Word Cloud
Running Head: Word Cloud
Using Word Clouds to Visually Present Q Methodology Data and Findings
Susan Ramlo, PhD
The University of Akron
[email protected]
Ramlo, Susan (2011). Using word clouds to present Q methodology data and findings. Human
Subjectivity, 9(2), 99-111.
Word Cloud
Abstract
Typical Q methodology studies include either tables of statements or sorting grids
with statement numbers representing the views of each factor that has emerged within the
study. Yet it could be useful to Q methodologists to have a simpler, more visual way to
present this data to present the “big ideas” of perspectives and consensus determined in
these studies. In this article, the benefits of visual representations of data, especially in
mixed research methods like Q methodology, are discussed. Word clouds are a method
for visually presenting text data, typically keywords on websites and in this paper we
present a data set as a series of word clouds and also tables and sort-grids that are more
typically seen in Q methodology presentations within this journal and elsewhere.
Keywords: Visual, image, presentation, word cloud, Q methodology
Word Cloud
Using Word Clouds to Visually Present Q Methodology Data and Findings
Pictorial representation of data can organize and summarize research data in a
way that tabular versions of the same data cannot (Dickinson, 2010). This is why
numerical data is often presented graphically such as columns or as an X-Y scatter graph.
Such alternative presentations of numerical data offer more descriptive way of presenting
the data and perhaps discovering trends or other interpretations of the data not easily
made from the table versions of the data (Wainer, 2005). Graphs of numerical data are
common and even school-aged children are well versed in creating and interpreting
graphs of numerical data.
In computer programming and in manufacturing, flowcharts are used to diagram
the flow of computer steps or product, respectively. Flowcharts, like the graphical
representation of data, are relatively common and assist in communicating ideas. But
graphical analyses and flowcharts are not helpful visual representations for the types of
data that is typically presented in Q methodology studies. In most Q studies, researchers
present text data in tabular form (Brown, 7/28/2011; McKeown, 7/22/2011). No
graphical / visual presentations of Q data or findings were found with a search of the
literature.
The author has had electronic discussions with several Q researchers about the
topic of presenting data/findings from Q methodology studies in journal articles. The
consensus is that alternatives that are more creative may improve communication and be
methodologically advantageous. Word Clouds are a method of visually representing text
data. Although they are most typically used as a summative way to display the frequency
of keywords in websites (Cui et al., 2010), other applications have included providing a
Word Cloud
visual of the most important topics in surgical journals over a decade (McGee & McGee,
2011) and enhancing reflection in an online course (Hamm, 2011).
Visual representations of data help organize and summarize research data. These
representations can enhance the clarity and support for research findings (Dickinson,
2010). In this article, the author presents an historical and pragmatic review of data
presentation including visual displays and then connects these ideas with the typical
presentation of text data in Q studies. Visual, grid, and tabular presentations of the same
Q data will be used to demonstrate the benefits of using word clouds to present Q data
and support research findings.
VISUAL PRESENTATIONS OF DATA, FINDINGS AND IDEAS
Ancient cultures conveyed ideas via pictorial representations, including those still
found on cave walls. During the period 1750 to 1850, the language of science shifted
from words to pictures partially due to William Playfair, (1759-1823) who founded most
well-known numerical graphics of the bar chart, the pie chart, and the time-series line
graph (Dickinson, 2010; Wainer, 2005). These graphical representations of data
expanded researchers’ abilities to examine patterns. Certainly this mode of visual display
changed the way people examine and represent data (Wainer, 2005).
Whether quantitative, qualitative or mixed methods are used, researchers seek
ways to summarize and display their data and findings such that they maximize
communication of their findings. Pictorial presentations of data and findings can allow
researchers to observe and communicate more effectively patterns and trends that might
not be apparent through tabular or other means of presentation (Dickinson, 2010; Wainer,
Word Cloud
2005). In short, visual displays provide a different way for researchers to tell the story of
their data (Dickinson, 2010).
Possible visual displays include frequency comparisons, comparison by time, and
identification of themes or trends (Dickinson, 2010). These types of displays are
frequently found in the presentation of numerical data (Wainer, 2005). Yet both
numerical and text data benefit from these visual displays. Dickinson (2010) discusses in
detail the benefits of using visual displays with quantitative and qualitative research
findings. She notes that although some have discussed ways of presenting procedure,
contemporary researchers need to consider the visual presentation of findings. Visual
presentation is especially important because patterns are more easily discerned and
themes magnified through visual data exploration and discovery. Mixed methods
research creates a need for new ways of communicating research data and findings
(Creswell, 2010; Greene, 2008). Combining mixed methods with visual display
optimizes research data investigation and communication (Dickinson, 2010).
