Communication Institute for Online Scholarship
Communication Institute for Online
Scholarship Continous online service and innovation
since 1986
Site index
 
ComAbstracts Visual Communication Concept Explorer Tables of Contents Electronic Journal of Communication ComVista

Q Methodology, Communication, and the Behavioral Text
EJC logo
The Electronic Journal of Communication / La Revue Electronique de Communication
***** McKEOWN ************** EJC/REC Vol. 1, No.1, 1990 ****


Q METHODOLOGY, COMMUNICATION, AND THE BEHAVIORAL TEXT


Bruce McKeown
Westmont College


        Abstract.  Q methodology offers an alternative
     approach to the study of human behavior through
     its emphasis on subjectivity as expressed through
     concourses of communication.  By providing
     rigorous methods of empirical research, Q method
j    and its technique allows the researcher to
     understand and interpret the subjective text of
     his or her respondents without confounding them
     with external categories of theoretical
     reflection.  The method is discussed in light of
     modern philosophy of science and hermeneutics, and
     the basic steps of conducting Q-method research
     are outlined and discussed.

        LA METHODE Q, LA COMMUNICATION ET LES TEXTES
     COMPORTEMENTAUX.  A cause de son insistance sur la
     subjectivite telle qu'exprimee dans divers
     "concours" (ensembles) de communication, la
     Methode Q offre une voie alternative pour l'etude
     des comportements humains.  La Methode Q constitue
     une methode empirique rigoureuse qui permet, avec
     ses techniques, de comprendre et d'interpreter les
     "textes" subjectifs des repondants sans les
     deformer par les categories externes de la
     reflexion theorique.  Cette methode est discutee a
     la lumiere de la philosophie contemporaine des
     sciences et de l'hermeneutique.  Les etapes
     fondamentales d'une recherche de type Q sont
     presentes et discutes.


     Q methodology is a perspective on and a technique for
the objective study of human subjectivity, commencing from a
philosophy of science reflecting developments in modern
physical and behavioral sciences.  As technique, it offers
operations that permit empirical examination of the meanings
and significance of self-referent communications.  Although
used in many disciplines, its hermeneutic and scientific
applications in communication studies are particularly
appropriate.  Q method avoids many of the problems inherent
to more widely used research strategies; its findings are
not methodological artifacts but reveal the patterns and
meanings of natural expressions of self-reference.
     These themes were incarnated in the work of Q
methodology's originator, William Stephenson, an Englishman
first trained in physics (Ph.D., 1926) when the implications
of quantum theory were being debated.  He also studied
psychology (Ph.D., 1929) and psychometrics under the
tutelage of Charles Spearman and Sir Cyril Burt.  His
professional career included appointments at British and
American universities where he pursued cross-disciplinary
studies in physical science, psychology, psychoanalysis, and
communication studies, including journalism and advertising.
     The method's origins can be traced to a brief letter
Stephenson published in NATURE (1935) in which he proposed
applying factor analysis to persons rather than to traits.
By way of this "inverted" use of the technique (to use the
langauge of his letter), Stephenson was advancing a new
concept as well as a different employment of a powerful
statistic.  Henceforth, subjectivity, understood on the
basis of intra-individual rather than inter-individual
differences, was amenable to factorization.  This
philosophic and statistical advance received extended
treatment in THE STUDY OF BEHAVIOR (Stephenson, 1953) and
THE PLAY THEORY OF MASS COMMUNICATION (Stephenson, 1967).
More recent texts that summarize the method, discuss the
statistical foundations, and provide numerous application
examples are POLITICAL SUBJECTIVITY by Brown (1980) and Q
METHODOLOGY by McKeown and Thomas (1988).  Also, of
particular note are the essays and studies covering science,
psychology and communication published in a festschrift in
Stephenson's honor (Brown & Brenner, 1972).

