Saldana-3784-Ch-01:Saldana-3784-Ch-01.

qxp 9/26/2008 7:10 PM Page 1

One
An Introduction to Codes and Coding

Any researcher who wishes to become proficient at doing qualitative analysis must
learn to code well and easily.The excellence of the research rests in large part on
the excellence of the coding.
(Anselm L. Strauss, Qualitative Analysis for Social Scientists, 1987, p. 27)

Purposes of the Manual

The three primary purposes of The Coding Manual for Qualitative

Researchers are:

• to briefly discuss the functions of codes, coding, and analytic memo writ-
ing during the qualitative data collection and analytic processes
• to profile a selected yet diverse repertoire of coding methods generally
applied in qualitative data analysis, and
• to provide readers sources, descriptions, examples, recommended applica-
tions, and exercises for coding and further analyzing qualitative data.

This manual does not address such matters as qualitative research design or how
to conduct interviews or participant observation fieldwork. These topics are
already masterfully discussed in other textbooks. The Coding Manual for
Qualitative Researchers is intended as a reference to supplement those existing
works.This manual focuses exclusively on codes and coding and how they play
a role in the qualitative data analytic process. For newcomers to qualitative
inquiry it presents a repertoire of coding methods in broad brushstrokes.
Additional information and extended discussion of the methods can be found
in most of the cited sources. Grounded theory (discussed in Chapter Two), for
example, is elegantly profiled, streamlined, and re-envisioned in Kathy
Charmaz’s (2006) Constructing Grounded Theory: A Practical Guide through
Qualitative Analysis; while Graham R. Gibbs’ (2007) Analyzing Qualitative Data
provides an elegant survey of basic analytic processes.
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 2

2 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

The Coding Manual does not maintain allegiance to any one specific
research genre or methodology. Throughout this book you’ll read a breadth
of perspectives on codes and coding, sometimes purposely juxtaposed to
illustrate and highlight the diverse opinions among scholars in the field. No
one, including myself, can claim final authority on the “best” way to code
qualitative data. In fact, there are a few instances where I take moderate lib-
erty with adapting and even renaming prescribed coding methods for clarity
or flexibility’s sake. This is not intended to standardize terminology within
the field, but simply to employ consistency throughout this particular man-
ual. My perspective acknowledges and promotes the pragmatist paradigm
(Patton, 2002), which chooses “the right tool for the right job” since all
research questions, methodologies, conceptual frameworks, and fieldwork
parameters are context-specific.
I also wrote this manual because I find it problematic (but not difficult) to
teach coding in my own qualitative research methods course. I provide students
with an array of readings about the process from multiple sources because I
have yet to find that single satisfactory book (to me) that focuses exclusively
on the topic. General introductory texts in qualitative inquiry are so numerous
and well-written that it becomes difficult not to find the best one to use, but
which one of such quality works to select as the primary textbook.This man-
ual supplements introductory works in the subject because most limit their dis-
cussions about coding to the writer’s prescribed, preferred, or signature
methods. I wanted to provide in a single resource a selected collection of var-
ious coding methods developed by other researchers (and myself) that provides
students and colleagues a handy reference for classroom exercises and assign-
ments, and for their own independent research for thesis and dissertation field-
work and future qualitative studies. But by no means is it an exhaustive
resource. I deliberately exclude such discipline-specific methods as Behavior
Coding (which notes problematic respondent and interviewer behaviors dur-
ing survey interviews [Singleton & Straits, 2002, p. 65]) and such signature
methods as the Davis Observation Code system (for medical interviews [Zoppi
& Epstein, 2002, p. 375]). If you need additional information and explanation
about the coding methods, check the References.
The Coding Manual is intended primarily as a reference work. It is not nec-
essarily meant to be read cover-to-cover, but it certainly can be if you wish to
acquaint yourself with all 29 coding methods profiles and their analytic possi-
bilities. There are, in fact, several principles related to coding matters not dis-
cussed in the first two chapters that are unique to some of the profiles. If you
choose to review all the contents, read selected sections at a time, not all of them
in one sitting, otherwise it can overwhelm you. If you’re scanning the manual to
see which coding method(s) might be appropriate for your particular study,
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 3

AN INTRODUCTION TO CODES AND CODING 3

read the profiles’ Description and Applications sections to see if further reading
of the profile is merited. It’s doubtful you’ll use every coding method included
in this manual for your particular research endeavors throughout your career,
but they are available here on an “as needed” basis for your unique projects.
Like an academic curriculum, the sequential order of the profiles has been
carefully considered. They don’t necessarily progress in a linear manner from
simple to complex, but are clustered generally from the fundamental to the
intermediate to the advanced.

What is a Code?

A code in qualitative inquiry is most often a word or short phrase that symbolically
assigns a summative, salient, essence-capturing, and/or evocative attribute
for a portion of language-based or visual data.The data can consist of inter-
view transcripts, participant observation field notes, journals, documents, lit-
erature, artifacts, photographs, video, websites, e-mail correspondence, and
so on. The portion of data to be coded during First Cycle coding processes
can range in magnitude from a single word to a full sentence to an entire
page of text to a stream of moving images. In Second Cycle coding
processes, the portions coded can be the exact same units, longer passages of
text, and even a reconfiguration of the codes themselves developed thus far.
Just as a title represents and captures a book or film or poem’s primary con-
tent and essence, so does a code represent and capture a datum’s primary
content and essence.

Coding examples
An example of a coded datum, as it is presented in this manual, looks like this
when taken from a set of field notes about an inner city neighborhood. The
one-word capitalized code in the right column is called a Descriptive Code,
which summarizes the primary topic of the excerpt:
1 1
I notice that the grand majority of homes have chain SECURITY
link fences in front of them. There are many dogs
(mostly German shepherds) with signs on fences that
say “Beware of the Dog.”

Here is an example of several codes applied to data from an interview tran-

script in which a high school senior describes his favorite teacher. The codes
are based on what outcomes the student receives from his mentor. Note that
one of the codes is taken directly from what the participant himself says and is
placed in quotation marks – this is called an In Vivo Code:
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 4

4 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

1 1
He cares about me. He has never SENSE OFSELF-WORTH
told me but he does. 2 He’s always 2
STABILITY
been there for me, even when my parents
were not. He’s one of the few things that
I hold as a constant in my life. So it’s nice.
3 3
I really feel comfortable around him. “COMFORTABLE”

Did you agree with the codes? Did other words or phrases run through your
mind as you read the data? It’s all right if your choices differed from mine.
Coding is not a precise science; it’s primarily an interpretive act.
Also be aware that a code can sometimes summarize or condense data, not sim-
ply reduce it.The introductory examples above were kept purposely simple and
direct. But depending on the researcher’s academic discipline, ontological and
epistemological orientations, theoretical and conceptual frameworks, and even
the choice of coding method itself, some codes can attribute more evocative
meanings to data. In the excerpt below, a mother describes her teenage son’s
troubled school years.The codes emerge from the perspective of middle- and
junior high school years as a difficult period for most youth.They are not spe-
cific types of codes; they are “first impression” phrases derived from an open-
ended process called Initial Coding:

1 1
My son, Barry, went through a really MIDDLE-SCHOOL HELL
tough time about, probably started the end
of fifth grade and went into sixth grade.
2 2
When he was growing up young in TEACHER’S PET
school he was a people-pleaser and
his teachers loved him to death.
3 3
Two boys in particular that he chose to BAD INFLUENCES
try to emulate, wouldn’t, were not very
good for him. 4 They were very critical of 4
TWEEN ANGST
him, they put him down all the time,
and he kind of just took that and really
kind of internalized it, I think, for a
long time. 5 In that time period, in the 5
THE LOST BOY
fifth grade, early sixth grade, they really
just kind of shunned him all together, and
so his network as he knew it was gone.

