Sei sulla pagina 1di 23

Qualitative Data Analysis Software

In: Handbook of Research Design & Social Measurement

By: Delbert C. Miller & Neil J. Salkind


Pub. Date: 2011
Access Date: February 21, 2018
Publishing Company: SAGE Publications, Inc.
City: Thousand Oaks
Print ISBN: 9780761920465
Online ISBN: 9781412984386
DOI: http://dx.doi.org/10.4135/9781412984386
Print pages: 165-179
©2002 SAGE Publications, Inc.. All Rights Reserved.
This PDF has been generated from SAGE Research Methods. Please note that the
pagination of the online version will vary from the pagination of the print book.
SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

Qualitative Data Analysis Software

In each systematic approach discussed thus far, we have highlighted procedures that
qualitative researchers might use in conducting their studies. This systematic procedure is
further encouraged by the use of computer packages that are helpful in developing a detailed
analysis of data in qualitative projects. In recent years, a number of books have addressed the
advantages and disadvantages of computer data analysis software programs (Fielding & Lee,
1998, Fisher, 1997; Tesch, 1990; Weaver & Atkinson, 1995; Weitzman & Miles, 1995).

These programs enable qualitative researchers to systematically analyze text or image files,
categorize and code information, build descriptions and themes, sort and locate important data
segments, and provide visual displays of codes and categories. With the refinements and
continual development of these software programs, qualitative researchers have little basis for a
comparative analysis of the programs and how they might be used in procedures of qualitative
approaches. To address this deficiency, we present here and evaluate seven programs using
eight key criteria that lie at the base of several approaches to qualitative data analysis.
Unquestionably, all the programs have features useful in conducting narrative research,
phenomenological inquiry, grounded theory studies, ethnographies, and case studies. In
addition, they all possess a rich history of software development by researchers and
demonstrate a commitment to future work in this area. The packages and companies who
produce them, including Web sites, are listed below.

ATLAS.ti—developed

by Scientific Software

Thomas Muhr

www.atlasti.de

ETHNOGRAPH5—developed by

Qualis Research

John Seidel

www.qualisresearch.com

HyperRESEARCH 2.5—developed by

ResearchWare

Page 2 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

Sharlene Hesse-Biber, T. Scott Kinder, Paul Dupuis, Ann Dupuis, and Richard Gaskin

www.researchware.com

Classic N4—developed by Qualitative Solutions and Research (QSR). (It is the new name for
what you may know as NUD∗IST or N4.)

Lyn and Tom Richards

www.qsrinternational.com

N5—developed by QSR

Lyn and Tom Richards

www.qsrinternational.com

NVIVO—developed by QSR

Lyn and Tom Richards

www.qsrinternational.com

WinMAX

Anne and Udo Kuckartz

www.winmax.de

The work of Tesch (1990), Weitzman and Miles (1995), Fielding and Lee (1998), Weitzman
(2000), and Fielding (2001) informs the structure and content of this section of Part 4. Our
concern is to present a balanced perspective and to examine each program as software offering
many possibilities for the user.

Our approach is not to declare a “best” program that covers the universe of qualitative data
analysis software but instead to select a few programs and allow you to assess the strengths
and weaknesses of each. Features that fit with a breadth of qualitative approaches shape the
discussion. We emphasize how qualitative researchers can integrate software into their desired
qualitative analysis approach. As shown in Table 4.2, we use a threefold classification scheme
—evaluations of how “flexible, fluid,” “comfortable,” and “helpful” each program is—to
summarize our assessment of how easily users can integrate each aspect of the programs into
their analysis styles. Think of the table as a usability scale.

Page 3 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

Flexible, fluid indicates that this feature of the program is easy to use, clear, and
designed in a way consistent with qualitative thinking and computer convenience
features. A flexible, fluid program generally is the most user-friendly.

Comfortable indicates that this aspect of the program is well designed and easy to
integrate into work with qualitative data. Compared to flexible, fluid it is not as carefully
fitted to qualitative approaches. A comfortable program, therefore, generally is not as
user-friendly as a flexible, fluid program.

Helpful indicates that the user can accomplish this goal with the software but in a style
that is more in line with the design of a particular software package as opposed to
common thinking in qualitative data methodology terms. A helpful program is the most
limited of the three categories.

The content of this review is based on work of ten consultants at ResearchTalk Inc., individuals
who use and teach about qualitative research techniques and software. From our field testing
of the packages and observations during consultation sessions with qualitative researchers, we
have solicited comments from users, and from these comments, we have assembled a picture
of what the qualitative process looks and feels like with different qualitative data analysis
software packages. Admittedly, this review is subjective and will change as new features are
added to the programs and as their tests and applications yield useful evaluative feedback.

Page 4 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

4.9.1 Making Informed Decisions

We begin with the need to make informed decisions before purchasing and using software. We
find that most scholars still have limited access to detailed, objective information about the
differences between major software packages. Marketing and word of mouth remain main
avenues for information. We find these sources to be less than reliable. Marketing material is
written and approved by producers of the product. Some companies are better versed in
marketing strategies than others. A recent advertisement from QSR, the creators of Classic N4,
N5, and NVIVO, claims that NVIVO signifies a “New Generation” in qualitative software, yet a
careful look at its features reveals little within the package that is new to the world of qualitative
software, with the ability to use rich text the primary novelty. The change from Classic N4 to
NVIVO makes the latter's features closely resemble features and approaches that have existed
in ATLAS.ti, ETHNOGRAPH5, HyperRESEARCH 2.5, and WinMAX. Our current generation of
qualitative software offers a range of attractive options for qualitative researchers seeking to
integrate software into their analysis without overwhelming their unique process.