                 Methodological Principles

Subjectivity and the Behavioral Text

     In Q methodology -- as distinguished from R
methodology, which derives from correlational studies
(Pearson's r) of interactions among traits and so forth --
subjectivity is demystified.  Q has little if anything to do
with theoretical or categorical conceptions of the self as
an inner "state of consciousness" typified by many
hermeneutical (e.g., phenomenological) or psychological
(e.g., psychoanalytic) constructs.  Rather, subjectivity is
a straightforward behavioral manifestation referring to
communications of self-reference (behavioral texts)
expressed, for example, by statements such as "In my opinion
...," "I believe ...," or "It seems to me...."  Through Q
method and its techniques (e.g., Q sorting, discussed
below), one attains access to, and maintains the integrity
of, subjective expressions from a personal frame of
reference.  Thus, as Brown (1986, p. 58) has written, "only
subjective opinions are at issue in Q" and these can "be
shown to have structure and form" which Q technique makes
manifest for observation and study.  The purpose of Q method
is to discover and display those structures (the scientific
component) in preparation for understanding and interpreting
their meanings (the hermeneutical component).
     Considerable attention has been given to joining
scientific and hermeneutic approaches to communication, but
with mixed results.  Typically, research is conducted
qualitatively along hermeneutical rather than empirical
lines.  Social texts (what people say about personal and
social reality) are often considered so incommensurate with
natural phenomena that empirical epistemology may be deemed
insufficient, if not irrelevant.  Human experience, we are
sometimes admonished, is not amenable to objective methods
and should be approached as "human science," "cultural
science," or "interpretative science" (Palmer, 1969).  Or,
communication is perceived fundamentally as
information-processing whereby objective facts reign and
subjectivity, statistically speaking, is disregarded as
"noise" or "psychometric slop." However, too frequently a
false dichotomy between empiricism and subjectivity is
accepted.  The disregard for or mistreatment of subjective
data, and the subsequent rejection of "subjective science,"
is unfortunate but is a conundrum to which Q methodology
offers a specific response.

Hermeneutical and Empirical Problems

     Studying subjectivity presents a methodological dilemma
inasmuch as there are always at least two subjective "texts"
involved -- the respondent's and the investigator's.
Furthermore, a researcher does not deal directly with
another person's experience (of a novel, a painting, a
movie, a political debate, a walk in the rain), but with
subjective expressions about those experiences which can be
examined.  To be sure, texts are not self explanatory or
self interpreting; people mean and people interpret.
Nonetheless, at the preliminary stages data should be
collected so that they remain the texts of the people being
studied rather than being a function of the meanings and
interpretations of the researcher.  As Stephenson (1967, p.
10) put it, "The focus has to be on the person looking,
viewing, reading, listening -- not as we observe him, but as
he observes it all himself."
     Therefore, a research question for communication
science is, How might we best prepare a text without unduly
contaminating it with our own subjectivity?  A partial
solution to the problems of the hermeneutical circle is
proposed in consideration of the mode of observation and the
techniques used for data collection.

Observational Perspectives: Operants and Artifacts

     The general interest of hermeneutics (e.g., Dilthey's
approach; see Palmer, 1969) is investigating "human
experience known from within."  Thus, the observational
standpoint for data collection is critical:  "The object
should not be to understand life in terms of categories
extrinsic to it but from intrinsic categories, ones derived
from life itself" (Dilthey, quoted in Palmer, 1969, p. 102).
Dilthey's point is well taken and is appropriated here to
mean that a respondent's self referent perspective -- for
example, his or her statements regarding a walk in the rain,
abortion or flag burning -- is what counts rather than what
the researcher may believe.  The researcher's view is
external; the respondent's view is the critical issue.
Therefore, of hermeneutical and empirical interest is
whether or not social research is conducted according to the
subjects' texts or with methodological artifacts, i.e., the
interpretive categories of the observer.
     Lasswell (1938) raised this point when he distinguished
between two modes of observation (see Note 1).  Lasswell
believed that intensive methods, e.g., the psychoanalytic
interview, were more likely than extensive ones to elicit
respondent meanings.  Intensive observational perspectives
are linked with the meanings intrinsic to the text; they
provide greater opportunity for self expressions from a
personal point of view.  Extensive approaches, on the other
hand, such as the public opinion survey, preclude personally
subjective operants because response categories are
predetermined (exemplified, e.g., by attitudinal scales).
Brown (1970) extends this thesis when he distinguishes
between "populist" and "elitist" research strategies:  the
former allow respondents to define the terms, whereas the
latter do not.  External/extrinsic methods are
interpretative from the outset.
     In Q methodology respondents provide their own
operations by behaviorally defining their points of view
according to their intentions and meanings; the researcher
then conceptualizes the operations.  The external view of
the scientist is involved (it cannot be avoided) but follows
from the subjects's operations.  In more orthodox research,
e.g., one employing a scale, a respondent receives a score;
implicit is the assumption there are correct answers.
Responses are right or wrong, liberal or not, alienated or
not.  In Q method the respondent assigns a score or value to
an expression according to its centrality to his or her
point of view.  Answers, therefore, are genuinely self
referent, and a cardinal rule of science is upheld:
concepts are kept as close as possible to actual operations
(Bridgman, 1927; Stephenson, 1977).
     Thomas' (1976) study of Tomkins' Polarity Scale is a
case in point.  Thomas tested Tomkins' (1963) theory of
personal ideology which posits a comprehensive and
generalized view of ideology which splits personality into
bipolar typologies.  Although Tomkins' sense of "Left-" and
"Right-wing" correlates with conventional political usage,
he believes the categories extend across many dimensions.
Thomas sought to determine if the bipolarity postulate held
true for a subject sample given opportunity to rank-order
the scale items in light of their own personal values (i.e.,
assign significance rather than receive it according to
their "score").  Congruence between the two texts (Tomkins'
and the respondents') existed only for the "left-wing";
"right-wing" respondents did not conform to Tomkins'
categorization.  Furthermore, bipolarity was discovered but
only in Tomkins' personal rank-orderings, a clear example of
methodological effect at odds with the respondents' own
categories of subjective evaluation (Note 2).