Note that when we reflect on a passage of data to decipher its core mean-
ing, we are decoding; when we determine its appropriate code and label it, we
are encoding. For ease of reference throughout this manual, coding will be the
sole term used. Simply understand that coding is the transitional process
between data collection and more extensive data analysis.
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 5

AN INTRODUCTION TO CODES AND CODING 5

Coding for patterns

In the examples presented thus far, each unit of data was assigned its own
unique code.This is due primarily to the short length of the excerpts. In larger
and complete data sets, you will find that several to many of the same codes
will be used repeatedly throughout. This is both natural and deliberate – nat-
ural because there are mostly repetitive patterns of action and consistencies in
human affairs, and deliberate because one of the coder’s primary goals is to find
these repetitive patterns of action and consistencies in human affairs as docu-
mented in the data. In the example below, note how the same Process Code (a
word or phrase which captures action) is used twice during this small unit of
elementary school classroom activity:
1 1
Mrs. Jackson rises from her desk and LINING UP FOR LUNCH
announces,“OK, you guys, let’s get lined
up for lunch. Row One.” Five children
seated in the first row of desks rise and
walk to the classroom door. Some of the
seated children talk to each other.
2 2
Mrs. Jackson looks at them and says, MANAGING BEHAVIOR
“No talking, save it for the cafeteria.
3 3
Row Two.” Five children seated in the LINING UP FOR LUNCH
second row of desks rise and walk
to the children already standing in line.

Another way the above passage could be coded is to acknowledge that MAN-
AGING BEHAVIOR is not a separate action or an interruption of the rou-
tine that disrupts the flow of LINING UP FOR LUNCH, but to interpret that
MANAGING BEHAVIOR is an embedded or interconnected part of the
larger social scheme that composes LINING UP FOR LUNCH.The coding
might appear thusly, using a method called Simultaneous Coding (which
applies two or more codes within a single datum):
1 1
Mrs. Jackson rises from her desk and LINING UP FOR LUNCH
announces,“OK, you guys, let’s get lined
up for lunch. Row One.” Five children
seated in the first row of desks rise and
walk to the classroom door. Some of the
seated children talk to each other.
1a 1a
Mrs. Jackson looks at them and says, MANAGING
“No talking, save it for the cafeteria.
1
Row Two.” Five children seated in the second BEHAVIOR
row of desks rise and walk to the children
already standing in line.

Take note of some important caveats when it comes to understanding pat-

terns and regularity: idiosyncrasy is a pattern (Saldaña, 2003, pp. 118–22) and
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 6

6 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

there can be patterned variation in data (Agar, 1996, p. 10). Sometimes we

code and categorize data by what participants talk about.They may all share
with you their personal perceptions of school experiences, for example, but
their individual value, attitude, and belief systems about education may vary
greatly from being bored and disengaged to being enthusiastic and intrinsi-
cally motivated. When you search for patterns in coded data to categorize
them, understand that sometimes you may group things together not just
because they are exactly alike or very much alike, but because they might also
have something in common – even if, paradoxically, that commonality con-
sists of differences.
For example, each one of us may have a strong opinion about who should
be leading our country.The fact that we each have an individual opinion about
that issue is what we have in common. As for whom we each believe should be
leading the country, that’s where the differences and variations occur.
Acknowledge that a confounding property of category construction in quali-
tative inquiry is that data within them cannot always be precisely and discretely
bounded; they are within “fuzzy” boundaries at best (Tesch, 1990, pp. 135–8).
That’s why a method called Simultaneous Coding is an option we have, when
needed. Finally, Hatch (2002) offers that you think of patterns not just as sta-
ble regularities but as varying forms. A pattern can be characterized by:

• similarity (things happen the same way)

• difference (they happen in predictably different ways)
• frequency (they happen often or seldom)
• sequence (they happen in a certain order)
• correspondence (they happen in relation to other activities or events)
• causation (one appears to cause another) (p. 155)

Coding filters
The act of coding requires that you wear your researcher’s analytic lens. But
how you perceive and interpret what is happening in the data depends on
what type of filter covers that lens. For example, consider the following state-
ment from an older male: “There’s just no place in this country for illegal
immigrants. Round them up and send those criminals back to where they
came from.” One researcher, a grounded theorist using In Vivo Coding to
keep the data rooted in the participant’s own language, might code the datum
this way:
1 1
There’s just no place in this country for illegal “NO PLACE”
immigrants. Round them up and send those
criminals back to where they came from.
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 7

AN INTRODUCTION TO CODES AND CODING 7

A second researcher, an urban ethnographer employing Descriptive Coding to

document and categorize the breadth of opinions stated by multiple partici-
pants, might code the same datum this way:

1 1
There’s just no place in this country for illegal IMMIGRATION ISSUES
immigrants. Round them up and send those
criminals back to where they came from.

And a third researcher, a critical race theorist employingValues Coding to cap-

ture and label subjective perspectives, may code the exact same datum this way:

1 1
There’s just no place in this country for illegal XENOPHOBIA
immigrants. Round them up and send those
criminals back to where they came from.

The collection of coding methods in this manual is a repertoire of possible

filters to consider and apply to your approaches to qualitative inquiry. But even
before that, your level of personal involvement as a participant observer – as a
peripheral, active, or complete member during fieldwork – filters how you per-
ceive, document, and thus code your data (Adler & Adler, 1987). So do the types
of questions you ask and the types of responses you receive during interviews
(Kvale, 1996; Rubin & Rubin, 1995), the detail and structuring of your field
notes (Emerson, Fretz, & Shaw, 1995), the gender and race/ethnicity of your par-
ticipants – and yourself (Behar & Gordon, 1995; Stanfield & Dennis, 1993), and
whether you collect data from adults or children (Greene & Hogan, 2005;
Zwiers & Morrissette, 1999).
Merriam (1998) states,“our analysis and interpretation – our study’s findings –
will reflect the constructs, concepts, language, models, and theories that structured
the study in the first place” (p. 48).And it is not only your approach to qualita-
tive inquiry (e.g., case study, ethnographic, phenomenological) and ontological,
epistemological, and methodological issues that influence and affect your cod-
ing decisions (Creswell, 2007; Mason, 2002). Sipe and Ghiso (2004), in their
revealing narrative about coding dilemmas for a children’s literacy study, note
that “All coding is a judgment call” since we bring “our subjectivities, our per-
sonalities, our predispositions, [and] our quirks” to the process (pp. 482–3).

Coding as a heuristic
The majority of qualitative researchers will code their data both during and
after collection as an analytic tactic, for coding is analysis (Miles & Huberman,
1994, p. 56). Differing perspectives, however, attest that “Coding and analysis are
not synonymous, though coding is a crucial aspect of analysis” (Basit, 2003, p. 145).
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 8

8 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

Coding is a heuristic (from the Greek, meaning “to discover”) – an exploratory

problem-solving technique without specific formulas to follow. Coding is only
the initial step toward an even more rigorous and evocative analysis and interpre-
tation for a report. Coding is not just labeling, it is linking:“It leads you from the
data to the idea, and from the idea to all the data pertaining to that idea”
(Richards & Morse, 2007, p. 137).
And, coding is a cyclical act. Rarely is the first cycle of coding data perfectly
attempted.The second cycle (and possibly the third and fourth, and so on) of
recoding further manages, filters, highlights, and focuses the salient features of
the qualitative data record for generating categories, themes, and concepts,
grasping meaning, and/or building theory. Coffey and Atkinson (1996) propose
that “coding is usually a mixture of data [summation] and data complication …
breaking the data apart in analytically relevant ways in order to lead toward fur-
ther questions about the data” (pp. 29–31).
Dey (1999) posits, though his original intent was to be critical,“With cate-
gories we impute meanings, with coding we compute them” (p. 95).To some,
code is a dirty “four-letter word.”A few research methodologists perceive a code
as mere shorthand or an abbreviation for the more important category yet to
be discovered. Unfortunately, some use the terms code and category interchange-
ably and even in combination when they are, in fact, two separate components
of data analysis. I advocate that qualitative codes are essence-capturing and
essential elements of the research story that, when clustered together accord-
ing to similarity and regularity – a pattern – they actively facilitate the devel-
opment of categories and thus analysis of their connections. Ultimately, I like
one of Charmaz’s (2006) metaphors for the process when she states that cod-
ing “generates the bones of your analysis. … [I]ntegration will assemble those
bones into a working skeleton” (p. 45).

