HOW TO APPLY Q METHODOLOGY

by Simon Moss

Introduction

Sometimes, researchers want to be able to characterize the variety of subjective opinions

individuals adopt on various issues. For example, researchers might want to describe the range of
opinions people have formed towards

• the best approaches to conserve energy and the environment

• the strategies they like to use to improve learning or memory
• how politicians or managers should prioritize funding
• the circumstances in which euthanasia should be legal
• specific people, such as the president or celebrities

Q methodology is a suite of techniques that can be applied to characterize these subjective

opinions. This approach can be traced to the work of Stephenson (1935). In essence, researchers

• first instruct participants to sort a range of statements—such as statements about the main
drawbacks of some proposal—from most favourable to least favourable
• then, subject these rankings to statistical techniques that resemble factor analysis, except each
item is a separate person instead of a separate question

After you skim the following illustrations, the benefits of Q methodologies will become more
apparent.

Collect statements about your topic

Before you begin your study, you need to develop a series of statements or insights about
your topic of interest—the statements that participants will rank. The following table clarifies the
activities you should initiate to generate these statements.

Activity Example or clarification

Review your topic. Conduct a literature You might, for example, conduct a literature
review or consult other sources to extend your review about various responses to crime, such
knowledge about your topic of interest. as prison and community service.
Construct the concourse. Attempt to collate You might, for example, collate about 200
and enumerate all the opinions about this
 topic you can—sometimes called a concourse. statements about this topic, such as

• encourage perpetrators to meet victims

• impose suspended sentences
• organize anger management
Develop a Q sample. Identify a subset of To identify this sample, you could
these opinions or statements that you want to
present to participants. This subset, called a Q • classify the original set of statements into
sample, might comprise around 30 statements themes—such as traditional practices,
progressive practices, and untried practices.
• ensure your Q sample represents each
theme proportionately
• include statements that are unambiguous
and perhaps interesting

Choose your participants

The second activity is to choose suitable participants—sometimes called the P set or persons
set. To choose suitable participants, ask yourself a series of questions like

• who would have developed strong attitudes towards these opinions or statements
• whose attitudes are most relevant to my goals and objectives

To illustrate, suppose your topic revolved around the allocation of health funding. Your
participants might include politicians, hospital executives, medical practitioners, and patients. In
contrast, suppose your topic revolved attitudes towards immigration. Your participants might
include anyone from the general public. Typically, the number of participants might range from 15
to 50.

Ask participants to sort the statements

The third activity is to invite these participant to sort these statements. The following table
illustrates how you could achieve this goal.

Activity Example or clarification

Write each statement on a separate piece of You could, for example, write each statement
paper on a small Post-it note or piece of cardboard
 Ask participants to classify these statements The three piles could represent the statements
into three rough piles they agree with to the greatest extent, to a
moderate extent, and to the least extent
respectively
To help participants sort these statements See the form below
more precisely, present a Q-sorting form—a
form that represents a quasi-normal
distributions

Here is a Q-sorting form. This form helps participants sort the statements more precisely. This
form is suitable if the number of statements is about 35. If the number of statements is less than 35,
you might need to delete rows or columns. If the number of statements exceeds 35, you might need
to add rows or columns.

-4 -3 -2 -1 0 1 2 3 4
Most Most
disagree agree

Enter the data into a spreadsheet

The fourth activity is to enter the data into a spreadsheet, such as Excel. If you use statistical
software such as SPSS, SAS, or Stata, you can enter the data into this software instead. To enter the
data properly

• Each column should represent one person. In contrast, for many other statistical techniques,
each row represents one person
• Each row should represent a different statement. The order is not important, although
researchers usually arrange these statements into clusters
• Each cell should represent the score that participants assigned the label, often ranging from -4 to
4. Below is an extract of this spreadsheet
 Person 1 Person Person 3 Person Person 5
2 4
Impose prison time -4 2 -3 2 0
Impose home detention -3 -1 3 -1 3
Impose a suspended sentence 2 3 -4 4 -4
Impose community work 0 4 4 1 4
Encourage perpetrators to meet victims -1 -4 0 2 -3
Encourage perpetrators to help victims -3 1 -1 0 1
Ask an elder to punish the perpetrator 3 -1 -3 -4 0
Organize anger managenement 2 -4 3 2 -1
Organize psychological counselling -4 1 2 4 3
Inject a tranquilizer 2 3 -1 -4 1

Compute correlations

The fifth, albeit optional, activity is to generate a correlation matrix. This matrix is merely a
table that represents the correlation between each pair of participants. To demonstrate, the
following table illustrates this correlation matrix.