Method: Science and Technique

     The hypothetico-deductive paradigm which governs most
behavioral research, such as Tomkins', is locked in a
hermeneutical dilemma; subjective behavioral texts are
constrained.  Operationalizations elicit responses
understood within the terms of the original interpretation.
To be sure, science is not free from research effects.
However, the relevant issue is whether the theoretical
musings of the scientist remain tentative throughout the
process (relying upon working and abductory hypotheses) or
if they are solidified in the operational deductions that
proceed from the initial hypothesis.
     The natural sciences, especially modern physics, have
recognized the consequences of the interactive nature of
observational perspective and results.  In quantum
mechanics, physicists realized that specific outcomes were
functions of the methods employed (see, e.g., Dirac, 1971;
Hoffmann, 1959).  A prime illustration was the startling
conclusion regarding electromagnetism and photoelectric
effects.  Depending upon method, light was either a wave or
a particle phenomenon.  Observation -- for example, attempts
to study subatomic particles (the behavior of electrons) --
produces physical artifacts; determining velocity influences
position, and vice versa.  In quantum theory, as Hoffmann
(1959, p. 152) writes, "A spirit of perversity is in the
air....  science has become more humble.  In the good old
days it could boldly predict the future.  But what of now?
To predict the future we must know the present, and the
present is not knowable, for in trying to know it we
inevitably alter it."
     The philosophic quandary these conclusions present is
summarized by Werner Heisenberg's uncertainty principle.
Scientists had to reconcile themselves to the fact that they
"cannot determine both the position and the velocity of a
particle with exactitude, even in imagination.  Now that the
quantum is here, we cannot know both q and p simultaneously.
When we measure q we disturb p" (Hoffmann, 1959, p. 149).
More generally, Heisenberg reminds us that science is a
human enterprise, the interplay of subject and object.
"What it describes is not nature as such, but nature as
exposed to man's method of questioning" (Peterson, 1968, p.
22) (see Note 3).  Method is an interpretation.  Quantum
theory and hermeneutics arrive at a similar conclusion:
"Progress in science has been bought at the expense of the
possibility of making the phenomena of nature immediately
and directly comprehensible to our way of thought"
(Heisenberg, 1952, p. 39).
     The behavioral sciences are not immune to these
principles.  Social psychology, for the most part, presumes
that attitudes, as an example, exist in quantity as well as
quality.  "Alienation" (or "liberalism", "machiavellianism",
etc.) is approached not unlike body temperature, height or
weight; an individual "contains" a certain amount of
alienation as determined by a validated external measure,
not unlike a thermometer, yardstick or bathroom scale which
can be used in common for all people.  As in physics, "One
was led to the tacit assumption that there existed an
objective course of events in space and time, independent of
observation ... completely independent of each other, and
thus represented an objective reality, which was the same to
all men" (Heisenberg, 1952, p. 11).
     The implications of these considerations are in line
with the purpose and approach of Q methodology.  The
prevailing paradigms of R methodological research, as is
true of classical physics, "extend just as far as the
conceptions which for its basis can be applied" (Heisenberg,
1952, p. 23).  These may be proper at the macro level;
Heisenberg, for example, did not deny the validity and
utility of classical science applied appropriately.
However, at the micro, subatomic level Newtonian physics may
be irrelevant.