Codifying and Categorizing

To codify is to arrange things in a systematic order, to make something part of

a system or classification, to categorize. When codes are applied and reapplied
to qualitative data, you are codifying – a process that permits data to be “segre-
gated, grouped, regrouped and relinked in order to consolidate meaning and
explanation” (Grbich, 2007, p. 21). Bernard (2006) succinctly states that analysis
“is the search for patterns in data and for ideas that help explain why those pat-
terns are there in the first place” (p. 452). Coding is thus a method that enables
you to organize and group similarly coded data into categories or “families”
because they share some characteristic – the beginning of a pattern (see the
examples in Pattern Coding and Focused Coding in Chapter Four). You use
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 9

AN INTRODUCTION TO CODES AND CODING 9

classification reasoning plus your tacit and intuitive senses to determine which
data “look alike” and “feel alike” when grouping them together (Lincoln &
Guba, 1985, p. 347).

From codes to categories

For example, in Harry, Sturges, & Klingner’s (2005) ethnographic study of the
overrepresentation of minorities in special education programs, data initially
coded as classroom MATERIALS, COMPUTERS, and TEXTBOOKS were
categorized under the major heading, Resources. As their study continued,
another major category emerged labeled Teacher Skills with the subcategories
Instructional Skills and Management Skills. The codes subsumed under these
subcategories – part of the overall hierarchical “coding scheme” (Lewins &
Silver, 2007) – were:

Category: Teacher Skills

Subcategory 1: Instructional Skills
Code: PEDAGOGICAL
Code: SOCIO-EMOTIONAL
Code: STYLE/PERSONAL EXPRESSION
Code: TECHNICAL
Subcategory 2: Management Skills
Code: BEHAVIORIST TECHNIQUES
Code: GROUP MANAGEMENT
Code: SOCIO-EMOTIONAL
Code: STYLE (overlaps with instructional style)
Code: UNWRITTEN CURRICULUM

As another example, in Basit’s (2003) study of the aspirations of teenage British

Muslin girls, analysis of interview data with the girls, their parents, and their teach-
ers brought forth 23 major categories that clustered under six major themes. One
major theme was IDENTITY, and its related categories were Ethnicity,
Language, and Religion. Under the theme CAREER ASPIRATIONS, the
categories were Career Choices, Unrealistic Aspirations, and Career Advice.
Rubin & Rubin (1995) recommend that you refine the contents of each
category (working within) from your data before you start comparing them
with each other (working across) (pp. 241, 251). Maykut & Morehouse (1994)
refine each category by developing a rule for inclusion in the form of a propo-
sitional statement, coupled with sample data. For example, if an emergent cat-
egory in a case study is labeled Physical Health, its rule for inclusion as a
propositional statement might read:

Physical Health: The participant shares matters related to physical health such as
wellness, medication, pain, etc.: “I’m on 25 milligrams of amitriptyline each night”;
“I hate going to the gym.”
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 10

10 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

Emergent categories might also evolve as conceptual processes rather than

descriptive topics such as:

Inequity: Participants perceive unfair treatment directed toward themselves and

favoritism directed toward others: “I’ve been working here for over 25 years and
some newcomers are making higher salaries than me.”

The categories’ propositional statements are then compared with each other to
discern possible relationships to create an outcome proposition based on their
combination.

Recoding and recategorizing

Rarely will anyone get coding right the first time. Qualitative inquiry demands
meticulous attention to language and deep reflection on the emergent patterns
and meanings of human experience. Recoding can occur with a more attuned
perspective using First Cycle methods again, while Second Cycle methods
describe those processes that might be employed during the second (and third
and possibly fourth … review of data.
As you code and recode, expect – or rather, strive for – your codes and cat-
egories to become more refined. Some of your First Cycle codes may be later
subsumed by other codes, relabeled, or dropped all together. As you progress
toward Second Cycle coding, there may be some rearrangement and reclassi-
fication of coded data into different and even new categories. Abbott (2004)
cleverly likens the process to “decorating a room; you try it, step back, move a
few things, step back again, try a serious reorganization, and so on” (p. 215).
For example, I observed and interviewed fourth and fifth grade children to
learn the ways they hurt and oppress each other (Saldaña, 2005b). This was
preparatory fieldwork before an action research project that attempted to
empower children with strategies, learned through improvised dramatic simu-
lations and role-playing, for dealing with bullying in the school environment.
I initially categorized their responses into Physical and Verbal forms of
oppression. Some of the codes that fell under these categories were:

Category: Physical Oppression

Code: PUSHING
Code: FIGHTING
Code: SCRATCHING
Category: Verbal Oppression
Code: NAME-CALLING
Code: THREATENING
Code: LAUGHING AT

As coding continued, I observed that a few oppressions were a combination of

both physical and verbal actions. For example, a child can EXCLUDE others
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 11

AN INTRODUCTION TO CODES AND CODING 11

physically from a game, accompanied with a verbal statement such as, “You
can’t play with us.” Hence, a third major category emerged: Physical and
Verbal Oppression.
As the study continued, more data were collected through other methods,
and gender differences in children’s perceptions and enactment of oppression
became strikingly apparent.To participants, oppression was not about the body
and voice; it was about “force” and “feelings.” The three initial categories were
eventually reduced to two, and renamed based on what seemed to resonate
with gender-based observations. The new categories and a few sample codes
and rearranged subcodes included:

Category: Oppression through Physical Force (primarily but not exclusively by

boys)
Code: FIGHTING
Subcode: SCRATCHING
Subcode: PUSHING
Subcode: PUNCHING
Category: Oppression through Hurting Others’ Feelings (primarily but not exclu-
sively by girls)
Code: PUTTING DOWN
Subcode: NAME-CALLING
Subcode: TEASING
Subcode: TRASH TALKING

See the Domain andTaxonomic Coding profile in ChapterThree for an extended

discussion of this case, and the Initial and Focused Coding examples in Chapters
Three and Four respectively to see how a series of codes gets categorized.

From codes and categories to theory

Some categories may contain clusters of coded data that merit further refinement
into subcategories.And when the major categories are compared with each other
and consolidated in various ways, you begin to transcend the “reality” of your data
and progress toward the thematic, conceptual, and theoretical. As a very basic
process, codifying usually follows an ideal and streamlined scheme illustrated in
Figure 1.1. Keep in mind that the actual act of reaching theory is much more com-
plex and messy than illustrated. Richards & Morse (2007) clarify that “categorizing
is how we get ‘up’ from the diversity of data to the shapes of the data, the sorts of
things represented. Concepts are how we get up to more general, higher-level, and
more abstract constructs” (p. 157). Our ability to show how these themes and con-
cepts systematically interrelate lead toward the development of theory (Corbin &
Strauss, 2008, p. 55), though Layder (1998) contends that pre-established sociolog-
ical theories can inform, if not drive, the initial coding process itself.The develop-
ment of an original theory is not always a necessary outcome for qualitative
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 12