Q METHODOLOGY AS A MIXED RESEARCH METHOD
William Stephenson specifically developed Q methodology to make subjectivity
operant via factor structure (Brown, 1980, 2008, 2010; McKeown & Thomas, 1988;
Stephenson, 1953). Yet this unique method commences with the selection of a concourse
via qualitative research methods such as interviews and focus groups. In this way, and
others, Q methodology uses an interactive blend of qualitative and quantitative research
methods (Newman & Ramlo, 2010; Ramlo & Newman, 2011a; Stenner & StaintonRogers, 2004). Stenner and Stainton-Rogers (2004) stated that such a unique approach
was deserving of a unique term, qualiquantology. Yet others (Newman & Ramlo, 2010;
Word Cloud
Ramlo & Newman, 2010; Ramlo & Newman, 2011a; Ramlo & Newman, 2011b; Ramlo
& Newman, 2011c) have simply described Q methodology as fitting into mixed methods
research, represented by the interactive qualitative-quantitative research continuum
described by Ridenour and Newman (2008).
When quantitative and qualitative techniques merge into mixed methods research,
the ability to address a variety of new research purposes and questions becomes enhanced
(Ridenour & Newman, 2008). Yet mixing research methods can create other issues
whether they include requirements for new techniques and new procedures (Creswell,
2010) or issues related to determining research quality and best ways to analyze and
present data (Greene, 2008).
Perhaps this issue of data presentation of findings that are qualitative and
quantitative has affected what some (Brown, 7/28/2011; McKeown, 7/22/2011) call the
normalization or standardization of the presentation of data and findings in Q
methodology journal articles and books. Some journal editors have insisted on certain
formats for data and findings in Q studies (Brown, 7/28/2011) and this may have also
affected this normalization. Efforts to use other types of presentation of data and findings
has met with resistance (McKeown, 7/22/2011).
Greene’s (2008)discussion about the challenges of mixing qualitative-quantitative
data and analyses are pertinent for Q methodologists as is Dickinson’s (2010)contention
that visual representation of data and analyses are beneficial for mixed methods research.
Dickinson gives a variety of visual displays of qualitative and mixed methods research
data and findings in her book chapter within the Handbook of Mixed Methods in Social
and Behavioral Research, Second Edition. Although Dickinson does not discuss word
Word Cloud
clouds as a way to present findings from mixed methods research, other research studies
(Bhowmick, Griffin, MacEachren, Kluhsman, & Lengerich, 2008; Cui et al., 2010;
Hamm, 2011; McGee & McGee, 2011)have used these visual displays to communicate
data and/or findings that are text based. The purpose of this paper is to demonstrate the
use of word clouds to present Q methodology data and findings and its benefits for
exploring themes and patterns..
WORD CLOUDS
Word clouds are popular for website and text analysis. In a typical text analysis,
words of interest (e.g. from a document, journal titles or tags from a website) are placed
in a rectangular form. The font size and color of the words that are placed into the word
cloud to represent frequency and usefulness, respectively. Other options such as fontstyle and layout are available for enhancing visual appeal of the word clouds.
In a typical word cloud, tags from a website (or words from a document) are
packed into a rectangular region in which font size indicates tag popularity (or word
frequency) and font color indicates other useful information (Cui et al., 2010). The more
prominent (larger text size) the word is in the word cloud, the more frequently it appeared
in the text provided. Examples and access to free word-cloud software are available from
Wordle, http://www.wordle.net/. The purpose of word-clouds is to summarize important
terms in a visual presentation that helps synthesize the “big ideas” present whether it is
related to the important course content (Hamm, 2011) or issues related to surgery
(McGee & McGee, 2011). Opposed to simply listing the most important terms with a
frequency count in another form such as a table, the word clouds in these and other
Word Cloud
situations offer a graphic that offers “semantically meaningful clusters with visually
appealing layouts” (Cui et al., 2010).
PRESENTATION STANDARDIZATION OF DATA AND FINDINGS IN Q
A review of recent issues of Operant Subjectivity revealed that predominantly
researchers present the findings from Q studies in specific tabular forms. In a real sense,
the presentation of Q methodology study data and findings have become normalized such
that the overall feeling is a sense of standardization to get published. Personal
communications with two leading Q researchers and OS reviewers support this
contention (Brown, 7/28/2011; McKeown, 7/22/2011). Typical tables include the highest
and lowest ranked statements for each factor/view, the entire Q-sample with factor scores
(as grid positions or z-scores) for each factor , and factor loadings. Other studies include
statements with grid positions within the discussion section of the article, without the use
of tables. In summary, presentations in Q articles typically consist of a listing of
statements with factor scores, a brief description of the factors and the ensuing label
(McKeown, 7/22/2011).
