QDA Miner User’s Manual 133
To retrieve text segments associated with a specific cluster, click anywhere on a cluster
to select it (the selected cluster is displayed with thicker black lines), and then click on
this button to retrieve the associated documents. This operation retrieves all pairs of
coded segments associated with the selected cluster.
Press this button to append a copy of the graphic in the Report Manager. A descriptive
title will be provided automatically. To edit this title or to enter a new one, hold down
the SHIFT keyboard key while clicking this button
This button allows storing the displayed dendrogram into a graphic file. QDA Miner
supports three different file formats: .BMP (Windows bitmap files), .PNG (Portable
Network Graphic compress files) and .JPG (JPeg compressed files).
Note: Clustering using other similarity or distance measures or agglomeration methods may be achieved
using Simstat and the MVSP multivariate analysis add-on module.
2D and 3D Concept Maps
The concept maps are graphic representations of the proximity values calculated on all included items
(codes or cases) using multidimensional scaling (MDS). When clustering codes, each point represents a
code and the distances between pairs of points indicate how likely these codes tend to appear together. In
other words, codes that appear close together on the plot usually tend to occur together, while codes that
are independent from one other or that do not appear together are located on the chart far from each
other. The interpretation of multidimensional scaling plot is somewhat different when analyzing cases. In
this case, each point represents a case and the distance between pairs of points indicate how similar the
two cases are. Cases with similar patterns of codes will tend to appear close each other, while dissimilar
cases will be plotted far from each other. Colors are used to represent membership of specific items to
different partitions created using hierarchical clustering. The resulting maps are useful to detect
meaningful underlying dimensions that may explain observed similarities between items.
Please note that since multidimensional scaling attempts to represent the various points into a two- or
three-dimensional space, some distortion may result, especially when this analysis is performed on a
large number of items. As a consequence, some items that tend to appear together or that are parts of the
same cluster may still be plotted far from each other. As well, performing a multidimensional scaling on
a large number of items usually produces a cluttered map that is hard to interpret. For these reasons,
interpretation of concept maps may be feasible only when applied to a relatively limited number of items.
2D and 3D Map controls
NO. CLUSTERS - This option is used to set how many clusters the clustering solution should
have. Different colors are used both in the dendrogram and in the 2D and 3D maps to indicate
membership of specific items to different clusters.
134 QDA Miner User’s Manual
The actual orientation of axes in the final solution is arbitrary. The map may be rotated in
any way desired; the distances between items remain the same. The rotating knob can be
used to adjust the final orientation of axes in the plane or space in order to obtain an
orientation that can be most easily interpreted.
Clicking this button enables you to zoom in on a plot. To zoom in on a area of the plot, hold
the left mouse button down and drag the mouse down and to the right. A rectangle indicates
the selected area. Release the left mouse button to zoom in.
Clicking this button restores the original viewing area of the plot.
Clicking this button performs another multidimensional scaling on a new random
configuration of points. This button is visible only when the initial configuration is set to
Press down this button creates a constrained multidimensional scaling. This clustering
algorithm now allows one to preserve the clustering structure in multidimensional scaling
plots, making the interpretation of 2D and 3D MDS maps a lot easier and more consistent
with the clustering solutions. Enabling this option allows one to use the MDS module to
create maps of concepts similar to those suggested by Trochim, in its Concept Mapping
Pressing down this button displays lines to represent relationships between data points of the
multidimensional scaling plot. When the button is down, a cursor will appear in a tool panel
below the plot, allowing you to select the minimum association strengths to be displayed.
This button is used to create a copy of the chart to the clipboard. When this button is clicked,
a shortcut menu appears in which you can select whether the chart should be copied as a
bitmap or as a metafile.
Press this button to append a copy of the graphic in the Report Manager. A descriptive title
will be provided automatically. To edit this title or to enter a new one, hold down the SHIFT
keyboard key while clicking this button.
This button allows editing of various features of multidimensional scaling plots such as the
appearance of value labels and data points, chart and axes titles, the location of the legend,
etc. (see Multidimensional Scaling Plot Options below)
This button allows storing the displayed multidimensional scaling plot in a graphic file.