12 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

Real Abstract

Code

Code Category

Code

Themes/
Theory
Concepts

Code

Code Category

Code Subcategory

Subcategory

Particular General

FIGURE 1.1 A streamlined codes-to-theory model for qualitative inquiry

inquiry, but acknowledge that preexisting theories drive the entire research
enterprise, whether you are aware of them or not (Mason, 2002).
In the example above of children’s forms of oppression, there were two
major categories that emerged from the study: Oppression through
Physical Force, and Oppression through Hurting Others’ Feelings. So,
what major themes or concepts can be developed from these categories? An
obvious theme we noticed was that, in later childhood, peer oppression is gen-
dered. One higher-level concept we constructed – an attempt to progress from
the real to the abstract – was child stigma, based on the observation that chil-
dren frequently label those who are perceived different in various ways
“weird,” and thus resort to oppressive actions (Goffman, 1963).We could not,
in confidence, formulate a formal theory from this study due to the limited
amount of fieldwork time in the classrooms. But a key assertion (Erickson,
1986) we did develop and put forth, based on the contexts of this study, was:

To artist and activist Augusto Boal, adult participation in theatre for social
change is “rehearsal for the revolution.” With ages 9–11 children, however,
their participation in theatre for social change seems more like an “audition”
for preadolescent social interaction. The key assertion of this study is:
Theatre for social change overtly reveals the interpersonal social systems and
power hierarchies within an elementary school classroom microculture,
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 13

AN INTRODUCTION TO CODES AND CODING 13

because the original dramatic simulations children create authentically reflect

their statuses and stigmas. It diagnostically shows which children are leaders,
followers, resisters, and targets; who is influential and who is ignored; which
children may continue to assert dominance in later grade levels; and which
children may succumb to those with more authority in later grade levels.
(adapted from Saldaña, 2005b, p. 131)

This key assertion, like a theory, attempts to progress from the particular to
the general by inferring transfer – that what was observed in just six elemen-
tary school classrooms at one particular site may also be observed in compa-
rable elementary school classrooms in other locations. This assertion also
progresses from the particular to the general by predicting patterns of what may
be observed and what may happen in similar present and future contexts.

The differences between codes and themes

Several qualitative research texts recommend that you initially “code for
themes.” That, to me, is misleading advice because it muddies the water. A
theme is an outcome of coding, categorization, and analytic reflection, not some-
thing that is, in itself, coded (that is why there is no “Theme Coding” method
in this manual, but there are references to thematic analysis and a section called
Themeing the Data).A datum is initially and, when needed, secondarily coded
to discern and label its content and meaning according to the needs of the
inquiry. Rossman & Rallis (2003) explain the differences: “think of a category
as a word or phrase describing some segment of your data that is explicit, whereas
a theme is a phrase or sentence describing more subtle and tacit processes” (p. 282,
emphasis added). As an example, SECURITY can be a code, but A FALSE
SENSE OF SECURITY can be a theme.
Qualitative researchers are not algorithmic automatons. If we’re carefully
reading and reviewing the data before and as we’re formally coding them,
we can’t help but notice a theme or two (or a pattern, trend, or concept)
here and there. Make a note of it in an analytic memo (see Chapter Two)
when it happens, for it can sometimes guide your continued coding
processes. A set of themes is a good thing to emerge from analysis, but at
the beginning cycles there are other rich discoveries to be made with spe-
cific coding methods that explore such phenomena as participant process,
emotions, and values.

What Gets Coded?

Richards & Morse (2007) humorously advise for analytic work, “If it moves,
code it” (p. 146). But what exactly gets coded in the data?
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 14

14 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

Units of social organization

Lofland, Snow, Anderson, & Lofland (2006) note that social life happens at
four coordinates, “the intersection of one or more actors [participants] engag-
ing in one or more activities (behaviors) at a particular time in a specific place”
(p. 121, emphasis in original). The authors first outline major units of social
organization into:

1 cultural practices (daily routines, occupational tasks, microcultural activ-

ity, etc.);
2 episodes (unanticipated or irregular activities such as divorce, championship
games, natural disasters, etc.);
3 encounters (a temporary interaction between two or more individuals such
as sales transactions, panhandling, etc.);
4 roles (student, mother, customer, etc.) and social types (bully, tight-ass,
geek, etc.);
5 social and personal relationships (husband and wife, party-goers, etc.);
6 groups and cliques (gangs, congregations, families, jocks, etc.);
7 organizations (schools, fast-food restaurants, prisons, corporations, etc.);
8 settlements and habitats (villages, neighborhoods, etc.); and
9 subcultures and lifestyles (the homeless, skinheads, gay leather bears, etc.)

But you won’t find in this manual any coding methods based on the major
units outlined above such as “Encounter Coding,” “Organization Coding” or
“Lifestyle Coding.” When the units above are combined with aspects listed
below, they then become topics for study and coding. Lofland et al.’s aspects
include:

1 cognitive aspects or meanings (e.g., ideologies, rules, self-concepts, identities);

2 emotional aspects or feelings (e.g., sympathy in health care, road rage, work-
place satisfaction);
3 hierarchical aspects or inequalities (e.g., racial inequality, battered women,
high school cliques)

Aside from examining the magnitude and frequency of social life outlined
above, Lofland et al. also recommend examining how participant agency inter-
acts and interplays with structures and processes, plus causes and consequences
observed in the data (2006, pp. 144–67).
Aspects in combination with units lend themselves to such First Cycle coding
methods (see Chapter Three) as Emotion Coding,Values Coding, and Versus
Coding. Structures and processes can be discerned through Descriptive
Coding, Process Coding, and Domain and Taxonomic Coding, while causes and
consequences can be discerned through Pattern Coding or grounded theory’s
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 15

AN INTRODUCTION TO CODES AND CODING 15

Second Cycle coding methods (see Chapter Four). But note that some question
qualitative research’s ability to assert causality: “the understanding of human
experience is a matter of chronologies more than of causes and effects” (Stake,
1995, p. 39).

Amounts of data to code

One related issue with which qualitative research methodologists disagree is
the amount of the data corpus – the total body of data – that should be coded.
Some (e.g., Lofland et al., 2006; Strauss, 1987; cf.Wolcott, 1999) feel that every
recorded fieldwork detail is worthy of consideration, for it is from the pat-
terned minutiae of daily life that we might generate significant social insight.
Others (e.g., Seidman, 2006), if not most, feel that only the most salient por-
tions of the corpus merit examination, and that even up to one half of the total
record can be summarized or deleted, leaving the primary half for intensive
data analysis. The danger is that the portions deleted might contain the as yet
unknown units of data that could pull everything together, or include the nega-
tive case that motivates a rethinking of a code, category, theme, concept, theory,
or assertion. Postmodern perspectives on ethnographic texts consider all docu-
mentation and reports partial and incomplete anyway, so the argument for main-
taining and coding a full or reduced data corpus seems moot. Amount
notwithstanding, insure that you have not just sufficient qualitative but sufficient
quality data with which to work that have been appropriately transcribed and
formatted (see Poland, 2002).
I have learned from years of qualitative data analysis that, only with experi-
ence, I now feel more secure knowing and feeling what is important in the
data record and what is not, and thus code what rises to the surface – “relevant
text” as Auerbach & Silverstein (2003) label it.The beginning of my fieldwork
career, however, was a major learning curve for me, and I coded anything and
everything that was collected. I advise the same for novices to qualitative
research.You, too, will eventually discover from experience what matters and
what doesn’t in the data corpus. (Of course, there will always be brief passages
of minor or trivial consequence scattered throughout interviews and field
notes. Code these N/A – not applicable.)
So, what gets coded? Slices of social life recorded in the data – participant
activities, perceptions, and the tangible documents and artifacts produced by
them. Your own reflective data in the form of analytic memos (discussed in
Chapter Two) and observer’s comments in field notes are also substantive mate-
rial for coding.The process does not have to be approached as if it were some
elusive mystery or detective story with deeply hidden clues and misleading red
herrings scattered throughout. If “human actions are based upon, or infused by,
social meanings: that is, by intentions, motives, beliefs, rules, and values”
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 16

16 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

(Hammersley & Atkinson, 1995, p. 7), then why not just code these actions and
social meanings directly (assuming they are represented in your data and your
inferential skills are working at an optimum)? The entire process and products of
creating data about the data in the form of codes, analytic memos, and graph-
ical summaries are “metadata activities” (MacQueen & Guest, 2008, p. 14).