In Q methodology, the focus is on measuring subjectivity which is made operant
via factor structure. The tables produced via the analyses of the Q sorts are imperative
for interpretation and description of the factors/views (Newman & Ramlo, 2010). But
perhaps other means of exploring these views exist beyond consideration of tables.
Dickinson (2010) offers that visual representations of data and findings, especially those
from mixed methods research, enhance the clarity and support for research findings.
Similarly, word clouds offer a visual way to communicate key words and phrases in a
variety of contexts (Cui et al., 2010; Hamm, 2011; McGee & McGee, 2011). Yet it
Word Cloud
seems that the best way to demonstrate the benefits of using visual representations of Q
methodology findings, such as word clouds, is to take study findings and present them in
both tabular and word cloud formats.
WORD CLOUD DEMONSTRATION
Recently, the author was involved with a needs-assessment related to caregiving
for aging adults [Ramlo & Berit – currently in review at a journal]. This study
investigated views of a specific population of Minnesotan caregivers at a small liberal
arts college. These caregivers were drawn from faculty, staff, and administrators via a
call for participants. In this study, the Q sample was developed from the open-ended
responses from an online survey about caregiving for aging adults. Eleven sorts resulted
in three factors/views: Dutiful Caregivers New to Caregiving, Nurturing and Prepared
Caregivers, and Loving and Fun Caregiving Relationship.
The third of these factors/views was bipolar. The written responses in
conjunction with the top five most like / most unlike statements were used to interpret
this factor as well as the others. The positive loader on the Loving and Fun Caregiving
Relationship was unique compared to the other views. This view had the most positive
experience of caregiving based upon his written comments and description of his current
caregiving situation and relationship with the aging adult he cares for. Therefore the
researchers determined that this view represented an ideal as far as a caregiver for an
aging adult.
The author has selected this view to use as an example for comparing different
presentation methods for describing the different views determined within a Q
methodology study. The Tables 1 and 2 contain the more typical presentation for a
Word Cloud
factor/view. The analyses of these tables, written comments, and the distinguishing
statements (not included here) led to the naming of the factor, Loving and Fun
Caregiving Relationship.
The factor/view focused on having a loving relationship with the aging adult(s)
(statement 46 at +5). The Factor 3 view sees being a caregiver as a way to get to know
the aging adult(s) in new ways (statement 9 at +5). They appear emotionally connected
and caring in relation to the aging adult. Although this is similar to Factor 2 (statements
10, 30, and 1 at +4), the Factor 3 view is having more fun with their aging adult
(statement 42 at -5) than the other views (same statement at 0 and 1, respectively, for
Factors 1 and 2; this is a distinguishing statement for Factor 3). They feel prepared to
care for the aging adult’s finances (statements 35 and 40 at -4). The positive loader on
this factor described a loving relationship with his aging parent who still lives
independently. The negative loader described her relationship with her aging parent as
stressful with an absence of love and caring.
Table 1: Factor 3’s most-like (+5, +4) statements
No.
Statement
Grid
Position
46
My caregiving experience feels very loving.
5
9
I like being a caregiver for an aging adult because it lets me get
to know that person in a new ways.
5
10
Being a caregiver for an aging adult makes me feel like I am
giving back to someone who cared for me.
4
30
I need to learn how to better discuss important issues with an
aging adult.
4
1
I feel my role as a caregiver is to provide emotional support to
that aging adult.
4
Word Cloud
Table 2: Factor 3’s most-unlike (-4, -5) statements
Grid
Position
No.
Statement
40
I feel unprepared to deal with financial issues associated with
caregiving.
-4
29
I need to know how to better interact / communicate with my
aging adult.
-4
35
I need to learn how to manage financial issues including
insurance.
-4
42
I feel like I need to be more relaxed and have more fun with my
aging adult.
-5
45
I need help picking up the slack at home – meals, cleaning, etc –
while I am tending to the needs of an aging adult.
-5
Further insight was gained by using word clouds to examine the top most-like and
most-unlike statements from Factor 3. The +4 and +5 statements, shown in Table 1, were
copied and pasted into the word cloud website www.wordle.com. Like the exploration
involved with hand rotation, the researcher tried several iterations of word clouds. The
words AGING and ADULT were in most of the statements and overwhelmed the other
text so many of the AGING ADULT words were eliminated and the word cloud was recreated. This did not result in a loss of meaning; these words were included within
statements to help the sorters focus on the caregiving of aging adults rather than other
caregiving situations such as caring for children. Also, phrases lost meaning as
individual words in the word cloud so these were hyphenated, e.g. get-to-know, to
preserve the phrase and enhance interpretation/meaning. Words such as “a”, “the”, and
“I” are not included in the word cloud. Other options include font, layout and color.