Several of your colleagues may be using qualitative software. Their advice certainly can be
useful but should not be used exclusively. As discussed here, all the packages addressed do
well in meeting the goals of qualitative researchers, but in very different ways. These variations
may suit the preferences of one person more than another for very quirky reasons. Consider
this seriously. Eventually the package should sit in the background as your energy is directed
by the content of what you read, observe, and note. In addition, your colleagues’ choices may
have been made with an eye toward a limited range of options and/or prior to the release of
several of the current versions and upgrades. We meet people who avoid ETHNOGRAPH5
because a colleague voiced discomfort with experiences with ETHNOGRAPH. However, we
frequently find that people who are disappointed with ETHNOGRAPH used an earlier version,
sometimes even the DOS version. Much progress has been made with ETHNOGRAPH5 since
that time. Several other products also have changed. N5 is new to the market. Scientific
Software frequently releases free upgrades that offer valuable features. ResearchWare will
have released HyperRESEARCH 2.5 before the appearance of this discussion in print, which
motivates our discussion of it rather than HyperRESEARCH 2.0.

4.9.2 Eight Criteria for Assessing Software Programs

We organize our comparison around eight key factors critical to analysis of qualitative data with
software. Important questions about interaction during each phase of qualitative analysis that
we address are presented here:

Ease of integration. A software package needs to be easy to use at all stages in the process

Page 5 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

of its use. This criterion relates to the ease of both “getting started” in using the software and
“working through” qualitative analysis with different software packages.
Type of data. Two popular forms of qualitative data are text data and audiovisual materials.
Each software package is assessed in terms of the use of “text” data and multimedia (e.g.,
videotape) data.
Read and review text. Finding specific text is an important step for a qualitative researcher.
For each software package, the process of highlighting and relating quotations and the ease
in searching for text and text strings are reviewed.
Memo writing. A process often involved in qualitative data analysis is to write memos to
yourself that you use in analysis. Each package will be assessed in terms of its memo writing
capability and the ease of retrieving memos for use.
Categorization. Categorizing data into codes and more broadly into themes is central to many
data analysis procedures. Software needs to be reviewed for the process of using a
codebook, applying codes, displaying codes, and reviewing and adjusting codes.
Analysis inventory and assessment. Various analysis procedures in qualitative research call
for using specific codes or making connections among the codes to facilitate data analysis.
Each software package can be reviewed for sorting and filtering capabilities, use of
connecting codes, assessing code combinations, mapping concepts, and interacting with
demographic information.
Quantitative data. The flexibility of qualitative analysis appears when you can integrate or
combine quantitative data with qualitative data. Each package is reviewed for its capability to
import and export quantitative data.
Merging projects. Although qualitative researchers often work alone during data analysis,
there are times when multiple researchers work at different computers and need to merge
their work. The software programs can be reviewed for their capability to merge data from
different projects.

4.9.2.1 Ease of Integration

A core set of operations driven around reviewing, categorizing, and labeling text and images
forms the foundation of both major qualitative traditions and qualitative software packages.
What we as qualitative researchers must do when we come to software is integrate the program
into our existing strategies for questioning and discovery. We can and should weigh the pros
and cons and decide what or what not to implement and how.

Before using software, consider the following questions: What is your qualitative research
knowledge and experience? What is your level of computer expertise? Do you even like
computers? These and other issues further argue for finding a package you can integrate well

Page 6 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

with your individual style. This review of how easily you can integrate a specific package into
your qualitative analysis approach focuses on three main points:

The language, logic, and layout of the package


The degree to which the package takes advantage of existing computer facilities such as
drag-and-drop, multiple select, and right mouse clicks
The supporting documentation that comes with the package

After mention of which packages work with which computer platforms, we apply the above
criteria to two phases of interaction with qualitative data, “getting started” and “working
through.”

HyperRESEARCH 2.5 and Classic N4 are the only packages we reviewed that are built for both
Windows or Macintosh platforms. HyperRESEARCH 2.5 files can simply be shared across both
platforms, and QSR provides a program that can convert files from one platform to the other.
Weitzman (2000) also reports that even the most powerful PC programs sold today run well
with the aid of PC emulators on Macintosh machines currently available.

Getting Started

One of the most common questions we are asked is “How long will it take me to get started with
my new qualitative software package?” This question is difficult to answer, and we find some
differences in how quickly people become comfortable with different packages. The best
packages in this regard tend to feature a clear interface, a simple and familiar language for
qualitative analysis, user-friendly options for performing tasks, and helpful supporting
materials.

The language and design of ETHNOGRAPH5 are very straightforward. The codebook is
accessed via a button that has a picture of a book on it. Text can be copied, formatted, and
entered into your project via the ETHNOGRAPH5 editor. Right mouse clicks and button options
appear in sensible places and provide access to key operations. ETHNOGRAPH5's manual is
very clearly written. An appendix provides an insightful overview of the qualitative analysis
process.

HyperRESEARCH 2.5 is built around case cards. All code and annotation work relevant to a
specific case (each unit of analysis) is linked to its case card. This setup allows easy review of
issues connected to one participant in your study or quick comparative exploration across
participants. The language is consistent with traditional qualitative analysis language, and both
the manual and help menus are carefully and clearly written. WinMax works with four main

Page 7 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

arenas for interaction: list of texts, list of codes, list of coded segments, and working text. Each
section can be open on its own or in combination with any other. Initial work is directed via
interaction with your list of texts and list of codes. Although the WinMAX manual can be hard to
follow, the help menus direct you to steps necessary to achieve your goals.