Science, Method and Communication Theory

     The uncertainty principle is applicable when
communication at the level of individual self-reference is
at issue.  As Brunner (1977) has noted, gross measures fail
to account for private meanings and intentions mediating
responses.  As in quantum theory, one cannot assume at the
"subatomic," personal level an objective reality "which is
the same to all men."  Therefore, methods selected for the
study of behavioral texts should allow subjectivity free
rein.  Indeterminate research strategies, such as Q
methodology and its reliance upon Q sorting and factor
analysis, are available.  The patterns ("quanta") of
subjective meaning are discoverable if they are intrinsic to
the data the respondent provides (Stephenson, 1982, 1983,
1988, 1988/1989).  Operant factors "have no critical
dependency on test 'construction' effects" (Stephenson,
1977, p. 8); they are "operant" inasmuch as they naturally
result from the operations of the respondent.  In this way,
hermeneutics and science can be conjoined:  a subjective
behavioral text is made available to others for
understanding prior to an interpretation.
     The hermeneutical dilemma can be translated into
methodological terms in line with Brown's (1980, p. 30)
statement that "in the human sciences ... as distinct from
the physical sciences, our subjects have their own
operational definitions and models of the world, and the
social scientist must avoid becoming so intrigued with his
own constructions that he becomes insensitive to those of
others."  This does not preclude metaphysics or metaphoric
usage (e.g., "psychic energy") as explanatory systems but
makes them subservient to method and retains Dilthey's
hermeneutical promise of examining "human experience known
from within" (Palmer, 1969, p. 103).
     The Q methodological equivalent is the concourse theory
of communication presented by Stephenson (1972, 1980, 1986).
If "experience known from within" is understood as referring
to any situation where self-reference is at issue, then the
metaphysical entity of "consciousness" can be replaced with
the propositions of communication theory (consciring) and,
for analytic purposes, with the structures of subjectivity
discovered through the use of appropriate empirical methods
(McKeown, 1984; Stephenson, 1968).  Essentialist traits or
entities ("ego," "consciousness" and so forth) are set aside
and replaced with acts of sharing knowledge (conscio)
according to one's point of view.  Subjectivity, therefore,
is public and available for scrutiny.
     The failure of hypothetico-deductive epistemology --
"objective science" (R methodological) -- is that its data
lack self-reference.  Subjective science, on the other hand,
preserves self-reference from beginning to end (Note 4).
And, as Stephenson (1983) has pointed out, it is guided by
the "law of affectability," i.e., "new ideas form from
concourse by way of confluences of feeling,"
"understandings, as new meanings, form in feeling" (pp. 93,
76).  Understanding refers to identifying the parts
comprising a personal world of experience and recognizing
how they fit in a whole.  In this fashion, meaning is
obtained.  The meaning of a whole is a "sense," an
"understanding," derived from the specific meanings given by
the respondent to the parts of his or her existence.  It is
attaining, as Brown (1987) has put it, a "feeling for the
organism."  Consequently, understanding is a function of
reciprocity among the components and the whole, and
discerning the relationships among the parts of the
self-referent text the respondent provides (what is meant by
his or her statements to the effect, "in my opinion..."?).
Q methodology's transactions are with statements of opinion
rather than statements of fact.
     This suggests, furthermore, that understanding is
contextual.  Parts are understood in light of the whole; the
whole is understood as an interaction of the parts; and, as
a hermeneutical system, understanding is dependent upon the
situation of its expression.  The meaning of an expression
in one instance may change in another.  Stephenson (1983,
1986) has made this point with the illustration "It is
raining."  Explanation of this exclamation would be an
account of meteorological physics (i.e., an accounting of
physical facts).  Understanding, on the other hand,
recognizes the personal experiences of the event.  In one
situation it could be a simple statement of fact; in
another, an expression of disappointment or depression; and
in yet another, a romantic feeling likened to "singing in
the rain," and so forth.  Many meanings can adhere to the
same event.  Explaining them must await their expression,
which is dependent upon the people who experience it.  In
operational terms, "everyone in a culture can understand
something of each statement in a concourse (of
communication).  Yet each statement may mean something
different to everyone, and something different to the same
person in different circumstances....  statements in
concourse shift their meanings with their company -- they
may have different meanings in different factors"
(Stephenson, 1983, pp. 75, 82).
     A communication concourse (Stephenson, 1986) is a
universe of "statements" for any context or situation, the
statements having to do with conversational, more so than
informational, possibilities.  Concourses occur in all
realms of human experience wherever declarative statements
are in made in line with "in my opinion":  "There is a
concourse for every concept, every declarative statement,
every wish, every object in nature when viewed subjectively,
in physics, philosophy, history, sociology, psychology, law,
art..."  (Stephenson, 1986, p. 44).  On the basis of these
possibilities, Q methodology provides a means for the
communication researcher to examine empirically the
subjective realities of the people engaged in the
discussion.  Through Q technique, one is able to obtain an
understanding of what people mean when they engage in these
conversations; additionally, it provides a means for
arriving at a theory of the self.

                  Technical Considerations

     A summary of technical procedures is outlined by
Stephenson (1986, pp. 44-45): "(1) A concourse is arrived at
empirically; it constitutes a Q universe; (2) Q samples are
drawn from it; (3) Q sorts are performed with these samples;
(4) these are factor analyzed; (5) the factors are
interpreted ... (and 6) It has to be shown next that the
factors are merely schematic for the conversational
possibilities of the individuals providing the factors; and
it is important to determine how the schemata ... relate to
the self of the individuals."