QDA Miner supports four different file formats: .BMP (Windows bitmap files), .PNG
(Portable Network Graphic compress files) and .JPG (JPeg compressed files) as well as
.WSX a proprietary file format (QDA Miner Chart file). Charts stored in the latter format
may be opened, further edited and customized using the external Chart Editor utility
Clicking this button prints a copy of the displayed chart.
Clicking this button closes the Dendrogram & Concept Map dialog box and returns to
QDA Miner's main window.
QDA Miner User’s Manual 135
3D Map buttons
This button shows or hides left, bottom and back walls.
Clicking this button draws anchor lines from the floor to the data point to better locate data
points in all 3 dimensions.
Clicking this button swithches the data of the X, Y and Z axes.
Clicking this button allows changing the viewing angle of the chart. To rotate the chart, make
sure this button is selected, click any area of the chart, hold the mouse button down and drag
the mouse to apply the desired rotation.
Locating a data point on the depth dimension of a 3D plot can be very difficult, especially
when the plot remains static. One often has to rotate this plot constantly on the various axes to
obtain an accurate idea of where the data point is located on this third axis. Clicking this
button forces QDA Miner to rotate the plot automatically. To disable automatic rotation, click
a second time.
Multidimensional Scaling Plot Options Dialog Box
The various options in this dialog box are used to customize the appearance of multidimensional scaling
plots. These options represent only a small portion of all settings available.
To further customize the chart, edit data points, value labels, etc., click the
button located at the right
side of the dialog box.
VIEW DATA POINTS - By default, multidimensional scaling plots display both the data points
and their associated labels. This option toggles on and off the display of the data points.
STYLE - This option defines the shape used to display the data points. Nine different pointer
shapes may be used to represent data point (e.g. square, triangles, dots, cross, etc.).
SHADOW - This option toggles on and off the shade on the sides of data points. This option
only affects square, triangle and down triangle data points when viewed in three-dimension.
SIZE - This option specifies the width and height of data points in pixels.
136 QDA Miner User’s Manual
TRANSPARENT LABELS - This option allows you to specify whether data labels are to be
displayed on an opaque background or whether the background should be invisible.
BACKGROUND - This option allows you to select the background color of data labels when not
BORDER - This option enables or disables the rectangle surrounding the label background.
VERTICAL GAP - The vertical gap property determines the vertical distance in pixels between
the top of the data point and the bottom of its label.
FONT - Click this button to adjust the font properties used to display data labels (i.e., style, size,
WALLS AND FRAME PAGE
WALL VISIBLE - Use this option to toggle the display of left, back and bottom walls to
simulate a 3D effect.
TRANSPARENT - The Transparent property controls whether the wall backgrounds will be
opaque or transparent.
DARK 3D - This option shades the sides of the walls.
SHOW TITLE - By default, multidimensional scaling plots have no title. To display a title,
enable the Show Title option and enter the desired title in the edit box to the right of this
check box. Enter several lines of text by pressing the <Enter> key at the end of a line before
entering the next line.
The FONT button at the bottom of this page is used to change the font size or style of this title.
QDA Miner User’s Manual 137
Cluster analysis and multidimensional scaling are both data reduction techniques and may not accurately
represent the true proximity of codes or cases to each other. In a dendrogram, while codes that co-occur
or cases that are similar tend to appear near each other, one cannot really look at the sequence of codes as
a linear representation of these distances. You have to remember that a dendrogram only specifies the
temporal order of the branching sequence. Consequently, any cluster can be rotated around each internal
branch on the tree without in any way affecting the accuracy of the dendrogram. The best analogy here is
to think of a Calder mobile. Different photos for this type of mobile will yield different distances
between hanging objects. While multidimensional scaling is a more accurate representation of the
distance between objects, the fact that it attempts to represent the various points in a two- or three-
dimensional space may result in distortion. As a consequence, some items that tend to appear together or
to be very similar may still be plotted far from each other.
When looking at coding co-occurrences, selecting an item enables the
button. Clicking this button
retrieves every pair of coded segments co-occurring, allowing one to further explore the factors that may
explain this co-occurrence.
The proximity plot is the most accurate way to graphically represent the distance between objects by
displaying the measured distance from a selected object to all other objects on a single axis. It is not a
data reduction technique but a visualization tool to help you extract information from the huge amount of
data stored in the distance matrix at the origin of the dendrogram and the multidimensional scaling plots.
In this plot, all measured distances are represented by the distance from the left edge of the plot. The
138 QDA Miner User’s Manual
closer an object is to the selected one, the closer it will be to the left.