ATLAS.ti can initially intimidate new users. We find that the primary source of this confusion is
that ATLAS.ti offers several different options for how to perform any function you need to use.
Once you figure out your most comfortable style and understand that ATLAS.ti allows you to
perform a similar set of key actions to key objects (documents, quotations, codes, and memos)
in your project, you can quickly find comfort with the program. For the most part, ATLAS.ti's
language is clear. The network diagram uses terms that can be uncomfortable, such as “import
neighbors,” used to bring linked quotes and codes into the diagram. Although the manual that
comes with the package can be frustrating, the extended manual (which you can print from the
CD that loads the software) and help menus are useful. ATLAS.ti takes the most advantage of
computer convenience features like drag-and-drop, right mouse click, and multiple select.

The good news is that once you adapt to a QSR product, you are probably very comfortable
with and loyal to it. We find that the major issue with QSR products is that initial and continued
use involves integration of a language and software structure that is less consistent with the
language and organization of qualitative traditions than any other product we feature here.
These concerns make QSR products the most difficult to integrate into your analysis and
computer use style. The foundation of each QSR product—Classic N4, N5, and NVIVO—is built
with a structure unique to QSR and a language that is often difficult for qualitative researchers
to adopt. Within NVIVO, memos are written as document or node links. “Nodes,” “Explorers,”
“Index Searches,” and “Assay Scope” are just a few terms that users find disconnected from
their normal language for qualitative analysis. In addition, codes and memos are organized into
a node structure within each package. Nodes can be either free or in an index tree. Nodes
within the index tree section are assigned addresses that are used for coding and questioning
throughout Classic N4 and to a lesser degree within N5 and NVIVO. In NVIVO and N5, project
pads are places where central functions are found.

The language and structure of QSR products add more distraction to new users than
enhancement to the research process. The vast majority of goals achieved with QSR products
are accomplished in other products with less cumbersome language and with interface
organization that is less awkward. The supporting material for all products meets mixed
reception, but it has improved. We find that once you are oriented to these products, you can
perform a sound, interesting, insightful analysis.

Page 8 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

Working Through

We discuss the fit of a package after initial orientation here. Does it get better or worse as you
use or work through the package? ETHNOGRAPH5 and HyperRESEARCH 2.5 are the most
consistent in convenient use and orientation throughout the life of a project. This strength
results from clean interfaces and from well-written and organized supporting materials.

The main dilemma for initial ATLAS.ti users concerns orientation to its layout and design. We
find that once an understanding of these factors is established, users achieve comfort and can
progress sensibly through the remainder of their interaction with the program. As your
understanding and interaction with the language and organization of QSR products increase,
so do your comfort and ability to grasp later features. You begin to see how the structure is
built to provide access to inventory and analysis tools used later in a project.

WinMAX users report initial disorientation when they begin to pursue issues of connection and
attempt to answer questions about the work they have done. This confusion is alleviated once
they realize that documents and codes about which they are curious must first be clicked and
activated before questions are initiated by clicking within the coded segment section and
reports requested via the file menu.

4.9.2.2 Type of Data

Text

It is important to know both how to format a document and the benefits and detriments to
following format suggestions offered by developers. All programs read ASCII text. You can
simply type your text and save as ASCII for any of them to read your text. As discussed below,
NVIVO can read rich text, and WinMAX has features that work well for researchers working with
open-ended responses to structured questions of a survey.

NVIVO allows you to use documents that contain color, bold, italic, underline, and other
formatting via its ability to read rich text documents. In addition, you can mark several levels of
structured sections that allow you to easily reference predefined sections of discussion such as
those that are common in structured interviews and meetings that follow the same form session
to session. Coloring documents, however, does not allow you to search for all “red” text, and
QSR has reported that researchers with larger databases can experience problems within
NVIVO. NVIVO offers character-based coding.

WinMAX will allow the use of rich text in its next version. Currently, it has a document feature
that aids researchers using open-ended responses to questionnaires. You can prepare your

Page 9 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

text in one word-processed document that contains each person's response. Markers you place
throughout the document allow WinMAX to recognize both all responses from each person in
your study and each person's response to each of your open-response questions. WinMAX
also offers character-based coding.

ATLAS.ti, ETHNOGRAPH5, and N5 read ASCII text and allow you to insert consistent labels on
structured sections of text that can be recognized for automatic sorting of each topic in each
document. Incorporation of text into your project is also straightforward. N5 lets you decide the
smallest character for coding as you enter the document: either lines, sentences, or
paragraphs. ETHNO-GRAPH5 allows you to code by line. ATLAS.ti offers character-based
coding. ATLAS.ti's next version will allow the use of rich text and other more dynamic document
formats.

Although HyperRESEARCH 2.5 has few requirements and challenges for document format, it
does not currently have the ability to recognize the end of key sections of discussion within
your data files. This limitation creates challenges for researchers working with structured data.
Hyper-RESEARCH 2.5 offers character-based coding. Classic N4 requires you to determine the
size of “text units,” the smallest amount of text to be coded, for your documents. Lines,
sentences, or paragraphs are the most common options for text units. Classic N4 requires you
to place hard returns at the ends of lines, sentences, or paragraphs upon document format
(this can be accomplished with relative ease via either “edit-replace” or “save as text only with
line breaks”). In addition, you have the option to insert headers (which are brief document
summaries) and/or sections (used for structured discussions) into the body of your document.
Users report that steps required to access these features can be more involved than they
desire.