(1) Communication Concourse (Q Universe)

     A communication concourse is determined by inviting
people into conversation about a topic of interest and
concern to them.  In Q methodology, a concourse is composed
of opinions rather than facts; thus, one is interested less
in the fact that the Supreme Court ruled on flag-burning
than in what people have said about that decision.  Given
the controversy over the ruling, its communication concourse
is extensive.  It is a simple matter to survey commentary on
it (e.g., newspaper and magazine articles, letters to the
editor, transcripts of Congressional debates, and so forth)
and note the range of issues involved (patriotism, national
honor, protecting First Amendment freedoms, etc.).  Thus,
the concourse includes statements of opinion such as the
following:

     o Punishing desecration of the flag dilutes the
       very freedom that makes this emblem so revered.
     o Amending the Constitution to protect the flag is
       not a matter of partisan politics.  It's an
       American issue.
     o The integrity of the symbol has been compromised
       by those leaders who seem to manipulate the
       symbol of national purpose into a pretext for
       partisan disputes about meaner ends.
     o I just don't like tampering with the Bill of
       Rights.
     o Freedom of speech does not include yelling
       "fire" falsely in a crowded theatre.  Burning
       the flag is the same thing and it, too, ought to
       be prohibited.

Each of these statements are expressions of feelings and
emotions (not necessarily knowledge) driven by immediate
experience and lived through personal experience.  The
concern is not with what is objective, such as counting how
many people believe one way or another, but with what people
think and how they feel when flag-burning, freedom of
speech, and the Constitution are being debated.

(2) Q Samples

     A communication concourse is assumed to be a
statistical population with a practically infinite number of
statements.  The universe of statements, therefore, is
subject to sampling; a sample of statements drawn from a
concourse is called a Q sample, the basic unit of Q
technique.  When constructing a Q sample, one includes a
variety of statements that, taken together, insures
representativeness of the range of opinions on the topic.
In "unstructured" Q samples, the selection of statements is
guided by a general effort to survey general discussion on
the issue (e.g., pro and con); an unstructured sample,
however, does not guarantee representativeness since some
aspects of the issue may be under- or oversampled.


________________________________________________________

A. Direction      (a) pro-amendment  (b) anti-amendment
B. Dimensions     (c) constitutional (d) political
C. Issue          (e) speech         (f) patriotism
________________________________________________________

Figure 1. Design of Q Sample for Flag Amendment Study


     An alternative sampling technique, which avoids these
problems, is a "structured" sample based on design
principles of factorial experimentation.  For example, a Q
sample of statements related to a proposed constitutional
amendment protecting the American flag could be developed by
selecting statements from the concourse according to the
following dimensions:  constitutional, political (partisan),
protecting freedom of speech, and protecting patriotic
symbols.  The pro and con direction could be included also.
Design of the Q sample, therefore, might take the form
presented in Figure 1.  Statements are selected that are
representative of each effect.  Thus, opinions are included
that are (a) in favor of the amendment for the reason of (c)
constitutional protection of (f) patriotic symbols; opinions
that (b) oppose the amendment due to (d) the political use
of (f) patriotic symbols; and so on.  The total number of
statements in the Q sample (which typically ranges from 30
to 60 items) is determined by the number of effects and the
number of replications of each condition.  In this case
there are eight possible combinations (A x B x C = 2 x 2 x
2); to ensure a variety of expressions for each combination,
each possibility might be replicated (m) five times (e.g.,
five statements representing an "ace" position) creating a Q
sample with 40 statement items (ABCm = 2 x 2 x 2 x 5 = 40).
One should remember, however, that the fundamental purpose
of the design is to insure representativeness of the Q
sample.  It remains to be seen what respondents mean by each
statement when given the opportunity.

(3) Q Sorting and Conditions of Instruction

     The fundamental behavioral act in Q methodology is the
Q sort whereby respondents model their opinions by
rank-ordering the Q sample items according to their
individual points of view (Note 5).  General practice in Q
method is to have respondents distribute the items along a
continuum in which a few items are placed at the extreme
ends, with a greater number of items in the middle.  This
convention follows from the understanding that in most cases
there are fewer issues people feel most strongly about (pro
or con); therefore, a greater number fall between the
extremes (Figure 2).  In any event, respondents determine
where items are placed (their point of view governs item
placement) and it has been demonstrated that shape of the
distribution has little if no effect on factor patterns
(Brown, 1971; Cottle & McKeown, 1981).


Most Disagree                   Most Agree
-5  -4  -3  -2  -1   0  +1  +2  +3  +4  +5
__________________________________________

 x   x   x   x   x   x   x   x   x   x   x
 x   x   x   x   x   x   x   x   x   x   x
         x   x   x   x   x   x   x
         x   x   x   x   x   x   x
                 x   x   x
                     x

Figure 2. Q Sort Distribution (N = 40). x indicates Q sample
          statement.


