-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathREADME.Rmd
168 lines (140 loc) · 11 KB
/
README.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
---
title: "Interpretive Clustering"
bibliography: inst/extdata/bibliography.bib
output:
html_document:
keep_md: yes
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
# NOTE: generate markdown for upload to github using:
# library(rmarkdown)
# render("README.Rmd", md_document(standalone = TRUE))
```
[](https://cran.r-project.org/package=OpenRepGrid.ic)
[](https://codecov.io/gh/markheckmann/OpenRepGrid.ic?branch=master) [](https://github.com/markheckmann/OpenRepGrid.ic/actions/workflows/R-CMD-check.yaml)
[](https://doi.org/10.5281/zenodo.7655618)
[](https://doi.org/10.21105/joss.03292)
# Interpretive Clustering
## Introduction
The [`OpenRepGrid.ic` R package](https://cran.r-project.org/web/packages/OpenRepGrid.ic/index.html)
is a browser-based software to perform *Interpretive Clustering* (IC) for repertory grid data. The
[repertory grid](https://en.wikipedia.org/wiki/Repertory_grid) (often abbreviated as *grid* or
*repgrid*) is a person-centered data collection method. It is primarily used in clinical psychology
and generates a mix of qualitative and quantitative data. IC is a method for clustering the
attributes (so called constructs) resulting from a repertory grid interview by finding relational
patterns between them. The IC method is described in more detail in our publication
[@burr_qualitative_2020](https://doi.org/10.1080/14780887.2020.1794088). An introduction to the
software is also available on [YouTube](https://youtu.be/f9oINYA22Rc). The package is part of the
[OpenRepgrid project](http://openrepgrid.org/), which contains several software packages for the
analysis of repertory grid data. In the following, we will briefly explain, why the software was
developed, describe the repertory grid technique and provide a short example of how a grid dataset
can be analyzed and interpreted using our
[`OpenRepGrid.ic`](https://cran.r-project.org/web/packages/OpenRepGrid.ic/index.html) software.
## Statement of need
Currently, the IC method is not implemented in any other existing repertory grid software. While IC
can also be conducted by hand, this is very time consuming, error-prone and only feasible for
small-sized grids. Hence, a software solution to support the IC procedure is needed. Without proper
software support that facilitates the IC steps, the method is too laborious and is thus likely to
become a methodological contribution which will rarely be used in research due to above mentioned
reasons.
## Repertory Grid Technique
The repertory grid technique (RGT) is a method which originated from *Personal Construct Theory
(PCT)* [@kelly_psychology_1955]. It was originally designed as an instrument for clinical psychology
but quickly spread to other disciplines like marketing, political, organizational, and educational
research in the decades after its inauguration [@fransella_manual_2004]. The RGT is a
person-centered method which focuses on understanding how an individual sees, or in constructivist
terms *construes*, the world. The data collected by the RGT is both, *qualitative* and
*quantitative*. The qualitative part of the data consists of a list of bipolar attributes (e.g.
*light-hearted vs. depressed*), which are elicited during the repertory grid interview. These
attributes are called *constructs* in PCT terminology. Each construct consists of two *poles*, e.g.
*light-hearted* and *depressed*, with the poles usually being opposites. In PCT, the constructs
constitute the templates a person uses to construe a set of objects under consideration. These
objects are called *elements* in PCT terminology and are, for example, persons, brands, countries,
companies etc. In our example in Figure 1, the elements are a set of persons relevant to the grid
interviewee, for example, *father*, *mother*, *self*, or *ideal self*. The quantitative part of a
grid is generated by letting the interviewee assign a score (e.g. 0/1, or 1 to 6) to each element on
each of his/her bipolar constructs. The result of the RGT is usually displayed as a constructs *x*
elements matrix with the cells containing the ratings of each element on each construct. Figure 1
shows a grid matrix where binary ratings (0/1) were used to assess the elements. For example, the
element *father* received a score of 1 on the *isolated = 0 vs. sociable = 1* construct, indicating
that the father is construed as sociable. The *mother*, in contrast, received a score of 0 and is
thus construed as isolated. The grid matrix is the central result of the RGT and yields a) a set of
bipolar constructs which are relevant to the person and b) an assessment of each element (persons in
Figure 1) on these dimensions. The white/red color coding in Figure 1 additionally indicates which
construct poles (white = left, red = right) apply and facilitates to identify similarities between
elements. In its original form [@kelly_psychology_1955] and in our example, grid ratings are binary
(0/1), i.e., each element is assigned to one construct pole. Nowadays, also scales with more grades
(e.g. 1 to 6) are common. A general and more comprehensive introduction to the RGT can be found in
@fransella_manual_2004.