These are purely aesthetic choices and do not affect the end result of the word cloud
Word Cloud
otherwise. The researcher selected the ChunkFive font, horizontal layout, and ghostly for
the color for both the most-like and most-unlike word clouds created from Table 1 for
comparison purposes. The word cloud is made on the Wordle website and screen capture
software, such as Jing 1, is needed to grab the image once it is created.
Figure 1 – Word cloud with text from the five most-like statements for Factor 3
Figure 1 presents the text and phrases from the most-like statements presented in
Table 1. Overall this figure represents a feeling of positive emotions; there is a sense of a
loving and caring relationship within this word cloud. Although the word “need” appears
in this figure, consultation with Table 1 indicates that this comes from statement 30.
Figure 2 contains the text from the five most-unlike statements from Factor 3
shown in Table 2. There is a more stressful feeling with this word cloud with the larger
text for NEED and FINANCIAL-ISSUES prominent in the image. Figure 2 allows the
1
Jing is available free via http://www.techsmith.com/download/jing/ for both PC and MAC.
Word Cloud
researcher to see an important aspect of this view that was missed initially from
consideration of the tables generated within the study. Although it was noted that the
positive loader commented on his preparation to be a caregiver (because of his
professional training) and the researcher examined the most-unlike statements for the
factor, Figure 2 symbolizes how almost carefree this view is relative to the other two
views that emerged within the study (which are not discussed here for brevity and to
maintain the focus of the article).
Because the other person on this factor loaded negatively, we can also use Figure
2, as well as Figure 1, to further understand the perspective of this negative-loader.
Figure 2 gives us a greater sense of how this negative-loader felt overwhelmed and
unprepared to be a caregiver. Figure 1 (which would contain her most-unlike statements)
also symbolizes her realization that her relationship with her aging parent lacked a sense
of love and rapport. Perhaps if she was less focused on financial-issues and needs, she
could have experienced something more like what is presented in Figure 1.
Word Cloud
Figure 2 – Word cloud with text from top five most-unlike statements for Factor 3
Finally, a word cloud can be used to examine the ideas presented in the Q-sample.
In a sense it is like performing a theme analysis of the Q-sample. It could be used to see
if there is balance among the ideas/themes within the Q sample, similar to how a Q-block
design is used.
In this study, a concourse of statements was selected primarily from the writings
of participants prior to their sorting session. Forty-eight statements were selected as the
Q-sample. Similar to the words used to create Figures 1 and 2, the phrase AGING
ADULT was removed from some of the Q-sample statements to create the Q-sample
word-cloud. Phrases were, again, hyphenated to preserve meaning that would be lost
otherwise. The result is shown in Figure 3. The themes of feel, need, issues, help, and
family emerge when we view the Q-sample as a word cloud. The Factor 3 view word
clouds can now be compared to the overall Q-sample to see what aspects are more-like
Word Cloud
and most-unlike this view. For instance, we can see the size of the font for the word
“NEED” in the entire Q-sample in Figure 3. Here it is somewhat smaller in font size than
the word “FEEL”. Yet in Figure 1, most like my view for Factor 3, the prominence/size
of the word “NEED” is reduced considerably and much smaller than the word “FEEL.”.
Yet in Figure 2, the word “NEED” is the most prominent with the largest font size of the
other text present including the word “FEEL.”
Figure 3 – Word cloud from the Q-sample’s 48 statements
CONCLUSIONS
Whether the research is qualitative, quantitative, or mixed, it can often be
difficult to generate a complete and overall Gestalt of meaning. The typically
presentations of data and findings in Q studies in journals are in tabular form. Yet
technology offers Q methodologists some new approaches for data presentation including
Word Cloud
those that are graphic/visual like word clouds. The benefits of visual representations of
data and findings were discussed, based upon the literature, and a demonstration using
word clouds was included within this article.
This researcher suggests using visual renditions of Q data (Q-sample) and
findings (tables of items with factor scores) in the form of word clouds. Word clouds
were designed to examine trends in text data that can be both fun and colorful (although
color was not selected in this demonstration). A sample from a data set from a recent
study was used to demonstrate the effectiveness of using word clouds to
represent/summarize the Q-sample as well as the most-like and most-unlike aspects of a
particular factor/view from this study. These word clouds provided insight that the
researcher initially missed when examining the more typical Q methodology tables and
also assisted her to describe the factor/view targeted here. Thus word clouds provide Q
researchers alternatives for exploration and communication. These types of visual
displays, and possibly others, can provide insight especially related to exploration of
patterns and communication of findings, especially to those not familiar with Q
methodology.