     It is this distribution, furthermore, that makes
subjectivity the central dynamic in Q technique and allows
for comparisons of Q sorts among people.  The center point
of a Q sort distribution ("0") indicates lack of
psychological significance; items placed there hold little
or no meaning for the individual.  Meaningful statements are
those to the right and left of the central neutral point.
In this fashion, there is "a basis for measurement of
feelings, attitudes, opinions, thinking, fantasy, and all
else of subjective nature; and it does so in relation to the
law of error.  All scores are pure numbers; all are standard
scores (mean 0, standard deviation 1.0)....  But, most
important, the scores given to the statements by different
individuals are comparable -- the zero on all scales is the
same absolute value for everyone" (Stephenson, 1967, p. 11).
Thus, every measurement made in Q method is subjective and
related to the self.
     Q sorts are performed according to a condition of
instruction which directs the respondents to rank-order the
Q sample statements (e.g., regarding a constitutional
amendment prohibiting desecration of the flag) according to
the purpose of the study.  In the flag-burning example, one
might want to know the types of opinions about the issue in
context of the dimensions related to it.  Therefore, a
likely condition of instruction would be:  Describe your
view on the flag-burning controversy by distributing the
statements from those with which you most agree (+5) to
those with which you most disagree (-5), as in Figure 2.
Respondents are free to place each item wherever they
choose, depending upon their point of view.  The same Q
sample can be used under a variety of conditions of
instruction in more extensive studies; thus, respondents
might be asked, in addition to their own point of view, to
describe a "liberal" position, a "conservative" position,
and so forth, each of these labels being defined by each
respondent.

(4) Data Analysis

     Data analysis consists of correlating the Q sorts of
all respondents and factor analyzing the correlation matrix.
Factorization reveals the patterns or schemata (factors) of
subjectivity; factors in Q methodology are considered
operant since they derive from the subjective expressions
displayed in the Q sorts.  Factor structure, in other words,
reveals what the data contain and not what the researcher
projects upon them as his or her own hypotheses or
categorizations.
     Factor analysis in Q is the same as that used in other
methodologies; the difference lies in the manner in which
the data were collected at the outset (discussed earlier).
Otherwise, there is nothing statistically distinct in
so-called "Q factor analysis."  Different factoring methods
have been employed, e.g., centroid and principal components;
Stephenson (1953) preferred the centroid due to its
indeterminacy (no mathematically correct solution) since it
allowed him to approach problems through numerous rotations
guided by hunches (abductory logic, working hypotheses) as
suggested by the theory or theories at hand.
     Data can be subjected to analysis of variance;
frequently this is used to determine whether or not they
validate the theory underlying the Q sample.  However, it
must be emphasized that in Q methodology ANOVA is an
ancillary statistical technique since its analysis engages
in terms of hypothetical and deductive categorizations
external to and imposed on the data.  Q method is a more
inductive approach, primarily grounding its interpretations
on factor structure and factor scores.

(6) Interpretation

     Factoring simplifies interpretation by providing a
structural overview of the patterns of association among the
people who performed the Q sorts.  Interpretation,
therefore, begins with analysis of the factor patterns
generated from the correlation matrix derived from the Q
sorts.  In extensive Q studies (i.e., those with many people
performing Q sorts under the same condition of instruction)
the factors will identify those who sorted the items
similarly (factor A, e.g.) but differently from others who
fall on other factors (B, C ... N).  A factor indicates
that, for the people who correlate significantly with it
(their factor loadings), they converse in similar terms --
believe pretty much the same thing -- but differently from
those loading on the other factors.  Factors can be thought
of as a model Q sorts summarizing the subjective
similarities among those who associate significantly with
them.
     A three-factor solution to the hypothetical
flag-burning study means that the conversation, at least
among those included in the person sample, breaks down into
three distinct points of view.  These perspectives might be
indicated by the demographics of the people associated with
each factor:  for example, factor A consists essentially of
those people who identified themselves as liberal Democrats,
factor B by conservative Republicans and factor C by jurists
and academics trained in law and constitutional
interpretation.  Demographic data of this sort can be
helpful in factor clarification.
     In addition to the factor structure (the patterns of
subjective response), one can calculate factor scores for
each of the Q sample statements that define each factor both
positively and negatively (factor arrays).  Thus, for factor
A there may be a number of statements at the positive end
(+3, +4, +5) primarily having to do with protecting the
freedom of speech clause of the First Amendment, and
statements rejected at the negative end (-3, -4, -5) having
to do with national pride, limitations on freedom of
expression, and so.  Thus, the most important
interpretations in Q methodology are based on the array of
factor scores of statements for each factor that give
content and context to the points of view the factors
summarize and represent.