The Mechanics of Coding

As you prepare text-based qualitative data for manual (i.e., paper and pencil)
coding and analyzing, lay out printed interview transcripts, field notes, and
other researcher-generated materials in double-spaced format on the left half
or left two-thirds of the page, keeping a wide right-hand margin for writing
codes and notes. Rather than keeping your data running together as long
unbroken passages, separate the text into short paragraph-length units with a
line break in-between them whenever the topic or subtopic appears to change
(as best as you can, because in real life “social interaction does not occur in
neat, isolated units” [Glesne, 2006, p. 150]). Gee, Michaels, & O’Connor (1992)
call these unit breaks and their rearrangement into poetic-like verses for dis-
course analysis “stanzas” of text, and emphasize that “formatting choices are
a part of the analysis and may reveal or conceal aspects of meaning and
intent” (p. 240). Unit divisions will also play a key role in formatting data for
CAQDAS – Computer Assisted Qualitative Data Analysis Software – programs
(discussed later).

Pre-coding
In addition to coding with words and short phrases, never overlook the
opportunity to “pre-code” (Layder, 1998) by circling, highlighting, bolding,
underlining, or coloring rich or significant participant quotes or passages
that strike you – those “codable moments” worthy of attention (Boyatzis,
1998). Creswell (2007, pp. 168–9) recommends that such quotes found in
data contained in a CAQDAS program file can be simultaneously coded as
QUOTES with their other codes to enable later retrieval. Selected programs
have areas dedicated to storing intriguing quotations for later access. These
data can become key pieces of the evidentiary warrant to support your
propositions, assertions, or theory (Booth, Colomb, & Williams, 2003;
Erickson, 1986; Lofland et al., 2006), and serve as illustrative examples
throughout your report.The codes or quotes may even be so provocative that
they become part of the title, organizational framework, or through-line of
the report. For example, in my study of theatre of the oppressed (i.e., theatre
for social change) with elementary school children, I was puzzled why young
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 17

AN INTRODUCTION TO CODES AND CODING 17

people continually employed combative tactics during improvisational dramatic

simulations to resolve imbalanced power issues, when I was trying to teach
them proactive peace-building efforts. A fourth-grade girl poignantly pro-
vided the answer when we discussed my concerns by explaining to me,
“Sometimes, you can’t be nice to deal with oppression” (Saldaña, 2005b, p. 117).
The quote was so powerful that it began my final research report as a datum
that would both capture the reader’s interest and later explain the through-
line of the study.

Preliminary jottings
Start coding as you collect and format your data, not after all fieldwork has
been completed. When you’re writing up field notes, transcribing recorded
interviews, or filing documents you gathered from the site, jot down any pre-
liminary words or phrases for codes on the notes, transcripts, or documents
themselves, or as an analytic memo or entry in a research journal for future ref-
erence. They don’t have to be accurate or final at this point, just ideas for ana-
lytic consideration while the study progresses. Be wary of relying on your
memory for future writing. Get your thoughts, however fleeting, documented
in some way.
Also make certain that these code jottings are distinct in some way from the
body of data – bracketed, capitalized, italicized, bolded, etc. Liamputtong &
Ezzy (2005, pp. 270–3) recommend formatting pages of data into three
columns rather than two.The first and widest column contains the data them-
selves – interview transcripts, field notes, etc. The second column contains
space for preliminary code notes and jottings, while the third column lists the
final codes. The second column’s ruminations or first impressions may help
provide a transitional link between the raw data and codes:

COLUMN 1 COLUMN 2 COLUMN 3

Raw Data Preliminary Codes Final Code
1 1
The closer I get to “retirement age” RETIREMENT ANXIETY
retirement age, the faster
I want it to happen. I’m
not even 55 yet and I would financial obligations
give anything to retire now.
But there’s a mortgage to
pay off and still a
lot more to sock away in
savings before I can even
think of it. I keep playing the
lottery, though, in hopes of dreams of early
winning those millions. No retirement
luck yet.
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 18

18 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

Virtually all methodologists recommend initial and thorough readings of

your data while writing analytic memos or jotting in the margins tentative
ideas for codes, topics, and noticeable patterns or themes. Write your code
words or phrases completely rather than abbreviating them to mnemonics or
assigning them reference numbers. Avoid such truncations as “PROC-AN
CD” or “122-A,” which just make the decoding processes of your brain work
much harder than they need to during analysis.

Questions to consider as you code

Auerbach & Silverstein (2003, p. 44) recommend that you keep a copy of your
research concern, theoretical framework, central research question, goals of the
study, and other major issues on one page in front of you to keep you focused
and allay your anxieties because the page focuses your coding decisions.
Emerson, Fretz, & Shaw (1995) advise a general list of questions to consider
when coding field notes (in chronological order), regardless of research
purpose:

• What are people doing? What are they trying to accomplish?

• How, exactly, do they do this? What specific means and/or strategies do
they use?
• How do members talk about, characterize, and understand what is going on?
• What assumptions are they making?
• What do I see going on here? What did I learn from these notes?
• Why did I include them? (p. 146)

I would add to this list the question I ask myself during all cycles of coding and
data analysis:“What strikes you?” Creswell (2007, p. 153) notes that a code can
emerge from data that is not only expected but even surprising, unusual, or
conceptually interesting.

Coding contrasting data

If you are working with multiple participants in a study, it may help to code
one participant’s data first, then progress to the second participant’s data.You
might find that the second data set will influence and affect your recoding of
the first participant’s data, and the consequent coding of the remaining partic-
ipants’ data.The same may hold true for a coding system applied to an inter-
view transcript first, then to a day’s field notes, then to a document. Bazeley
(2007) recommends that the second document coded should contrast “in some
important way with the first … to maximize the potential for variety in con-
cepts (or in their forms of expression) early in the process” (p. 61). Be aware
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 19

AN INTRODUCTION TO CODES AND CODING 19

that, depending on the coding method(s) chosen, some codes may appear more
frequently in selected types of data than others. Selected CAQDAS program
functions can keep you abreast of the codes and their frequencies as analysis
progresses.

The Numbers of Codes

The actual number of codes, categories, themes and/or concepts you generate
for each project will vary and depend on many contextual factors, yet one ques-
tion students ask most is how often codes “should” get applied to qualitative
data. The answer depends on the nature of your data, which particular coding
method you select for analysis, and how detailed you want or need to be – in
other words, more filters to consider.

“Lumping” and “splitting” the data

For example, the following data excerpt is from a speech by a second year,
inner city, grades K–8 school teacher speaking to pre-service education majors
enrolled in a university teaching methods course (Saldaña, 1997). She has just
completed several poignant vignettes about some of her most difficult students.
Notice that just one In Vivo Code is applied to capture and represent the
essence of this entire 145-word excerpt – a broad brush-stroke representation
called Holistic Coding:

1 1
I’m not telling you this to depress you or scare you “A LOT TO LEARN”
but it was a reality for me. I thought I was so ready
for this population because I had taught other groups
of kids. But this is such a unique situation, the
inner city school. No, I should take that back: It’s not
as much of a unique situation anymore. There are
more and more schools that are turning into inner
city schools … I really had to learn about the kids.
I had to learn about the culture, I had to learn the
language, I had to learn the gang signals, I had to
learn what music was allowed, what t-shirts they
could wear on certain days and not on other days.
There was just a lot to learn that I had never even
thought about.

