkins, epistemological views related to their learn-
Adams, Pollock, Finkelstein, & Wieman, ing of force and motion concepts. Thus, a
2005; Schommer, 1990). follow-up study was done with a slightly
Yet the concourse of statements for the revised Q sample to better investigate
Q methodology study on student epistemol- students’ experiences in the physics class-
ogy began with the popular Likert survey room and laboratory (Ramlo, 2008a,
developed by Schommer (1990), which she 2008c). This Q sample was not used in
developed from interviews. A Q sample of other courses, which allowed the researcher
32 statements was selected from this 72-item to change the Q sample such that it targeted
questionnaire. Students were required to sort learning in the first semester of physics only.
the Q sample statements; analyses of the This subsequent study demonstrated how
Q sorts were performed on each class. The Q methodology can be used to investigate
results and students’ written comments from college physics students’ views of their
this initial epistemological study indicated learning (Ramlo, 2008a, 2008c) and how
that students typically sorted the Schommer those views affected their learning of impor-
statements—not based on their personal tant physics concepts in a first-semester
epistemological views but, instead, based on physics course (Ramlo, 2008c). In addition,
their public epistemology. Lising (2005) dif- this study contributed to the already large
ferentiated these two views by describing body of literature on students’ learning of
personal epistemology as how someone per- force and motion concepts, which has
ceives their own learning and knowledge. established students’ difficulty gaining
Alternatively, someone’s public epistemology Newtonian-based understanding of force
represents how they view others’ epistemol- and motion concepts (Ramlo, 2008d;
ogy such as scientists or other authorities. Redish, Saul, & Steinberg, 2000; Thornton
Thus, the preliminary results led Ramlo & Sokoloff, 1998).
(2006/2007) to change the wording of these Although this study further confirmed
initial Q sample statements to make them the connection between learning in physics
more personal. For example, “Learning and students’ epistemologies (Halloun &
something really well takes a long time” Hestenes, 1998; Hammer & Elby, 2003;
was changed slightly to “Learning some- Lising & Elby, 2005), the focus of the
thing really well takes me a long time in this remainder of this section is to further
course.” These changes were intended to describe the sorting process and introduce
stress to students that they were to reflect the reader to the type of results produced
on their own personal epistemology relative within a Q methodology study. The four
to the course they were taking. Student basic types of tables that are generated are
interviews also were used to develop an (1) factor scores, (2) rank-ordered list of the
additional 22 statements, some replacing Q sample statements with z-scores to create
Schommer survey items, such that the Q a representative sort for each factor, (3) the
sample increased from 32 to 44 statements. list of statements that distinguish each fac-
Although other courses and instructors tor from the other, and (4) the list of con-
used this revised Q sample, only the results sensus statements that represent agreement
of the physics course’s portion of the study among all the factors (Brown, 1980;
are currently available as a journal article McKeown & Thomas, 1988). Each of these
(Ramlo, 2006/2007). This aspect of the tables will be described related to the
study allowed the instructors to investigate Ramlo 2008 study (Ramlo, 2008a, 2008c)
how their perceptions of students’ views of In this study (Ramlo, 2008a, 2008c), 18
learning in their courses actually compared students sorted the 44 statements into the
to the students’ views. distribution shown in Figure 20.1. Each of
Using Q Methodology and Q Factor Analysis–––◆–––513
these statements was on a separate slip of types of tables produced are specifically for
paper for ease in sorting. The participants interpreting the views of those represented
sorted these items based on how they by each of the perspectives (factors). These
viewed their learning in this first-semester tables are for the representative sort for
physics course. each factor, distinguishing statements for
Each individual sort was entered into PQ each factor, and statements representing
Method, one of several software packages consensus among the factors. Each of these
designed specifically for analysis of Q sorts tables is described below in the context of
(Schmolck, 2002).4 Only these types of our example on studying students’ episte-
packages provide the types of output mology in a first-semester physics course.
reports required to interpret the partici-
pants’ view on a topic. As Bazeley (2010
Table of Factor Scores
[this volume]) suggests, software such as
PQ Method is necessary for mixed methods The first table, Table 20.1, is the factor
research to effectively integrate different matrix for the Ramlo (2008a, 2008c) study
data elements and analyses. Although fac- discussed here. This table illustrates that four
tor analysis more often fits into the more perspectives on learning in this physics
conventional standard statistical packages course, each represented by a factor, were
such as SPSS, the analysis of the Q sorts found in this study. In qualitative research,
requires software that allows the researcher these factors would be called typologies
to combine the qualitative and quantitative because they group people of similar views.
aspects of such studies. We agree with The table has rows for each sorter that
Onwuegbuzie and Combs (2010 [this vol- include the sorter’s identification and that
ume]) that one of the difficulties of per- person’s loadings (correlations) with each of
forming mixed methods research is that the the four factors that were retained. If a sorter
researcher must be competent in analyzing is primarily represented by one factor, that is
both qualitative and quantitative data. indicated by placing in X next to that
However, Q methodology and Q factor person’s factor score for that particular factor.
analysis actually represents the integration For instance, in Table 20.1, the sorter in Row
of qualitative and quantitative methods. As 1, CJ24D14, has an X placed next to the fac-
Bazeley suggests, this integration allows the tor score of .45 in the Factor 2 column. Thus,
researcher to produce findings that are of as one can see from Table 20.1, Typology /
greater use and to better address the Factor 1 is made up of persons 5, 8, 10, 12,
research purpose (Newman, Ridenour, 13, 17, and 18, and Typology / Factor 2 is
Newman, & DeMarco, 2003). This think- made up of participants 1, 6, and 11. That
ing is supported by the work of a variety of means that Factor 2 represents the view held it
mixed methods researchers (Bazeley, 2010; by CJ24D14 along with two other sorters
Creswell, 2010; Onwuegbuzie & Combs, (Row 6, CJ21C28, and Row 11, MF22D7).