(6) Factor Structure and the Single Case

     The final step in Q methodology is the analysis of
single cases.  Data analysis to this point has been
essentially extensive:  what are the points of view on the
topic of flag-burning?  We have determined what the issue
means for the people in the respondent sample (which may or
not be stated in accord with expert, legislative and
journalistic, opinion).  The main consideration is not with
how many people hold a point of view, but rather, how and
why people believe as they do.  Thus, it remains to be seen
how individual people reflect upon and respond to the issue.
     At this point, Q method presents a significant approach
for the analysis of single cases (Stephenson, 1974); we are
able to determine whether or not the factors actually
represent their conversational modes.  At this point it is
less important what we label a factor ("civil liberties,"
"national pride," "legal jurisprudence"); "What is more
profoundly at issue is the communicability of the person"
(Stephenson, 1986, pp. 47-48).  Therefore, how is the
flag-burning issue understood and interpreted according to
an individual's point of view?  Also, one can bring to bear
on the analysis personality or other theory in order to
understand the dynamics of the "communicability of the
self."
     In principle any person loading on a factor is a
candidate for single case analysis since a significant
factor loading indicates a shared point of view among all
"members" of a factor.  One can be even more precise and
systematic, however; a "specimen" can be selected for
intensive analysis, that is, an individual who is "saturated
with" the kind of subjectivity one wishes to study.
"Saturation" refers to somebody with the purest factor
loading (highest correlation on the factor and with little
or no correlation with the remaining factors).
     In single case studies, individuals are again engaged
in Q sorting exercises, but under differing conditions of
instruction.  A representative of factor A, therefore, may
be asked to use the original Q sample and describe his or
her view of what President Bush believes, of the Supreme
Court's majority point of view, of Senator Jesse Helms'
perspective, his or her mother's and father's views, and so
forth.  Or, another Q sample may be created, based on the
individual's own concourse of conversation regarding the
topic, and then applied under differing conditions of
instruction:  describe your point of view, describe a
patriot's view, describe a liberal's and a conservative's
views, describe a religious view, describe an atheist's
view, etc.  These Q sorts are correlated, factor analyzed,
and interpreted as discussed above.
     In the intensive case, however, one is able to explore
in greater depth the relationships among and dynamics of the
variables contributing to an individual's subjectivity and
arrive at an understanding of the impact of personal history
and experience in its formation.  The potential of the study
is limited only by the number of conditions of instruction
and the desire of the researcher to consider alternative
interpretations of the data.  It does require separating
opinion from fact, communication from information.  "But
there is nothing mysterious at issue, merely the ordinary
day-by-day communication of possibilities of the individual
... in relation to concourses" (Stephenson, 1986, p. 51).


                           Notes

1.  Given the specifically psychodynamic nature of his early
work, this emphasis is understandable.  One can argue that
in its origins psychoanalysis was prototypically an operant
hermeneutical science.  The behavioral foundations of
psychoanalytic hypotheses were patients' self-referent
narrative histories from which Freud discerned the intrinsic
psychological patterns.  As the field progressed, however,
Freud and his colleagues became more distant from case
histories and reliant upon their own deductions, forcing any
given historical (therapeutic) situation into preconceived
theoretical interpretations.  The alienation that developed
between Freud and his disciples was the former's insistence
that established theoretical constructs be adhered to and
employed even in those circumstances where evidence seemed
contrary to or divergent from them (see Freud & Jung, 1974).

2.  Thomas was fortunate to have Tomkins provide Q sort
descriptions of his views of left and right types.  Tomkins'
description of the right-winger was isolated from the other
right-wingers, positioned alone in the expanse of bipolar
factor space (cf. McKeown & Thomas, 1988, pp. 67-74).

3.  This is congruent with Brown's (1970) point in his
critique of Converse's (1964) conclusions regarding belief
systems of nonelites.  Converse's methods determined a
particular outcome, that is, nonelites are not ideological
since their beliefs are neither consistent (constrained) nor
persistent.  Brown, on the other hand, using a different
approach which permitted respondents' to define their own
positions apart from any elite notions of ideology,
discovered that nonelites' opinions about politics were
consistent and persistent.  In physics, light is a wave or a
particle depending on how one examines it.  Apparently, the
same holds true of ideologies.  Attitudes, like light,
probably ought to be considered "wavicles."

4.  A recent example of the scientific model Dilthey rails
against is efforts in artificial intelligence to model human
thinking.  Winograd and Flores (1985), applying a
hermeneutical critique, restate the point that AI is doomed
to failure because it is based entirely upon
information-processing precepts and does not account for
self-reference inherent in language use.

5.  A person sample (P sample) likewise can be selected
according to design principles.  An unstructured P sample
could include anybody available and willing to participate
in the study.  However, more deliberate sampling might be
preferred to insure representativeness of opinions according
to people's positions in the debate.  Thus, one might select
an equal number of people in the following conditions:
those expressing (a) pro and (b) con opinions and who are
(c) political elites (Representatives, Senators, state
legislators, judges, etc.), (d) opinion leaders (editors,
commentators, journalists), (e) academics (law professors),
and (f) lay people ("man-in-the-street").