Multimedia Data

The use of multimedia data is a key area that will continue to develop in the coming years. In
particular, you may soon witness the ability to see or have your video or your audio file run with
concurrent text streaming. In the meantime, ATLAS.ti, HyperRESEARCH 2.5, and NVIVO
currently allow you to work with multimedia files. ATLAS.ti and HyperRESEARCH 2.5 offer a
greater degree of interaction with these files.

ATLAS.ti and HyperRESEARCH 2.5 provide flexible options for use of graphic, audio, and video
files. Neither program contains facilities for you to digitize (create computer files of) your data,
but they read a range of file types for each file format. After your multimedia data are entered
into the program, you can code entire files or parts of the graphic, audio, or video file. Coded
segments can be reviewed during an examination of items coded to one category or while

Page 10 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

exploring a range of cooccurrence options such as overlapping segments or one topic inside
another.

NVIVO also allows you to use graphic, audio, or video files, but with a much less flexible
requirement. Multimedia files must first be broken into “data-bites.” Before you enter the file into
your project, you decide the size and break points of larger graphic files and enter those parts
into the program. Each data-bite must be coded in its entirety, resulting in a loss of fine coding
ability. ETHNOGRAPH5, Classic N4, N5, and WinMAX do not currently offer the ability to
introduce multimedia files.

4.9.2.3 Read and Review Text

Upon review of your documents, you may find quotations that deserve highlighting, sometimes
without any link to specific code categories or memos. This practice can be particularly useful
early in an analysis to simply recognize powerful quotations, but it can be adopted throughout
your data review. Think of this exercise as marking participant statements with a highlighter
pen.

Highlight and Relate Quotations

ATLAS.ti and NVIVO offer explicit functions that provide for this task. ATLAS.ti's options for
“quote-work” are particularly powerful, as they easily link to its network diagram feature.
ATLAS.ti considers quotes you mark as equal in status to your data documents, codes, and
memos. You can mark a segment of text as a “free quote.” Free quotes can be reviewed in
isolation and left “free” or can be linked to a code or memo later in a project. In addition,
ATLAS.ti's hyperlink feature lets you link connected quotes from a discussion that can be
displayed in a network diagram. In a study of life satisfaction, one of our male respondents
made a range of statements about his philosophy on how to treat people in his life. Some of his
statements were about why treating everyone with respect comes back to benefit you later in
life. At other points in his interview, he discussed life events that provided evidence of less than
decent treatment of people in his life. At one point he discussed how his extramarital affair was
a good thing for his wife. When you hyperlink these statements, you can label the connection
between them as discussions or criticisms of each other. Once labeled, each quote can be
displayed in a network diagram to visually represent the ebbs and flows of the discussion.

NVIVO supports non-code marking of text through a link feature. “Node links” connect quote
segments from different documents. They are contained in what NVIVO calls a node extract,
which is essentially an unnamed code. These extracts can be named at any point in a project,
and the status can be changed to make the original node extract a code. ETHNOGRAPH5,

Page 11 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

HyperRESEARCH 2.5, Classic N4, N5, and WinMAX contain easy options to begin to emulate
these features. The trick is to simply name a code category “interesting quotes” or
“contradictory statements” and then code quotes to these categories. Although this is not a
perfect option, this practice ensures that you do not lose the observations about these quotes.

Text Search

One of the things computers do well is find patterns of text strings (words). All the packages we
discuss can do that. The packages that do it better can not only search but also store results,
mark results in the same way a code is marked, allow these results to be included in questions,
allow you to store and explore context, and give you counts of word finds.

ATLAS.ti, HyperRESEARCH 2.5, the QSRs, and WinMAX accomplish the major goals we set for
text searching. Classic N4 and N5 cannot give pure counts of word instances because they can
only mark a text unit (QSR term for the smallest unit to be coded, either a line, paragraph, or
sentence). If text units contain more than one instance of a search string, counts would register
only one instance.

ETHNOGRAPH5 is primitive in comparison to the other packages in this regard. ETHNO-


GRAPH5 can search for words only one at a time within an individual data document. If you
elect to code “found” search strings, you must do so manually. The positive aspect here is that
you are forced to read and consider each instance. The other packages can all mark codes
automatically via a process commonly called autocoding. You can also opt to decide what
instances should be coded one by one in the other packages.

4.9.2.4 Memo Writing

Note writing is a critical aspect of several qualitative analysis approaches. We consider two
major issues in regard to memo writing. First, we discuss how well integrated the memo system
is within each program. Second, we discuss how readily you can retrieve memos you have
written.

Memo System Integration

A well-integrated memo system invites notes about an entire research project, about the
analysis process, on individual data documents, on codes, on specific text, on conceptual maps
—basically anywhere that written insight can benefit a project. ETHNOGRAPH5 offers a wide
range of flexibility for memo writing. Free-standing memos on any topic can be composed, and
there are opportunities to write memos on any document and referencing any text selection.
Memos written on specific text sit in the data document with an icon showing their presence.

Page 12 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

Clicking on the icon allows access and editing. ATLAS.ti offers “comments” that can be written
to any object in the project and memos that can be free-standing or linked to any quotation in
your data. As in ETHNOGRAPH5, margin icons for memos allow access for editing. Users
would like to see more direct pathways to link more than one quotation to the same memo and
to classify memo types.

HyperRESEARCH 2.5 handles memos as annotations. Annotations can be linked to codes


applied to text segments. There is no dedicated memo list. The link to codes on case card
design initially appears limiting, but easy work-around strategies such as creation of a “general
annotations” card for general topic memos and a document annotation memo at the top of each
document case card can satisfy most memo writing needs. WinMAX allows memo placement
on code categories and linked to any text instance. Similar to HyperRESEARCH 2.5, this
apparent limitation can be overcome with work-around tricks such as creating mock codes to
hold discussions of specific topics. Document memos can be written as text memos at the top
of a document and linked to a code called “document memos.”