2010; Tashakkori & Teddlie, 2009) This is an important table in that allows us to
Within Q methodology, the analyses see that the 18 sorters can be identified as
produce an extensive report with a variety four types via data reduction. However, this
of tables. Four basic types of tables are pro- table does not explain these types. Other
duced, however. These tables are the basis tables must be examined to identify the vari-
of the interpretation of the Q sorts and, ous views determined and to name these dif-
therefore, the participants’ views. The first ferent perspectives. Therefore, to describe
table is a listing of the factor scores that are these world views, we must examine other
used to determine which participants are tables produced from the analyses.
represented by which factors (views). Each To learn more about those represented
factor represents a similar perspective or by Factor 1, for instance, their Q sort data
world view of the topic. The other three must be analyzed. The remaining analyses
514–––◆–––Issues Regarding Methods and Methodology
are based on the individual’s sorts selected The most extreme z-scores (represent-
for each factor, shown with the Xs. ing the positions toward the outside of
Participants who do not have an X placed the sorting grid) are most useful for inter-
next to a factor score are not represented by preting the factor. Table 20.2 lists the top
any of the four factors and, thus, do not and bottom five statements for Factor 1
have their sorts included in the analyses. In in this study. The Q sorts for each person
this way, the factor interpretation follows who is represented by Factor 1 were used
evaluating the factor scores (McKeown & to create this statement list, which can be
Thomas, 1988). used to create a Q sort representing that
factor. The ranking of the z-scores was
used to determine the grid position,
Representative Sorts for Each Factor
which also is given in the table. Although
One sort that represents the views of the only the extreme ends of the grid are
people on each factor is created through the reported here, the researcher can use the
analyses. This representative sort is created z-scores to create a complete representa-
from the Q sorts of those who were selected tive sort for this factor, as already men-
as being represented by that factor. This tioned. Similarly, the remaining three
representative sort is created via the listing factors in this study have statement rank-
of all the statements, in rank order of ings, based on z-scores, which can be
largest positive to largest negative z-score. It used to create the additional Q sorts that
is these z-scores that represent each state- represent these three views.
ment’s position in the sorting grid (Brown, Representative Q sorts, with focus on
1980; McKeown & Thomas, 1988). those statements on the ends of the grid,
Table 20.2 Factor 1 Tops Five Most Like / Most Unlike Statements
5 I have very little control over how much I learn in this −1.310 −3
course.
were used, in part, for the interpretation describes the use of distinguishing state-
of the epistemological views of the ments in this Q methodology study.
students in this study. Because of the impor-
tance of the Factor 1 view, we will focus
Distinguishing Statements
on its interpretation here. From Table 20.2,
the most extreme positioned statements As previously mentioned, the extensive
for Factor 1 indicate that those repre- report produced within Q methodology
sented by this view were reflective, help studies, includes tables of distinguishing
seeking, and enjoyed math or problem statements. These tables are created for
solving. They also saw the relevance of each factor in the study. Such statements
this course to other courses they would distinguish, here, each factor from the other
take. These students also did not see three at a significance level of .05. Again,
learning as immediate (disagree with our primary interest here concerns Factor 1
Statements 7, 8, and 16) but did see that so we will only focus on those statements
they have control over their learning (dis- that distinguish Factor 1 from the other
agree with Statement 5). These state- three factors found in the study. Table 20.3
ments allowed us to consider names for contains these statements for Factor 1.
this factor or world view, such as reflec- Although the representative sort results,
tive-learners. However, examining the such as those given in Table 20.2, are help-
distinguishing statements, which allowed ful in describing a particular factor’s view,
us to differentiate this Factor 1 view from the distinguishing statement results provide
the remaining three, led to other possible additional and often insightful informa-
names for this view. The next section tion for the researcher. The statements that
differentiate Factor 1 from the other three in the physics class. Consensus has allowed
views are contained in Table 20.3. It is this researchers to focus on agreement among dif-
table that reveals that Factor 1 students ferent views, which can be used to start a dia-
indicated that they sought a coherent view logue related to commonality, a key idea in
of force and motion (Statement 1) and organizational change (Ramlo, 2005).
believed that their learning would take time In this way, the consensus yielded addi-
(Statements 8 and 16). These three state- tional information related to students’ epis-
ments distinguish this view from the others temologies. Here, two consensus statements
at the .01 level. In other words, these three were determined. One of these consensus
statements distinguish Factor 1 from the statements indicated that the physics
other factors; it is more certain, therefore, students agreed that they ask their peers for
that this difference is not due to chance. help in understanding the lab activities.