                         References

Bridgman, P.W. (1927).  The logic of modern physics.  New
   York: Macmillan.
Brown, S.R. (1970).  Consistency and the persistence of
   ideology: Some experimental results.  Public Opinion
   Quarterly, 34, 60-68.
Brown, S.R. (1971).  The forced-free distinction in Q
   technique.  Journal of Educational Measurement, 8,
   283-287.
Brown, S.R. (1980).  Political Subjectivity.  New Haven:
   Yale University Press.
Brown, S.R. (1986).  Q technique and method.  In W.D. Berry
   & M.S. Lewis-Beck (Eds.), New tools for social
   scientists.  Beverly Hills: Sage.
Brown, S.R. (1987, July).  A feeling for the organism:
   Understanding and interpreting political subjectivity.
   Paper presented at the annual meeting of the
   International Society of Political Psychology, San
   Francisco.
Brown, S.R. & Brenner, D.J. (Eds.). (1972).  Science,
   psychology and communication.  New York: Teachers College
   Press.
Brunner, R.D. (1977).  An "intentional" alternative to
   public opinion research.  American Journal of Political
   Science, 21, 435-464.
Converse, P.E. (1964).  The nature of belief systems in mass
   publics.  In D.E. Apter (Ed.), Ideology and discontent.
   New York: Free Press.
Cottle, C.E. & McKeown, B.F. (1981).  The forced-free
   distinction in Q technique: A note on unused categories
   in the Q sort continuum.  Operant Subjectivity, 3, 58-63.
Dirac, P.A.M. (1971).  The development of quantum theory.
   New York: Gordon and Breach Science Publishers.
Freud, S. & Jung, C. (1974).  The Freud/Jung Letters:
   Correspondence Between Sigmund Freud and Carl Jung.
   (Bollingen series, Vol. 94.)  (W. McGuire, Ed.).
   Princeton: Princeton University Press.
Heisenberg, W. (1952).  Philosophical problems of quantum
   physics.  Woodbridge, CT: Ox Bow Press.
Hoffmann, B. (1959).  The strange story of the quantum.  New
   York: Dover.
Lasswell, H.D. (1938).  Intensive and extensive modes of
   observing the personality-culture manifold.  Yenching
   Journal of Social Studies, 1, 72-86.
McKeown, B. (1984).  Q methodology in political psychology:
   Theory and interpretation in psychoanalytic applications.
   Political Psychology, 5, 415-436.
McKeown, B. & Thomas, D. (1988).  Q Methodology.
   (Quantitative Applications in the Social Sciences series,
   Vol. 66).  Newbury Park, CA: Sage.
Palmer, R.E. (1969).  Hermeneutics.  Evanston, IL:
   Northwestern University Press.
Petersen, A. (1968).  Quantum physics and the philosophical
   tradition.  Cambridge, MA: MIT Press.
Stephenson, W. (1935).  Technique of factor analysis.
   Nature, 136, 297.
Stephenson, W. (1953).  The study of behavior.  Chicago:
   University of Chicago Press.
Stephenson, W. (1967).  The play theory of mass
   communication.  Chicago: University of Chicago Press.
Stephenson, W. (1968).  Consciousness out -- subjectivity
   in.  Psychological Record, 18, 499-501.
Stephenson, W. (1972).  Applications of communication
   theory: I. The substructure of science.  Psychological
   Record, 22, 17-36.
Stephenson, W. (1974).  Methodology of single case studies.
   Journal of Operational Psychiatry, 5(2), 3-16.
Stephenson, W. (1977).  Factors as operant subjectivity.
   Operant Subjectivity, 1, 3-16.
Stephenson, W. (1980).  Consciring: A general theory for
   subjective communicability.  In D. Nimmo (Ed.),
   Communication Yearbook 4.  New Brunswick, NJ: Transaction
   Books.
Stephenson, W. (1982).  Q methodology, interbehavioral
   psychology, and quantum theory.  Psychological Record,
   32, 235-248.
Stephenson, W. (1983).  Against interpretation.  Operant
   Subjectivity, 6, 73-125.
Stephenson, W. (1986).  Protoconcursus: The concourse theory
   of communication.  Operant Subjectivity, 9, 37-58, 73-96.
Stephenson, W. (1988).  William James, Niels Bohr, and
   complementarity: IV -- the significance of time.
   Psychological Record, 38, 19-35.
Stephenson, W. (1988/1989).  The quantumization of
   psychological events.  Operant Subjectivity, 12, 1-23.
Thomas, D.B. (1976).  Exploring the personality-ideology
   interface: Q sort consideration of Tomkins' polarity
   theory.  Experimental Study of Politics, 5(3), 34-89.
Tomkins, S.S. (1963) Left and right: A basic dimension of
   ideology and personality.  In R.W. White (Ed.), The study
   of lives.  New York: Springer.
Winograd, T. & Flores, F. (1985).  Understanding computers
   and cognition.  Norwood, NJ: Ablex.

------------------------------------------------------------
     Bruce F. McKeown is in the Department of Political
Science, Westmont College, 955 La Paz Road, Santa Barbara,
CA 93108, U.S.A.

                  Copyright 1990
Communication Institute for Online Scholarship, Inc.