The above has been colloquially called “lumper” coding. The opposite is
someone who codes as a “splitter,” or, one who splits the data into smaller cod-
able moments. Thus, more detailed In Vivo Coding of the exact same passage
might appear thusly:
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 20

20 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

I’m not telling you this to depress you

or scare you but it was a 1 reality 1
“REALITY”
for me. 2 I thought I was so ready 2
“I THOUGHT IWAS SO READY”
for this population because I had
taught other groups of kids. But this
is such a 3 unique situation, the 3
“UNIQUE SITUATION”
inner city school. No, I should take that
back: It’s not as much of a unique
situation anymore. There are more and
more schools that are turning into 4 inner 4
“INNER CITY SCHOOLS”
city schools … 5 I really had to learn 5
“I REALLY HAD TO LEARN”
about the kids. I had to learn about
6 6
the culture, I had to learn the language, “THE CULTURE”
I had to learn the gang signals, I had to
learn what music was allowed, what
t-shirts they could wear on certain days
and not on other days. There was just
7 7
a lot to learn that I had never even “A LOT TO LEARN”
thought about.

Now this 145-word excerpt is represented with seven codes rather than one. I
state the numbers not to suggest that more is better or that less is more, but to
highlight that lumping is an expedient coding method (with future detailed
subcoding still possible), while splitting generates a more nuanced analysis from
the start. Each approach has its advantages and disadvantages aside from the
obvious factors of time and mental energy required. Lumping gets to the
essence of categorizing a phenomenon while splitting encourages careful
scrutiny of social action represented in the data. But lumping may lead to a
superficial analysis if the coder does not employ conceptual words and phrases,
while fine-grained splitting of data may overwhelm the analyst when it comes
time to categorize the codes. During Second Cycle coding, you might collapse
the original number of First Cycle codes into a smaller number as you reana-
lyze the data and find that larger segments of text are better suited to just one
key code rather than several smaller ones. It is only from experience that you’ll
discover which approach works best for you, your particular study, and your
particular research goals.

The quantities of qualities

Lichtman (2006) projects that most qualitative research studies in education will
generate 80–100 codes that will be organized into 15–20 categories which even-
tually synthesize into five to seven major concepts (pp. 164–5). Creswell (2007)
begins his analyses with a short-list of five to six Provisional Codes to begin the
process of “lean coding.”This expands to no more than 25–30 categories that
then combine into five to six major themes (p. 152). Other disciplines and
varying approaches to qualitative inquiry may prescribe different sets of numbers
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 21

AN INTRODUCTION TO CODES AND CODING 21

as general guidelines for analysis.The final number of major themes or concepts

should be held to a minimum to keep the analysis coherent, but there is no stan-
dardized or magic number to achieve. Unlike Lichtman’s five to seven central
concepts and Creswell’s five to six major themes, anthropologist Harry F.Wolcott
(1994, p. 10) generally advises throughout his writings that three of anything
major seems an elegant quantity for reporting qualitative work.

The codebook or code list

Since the number of codes can accumulate quite quickly and change as analy-
sis progresses, keep a record of your emergent codes in a separate file as a code-
book – a compilation of the codes, their content descriptions, and a brief data
example for reference. CAQDAS programs, by default, will maintain a list of
codes you have created for the project and provide space to define them. This
can be reviewed periodically – both on the monitor screen and on hard copy –
as coding progresses to assess its current contents and possible evolution.
Maintaining this list provides an analytic opportunity to organize and reorga-
nize the codes into major categories and subcategories.This management tech-
nique also provides a comparative list if you are working with multiple
participants and sites. One school site’s data, for example, may generate a list of
codes significantly different from another school site.
Codebooks or CAQDAS code lists become especially critical as a set of cod-
ing standards when multiple team members work together on the same pro-
ject’s data. MacQueen et al. (2008) strongly recommend that each codebook
entry should contain “the code, a brief definition, a full definition, guidelines
for when to use the code, guidelines for when not to use the code, and exam-
ples” (p. 121). Also note that a codebook differs from an index, the latter being
a coded composite of the data corpus, organized alphabetically, hierarchically,
chronologically, categorically, etc. CAQDAS programs are superior for index-
ing functions with a qualitative data corpus.

Manual and CAQDAS Coding

Some instructors of statistics and quantitative data analysis require that their
students first learn how to “crunch the numbers” manually using only a
pocket/hand calculator to provide them with cognitive understanding and
ownership of the formulas and results. Once a statistical test has been adminis-
tered this way, they can then use computers with software specifically designed
to calculate numeric data.
Coding and qualitative data analysis have their equivalent trial. I am one of
those instructors who require that my students first perform “manual” coding
and qualitative data analysis using paper and pencil on hard copies of data
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 22

22 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

entered and formatted with basic word-processing software only.The reason is

that each class assignment of data gathering is relatively small-scale and thus a
manageable project to analyze in this manner. But if a student’s dissertation pro-
ject or my own independent research studies will require multiple participant
interviews or extended fieldwork and extensive field note-taking, then CAQ-
DAS becomes a vital and indispensable tool for the enterprise. Basit (2003)
compared personal experiences between manual and electronic coding and
concluded, “the choice will be dependent on the size of the project, the funds
and time available, and the inclination and expertise of the researcher” (p. 143).

Coding manually
Trying to learn the basics of coding and qualitative data analysis simultane-
ously with the sometimes complex instructions and multiple functions of
CAQDAS programs can be overwhelming for some, if not most.Your mental
energies may be more focused on the software than the data. I recommend
that for first-time or small-scale studies, code on hard-copy printouts first, not
via a computer monitor (cf. Bazeley, 2007, p. 92). There is something about
manipulating qualitative data on paper and writing codes in pencil that give
you more control over and ownership of the work. Perhaps this advice stems
from my admitted lack of technological expertise and old-fashioned ways of
working that have become part of my “codus” operandi. Nevertheless, there is
something to be said for a large area of desk or table space with multiple pages
or strips of paper spread out to see the smaller pieces of the larger puzzle – a
literal perspective not always possible on a computer’s monitor screen. After
you feel the codes are fairly well set from your initial hard-copy work, then
transfer your codes onto the electronic file. But first, “Touch the
data.…Handling the data gets additional data out of memory and into the record.
It turns abstract information into concrete data” (Graue & Walsh, 1998, p. 145).
Even proponents of CAQDAS recommend that hard-copy printouts of code
lists and coded data be generated occasionally to permit you to work with tra-
ditional writing materials such as red pens and highlighters to explore data in
fresh ways.

Coding electronically
After you have gained some experience with hard-copy coding and have
developed a basic understanding of the fundamentals of qualitative data analy-
sis, apply that experiential knowledge base by working with CAQDAS. Keep
in mind that CAQDAS itself does not actually code the data for you; that task
is still the responsibility of the researcher.The software efficiently stores, orga-
nizes, manages, and reconfigures your data to enable human analytic reflection.
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 23

AN INTRODUCTION TO CODES AND CODING 23

Some programs even enable coding of digital audio and video documents
stored in their files. I advise that you work with a smaller portion of your data
first, such as a day’s field notes or a single interview transcript, before import-
ing the data corpus into the program. As with all word-processed work on a
computer, backup your original files as a precautionary measure.
Three major CAQDAS programs to explore, whose commercial websites
provide online tutorials and demonstration software/manual downloads of
their most current versions, are:

• ATLAS.ti: www.atlasti.com
• MAXQDA: www.maxqda.com
• NVivo: www.qsrinternational.com

Refer to Lewins & Silver (2007) and Bazeley (2007) for accompanying lit-
erature on these programs. Also see Richards & Morse (2007, pp. 85–90) for
what selected CAQDAS programs can and cannot do; Hahn (2008) and La
Pelle (2004) for qualitative data analysis with basic word-processing soft-
ware; and Brent & Slusarz (2003) for advanced computational strategies
with software. Other CAQDAS programs, such as HyperRESEARCH and
QDA Miner, are discussed and reviewed at an online forum for users: http://
caqdas.soc.surrey. ac. uk/.