It is important to note that only those Instructors’ observations substantiated this
represented by this view agreed that they finding. These same students disagreed that
sought coherence for the force and motion they tried to combine ideas across the lab
concepts and disagreed that learning needed activities. It is worth noting, however, that
to be immediate (Statement 16). This addi- the lab activities frequently refer to previous
tional information, in conjunction with the activities from earlier labs and that these
representative sorts from the other views, activities including having students com-
allowed the researcher to better reveal the pare their current findings to the findings
epistemology of this Factor 1 perspective on from earlier activities. Thus the consensus
learning in this first-semester physics course. statements here supported the importance
This additional insight allowed the research of peer interaction among the lab students
to select the name of “Reflective learners but indicated that lab materials and instruc-
who sought coherence and found it” for this tors needed to focus more on having
factor, or view. Describing this view is espe- students combine ideas across the various
cially important, given that the Factor 1 laboratory activities during the semester.
view also contains those students who
scored highest on the Force and Motion
Correlation of the Epistemological
Conceptual Evaluation (FMCE) posttest,
Views With Conceptual Understanding
thus the additional “and found it” in the
view’s name. Thus the Factor 1 epistemo- The true purpose of the Ramlo (2008c)
logical view correlates with the only group epistemology study, however, was to inves-
that had a Newtonian view of force and tigate the relationship between students’
motion, as will be discussed further in a sub- views about learning and knowledge and
sequent section. However, the consensus their understanding of force and motion
among the four factors also revealed impor- concepts. In other words, discovering the
tant insight into the physics students’ episte- four different views is not the end of our
mological views about the course. story. Instead, we can relate these world
views to the learning of physics concepts.
Thus, we had to ask how well this quasi-
Consensus Statements
qualitative outcome, via Q methodology,
Consensus statements do not distinguish relates to an important research question
between any of the factors (Brown, 1980). within the physics education research com-
Thus, in addition to the tables of distinguish- munity that could not have easily been
ing statements for each factor, a table of con- detectable without Q methodology.
sensus statements for the factors also is The understanding of force and motion
identified in the Q methodology analyses’ concepts was determined by using the
report. Thus in this study, agreement existed FMCE, an instrument that has been shown
across the four different views about learning to have strong estimates of both validity
citation: (Ramlo &
McConnell, 2008)
Using Q Methodology and Q Factor Analysis–––◆–––517
and reliability (Ramlo, 2008d), at the end enabled discussions about organizational
of the semester. Thus, the Q methodology change (Ramlo, 2005). The effectiveness of
aspect of the study allowed the researcher a reading circle professional development
to identify four different factors that repre- experience to help university faculty reflect
sented the four differing epistemologies on their teaching also was determined
held by students in the course. The correla- through a Q methodology study. Thus,
tions between these factors and the FMCE using Q methodology to reveal the differing
posttest scores are presented in Table 20.4. perspectives, as well as consensus, within
These statistical results indicate that only groups of people can address a large
the Factor 1 epistemological view had a pos- number of research purposes.
itive correlation (.46) with the FMCE posttest
scores. Thus, this view is most interesting in
that only this epistemological view represents EXAMPLE 2: CLASSIFICATION OF
students who have a Newtonian understand- INJURED WORKERS IN RELATION
ing of force and motion. The remaining three TO VOCATIONAL TRAINING WITH
views are negatively correlated with the Q FACTOR ANALYSIS
posttest, and are held by students who did
not have a Newtonian understanding of force Although Q factor analysis is a by-
and motion. Therefore, these correlations person factor analysis similar to Q method-
suggest more about the epistemological views ology, it is not the same as Q methodology.
revealed by the Q methodology aspect of the Instead, Q factor analysis only employs
study. This investigation further demon- one aspect of Stephenson’s procedure: the
strated the strength of Q to investigate views grouping of people with factor analysis.
in the research area of conceptual under- Yet this grouping is not based on partici-
standing in physics. In addition, it demon- pants’ sorting of items as it is in Q method-
strates how the perspectives revealed within a ology (Stephenson, 1953; Watts & Stenner,
Q study can be used to investigate how those 2005). Q methodology groups people
views are related to other student attributes. based on subjective data (from the sorts)
The usefulness of Q methodology has and Q factor analysis groups people using
become apparent in other studies with dif- data that may come from a variety of
ferent purposes. Consensus and differing sources including interviews, observations,
views about the creation of a school of tech- surveys, and demographic information.
nology at a large Midwestern university These differences and similarities of Q
were determined using Q methodology that methodology and Q factor analysis are
Table 20.4 Correlations Between the Factors (Views) and the FMCE Posttest Scores
F1 0.463 31 7 7
F2 −0.393 16 11 3
F3 −0.171 21 3 3
F4 −0.318 12 N/A 1
NOTES: The FMCE was used for the posttest and has a maximum of 47 points possible. Only Factor 1 had a
positive correlation with the posttest.
518–––◆–––Issues Regarding Methods and Methodology
demonstrated in Figure 20.2. Specifically, scores overall, they do not represent the
for Q factor analysis, the groupings of same type or profile. Instead, Persons 1 and
people into types are based on the shape of 3 have similar shapes to their responses;
the responses to the various items, as shown therefore, the Q factor analysis would con-
in Figure 20.3. In this figure, although sider these two persons as representing the
Persons 1 and 2 have similar absolute same type or profile.
Factor Q Factor
Q methodology
analysis analysis uses
uses Q sorts
groups data (non-Q
of items.
people. sort).
1
Z-scores
−1
−2
−3
Variables
NOTE: The plots of Z-scores, called profiles, for three different people are shown here to demonstrate that
Persons 1 and 3 have similar profiles and therefore represent one type. Person 2 has a different profile and
therefore represents a different type.
Using Q Methodology and Q Factor Analysis–––◆–––519
NOTE: This table is a subset of items based upon the study by Waechter and colleagues (1998). This specific
example of data is useful in demonstrating how Q factor analysis can be used to differentiate between types. Please
note that the data presented here is not exactly the same as that presented in Waechter and colleagues (1998).