Person 1 Person 2 Person 3 Person 4 Person 5

Person 1 1
Person 2 0.32 1
Person 3 0.01 -0.03 1
Person 4 -0.03 0.02 0.41 1
Person 5 0.24 -0.18 0.01 -0.27 1
Person 6 -0.39 0.32 0.24 0.32 0.01
Person 7 0.02 0.03 -0.03 -0.18 -0.39
Person 8 -0.27 0.01 -0.39 0.02 0.24
Person 9 0.01 0.04 0.32 -0.03 -0.02
 Person 10 -0.02 -0.03 -0.27 0.24 0.32

In this table

• High positive numbers represent pairs of participants who assigned similar scores and thus
expressed similar opinions
• High negative numbers represents pairs of participants who assigned very different scores and
thus expressed conflicting opinions
• Numbers that approach zero represent pairs of participants who assigned moderately different
scores; their opinions were unrelated to each other
• To illustrate, in this example, Persons 1 and 2 assigned similar scores. Persons 1 and 6 assigned
conflicting scores.

Excel, SPSS, Stata, R, or indeed most statistical software can be utilized to generate this
correlation matrix. For example, in SPSS, you would merely need to

• Choose the “Analyze” menu. Select “Correlate” and then “Bivariate”

• Transfer the column labels, such as “Person 1” and “Person 2”, into the box called “Variables”
• Press OK.

Subject the data to factor analysis

The sixth activity is to conduct a factor analysis. If you are familiar with factor analysis
already, this activity will be straightforward. If you are unfamiliar with factor analysis, you should
still be able to understand this illustration. But, to conduct this technique, you might need to read a
simple textbook on factor analysis or seek some advice. To illustrate, if you wanted to conduct this
factor analysis in SPSS, you would merely need to

• Choose the “Analyze” menu. Select “Data reduction” and then “Factor analylsis”
• Transfer the column labels, such as “Person 1” and “Person 2”, into the box called “Variables”
• Press the “Extraction” button and perhaps choose the option “Principal Axis Factoring”, although
other options are equally suitable. Press continue.
• Press the “Rotation” button and perhaps choose the option “Varimax rotation”, although other
options are equally suitable. Press continue.
• Press OK to generate the following output.
• For Stata and other software, the procedure should be similar. For R, you need to write the
appropriate code

You now need to interpret the output this factor analysis generates. First, skim the table that
resembles the following—a table that includes information called “Eigenvalues”. To use this table,
simply

• Locate the column that lists all the eigenvalues—in this instance, the leftmost blue column
• Determine the number of eigenvalues that exceed 1—in this instance 3
• This finding merely indicates the participants can be divided into three clusters. Participants in
the same cluster assigned similar values to each statement. Participants in different clusters
assigned different or even conflicting values to each statement.

Initial eigenvalues Extraction sum of square loadings

Factor Total % variance cumulative Total % variance cumulative
% %
1 2.87 31 31 2.87 31 31
2 2.54 28 59 2.54 28 59
3 1.32 18 77 1.32 18 77
4 .87 7 85
5 .65 5 89
6 .26 4 93
7 .21 3 96
8 .10 2 98
9 .08 1 99
10 .05 1 100

Next, locate and interpret a table called the rotated factor matrix or pattern matrix. The
following table resembles this matrix. To utilize this table

• Circle or embolden the numbers that are higher than other numbers—usually numbers that
exceed .32
• These numbers indicate which participants correspond to each cluster, called a factor
• For example, in this instance, Persons 1, 2, and 3 correspond to the first cluster or factor
• Persons 4, 5, and 6 correspond to the second cluster or factor
• Persons 7 to 10 correspond to the third cluster or factor
• Unfortunately, in practice, and in contrast to this example, the factor to which person belongs is
not always as obvious.

Factor
Factor 1 2 3
Person 1 0.37 0.02 0.02
Person 2 0.41 0.04 0.01
Person 3 0.51 0.03 0.03
Person 4 0.04 0.54 0.04
Person 5 0.09 0.39 0.09
Person 6 0.02 0.61 0.02
Person 7 0.07 0.07 0.47
Person 8 0.01 0.01 0.51
Person 9 0.02 0.02 0.42
Person 10 0.02 0.32 0.62

Interpret the results

The final activity is to interpret the factors or clusters. In essence, your task is to clarify how
the responses of participants in the same cluster are similar. To illustrate

• You might notice that Persons 1 to 3—the participants in the first factor—tended to recommend
harsh punishments such as custodial sentences. You would thus conclude the first factor
represents harsh punishments
• You might notice that Persons 4 to 6—the participants in the second factor—tended to
recommend progressive practices, such as restorative justice. You would thus conclude the first
factor represents progressive approaches.
• You might notice that Persons 7 to 10—the participants in the third factor—tended to
recommend a blend of custodial sentences and counselling.

You would thus conclude that participants can be divided into one of three constellations: harsh
sentences, progressive approaches, or a blend of harsh sentences with counselling. You might even
conduct additional data analysis to clarify the characteristics of each cluster. You might reveal, for
example, that participants who prefer harsh punishments tend to be older. In contrast, participants
who prefer a blend of custodial sentences and counselling might tend to be more educated, and so
forth.
 Variations

Besides the topic, most participants utilize similar procedures to implement a Q

methodology. However, some features vary across research projects.