In Classic N4 and N5, memos are attached to either a document or a node. Nodes should be
considered containers that hold codes, memos, both, or neither. This use of nodes allows you
to make a node that can hold a memo of any type, like a grounded theory analysis audit trail.
Annotations can also be written on any selection of text; however, the annotation actually
becomes text that exists in your data document, thus altering the original content and structure
of your data. At times, decision making on where to put a memo can be challenging.

NVIVO memo strategies are either liked or not liked. Every memo is considered its own
document, equal in status to your data, which can be coded. Some researchers find this
liberating, while others find it to be circular and unnecessary. Memos can be included as links
within documents. Icons for memos appear in the body text and can be clicked for access and
editing. Our clients who use NVIVO ask for appearance of the word “memo” and a dedicated
list of memos like those that appear in ETHNOGRAPH5, ATLAS.ti, and WinMAX.

Memo System Access

We find that researchers’ major source of frustration with memo writing is retrieving memos they
have written. Dedicated memo lists facilitate memo access. The ability to retrieve memo text as
it links to codes and code combinations, without first having to code memo text, provides
recovery for those who find themselves asking “Where did I put that note?”

ETHNOGRAPH5 offers the most thoroughly linked memo system. It contains an isolated, sort-
able memo list and the ability to request memo text written in concert with any code or code

Page 13 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

combination without document text. This facility permits focused examination of written notes.
ATLAS.ti offers a memo list you can sort and filter, and it can show memo notes along with code
and code combination reports. WinMAX has a dedicated memo list that can be filtered by any
document, code, or code combination. Memos written to specific text can be accessed when
that text is retrieved via code review in both ATLAS.ti and WinMax.

HyperRESEARCH 2.5, Classic N4, N5, and NVIVO do not have dedicated memo lists. Classic
N4 and N5 have memo buttons that can be clicked with any highlighted code or document.
Annotations are stored within documents and have their own home in the node explorer.
HyperRESEARCH 2.5 allows isolated retrieval of annotations as a single click option in a report
request. NVIVO offers options to group and focus on only memos, and it shows memo icons
along with text retrieval at any code or code combination report.

4.9.2.5 Categorization

We use the term “categorization” to address work that qualitative researchers do to identify
categories, themes, codes, and “bucket categories.” You should expect to see the word “code”
within each package. Qualitative software “codes” can be used to represent any type of
categorization you need to do to your data, even if you simply want to mark the most interesting
examples of what is coded to a category. You can make a “code” called “interesting stuff.” This
example demonstrates the kind of flexibility you should feel when working with your codes,
whether you are in a codebook, applying codes, working with code displays, or reviewing and
adjusting coded text.

Codebook

Codebooks should be built to accommodate adjustments. Changes to code names, definitions,


and location should be easy to make (Fielding, 2001). Codebooks should provide the ability for
focused work on the code list itself. The look of the codebook is also critical, especially during
code application. ATLAS.ti and ETHNOGRAPH5 both incorporate flexible views of your
codebook when taking an isolated look at your codes and displaying your code list while you
apply codes. It is easy to display and edit definitions of codes. Codes can be combined easily.
Changes to code names, definitions, and hierarchical relationships between codes can be
made very easily. ATLAS.ti adds detailed, straightforward, and accessible sorting and filtering of
your code list. ETHNOGRAPH5 offers options for sorting your codebook.

HyperRESEARCH 2.5's code list editor serves as its codebook. It is easy to use, and you can
easily make adjustments such as renaming, deleting, and merging codes with a right mouse or
control click. The main drawback of the code list editor is that it does not provide an option for a

Page 14 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

structured code list. If you want to organize a list of all actors in a study, you need to begin the
name of each category with the word “actor” (“actor-friends,” “actor-parents”) in order to group
the codes in the same location. Codebooks in both NVIVO and WinMAX are built to arrange
codes hierarchically. Codes can be moved to become a subcategory of a different code by
simply dragging and dropping. Making changes to names and definitions and combining codes
is straightforward, although WinMAX does not have a specific code definition box—instead, you
can write definitions in code memos, which might be awkward for some. Our clients report that,
compared to ETHNOGRAPH5 and ATLAS.ti, code work and code display in both WinMAX and
NVIVO are too structured.

Classic N4 and N5 provide a helpful, basic environment for building and presenting codes.
Code names and definitions are easy to create and alter. Combining codes is done via cut-and-
paste activities, rather than a simple merge or combine option. As in NVIVO and WinMAX, the
code list is hierarchical in nature, but it is more uncomfortable to use because node addresses
accompany names. You cut and attach codes to move them between subcategories rather than
performing a more straightforward drag-and-drop as in WinMAX and NVIVO.

Code Application

Qualitative software's life is long enough for us to now be fairly demanding about basic tasks
like code application and deletion. Codes should be easily visible alongside document text.
Different users should be able to find different options for how they decide to apply codes: right
mouse clicks, drag-and-drop, multiple select, and so on. Removing code applications should
require only a few mouse clicks.

ATLAS.ti offers a wide range of easy-to-use and flexible options for applying either newly
created or existing codes. You can drag-and-drop from a code list to a highlighted section. You
can highlight text and right mouse click or use buttons or menu options. To remove a code
application, you can right click the code in the margin display and “unlink” it.

ETHNOGRAPH5's quick code and code set options are easily used and accessed after you
highlight text via buttons or a right mouse click. Code applications can be removed easily by
clicking on a code marker and eliminating the code from the code set.