520–––◆–––Issues Regarding Methods and Methodology
data were used to classify people creating In the injured worker study, two types of
typologies, sometimes also referred to as persons were identified with 59 of the 67 par-
profiles or dimensions (Nunnally, 1978). ticipants represented by one of the two types.
The purpose of the injured workers The patterns (shapes) for the two types found,
study was to develop a process of classify- for each of the variables, are represented in
ing people who experienced work-related Figure 20.4, which shows the z-scores for
injuries and who were identified for voca- each variable for Types I and II plotted as a
tional job retraining. Sixty-seven injured traditional X–Y graph. Table 20.5 contains
workers, 37 males and 30 females, who had the data used to create figures 20.4 and 20.5.
their injuries verified, were involved in this The patterns found were used to interpret the
study. The data were Q factor analyzed two types. This table lists the descending array
with QUANAL in order to determine simi- of z-scores for each of these two types of
lar patterns in the data (see Figure 20.4) injured workers by variable type. The differ-
that were used to identify the different ence between the z-scores listed in the fourth
typologies within the sample. Thus, the column was used to determine which items
analyses are grouping people based on their differentiated one type from the other and to
similar patterns. indicate where there were similarities.
2
1.5
1
Z-scores
0.5
0
−0.5
−1
−1.5
Motor coordination
Finger dexterity
Manual dexterity
Eye-hand-foot coordination
Back injury
Knee injury
Finger/hand injury
Male/Female
Number of injuries
Neck injury
Clerical perception
Verbal
Spatial
Intelligent
Numerical
Age
Variable
Type I Type II
NOTE: Visual representation of the two types, I and II, of injured workers found in the vocational education study
using the z-score data from Table 20.5.
Using Q Methodology and Q Factor Analysis–––◆–––521
Type_II
1.31 0.75 −1.07 0.99 0.3 0.84 −1.08 −1 −1.08 0.38 −0.54 −1.05 −1.05 1.44 1.02 1.05
Type_I
0.91 −0.28 −1.1 0.81 0.7 1.6 −1.12 −1.3 −1.12 0.9 0.743 −1.12 −1.1 0.97 0.71 0.81
I a b C e f f g k m m n n n s v
Q g a l y i i e n a o e u u p e
s e c e e n n n e n t c m m a r
c k r _ g g d e u o k _ e t b
a i h e e e a r i r i a
l c a r r r l n i a l
e a n _ _ _ j c l
l d d h d u a
_ e a e r l
f x n x y
t d
Factor
NOTE: Alternative representation of the Table 5 z-score data using this bubble plot presentation provided by
Dickinson (2010).
numerical, and spatial reasoning. The sub- notoriously unstable. We strongly recom-
jects’ files also were reviewed to help inter- mend that when doing a Q or R factor
pret these findings. This part of the study analysis, the sample be split in half to see
indicated that Type II individuals were what types emerge and then to cross vali-
referred to the testing by their attorneys as date the types. For example, if four types
part of the litigation process against their emerge in the first Q factor analysis and we
employers. Since they were older workers, get five types from the second sample, we
it is not surprising that they also had longer could correlate the types from Sample 1
work histories. In addition, these workers with the types from Sample 2. To the extent
refused opportunities to participate in that this results in similar types in both sam-
retraining programs. They also had stopped ples, we have reason to believe that these
receiving Worker’s Compensation, whereas types are more stable. From these analyses,
the Type I workers were still receiving assis- we may see, for instance, that only three of
tance. Point biserial correlations also indi- the types replicate between the two samples.
cated that Type II workers scored lower on These three may then be more likely to be
several of the dexterity and coordination more stable and this may warrant further
scales of the Apticom than their Type I investigation. The other types may be sam-
counterparts. These results and method of ple specific. Thus, depending on the research
classifying injured workers may allow voca- question these may or may not also warrant
tional evaluations to be better informed further investigation. With R factor analy-
about the types of injured workers and may sis, other methods tend to be used such as
improve decision making about screening confirmatory factor analysis or Kaisers fac-
applications (Waechter et al., 1998). tor matching techniques (Newman,
The method is identical to, but the Dimitrov, & Waechter, 2000).
topic is different from, an earlier Q factor
analysis study by Newman (1971) where
the views of both Black and White bas- FUTURE RESEARCH: EXTENDING
ketball team members were evaluated by THE USE OF GROUPS OF PEOPLE
grouping the team members and the
coach into factors. The data used were This chapter has demonstrated how
from the Subjective Perception Rating sophisticated statistical techniques that
Scales, a semantic differential scale, and a are employed by Q factor analysis and
behavioral differential scale. Profiles Q methodology can effectively reduce large
identified for White and Black subjects amounts of data that are frequently used
were then determined from the patterns in mixed method research. Such data
of responses to these 130 variables. Basically, reduction techniques may broaden the
two profiles or typologies emerged, each researchers’ ability to interpret the data in a
representing a different type across these more efficacious manner. In addition, we
130 items. These two types were named also suggest that coupling the groupings
White Typology and Black Typology with other statistical techniques allows
since the most discriminating factor researchers the ability to extend beyond the
between the two factors was race. It information they would get by simply
turned out that everyone on Type I was grouping. For instance, we could use Chi-
Black and everyone on Type II was squares to compare the groups to investi-
White, except for one person who loaded gate differences. We also could use these
similarly on both Types. This person was groupings as predictor variables in linear
the coach (Newman, 1971). regression models.