Data formatting for CAQDAS

The heading and paragraph formats of qualitative data such as field notes and,
in particular, interview transcripts, need to conform consistently with the par-
ticular software package’s prescriptions for text layout. This becomes vital for
its coding and retrieval functions to work consistently and reliably. ATLAS.ti,
MAXQDA, and NVivo all import and handle documents saved in rich text
format, enabling you to employ supplemental “cosmetic” coding devices such
as colored fonts, bolding, and italicizing in your data (Lewins & Silver, 2007, p. 61).
One of the best features of some CAQDAS programs is their ability to display
code labels themselves in various user-assigned colors for “at a glance” refer-
ence and visual classification. Figure 1.2 illustrates a sample screen shot from
the most current version of NVivo. Note how the video data and its transcript
are accompanied with codes and “coding stripes,” which delineate which por-
tion of data is assigned a particular code.

Coding capabilities with CAQDAS

Selected qualitative data analysis programs permit you to do what you can do
manually, such as: apply more than one code to the same passage or sequential
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 24

24 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

FIGURE 1.2 A Video Coding screenshot from NVivo, v. 8 (NVivo 8 is designed

and developed by QSR International Pty Ltd. NVivo is a trademark or registered
trademark of QSR International. Patents pending. www.qsrinternational.com)

passages of text (variously labeled in the methods literature as “simultaneous

coding,” “double coding,” “co-occurrence coding,” “multiple coding,” or
“overlap coding”); code a smaller portion of text within a larger portion of
coded text (“subcoding,” “embedded coding,” or “nested coding”); and sub-
sume several similarly coded passages under one larger code (“pattern coding,”
“meta coding,” “umbrella coding,” or “hierarchical coding”); along with the
ability to instantly and conveniently insert analytic memos related to a specific
datum or code. Each CAQDAS program will employ its own distinct set of
terms for its coding functions and operations, so refer to the user’s manual for
specific ways of working.
CAQDAS, unlike the human mind, can maintain and permit you to orga-
nize evolving and potentially complex coding systems into such formats as
hierarchies and networks for “at a glance” user reference. Figure 1.3 illustrates
a sample window excerpt from ATLAS.ti’s Code Manager function, which lists
each code by name and provides descending frequency counts of the codes in
progress (serendipitously, the codes are extracted from work on a document
about grounded theory).
Though I stated above that software does not code for you, there is an inter-
esting utilitarian function called “auto coding” available in most CAQDAS
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 25

AN INTRODUCTION TO CODES AND CODING 25

FIGURE 1.3 A Code Manager screenshot excerpt from ATLAS.ti, v.5.2 (courtesy
of ATLAS.ti)
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 26

26 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

programs, which can alleviate some of the repetitiveness of manually coding

similar passages of text. Passages have to be formatted in prescribed ways and
contain the exact same word or phrase, however, for this function to work
accurately. The ATLAS.ti handbook strongly recommends a manual review
after auto coding has been performed to verify the software’s coding assign-
ments, and Lewins & Silver (2007) suggest that researchers should not feel
“compelled to make use of auto coding just because it is available” (p. 21).

Searches and queries with CAQDAS

Another one of CAQDAS’s advantages over manual paper-and-pencil coding
and analysis is its search and querying abilities to quickly collect and display key
words and phrases and similarly-coded data for examination. Searches or
queries of coded passages can even find where particular codes co-occur, over-
lap, appear in a sequence, or lie in proximity to each other.These search func-
tions can perform such actions as retrieve, filter, group, link, and compare,
enabling the researcher to perform such human actions as infer, make connec-
tions, identify patterns and relationships, interpret, and build theory with the
data (Lewins & Silver, 2007, p. 13). Figure 1.4 illustrates a sample MAXQDA
Code Relations Browser window, which enables you to determine possible
interrelationships among coded data (Kuckartz, 2007).The varying sizes of the
squares within the matrix indicate the relative frequency of such matches.
CAQDAS also permits the researcher to shift quickly back and forth between
multiple analytic tasks, such as coding, analytic memo writing, and exploring pat-
terns in progress.Add to this the software’s ability to recode, uncode, rename, delete,
move,merge,group,and assign different codes to shorter and longer passages of text
with a few mouse clicks and keystrokes during Second Cycle Coding, and the
advantages of CAQDAS over paper and pencil soon become apparent.
Rather than presenting in this section an extended discussion of CAQDAS’s
specific applications with coding and data analysis, additional references will be
made on an “as relevant” basis throughout the rest of this manual. Since most
readers of this book are more than likely newcomers to qualitative data analysis,
I assume that manual coding will be the first method you employ.Thus, I pre-
sent the coding profiles with that assumption in mind.Those with experience or
expertise in CAQDAS programs can adapt the coding principles described in
this manual into their particular software package’s active files and documents.

Solo and Team Coding

Coding in most qualitative studies is a solitary act – the “lone ethnographer”

intimately at work with her data (Galman, 2007) – but larger fieldwork pro-
jects may involve a team.
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 27

AN INTRODUCTION TO CODES AND CODING 27

FIGURE 1.4 A Code Relations Browser screenshot from MAXQDA, v. 2007

(courtesy of MAXQDA/VERBI Software)

Coding collaboratively
Writers of joint research projects advocate that coding in these cases can and should
be a collaborative effort (Erickson & Stull, 1998; Guest & MacQueen, 2008).
Multiple minds bring multiple ways of analyzing and interpreting the data: “a
research team builds codes and coding builds a team through the creation of
shared interpretation and understanding of the phenomenon being studied”
(Weston et al., 2001, p. 382). Provocative questions are posed for consideration
that could possibly generate new and richer codes (Olesen et al., 1994).
Ultimately, team members must coordinate and insure that their sometimes indi-
vidual coding efforts harmonize, particularly if a central data base and CAQDAS
system are employed. MacQueen et al. (2008, p. 132) strongly advise that one
member of the team be assigned primary responsibility as “codebook editor” –
the one who creates, updates, revises, and maintains the master list for the group.
Those conducting action or community-based research can invite the
study’s participants/stakeholders themselves into the analytic process as a collab-
orative venture to provide a sense of ownership and investment in data analysis
and its consequent recommendations for social change (Stringer, 1999).
Northcutt & McCoy (2004) label focus group development of their own cat-
egories of interest “affinities.” Children and adolescents, too, can be taught to
investigate and analyze issues that relate to their social worlds (Alderson, 2000;
Heiligman, 1998;Warren, 2000).
Team members can both code their own and others’ data gathered in the
field to cast a wider analytic net and provide a “reality check” for each other.
For these types of collaborative ventures, intercoder agreement or interpretive con-
vergence – the percentage at which different coders agree and remain consistent
with their assignment of particular codes to particular data – is an important part
of the process (see Bernard, 2006, pp. 512–15; Boyatzis, 1998, pp. 144–59;
Hruschka et al., 2004; and Miles & Huberman, 1994, p. 64 for simple formulas).
There is no standard or base percentage of agreement among qualitative
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 28

28 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

researchers, but the 85%–90% range seems a minimal benchmark to those most
concerned with an evidentiary statistic. Other research teams may wish to dis-
pense with such quantitative measures all together and rely on intensive group
discussion and simple group “consensus” as an agreement goal (Harry, Sturges, &
Klingner, 2005, p. 6).

Coding solo
If you’re working as a lone ethnographer, shop talk with a colleague or men-
tor about your coding and analysis as you progress through them. Both solo
and team coders can even consult the participants themselves during analysis
(a process sometimes called “member checking”) as a way of validating the
findings thus far. Even if you and other members of a research seminar are each
working on different projects, sharing coded field note excerpts and discussing
your “dilemmas” about coding and analysis generate peer support and may
even help you and others find better connections between categories in
progress (Burant et al., 2007; Strauss, 1987). Discussion provides not only an
opportunity to articulate your internal thinking processes, but also presents
windows of opportunity for clarifying your emergent ideas and possibly mak-
ing new insights about the data.
Ezzy (2002, pp. 67–74) recommends several strategies for checking the
progress of your analysis while still in the field. Though applicable for team
researchers as well, the lone researcher can benefit most from these recommen-
dations to assess the trustworthiness of her account: (1) check your interpreta-
tions developed thus far with the participants themselves; (2) initially code as
you transcribe interview data; and (3) maintain a reflective journal on the
research project with copious analytic memos.