HyperRESEARCH 2.5 code application is very straightforward, flexible, and fast. You can create
new codes easily and multiple select to apply more than one code at a time. Right mouse and
control click options for MAC make code applications more convenient. Removing code
references is less straightforward. You have to move off the document text to the case card for
the person whose text you are reading.

Page 15 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

WinMAX code references can be added and deleted quickly within WinMAX. With text
highlighted, you can click on categories one by one to quickly apply codes. You can delete
codes by right clicking on the code indicator in the margin area. Coding could be even more
user friendly if drag-and-drop and right mouse options were available for coding. WinMax is
also the only program to offer a “code-with-weight” option. You can apply a weight score of 1-
100 to any code application. Weight scores are reported back upon code retrieval.

NVIVO's coding facilities are the easiest of all QSR products. You can drag-and-drop from a
code list, and new codes are easily created. Initial negotiation with NVIVO's coder can be
clumsy, but users adapt to it quickly. Strategies to use multiple selection of codes before
application are disappointing.

Coding in N5 is more convenient than in Classic N4 but still awkward compared to ATLAS.ti,
ETHNOGRAPH5, HyperRESEARCH 2.5, NVIVO, and WinMAX. Applying and deleting codes in
both Classic N4 and N5 is impaired without an active on-screen display of codes, a reliance on
node addresses, and the fact that the programs still takes less advantage of drag-and-drop,
multiple select, and other convenience factors than they could.

Code Display

One strategy researchers use when coding transcripts and conducting fieldwork without
computers is to use colored pencils to mark a bracket around text to signify where a code
application would begin and end. Then the researcher writes the name of the code(s) that the
section addressed next to the bracket. Qualitative software packages offer different versions of
code displays that are available either as you work or on request after you have coded
segments of text. ATLAS.ti and ETHNOGRAPH5 combine easy-to-read and easy-to-understand
code displays with the ability to activate each code reference to reflect your thinking about
specific code applications and/or categories. This ability to activate, query, and alter your
recorded thoughts on code work captures the ongoing flow of qualitative thought on concepts
integral to your project. This feature is what we point to as “state of the art” in this area.

ATLAS.ti offers a margin area display of codes to the right side of document text. You can
choose font type and size for the margin display. When you apply a code to a segment of text,
two things happen. First, a bracket appears that shows the start and stop point ofthe section.
Second, the code(s) you selected appear(s) next to the bracket.

You can right click on each code in the margin area to access information about the category
and record notes about it. An information display will show when a code was made, when it
was last modified, and its current comment. Editing a comment lets you build running

Page 16 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

commentary about a specific category. The same right mouse click on a code in the margin
area will allow you to view and work within a network diagram that contains that code. You can
also view all other quotes coded to that category or remove a code reference with the unlink
feature.

ETHNOGRAPH5 displays code names above the code segments where they are applied and
shows the range of lines the code is applied to in the right margin. You can double click a
displayed code to edit the code in its code set window, which allows you to delete the code
from the segment or add another code to this section of text. You can change or expand the
definition of the code or view it in the code family tree to group the code in relationship to other
codes. You also have the option to display all code sets for a given document while you work
with or review its codes. This feature provides a window to co-occurrences that may inform your
analysis.

HyperRESEARCH 2.5 “Codes in Context” appear to the left of document text. They appear at
the beginning of a code segment and display exact start and stop points of codes segments
only when a code is selected. The code display is visual only; you cannot click on the code
markers for more information or to make changes to information about the code. Code
instances also display on individual case cards to show sequences and frequency of codes for
each case in your study.

NVIVO displays codes as “code stripes” to the right of document text. Colored brackets mark
the beginning and end of code segments. Codes applied to the same segments appear in a
horizontal line. As in HyperRESEARCH 2.5, the code display is visual only, so it cannot be used
interactively. NVIVO's code stripes also do not scroll up and down in sync with the document
text.

WinMAX shows the start and end points of code segments, and it displays code names only
when you move to a code segment indicator. You can click on a code segment indicator to
delete that code reference, but no further work with or information about the category is
available.

Classic N4 and N5 do not offer code displays that update and build as you work. You can
request “examine coding” while you work; however, this feature does not indicate start and stop
points of coded segments. You can also request a report with “code stripes” that allows you to
select up to 26 codes to show in display. “Code stripes” appear in the right margin. Stripes are
not code names but alphabetic codes that are defined in a key at the top of the code stripe
report. Code indicators in “Code Stripe Reports” and “Examine Coding” are inactive.

Page 17 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

Review and Adjust Codes

Taking multiple passes at coded data and adjusting work as you move along is critical to careful
qualitative analysis. When reviewing and adjusting codes, you should be able to easily move
back to original text and have full access to the same features you had while first reviewing
documents. Most programs accommodate this process well.

ATLAS.ti, HyperRESEARCH 2.5, QSRs, and WinMAX all allow you to move back to your
original text while you review the contents of a code category. This direct access lets you easily
think more deeply about the logic behind your codes and coding and to adjust codes as you
deem necessary. You can remove code references, add new codes to sections you review, and
use the memo features of the program.

ETHNOGRAPH5 does not let you adjust code applications while you review code category
contents. There is, however, an easy facility to add and delete coding from a document as you
decide on changes. ETHNOGRAPH5 has another review feature worth noting. As you review
each segment coded to a category, you can “mark” instances. “Marking” serves to distinguish
select segments. After you mark segments, you can print or transfer to a word processor only
the marked segments.