With factor analysis, factors are not To further illustrate this concept, pre-
stable—they are sample specific. They are vious studies investigating conceptual
Using Q Methodology and Q Factor Analysis–––◆–––523
understanding have shown that numer- Price, & Jones, 2007; Halloun & Hestenes,
ous student characteristics play roles in 1998; Perkins et al., 2005; Perkins, Gratny,
students’ learning of force and motion Adams, Finkelstein, & Wieman, 2006).
(Ates & Cataloglu, 2007; Dykstra, Boyle, Q factor analysis offers researchers a
& Monarch, 1992; Ramlo, 2003, 2007a, way to create profiles or groupings of
2007b; Rowlands, Graham, & Berry, people based on patterns of data. These
1998). Thus, to further study the poten- profiles can be used to identify the underly-
tial impact of the epistemology of physics ing constructs that can assist in classifying
students described in Ramlo (Ramlo, people in a meaningful way. Such profiles
2008a, 2008b), we could control vari- also can be used for evaluation purposes,
ables shown to influence the learning of such as the example given about vocational
force and motion concepts. Thus, we training (Waechter et al., 1998). In that
could predict the learning of the physics example, further research also may have
concepts at the end of the course (FMCE included using the profiles within linear
posttest scores) in a linear regression regression analyses. The researchers could
model that included the epistemology have extended their study by identifying
views (Q methodology factors) along implications drawn from their data that
with students’ FMCE pretest scores and could suggest further direction for their
previous physics course experience. We work, based on their analyses. In such a
could then “test” the effect of the episte- study, a full linear regression model could
mological views by comparing, statisti- have the completion of vocational training
cally, the full model (the one described predicted by each of the worker types
above) to the model that includes only (Type 1 or Type 2) and the intervention.
the FMCE pretest scores and previous The restricted model could “test” the inter-
physics course experience, without the vention by removing that variable and
epistemology views, to better understand using only the injured worker type to pre-
the influence of student epistemologies dict completing vocational training. This
on their learning while “controlling” the new study would then have fewer variables
other variables (FMCE pretest and previ- in the models because only the injured
ous physics course work). It is important worker type is included, omitting all of the
to remind the reader that although sam- variables used to determine these two types.
ple size in Q methodology is related to Thus, the number of variables in the mod-
the number of statements sorted, studies els is reduced, which increases the statistical
using statistical techniques such as linear power (Cohen, 1988) and potentially
regression require the researcher to have increases the conceptual understanding of
a sufficiently large sample of people in the models.
order to have sufficient statistical power Profiles and perspectives determined
(Cohen, 1988). Thus a larger student with Q factor analysis or Q methodology,
sample would be very desirable to con- respectively, can be used in studies that have
duct the investigation suggested here. purposes beyond simply classifying people
However, the current Q methodology into groups. It makes sense to use the types
study and its suggested further investiga- (concepts, constructs, factors) instead of all
tion offers the potential of gaining greater the individual variables that make up the
insight into physics students’ thinking type, when developing the linear regression
about their learning and knowledge and models, or when using other statistical tech-
its potential relationship to student learn- niques. In addition, these profiles could be
ing, beyond other epistemology studies in used to disaggregate data so that
physics education that have used Likert- researchers could further study a particular
type scales (Adams et al., 2005; Gire, group, such as the Type II workers in the
524–––◆–––Issues Regarding Methods and Methodology
vocational education study. More data, Newman & Benz, 1987; Onwuegbuzie &
such as interview responses, could be col- Combs, 2010; Tashakkori & Teddlie, 2009).
lected on others who fit that profile. This
could potentially enhance the insight and
understanding of the initial findings ♦ Use of Other Multivariate
(Waechter et al., 1998). Techniques to Facilitate
Thus, investigations that use Q factor Interpretation
analysis or Q methodology allow researchers
to better study the stakeholders from differ-
ent perspectives. This is frequently impor- Certainly, data for grouping can be either
tant in program evaluation, where there is quantitative or qualitative. Whichever is
often value in addressing the various stake- used, there always will be some similarity
holder groups differently to ascertain their coefficient (such as correlations, distance,
needs, in an attempt to improve the effec- z-scores, variability estimates, density func-
tiveness of the program (McNeil, Newman, tions, etc.) to decide how to group, but one
& Steinhauser, 2005). To the extent that the has to understand the strengths and weak-
evaluation identifies and communicates nesses of such techniques. There is really
with the relevant stakeholder, the more no right or wrong approach but it is impor-
likely it is to be useful and the recommen- tant to understand the approach that is
dations implemented. Too often, though, chosen. Different procedures for grouping
stakeholder groupings are simply based on people are based in different assumptions.
demographic characteristics such as ethnic- In other words, one has to determine if the
ity or socioeconomic status. We are sug- principle used for grouping makes logical
gesting that such variables may not be the sense and is aligned with the purpose of the
most appropriate way to group stake- study and the research question. Thus, a
holders. Instead, Q factor analysis and researcher would not use height and
Q methodology can be used to group weight as characteristics to investigate
people more effectively using profiles that student motivation. Instead, there must be
go beyond such surface characteristics. a logical, theoretical link between the char-
Therefore, these types of profile analyses acteristics of the individuals studied and
can provide the evaluators with additional the purpose of the study.