To select a code or a case that will be used as the point of reference, you can choose it from the CODES
or CASE drop-down list located at the top of the page. You can also freely browse through different
codes or cases by double-clicking its line in the proximity plot.
The table view allows you to examine in more detail the numerical values behind the calculation of these
plots. When the distance measure is based on co-occurrences, the table provides detailed information
such as the number of times a given code co-occurs with the selected one (CO-OCCURS) and the
number of times that it appears in the absence of this selected code (DO NOT). This table also includes
the number of times the selected code appears in the absence of the given code (IS ABSENT). In the
example below, we can see on the first line of the table that the INSTRUMENTAL BEHAVIOR code
co-occurs 3727 times with SOCIAL BEHAVIOR, but on 3376 occasions, this code is encountered
without the presence of SOCIAL BEHAVIOR, while this last code is found to occur 1182 times in the
absence of INSTRUMENTAL BEHAVIOR.
QDA Miner User’s Manual 139
Code Sequence Analysis
While coding co-occurrences analysis looks at the concomitant presence of codes regardless of their
order of appearance in the documents, the Code Sequences finder is used to identify codes that not only
co-occur, but do so in a specific order and under specific conditions. This command list the frequency of
all sequences involving two selected sets of codes as well as the percentage of time one code follows or
is followed by another one.
To search for code sequences, select the CODING SEQUENCES analysis from the ANALYSIS menu. A
dialog box similar to the one below will appear:
The first page of the dialog box is used to restrict the analysis to specific document variables or specific
codes and to set the conditions that must be met to determine whether a code follows another one.
SEARCH IN - This option allows you to specify the document variable to search. If the current
project contains more than one document variable, you will have a choice of selecting either one
or more of them. By default, all document variables are selected. To restrict the analysis to a few
of them, click the down arrow key at the right of the list box. A list of all available document
variables will appear. Select all variables on which you want the search to be performed.
FIRST CODES and SECOND CODES - These two options allow you to select which specific code
sequences will be analyzed. By default, the analysis is performed on all codes in the codebook,
providing you with a comprehensive list of all sequences found in the documents. The list of
sequences can however be limited by restricting either the leading or the following codes. For
example, if someone is interested in the reactions of a therapist to client verbalizations, he may set
the First Codes option to include only the codes used to describe client verbalizations and limit
140 QDA Miner User’s Manual
the codes in the Second Codes option to those used to describe the therapist's reaction. You can
select specific codes for these two options either from the corresponding drop-down checklist by
clicking the arrow button or from a tree representation of the codebook by clicking the
MINIMUM DISTANCE - This option is used to specify whether two codes will be considered to
form a sequence when they partially overlap. If the first option is selected, a code can be
considered as following the first code despite the fact that the text segment to which it was
assigned starts before the end of the segment associated with the first code. However, if the
second radio button is selected, QDA Miner will not consider these overlapping codes to be code
sequences. For example, in the following example:
If the option is set to Start of second segment can overlap the first segment, then QDA Miner
will treat the Local Economy and Free Trade codes as a valid sequence. If the Minimum
Distance option is set so that the start of the second code must follow the end of the first one, then
the same pair of codes will not be considered as a sequence and will not be included in the results
MAXIMUM DISTANCE - This option is used to specify what should be the maximum distance
separating two codes in order to consider them to form a sequence. If disabled, QDA Miner will
consider a code to follow another one as long as it is in the same document and meets the
minimum distance criteria (see above). When enabled, code sequences are restricted to codes
separated by no more than a predefined distance. This distance is calculated from the end of the
first coded segment to the beginning of the second coded segment and can be measured either in
number of characters, words, or paragraphs, or alternatively by setting a maximum number of
codes that can occur between them. If the minimum distance is set to zero code, then only the
codes following immediately the first code will be considered to be a sequence.
To store the search expression and options, click the
button and specify the name under which those
query options will be saved.
To retrieve a previously saved query, click the
button and select from the displayed list the name of
the query you would like to retrieve (for more information, see Saving and Retrieving Queries on page
Once the criteria have been set properly, click the
button to find all sequences meeting those
criteria. If at least one sequence if found, the Frequency Matrix and Search Hits page become enabled.
Documents you may be interested
Documents you may be interested