References
Bhowmick, T., Griffin, A. L., MacEachren, A. M., Kluhsman, B. C., & Lengerich, E. J.
(2008). Informing geospatial toolset design: Understanding the process of cancer
Word Cloud
data exploration and analysis. Health & Place, 14(3), 576-607.
doi:10.1016/j.healthplace.2007.10.009
Brown, S. R. (1980). Political subjectivity: Applications of Q methodology in political
science. New Haven, CT: Yale University Press.
Brown, S. R. (2008). Q methodology. In L. M. Given (Ed.), The Sage encyclopedia of
qualitative research methods (pp. 700-704). Thousand Oaks, CA: Sage Publications.
Brown, S. R. (2010). Q methodology. In N. J. Salkind (Ed.), Encyclopedia of research
design (pp. 1149-1155). Thousand Oaks, CA: Sage.
Brown, S. R. (7/28/2011). Personal communication re: Data reporting
Creswell, J. W. (2010). Mapping the field of mixed methods research. In A. Tashakkori,
& C. Teddlie (Eds.), Handbook of mixed methods in social & behavioral research
(Second ed., pp. 45-68). Thousand Oaks, Calif.: SAGE Publications.
Cui, W., Wu, Y., Liu, S., Wei, F., Zhou, M., & Qu, H. (2010). Context-preserving,
dynamic word cloud visualization. IEEE Computer Graphics & Applications, 30(6),
42-53. doi:10.1109/MCG.2010.102
Dickinson, W. B. (2010). Visual displays for mixed methods findings. In A. Tashakkori,
& C. Teddlie (Eds.), Handbook of mixed methods in social & behavioral research
(Second ed., pp. TBD). Thousand Oaks, Calif.: SAGE Publications.
Word Cloud
Greene, J. C. (2008). Is mixed methods social inquiry a distinctive methodology? Journal
of Mixed Methods Research, 2(1), 7-22. doi:10.1177/1558689807309969
Hamm, S. E. (2011). Using word clouds for reflection and discussion in an online class.
Teaching Theology & Religion, 14(2), 156-156. doi:10.1111/j.14679647.2011.00692.x
McGee, R. G., & McGee, L. M. (2011). A picture is worth a thousand words. American
Journal of Transplantation, 11(4), 871-872. doi:10.1111/j.1600-6143.2011.03459.x
McKeown, B. (7/22/2011). Personal communication re: Operant subjectivity
McKeown, B., & Thomas, D. (1988). Q methodology. Newbury Park, CA: Sage
Publications.
Newman, I., & Ramlo, S. (2010). Using Q methodology and Q factor analysis to facilitate
mixed methods research. In A. Tashakkori, & C. Teddlie (Eds.), Handbook of mixed
methods in social & behavioral research (2nd ed., pp. 505-530). Thousand Oaks,
CA: Sage Publications.
Ramlo, S., & Newman, I. (2010). Classifying individuals using Q methodology and Q
factor analysis: Applications of two mixed methodologies for program evaluation.
Journal of Research in Education, 21(2), 20-31.
Ramlo, S., & Newman, I. (2011a). Q methodology and its position in the mixed methods
continuum. Operant Subjectivity: The International Journal of Q Methodology,
34(3), 173-192.
Word Cloud
Ramlo, S., & Newman, I. (2011b). Response to Gourlay. Operant Subjectivity: The
International Journal of Q Methodology, 34(3), 213-214.
Ramlo, S., & Newman, I. (2011c). Response to Stenner. Operant Subjectivity: The
International Journal of Q Methodology, 34(3), 205-208.
Ridenour, C. S., & Newman, I. (2008). Mixed methods research : Exploring the
interactive continuum. Carbondale: Southern Illinois University Press.
Stenner, P., & Stainton-Rogers, R. (2004). Q methodology and qualiquantology: The
example of discriminating between emotions. In Z. Todd, B. Nerlich, S. McKeown
& D. D. Clarke (Eds.), Mixing methods in psychology (pp. 101-120). Hove, NY:
Psychology Press.
Stephenson, W. (1953). The study of behavior: Q-technique and its methodology.
Chicago: University of Chicago Press.
Wainer, H. (2005). Graphic discovery : A trout in the milk and other visual adventures.
Princeton, NJ: Princeton University Press.