insight into the participants, as demon- This chapter has focused on two specific
strated by Ramlo and others (Newman & types of multivariate analyses that we believe
Benz, 1987; Ramlo, 2005; Ramlo & demonstrate ways to effectively reduce the
McConnell, 2008; Ramlo, McConnell, huge amount of qualitative variables into
Duan, & Moore, 2008). In addition, this profiles that can be used to better investigate
type of profile analysis can improve com- research questions. When using multivariate
munication with the various groups or sub- techniques for data reduction, one has to be
groups. By identifying the specific needs of sensitive to how data are being aggregated or
the various stakeholder groups, evaluators disaggregated to better understand the out-
can tailor-fit their services and recommen- comes. For example, factor analysis can be
dations, as suggested by McNeil and col- used to aggregate data that was initially qual-
leagues (2005). This is likely to make itative, to facilitate the interpretation concep-
evaluation services more effective and more tually. Let’s assume there are a number of
meaningful to the stakeholders. The bene- subjects that have been interviewed and that
fits of integrating qualitative and quantita- the data from each subject has been initially
tive methods in evaluation is a pragmatic coded, and let’s further assume that four ini-
approach that is not new and is gaining tial codes have emerged. That is, all subjects
more widespread use and acceptance interviewed talk about clarity, flow, persua-
(Creswell, 2010; McNeil et al., 2005; siveness, or resources. As one can see from
Using Q Methodology and Q Factor Analysis–––◆–––525
Figure 20.6, the coding of the interview of matrix code of 0 0 1 1. All the subjects can be
Subject 1 indicates that this subject identified qualitatively coded in a similar manner; these
clarity, flow, and persuasiveness, but not qualitative codings can then be quantitatively
resources, yielding a matrix code of 1 1 1 0. coded and placed into a matrix made up of 1s
The coding of the interview of Subject 2 iden- and 0s. The following demonstrates how this
tified persuasive and resources, yielding a could be done.
Figure 20.6 Qualitative Coding “Matrix” From Faculty Interviews Related to Writing for
People “subj1” to “subj4”
A B C D ...
1 1 1 0 ...
subj1 clarity flow persuasive --- ...
0 0 1 1 ...
subj2 --- --- persuasive resources ...
1 0 0 1 ...
subj3 clarity --- --- resources ...
0 0 0 1 ...
subj4 --- --- --- resources ...
...
... ... ...
NOTE: Resulting matrix from turning the codes into 1s and 0s based on whether the subject spoke of, for
instance, clarity (yes = 1, no = 0).
From this initial coding, qualitative (Berg-Schlosser, De Meur, Rihoux, & Ragin,
researchers would typically identify differ- 2009; Ragin, 1987).
ent concepts or categories that occur based We are suggesting that researcher can go
on how the initial codes logically group back to the 1–0 matrix shown at the right
together; and the logical groupings of these side of Figure 20.6, and factor analyze it to
items is frequently referred to as the determine what factors emerge from this
themes. This is demonstrated in Figure numerical coding, as shown in Figure 20.7.
20.6 where we have taken the text-codes
from Figure 20.6 (A through D), which can
be thought of as items, and used them to Figure 20.7 The Theme Analysis
construct the themes, which can be con- Determined Three Themes From
ceptually thought of as factors. Additionally, our Simulated Qualitative Data
we can take these text-codes and convert
them into the 1–0 matrix similar to the one
shown on the right in Figure 20.6. This is A Theme 1 --- --- 1 0 0
very much what a qualitative researcher
B --- Theme 2 ... 0 1 0
might do during the coding process and is
where they may end with the development C --- Theme 2 ... 0 1 0
of the emerging themes. This type of table D --- ... Theme 3 0 0 1
looks similar to a truth table, which an ... ... ... ...
electrical engineer might use to address
computer logic or a sociologist might use
to investigate complex human phenome- NOTE: These themes can be represented as a matrix,
non. However it is more similar to Qualita- with different items (A, B, C, etc.) forming Themes 1, 2,
tive Comparative Analysis, known as QCA and 3.
526–––◆–––Issues Regarding Methods and Methodology
1. Discuss data reduction techniques; include some of the advantages and disadvantages
of using them.
2. Discuss univariate and multivariate analysis as data reduction techniques.
3. Describe how Q methodology and Q factor analysis are related and how they are different.
4. Discuss how Q factor analysis and Q methodology may be used to facilitate the inter-
pretation of qualitative research.
2. See Brown (1980) for a greater explana- techniques (pp. 1–18). Thousand Oaks,
tion of different extraction methods. Suffice it to CA: Sage.
say here that Centroid extraction is the preferred Block, J. (2008). The Q-sort in character
method for Q methodology based on philosoph- appraisal: Encoding subjective impressions
ical considerations related to the type of rotation of persons quantitatively. Washington, DC:
promoted by Stephenson (Brown, 1980; American Psychological Association.
Stephenson, 1953). Brown, S. R. (1980). Political subjectivity:
3. Hand rotation, via a graphical interface, Applications of Q methodology in political
is preferred in cases where it is important to science. New Haven, CT: Yale University
ensure a specific participant is represented by a Press.
factor such as an instructor or leader within a Brown, S. R. (1986). Q technique and method:
group (McKeown & Thomas, 1988). Brown Principles and procedures. In W. D. Berry
(1986) explained that Centroid extraction of the & M. S. Lewis-Beck (Eds.), New tools for
factors, followed by hand rotation, allows the social scientist: Advances and applications
investigator the opportunity to rotate based upon in research methods (pp. 57–76). Beverly
hunches and to examine the data from a theoret- Hills, CA: Sage.
ical standpoint. See Brown (1980) for a detailed Brown, S. R. (1998). The history and principles
explanation of hand-rotation procedures. of Q methodology in psychology and the
4. This software is available for free down- social sciences. Unpublished manuscript.
load at http://www.lrz-muenchen.de/~schmolck/ Retrieved May 5, 2009, from http://
qmethod/downpqx.htm facstaff.uww.edu/cottlec/QArchive/Bps.htm
5. This program was previously available Brown, S. R. (2008). Q methodology. In
only for mainframe computers. An alternative L. M. Given (Ed.), The Sage encyclopedia
for QUANAL is PQ Method, mentioned previ- of qualitative research methods (pp. 700–704).
ously within this chapter. Thousand Oaks, CA: Sage.