Repetitions with one participant

For example, sometimes, only one participant completes the Q sort. However, this
participant completes the Q sort many times under different instructions. To illustrate

• Each statement could represent one characteristic, such as extraverted, diligent, or moody
• The participant might first be asked to complete several Q sorts in which they order the
characteristics from most suitable to least suitable
• However, each Q sort would correspond to a different occupation
• For example, during the first Q sort, they would order the characteristics from most suitable for
doctors to least suitable for doctors
• During the second Q sort, they would order the characteristics from most suitable for
accountants to least suitable for accountants and so forth

Other variations

In addition, you should be aware of other variations. In particular

• Some researchers do not use a Q sort form. Participants might simply assign a number, perhaps
ranging from 1 to 9, to each statement. This approach is simpler, although the numbers are not
always dispersed enough to generate compelling results
• Sometimes, you might not be able to conduct Q sorts in person. You could send the statements
and Q sort forms to participants. You could send the participants a Q sort in Powerpoint—and
participants could be asked to move the statements into suitable spots in the Powerpoint slide.
Or you could use online software to enable participants to complete these Q sorts on computer.

Software
 People have developed software to facilitate Q sorts. Some of the software enables you to
collect data online. Other software facilitates the analysis. Often, but not always, this software is
free. For an example, google Ken-Q data and Ken-Q Analysis.

References and examples

Akhtar-Danesh, N., Baumann, A., & Cordingley, L. (2008). Q-methodology in nurs- ing research: A
promising method for the study of subjectivity. Western Journal of Nursing Research, 30, 759-773.

Baker, R., van Exel, J., Mason, H., & Stricklin, M. (2010). Connecting Q & sur- veys: Three methods to
explore factor membership in large samples. Operant Subjectivity, 34(1), 35-58.

Barbosa, J. C., Willoughby, P., Rosenberg, C. A., & Mrtek, R. G. (1998). Statistical methodology: VII. Q-
methodology, a structural analytic approach to medical subjectivity. Academic Emergency Medi-
cine, 5(10), 1032–1040.

Barry, J., & Proops, J. (1999). Seeking sustainability discourses with Q methodology. Ecological
Economics, 28, 337-345.

Berkhout, J. J., Teunissen, P. W., Helmich, E., van Exel, J., van der Vleuten, C. P. M., & Jaarsma, D. A.
D. C. (2017). Patterns in clinical students’ self-regulated learning behavior: A Q-methodology study.
Advances in Health Sciences Education, 22(1), 105–121

Brown, S. R. (1993). A primer on Q methodology. Operant Subjectivity, 16(3), 91-138.

Chinnis, A. S., Summers, D. E., Doerr, C., Paulson, D. J., & Davis, S. M. (2001). Q methodology: A new
way of assessing employee satisfaction. Journal of Nursing Administration, 31, 252-259.

Cross, R. (2005). Exploring attitudes: The case for Q methodology. Health Education Research, 20,
206-213.

Ho, G. W. K. (2017). Examining perceptions and attitudes: A review of Likert-type scales versus q-
methodology. Western Journal of Nursing Research, 39(5), 674–689.
Jeffares, S., & Dickinson, H. (2016). Evaluating collaboration: The creation of an online tool
employing Q methodology. Evaluation, 22, 91-107.

Kim, B. (1999). Understanding Q methodology and application in consumer studies. Asian Marketing
Journal, 1, 120 – 140.

McKeown, B. (2001). Loss of meaning in Likert scaling: A note on Q methodological alternative.

Operant Subjectivity, 24, 201-206.

Meshaka, R., Jeffares, S., Sadrudin, F., Huisman, N., & Saravanan, P. (2017). Why do pregnant women
participate in research? A patient participation investigation using Q-Methodology. Health
Expectations: An International Journal of Public Participation in Health Care & Health Policy, 20(2),
188–197.

Paige, J. B., & Morin, K. H. (2016). Q-sample construction: A critical step for a Q-methodological
study. Western Journal of Nursing Research, 38, 96-110.

Ramlo, S. E., & Newman, I. (2011). Q methodology and its position in the mixed-
methods continuum. Operant Subjectivity, 34(4), 172-191.

Song, S., & Ko, E. (2017). Perceptions, attitudes, and behaviors toward sustainable fashion:
Application of Q and Q‐R methodologies. International Journal of Consumer Studies, 41(3), 264–273.

Stephenson, W. (1935). Correlating persons instead of tests. Character and

Personality, 4, 17-24.

ten Klooster, P. M., Visser, M., & de Jong, M. D. T. (2008). Comparing two image
research instruments: The Q-sort method versus the Likert attitude questionnaire.
Food Quality and Preference, 19, 511-518.

Thompson, A. W., Dumyahn, S., Prokopy, L. S., Amberg, S., Baumgart-Getz, A.,
Jackson-Tyree, J., . . . Mase, A. S. (2012). Comparing random sample Q and R methods for
understanding natural resource attitudes. Field Methods, 25, 25-46.

Watts, S., & Stenner, P. (2012). Doing Q methodological research: Theory, method & interpretation.
Beverly Hills, CA: Sage.