4.9.2.6 Analysis Inventory and Assessment

Qualitative research requires easy access to information on what has been done to date.
Ideally, you should regularly assess the status of code lists, items coded to codes, and written
notes. This practice teaches you about each item and enables you to question and perhaps
change your current analysis strategies. Qualitative software packages contain several features
to facilitate satisfying your curiosity about what your work with your data means. Here we
discuss sort and filter abilities, concept combination tools, conceptual maps, and the ability to
do demographic comparisons.

Sort and Filter

This section details the options and uses of sort and filter facilities in qualitative data analysis
packages. Qualitative analysis often requires looking at information from different angles. You
might elect to view only statements by women or focus on only a specific set of codes. Sorting
codes or memos can help provide an inventory of what exists in your project and changes to
these items across the life of your analysis. Fortunately, sorting and filtering is something
computers do well.

ATLAS.ti has convenient sort and filter options for each of its main items in your project. You

Page 18 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

can sort documents, quotes, codes, and memos alphabetically, by recent usage, by date of
creation, by volume of text, or by inclusion in network diagrams to gain perspective on
relationships among these items. Filtering of the same set of items invites careful focus on key
points of inquiry.

HyperRESEARCH 2.5 sorts codes on case cards and annotations attached to them. Although
not as thoroughly integrated as the sort and filter options within ATLAS.ti, HyperRESEARCH 2.5
does provide convenient exploration of connections between documents and codes through
filtering. You can literally turn on documents and/or codes to focus on only one set of items at a
time.

NVIVO and WinMax permit filtering of codes and documents in similar ways to ATLAS.ti and
HyperRESEARCH 2.5. WinMax allows you to activate any number of codes, documents, and
demographic variables and to filter memos to focus only on items of interest to you. The filtering
tools in NVIVO are similar and useful, but they are difficult to find, buried under “set” sections of
document and node tabs of your main project pad. Sort options are not as extensive in either
NVIVO or WinMax as in ATLAS.ti and HyperRESEARCH 2.5.

Filters in ETHNOGRAPH5 are straightforward. Although they do not literally turn items on and
off as in ATLAS.ti, HyperRESEARCH 2.5, NVIVO, and WinMAX, they are very easy to use when
you examine items referenced at any one code or combination of codes. Access to memos and
work with codes in your codebook is made more dynamic by detailed sort options. Classic N4
and N5 do not have extensive sorting facilities within either package. Filtering is accommodated
via a restrict option on searches of code relationships but does not turn project items on or off.

Concept Combination Tool

We use the term “connecting codes” to discuss the range of tools within qualitative software
that allow you to make and assess connections among different codes. Each program has its
own method that basically assesses how your applied codes relate to one another. Query
Tools, Search Procedures, Hypothesis Testers, Index Search Operators, and Logic Machines all
let you pursue the ways codes combine to inform your analysis. Our evaluation of these tools
reveals a solid group of performers that provide you with powerful aids to analysis. Dynamic
code results in ATLAS.ti, iterative questioning (“system closure” according to Richards and
Richards [1994]), and matrix searches where you can pursue connection between and among
several subcategories of two different codes create a strong field of concept combination tools.
Although not readily accommodating matrix and iterative searching, ETHNOGRAPH5's search
procedure tool deserves recognition for its clear organization and direct access.

Page 19 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

ATLAS.ti contributes super-codes to the arena of concept combination. Super-codes are


dynamic results of searches. If you ask ATLAS.ti to show the places where a code for friends
appears within a code for issues of self and save that result as a super-code, you never need to
run that search again. Super-codes should be considered live results. Coding updates that
create more results to your original query are kept track of in your code list for easy retrieval.
ATLAS.ti lets you question connections between subcategories of codes and supports iterative
questioning.

HyperRESEARCH 2.5's “Select by Criteria” for codes and cases lets you easily pursue a range
of connections between codes and cases. The hypothesis tester lets you ask involved
questions of connection that can be pursued via the iterative questioning facilities of the other
programs, but not with the explicit language of if-then statements and reports that goals are
attained.

N5 offers a powerful range of question options that afford rich exploration of concept
connections, including iterative searching and matrix connections, and provides a new
organization and graphic diagrams to clear confusion that existed in Classic N4. Like ATLAS.ti,
ETHNOGRAPH5, and WinMAX, NVIVO condenses search options into a single window that lets
you assess code connection and text search at once for an entire data set or a focused look at
only part of a data set. The logic machine of WinMAX is convenient and flexible. It invites
iterative questioning and searching connections between subcategories in an easy-to-use
single window.

ETHNOGRAPH5's search procedures do not accommodate matrix searches of the members of


two different code families. Iterative questioning is possible only by saving results of a first
search as a new data document and coding the entire text to a code named for the search you
performed, which is a bit cumbersome. ETHNOGRAPH5's search procedure window, however,
is the least difficult to learn. Its straightforward design and accessibility score high points.
Classic N4's Index Search operators are powerful and incorporate a range of question options,
including iterative questions and matrix searches; however, its organization and language are
challenging.

Conceptual Map

As your analysis progresses, the ability to diagram observed relationships and to pursue
connections between documents, quotes, codes, and memos can assist theory building
(Weitzman, 2000). ATLAS.ti, HyperRESEARCH 2.5, and NVIVO are the only programs we
discuss that offer conceptual mapping facilities. Here we discuss the degree to which the
conceptual map feature is integrated throughout the program, how easy it is to use, and how

Page 20 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

attractive the display is.