Burt, C. L. (1941). The factors of the mind: An
introduction to factor-analysis in psychol-
♦ References ogy. New York: Macmillan.
Cattell, R. B. (1978). The scientific use of factor
analysis in behavioral and life sciences.
Adams, W. K., Perkins, K. K., Dubson, M., New York: Plenum.
Finkelstein, N. D., & Wieman, C. E. (2005). Chan, K., & Elliott, R. G. (2004). Relational
The design and validation of the Colorado analysis of personal epistemology and con-
learning attitudes about science survey. AIP ceptions about teaching and learning.
Conference Proceedings, 790(1), 45–48. Teaching and Teacher Education, 20(8),
Alston, R. J., & Mngadi, P. S. (1992). A study of 817–831.
APTICOM’s effectiveness in assessing Cohen, J. (1988). Statistical power analysis for
level of physical functioning. Journal of the behavioral sciences (2nd ed.). Hillsdale,
Rehabilitation, 58(3), 35. NJ: Erlbaum.
Ates, S., & Cataloglu, E. (2007). The effects of Duell, O. K., & Schommer-Aikins, M. (2001).
students’ cognitive styles on conceptual Measures of people’s beliefs about knowl-
understandings and problem-solving skills in edge and learning. Educational Psychology
introductory mechanics. Research in Science Review, 13(4), 419–449.
& Technological Education, 25(2), 167–178. Dykstra, D. I., Boyle, C. F., & Monarch, I. A.
Berg-Schlosser, D., De Meur, G., Rihoux, B., & (1992). Studying conceptual change in
Ragin, C. C. (2009). Qualitative compara- learning physics. Science Education, 76(6),
tive analysis (QCA) as an approach. In 615–652.
B. Rihoux, & C. C. Ragin (Eds.), Configu- Elby, A., & Hammer, D. (2001). On the sub-
rational comparative methods: Qualitative stance of a sophisticated epistemology.
comparative analysis (QCA) and related Science Education, 85(5), 554–567.
528–––◆–––Issues Regarding Methods and Methodology
Gire, E., Price, E., & Jones, B. (2007). Charac- Newman, I., Dimitrov, D., & Waechter, D.
terizing the epistemologicai development of (2000). Factor structure of perceived indi-
physics majors. AIP Conference Proceedings, vidualization of instruction: Argument for
883(1), 65–68. multiple perspective. Educational Research
Halloun, I., & Hestenes, D. (1998). Interpreting Quarterly, 24, 20–29.
VASS dimensions and profiles for physics Newman, I., Ridenour, C. S., Newman, C., &
students. Science and Education, 7(6), DeMarco, G. M. P. Jr. (2003). A typology
553–577. of research purposes and its relationship
Hammer, D., & Elby, A. (2003). Tapping epis- to mixed methods. In A. Tashakkori &
temological resources for learning physics. C. Teddlie (Eds.), Handbook of mixed
Journal of the Learning Sciences, 12(1), methods in social & behavioral research
53–90. (pp. 167–188). Thousand Oaks, CA:
Hofer, B. K., & Pintrich, P. R. (2002). Personal SAGE.
epistemology: The psychology of beliefs Nunnally, J. C. (1978). Psychometric theory
about knowledge and knowing. Mahwah, (2nd ed.). New York: McGraw-Hill.
NJ: Erlbaum. Perkins, K. K., Adams, W. K., Pollock, S. J.,
Howard, D. A. (2005). Albert Einstein as a Finkelstein, N. D., & Wieman, C. E. (2005).
philosopher of science. Physics Today, Correlating student beliefs with student
58(12), 34–40. learning using the Colorado learning atti-
Lising, L. (2005) The impact of epistemology on tudes about science survey. AIP Conference
learning: A case study from introductory Proceedings, 790(1), 61–64.
physics. American Journal of Physics, Perkins, K. K., Gratny, M. M., Adams, W. K.,
73(4), 372–382. Finkelstein, N. D., & Wieman, C. E.
Lising, L., & Elby, A. (2005). The impact of epis- (2006). Towards characterizing the rela-
temology on learning: A case study from tionship between students’ interest in and
introductory physics. American Journal of their beliefs about physics. AIP Conference
Physics, 73(4), 372–382. Proceedings, 818(1), 137–140.
McKeown, B. (2001). Loss of meaning in Ragin, C. C. (1987). The comparative method:
Likert scaling: A note on the Q method- Moving beyond qualitative and quantitative
ological alternative. Operant Subjectivity, strategies. Berkeley: University of California
24, 201–206. Press.
McKeown, B., & Thomas, D. (1988). Q method- Ramlo, S. (2003). A multivariate assessment of
ology. Newbury Park, CA: Sage. the effect of the laboratory homework com-
McNeil, K. A., Newman, I., & Steinhauser, J. ponent of a microcomputer-based labora-
(2005). How to be involved in program eval- tory for a college freshman physics course.
uation: What every administrator needs to Unpublished doctoral dissertation, The
know. Lanham, MD: Scarecrow Education. University of Akron, Akron, OH.