Necessary Personal Attributes for Coding

Aside from such cognitive skills as induction, deduction, abduction, synthesis,

evaluation, and logical and critical thinking, there are seven personal attributes
all qualitative researchers should possess, particularly for coding processes.
First, you need to be organized. This is not a gift that some people have and
others don’t. Organization is a set of disciplined skills that can be learned and cul-
tivated as habits.A small-scale qualitative study’s word count of data will range in
the tens- and sometimes hundreds-of-thousands of words. The multiple codes
you generate will need an organized framework for qualitative analysis. And
despite the electronic filing systems of hard drives and CAQDAS, you will still
encounter and manipulate many pages of paper in qualitative work. Date and
label all incoming data and keep multiple digital and hard copies as backup.
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 29

AN INTRODUCTION TO CODES AND CODING 29

Second, you need to exercise perseverance.Virtually every writer of qualitative

research methods literature remarks that coding data is challenging and time-
consuming. Some writers also declare how tedious and frustrating it can be.Take
breaks from your work when you need to, of course – this will keep you
refreshed and alert. But cultivate a personal work ethic and create an environ-
ment and schedule that enable you to sustain extended periods of time with
analytic tasks requiring your full concentration.
Third, you need to be able to deal with ambiguity.The acts of coding and cod-
ifying are not precise sciences with specific algorithms or procedures to follow.
Yes, occasionally answers may suddenly and serendipitously crystallize out of
nowhere. But at other times, a piece of the analytic puzzle may be missing for
days or weeks or even months. Rich ideas need time to formulate, so have trust
and faith in yourself that these may emerge in due time. But remember that
you can accelerate the process through analytic memo writing.
Fourth, you will need to exercise flexibility. Coding is a cyclical process that
requires you to recode not just once but twice (and sometimes even more).
Virtually no one gets it right the first time. If you notice that your initial meth-
ods choices may not be working for you or not delivering emergent answers
you need, be flexible with your approach and try a modified or different
method all together. Virtually all researcher-developed coding schemes are
never fixed from the beginning – they evolve as analysis progresses.
Fifth, you need to be creative. There’s a lot of art to social science. Noted
ethnographer Michael H. Agar (1996) asserts that the early stages of analysis
depend on “a little bit of data and a lot of right brain” (p. 46). We generally
advocate that qualitative researchers remain close to and deeply rooted in their
data, but every code and category you construct or select are choices from a
wide range of possible options. Creativity also means the ability to think visu-
ally, to think in metaphors, and to think of as many ways possible to approach
a problem. Creativity is essential for your data collection, data analysis, and even
for your final written report.
Sixth, you need to be rigorously ethical. Honesty is perhaps another way to
describe this, but I deliberately choose the phrase because it implies that you
will always be: rigorously ethical with your participants and treat them with
respect; rigorously ethical with your data and not ignore or delete those seem-
ingly problematic passages of text; and rigorously ethical with your analysis by
maintaining a sense of scholarly integrity and working hard toward the final
outcomes.
The seventh and arguably most important skill you need for coding is an
extensive vocabulary. Quantitative research’s precision rests with numeric accu-
racy. In qualitative research, our precision rests with our word choices. An
unabridged dictionary and thesaurus become vital reference tools to find just
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 30

30 THE CODING MANUAL FOR QUALITATIVE RESEARCHERS

the right words for your codes, categories, themes, concepts, and theories.
Explore the origins of key words in an unabridged dictionary to find surprising
new meanings (for example, did you know that the root word of hypocrite is
“actor”?). A thesaurus review of a key word chosen as a code or category may
introduce you to an even better – and more precise – word for your analysis.
For an applied introduction to the cognitive skills and personal attributes
necessary for coding and qualitative data analysis, see Appendix A’s exercises
and simulations.

On Method

Thorough – even cursory – descriptions about the researcher’s code develop-

ment and coding processes rarely make it into the methods section of a final
report (but a dissertation writer should consider including his or her codebook
as an appendix to the study).The majority of readers would most likely find the
discussion tedious or irrelevant compared to the more important features, such
as the major categories and findings. Plus, scholarly journals place length restric-
tions on article manuscripts, so some components of the research story must be
left out and, more often than not, codes and coding fall by the wayside. But in
all honesty, I don’t think most of the academic community minds (cf. Stewart,
1998). I’m not advocating that published research should include what most feel
is a behind-the-scenes matter. Just acknowledge that the long time and rigor-
ous effort you put into, and joyous personal analytic growth you experience
through, coding and analytic memo writing are private affairs between you and
your data (cf. Constas, 1992).When you invite important guests to your home
for dinner, you don’t ask them to appear two or three hours before the sched-
uled serving time to watch you cook in the kitchen.They arrive just before the
meal to feast on and enjoy what you’ve worked so hard to prepare.
Yet, analogy aside, please don’t refer to or consider this manual a “cookbook”
for your raw data. That suggests that the methods profiled here are like tested
recipes guaranteed to produce successful dishes every time. Most methodolo-
gists concur that coding is “an idiosyncratic enterprise” (Glesne, 2006, p. 153)
and the “search for one perfect method of data analysis is fruitless” (Coffey &
Atkinson, 1996, p. 2). Each qualitative study is context-specific and your data are
unique, as are you and your creative abilities to code them. I don’t have the
answers to your questions, but you and your data do. In good faith, I guarantee
you some guidance and, if we’re both lucky, perhaps some insight.
(I jokingly mused to myself whether this manual might be disparagingly
tagged by some as “the Cliff’s Notes of qualitative data analysis” or Coding for
Dummies. Either way, as a pragmatist I’ll take that as a compliment about the work.)
Saldana-3784-Ch-01:Saldana-3784-Ch-01.qxp 9/26/2008 7:10 PM Page 31

AN INTRODUCTION TO CODES AND CODING 31

Coding as craft
I am very well aware of the interpretivist turn in qualitative inquiry and the
movements toward narrative presentation and emancipatory social action
through ethnographic fieldwork (Denzin & Lincoln, 2005). My own qualita-
tive research projects, in fact, have ranged from the realist to the literary and
from the confessional to the critical (van Maanen, 1988). But as a theatre prac-
titioner, my discipline acknowledges that we must attend to both the art and
craft of what we do to make our stage production work successful. And as a
teacher educator, it’s my job to teach how to teach. Hence, I must have an
attunement to various methods of classroom practice because my professional
responsibilities require that I do. Some methods are organizational, managerial,
time-efficient, and related to carefully planned curriculum design.Yet I empha-
size to my students that such processes as the creative impulse, trusting your
instincts, taking a risk, and just being empathetically human in the classroom
are also legitimate methods of teaching practice. Education is complex; so is
social life in general and so is qualitative inquiry in particular.
This heightened, ever-present awareness of craft, of “how to,” transfers into
my research work ethic. I have become both humbly and keenly aware not
only of what I’m doing but why I’m doing it. A metacognition of method,
even in an emergent, inductive-oriented, and socially conscious enterprise such
as qualitative inquiry, is vitally important.This awareness comes with time and
experience (and trial and error), but development can be accelerated if you
have some preparatory knowledge of “how to.” I hope this manual smoothes
your learning curve a bit and assists with your professional and personal growth
as a researcher.
This introduction focused on codes and coding. There is an accompany-
ing heuristic with this process – writing analytic memos, the subject of the
next chapter.