ATLAS.ti features a network diagram that is well integrated throughout the program and
incorporates major objects of your analysis. Network diagrams can be built from within their
own window or through menu prompts. Options for creation of diagrams prove fruitful. You can
build connections while reviewing text by using the hyperlink feature or while working within
document, quote, code, and memo lists. For the individual who does not care about diagrams,
this avenue to network building is easier to access and understand. Connections between
items in a diagram can be labeled with a selection of preset options within ATLAS.ti or
personalized options created by the user. You can also change the color and shape of icons in
the model. The introduction of quotations onto tiles within the diagram encourages strong links
between concepts and real data. It also presents opportunities for visually shifting quotations
referenced at a code to further pursue what you can learn from that category. To use the
network feature of ATLAS.ti, you will need to be familiar with the language of the network tool.

NVIVO's models invite diagramming connections between codes, memos, documents, and
attribute characteristics. The conceptual map display is attractive and flexible. You can label
arrows that connect items in the model and can change the color and shape of icons in the
model. You cannot include individual quotes in the model, but you have direct access to the
text referenced at codes within your model via a right mouse click. Access to models is notas
thoroughly integrated as in ATLAS.ti.

HyperRESEARCH 2.5 includes a “code map” that lets you draw connections between code
categories. Its appearance is unappealing, however, and interaction with it is a bit unwieldy.
Action buttons that direct what you want to do, such as “move” a code, must be clicked before
you can perform simple functions such as shifting a code in the map from one area to another.
In addition, the code map is notas deeply integrated as those in ATLAS.ti and NVIVO.
Documents, memos, and quotations are not included as items that can appear in the diagram
and, diagram records can be built only from within the code map. ETHNOGRAPH5, Classic N4,
N5, and WinMAX do not contain conceptual mapping-network diagram features. In Classic N4
and N5, you can transfer code items to conceptual mapping programs Inspiration and Decision
Explorer.

Demographic Comparisons

You can assess patterns of discussion along demographic lines within qualitative software. At a
minimum, programs let you record demographics by coding all the text of a document, such as
“female” or “male.” Some programs have easy-to-understand entry windows. Most programs

Page 21 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

also hold the results of demographic searches for later query.

ETHNOGRAPH5's face sheet is ideal. After easily entering demographics into the program, you
can ask for comparison along demographic lines and save the results. NVIVO's attribute entry
window is designed in similar ways to those of ETHNOGRAPH5 and WinMAX and is very easy
to use. All three packages allow you to easily focus questions on specific values of
demographic variables and save these results.

HyperRESEARCH 2.5, Classic N4, and N5 do not have dedicated windows for entry of
demographic characteristics, but they do offer table import options and manual procedures that
are not difficult to follow. In all three programs, demographic results are saved for later
questioning only when demographics are entered as codes. You can record demographics in
ATLAS.ti via its “family” feature and narrow question results with a textbase selection directed to
only one demographic category, but the results of queries focusing on demographic families
are not saved.

4.9.2.7 Quantitative Data

Work with quantitative data within qualitative software has moved onto relatively equal footing
rather quickly. All programs allow import and export of quantitative data in table formats. You
can import demographic-attribute characteristics for your sample to compare how code
applications differ by groups of interest. You can export code counts and/or absence or
presence of codes and even transfer these tables to quantitative packages for more involved
statistical analysis, if desired.

ETHNOGRAPH5, HyperRESEARCH 2.5, Classic N4, N5, NVIVO, and WinMAX all allow you to
move to and from quantitative and spreadsheet packages with direct links to document IDs in
these programs. With the exception of Classic N4 and N5, frequencies of codes can also be
easily displayed within each program. NVIVO's “profile of codes,” which brings this report to
your screen, can be slow to produce.

ATLAS.ti's quantitative option is strong and well structured. The code-document table within
ATLAS.ti is easy to understand and quick to produce. However, ATLAS.ti links documents within
your hermeneutic unit to quantitative programs by using ATLAS.ti's internal document number
rather than document IDs you set within either a spreadsheet or a statistical package. This
mismatch means you have to do manual matching of ATLAS.ti PD numbers and document IDs
in your stats package. Classic N4 and N5 both offer options for table outputs. They do not
provide pure code instance reports and require specific organization of codes within the “node
explorer” to produce absence or presence of coding reports.

Page 22 of 23 Handbook of Research Design & Social Measurement


SAGE SAGE Research Methods
©2002 SAGE Publications, Ltd.. All Rights Reserved.

4.9.2.8 Merging Projects

Merging projects is an area where user desires continue to exceed current technology. Users
want merge procedures that are lower maintenance than what is currently available. Currently,
you must carefully monitor who works on what sections of a project and ensure that exact
language is used for like items in each project—a practice that is recommended. Some
programs allow different strategies for teamwork with their merge design; others are less
flexible in this regard. As progress is made in this area, we will see more straightforward
procedures that allow a more flexible work strategy for different researchers on the same
project and more precise procedures for ensuring that the work of each user is properly tagged
in each project.

ATLAS.ti, Classic N4, and N5 all offer merge options that provide flexibility for team members’
work strategy. You can either divide documents among team members using the same set of
codes or have all members work with the same documents and different sets of codes.
However, multiple merges and changes to division-of-labor strategies can prove problematic.
HyperRESEARCH 2.5 lets you copy and paste selected parts of one project to another but is
less efficient when you attempt to merge a large portion or the entire project.

Both ETHNOGRAPH5 and WinMAX have easy-to-use merge facilities; however, having multiple
people work on the same documents, each with her or his own set of unique codes, is not
supported, although it is a common strategy for research teams. QSR has just introduced
MERGE for NVIVO as a separate product. Although the timing of its release did not allow us to
test it extensively, we can tell you that its functions would earn it the classification “flexible,
fluid.”

http://dx.doi.org/10.4135/9781412984386.n38

Page 23 of 23 Handbook of Research Design & Social Measurement

Potrebbero piacerti anche