Newman, I. (1971). A multivariate approach to Ramlo, S. (2005). An application of Q method-
the construction of an attitude battery. ology: Determining college faculty perspec-
Unpublished doctoral dissertation, Southern tives and consensus regarding the creation
Illinois University at Carbondale. of a school of technology. Journal of
Newman, I., & Benz, C. R. (1987). Multivariate Research in Education, 15(1), 52–69.
evaluation design: A suggested technique for Ramlo, S. (2006). A physicist’s reflection on
mental health systems. Unpublished manu- Q methodology, quantum mechanics &
script. Paper presented to the Ohio Academy Stephenson. Operant Subjectivity, 29(2),
of Science, Malone College, Canton, OH. 81–86.
Newman, I., & Benz, C. R. (1998). Qualitative- Ramlo, S. (2006/2007). Student views of learn-
quantitative research methodology: Explo- ing in an introductory college physics
ring the interactive continuum. Carbondale: course: A study using Q methodology.
Southern Illinois University Press. Operant Subjectivity, 30(1 / 2), 52–63.
Using Q Methodology and Q Factor Analysis–––◆–––529
Ramlo, S. (2007a). Critical thinking and the learn- Sauer, T. (2008). Einstein’s struggles with
ing of force and motion concepts. Paper pre- quantum theory. Physics Today, 61(5),
sented at the American Association of Physics 56–57.
Teachers, Summer Meeting, Greensboro, NC. Schmolck, P. (2002). PQMethod manual mirror.
Ramlo, S. (2007b). Physics lab renovation 101. Unpublished manuscript. Retrieved April 29,
Physics Teacher, 45(4), 228–231. 2004, from http://www.rz.unibw-muenchen
Ramlo, S. (2008a). Determining the various per- .de/~p41bsmk/qmethod/
spectives and consensus within a classroom Schommer, M. (1990). Effects of beliefs about
using Q methodology. Physics Education the nature of knowledge on comprehen-
Research Conference Proceedings, 1064(1), sion. Journal of Educational Psychology,
179–182. 82, 498–504.
Ramlo, S. (2008b). Student perspectives on Schraw, G., Bendixen, L. D., & Dunkle, M. E.
learning physics and their relationship with (2002). Development and validation of the
learning force and motion concepts: A study epistemic belief inventory (EBI). In B. K. Hofer
using Q methodology. Paper presented at & P. R. Pintrich (Eds.), Personal epistemology:
the Paper Presented at the International The psychology of beliefs about knowledge
Society for the Scientific Study of Subjectivity/ and knowing (pp. 261–277). Mahwah, NJ:
Q Methodology Conference, Hamilton, Erlbaum.
Ontario, Canada. Siegesmund, R. (2008). Subjectivity. In L. M. Given
Ramlo, S. (2008c). Student perspectives on (Ed.), The Sage encyclopedia of qualitative
learning physics and their relationship with research methods (pp. 843–845). Thousand
learning force and motion concepts: A Oaks, CA: Sage.
study using Q methodology. Human Stainton-Rogers, R. (1995). Q methodology. In
Subjectivity, 2(1), 73–90. J. A. Smith, R. Harré & L. van Langenhove
Ramlo, S. (2008d). Validity and reliability of the (Eds.), Rethinking methods in psychology
force and motion conceptual evaluation. (pp. 178–192). London; Thousand Oaks,
American Journal of Physics, 76(9), 882–886. CA: Sage.
Ramlo, S., & McConnell, D. (2008). Perspectives Stenner, P., & Stainton-Rogers, R. (2004).
of university faculty regarding faculty read- Q methodology and qualiquantology: The
ing circles: A study using Q methodology. example of discriminating between emotions.
Journal of Faculty Development, 22(1) In Z. Todd, B. Nerlich, S. McKeown, &
Ramlo, S., McConnell, D., Duan, Z., & Moore, F. D. D. Clarke (Eds.), Mixing methods in psy-
(2008). Evaluating an inquiry-based bioin- chology (pp 101–120). Hove, UK; New York:
formatics course using Q methodology. Psychology Press.
Journal of Science Education and Stephenson, W. (1935). Technique of factor
Technology, 17(3), 219–225. analysis. Nature, 136, 297.
Redish, E. F., Saul, J. M., & Steinberg, R. N. Stephenson, W. (1953). The study of behavior:
(2000). On the effectiveness of active-engage- Q-technique and its methodology. Chicago:
ment microcomputer-based laboratories. University of Chicago Press.
American Journal of Physics, 65, 45–54. Stephenson, W. (1982a). Newton’s fifth rule
Ridenour, C. S., & Newman, I. (2008). Mixed and Q methodology: Application to psy-
methods research: Exploring the interactive choanalysis. Operant Subjectivity, 5(4),
continuum. Carbondale: Southern Illinois 127–147.
University Press. Stephenson, W. (1982b). Q-methodology, inter-
Rowlands, S., Graham, T., & Berry, J. (1998). behavioral psychology, and quantum theory.
Identifying stumbling blocks in the devel- Psychological Record, 32(2), 235–248.
opment of student understanding of Stephenson, W. (1987). Q methodology: Inter-
moments of forces. International Journal behavioral and quantum theoretical con-
of Mathematical Education in Science & nections in clinical psychology. New York;
Technology, 29(4), 511. England: Greenwood.
530–––◆–––Issues Regarding Methods and Methodology