{width="90%"}
## Interpretive Clustering
IC is an idiographic method of interpretation which makes use of both, the qualitative and the
quantitative grid data. In contrast to thematic analysis or content analysis, which could be used to
identify themes in person's elicited constructs [see @braun_using_2006], IC identifies the relation
between the constructs via quantitative assessment. By assessing the quantitative construct
relations, implications which constructs hold for each other are identified. IC essentially
identifies clusters of constructs which hold implications for each other. These clusters (or
cliques) of constructs form the basis for a subsequent qualitative interpretation. We will provide a
brief interpretation example below. The example builds on the analysis results the software
generates as shown next. However, for a more thorough introduction to the IC method and
comprehensive illustrations of interpretations, the reader is referred to our publication
[@burr_qualitative_2020](https://doi.org/10.1080/14780887.2020.1794088). Please note that the IC
method only allows for binary grid ratings.
## Installing and running the software
The package can be downloaded from the CRAN repository via the R command
`install.packages("OpenRepGrid.ic")`. The software is then started as follows.
```{r eval=FALSE}
library(OpenRepGrid.ic)
ic()
```
Also, we provide a docker image under
<https://hub.docker.com/repository/docker/markheckmann/openrepgrid.ic> with a short instruction on
how to run it locally. A video of the software is displayed in Figure 2. It shows the various tabs
and processes a grid dataset from an interview with *Sylvia* which is also contained in the package.
The results of the IC analysis are not displayed interactively but are included in an MS Excel file
that can be generated and downloaded. Also the results of the intermediate IC steps as described in
[Burr, King, and Heckmann (2020)](https://doi.org/10.1080/14780887.2020.1794088) are contained. The
main purpose of the software is to automate the cluster identification step of the IC procedure,
which is a cumbersome and error-prone task if performed manually. In Figure 3, an extract of the
analysis results for Sylvia's grid and corresponding analysis settings as shown in Figure 2 are
displayed and subsequently discussed.
{width="100%"}
## Interpretation
Psychologically relevant information can be obtained from the interpretation of the network graphs.
What follows is a shortened example. More comprehensive examples are outlined in our publication.
In the resulting diagram for Sylvia's grid in Figure 3, a construct is indicated by a circle, with
(+) denoting the preferred and (-) the non-preferred pole. The diagram shows three clusters (also
called cliques), indicated by the colored hulls around several constructs. In Sylvia's case, the
three clusters are highly overlapping. Two of these are of particular interest, sharing a 'core' of
three constructs -- '(+) Wild, free *vs* (-) controlled, contained', '(+) Massive sense of space,
expansive *vs* (-) railed-off, small world' and '(+) Freedom, wildness *vs* (-) conventional', with
(+) indicating the preferred and (-) the non-preferred pole. In one cluster, these three constructs
are strongly associated with '(+) Verdant *vs* (-) dead, nothing thriving'; the association between
her preferred poles suggests that she is drawn to places that are thriving and green, wild and
expansive, as opposed to those which lack life, are small-scale, controlled and conventional.
However, these three constructs share another cluster with the construct '(+) Cosy *vs* (-)
Chaotic', where 'cosy' is her preferred pole. In this cluster, however, her desires for the wild,
free and expansive appear to be in tension with her desire for the 'cosy', as they are aligned with
her non-preferred pole 'chaotic'. The attraction of wild, free spaces for Sylvia is therefore not
straightforward.
The third cluster includes the '(+) Verdant *vs* (-) dead, nothing thriving' construct, which is
here associated with '(+) Exciting, a lot going on *vs* (-) flat, unvarying, depressed,
unenergetic', '(+) Dramatic *vs* (-) unvarying, goes on and on' and '(+) Variable *vs* (-) doesn't
change'. This suggests that to Sylvia 'verdant' spaces are also full of excitement, drama and
variability- they are full of life in these ways. However, the fact that these three constructs do
not cluster with the wild/expansive/freedom constructs indicates that they constitute a somewhat
separate idea for her. A 'wild' space for her need not be 'exciting', for example, although a
'verdant' space is likely to be both exciting and wild. Interpretive clustering therefore gives us
insight into some of the complexity of Sylvia's construing.
## Contributing
In order to maximize the package's usefulness for the research community, we welcome participation
in the package's development. Experienced R programmers are asked to make pull requests to the
[`OpenRepGrid.ic` GitHub repository](https://github.com/markheckmann/OpenRepGrid.ic), [report
issues](https://github.com/markheckmann/OpenRepGrid.ic/issues), or commit code snippets.
Non-technical oriented researchers are invited to send us feature requests or suggestions for
improvement.
## References