volume 13 number 2 construct validation using 2010 sage
TRANSCRIPT
Construct Validation UsingComputer-Aided Text Analysis(CATA)
An Illustration Using Entrepreneurial Orientation
Jeremy C. ShortTexas Tech University
J. Christian BrobergWichita State University
Claudia C. CogliserKeith H. BrighamTexas Tech University
Construct validity continues to pose challenges in the organizational sciences. To capture
difficult-to-measure constructs of interest, researchers have often relied on content analysis.
One content analysis technique, computer-aided text analysis (CATA), is particularly
attractive because of the ability to process large samples with high speeds and reliabilities.
Unfortunately, inconsistent guidance exists to guide researchers through the use of this tool
in a manner compatible with accepted methods used to validate constructs in a rigorous
manner. The authors review research using content analysis to examine the extent to which
such studies integrate methods for assessing content, external, discriminant, and predictive
validity. To provide direction for organizational researchers interested in using CATA to
measure theoretically based constructs relevant to the management field, they suggest a
number of possible procedures to enhance construct validity. They illustrate these
procedures using the construct of entrepreneurial orientation.
Keywords: computer-aided text analysis; construct validation; content analysis; DICTION;
entrepreneurial orientation
Validity refers to evaluating inferences drawn from measures of concepts of interest and
considers the extent to which a measure accurately represents that focal concept
(Cronbach, 1971). The importance of rigor in construct measurement cannot be minimized.
Kerlinger and Lee (2000) argued that the ability to demonstrate appropriate measures of
theoretical notions of interest through construct validity ‘‘is one of the most significant
advances of modern measurement theory and practice’’ (p. 670).
Despite advancements in the organizational sciences vis-a-vis construct validation, reviews
of construct validity in the management field have yielded generally bleak conclusions about
Authors’ Note: The authors would like to thank Tyge Payne for his thoughtful comments on previous drafts of this
manuscript. Please address correspondence to Jeremy C. Short, Area of Management, Texas Tech University,
Lubbock, TX 79408; e-mail: [email protected].
Organizational
Research Methods
Volume 13 Number 2
April 2010 320-347
# 2010 SAGE Publications
10.1177/1094428109335949
http://orm.sagepub.com
hosted at
http://online.sagepub.com
320
its adequacy (e.g., Bagozzi, Yi, & Phillips, 1991; Boyd, Gove, & Hitt, 2005; Scandura &
Williams, 2000). One reason that management scholars use measures that are psychometri-
cally inadequate may be because of the difficulty encountered when implementing survey
methods with samples that typically receive a less-than-ideal response rate in certain business
settings (Baruch, 1999). This is particularly problematic in fields such as entrepreneurship,
where response rates have often been much lower than the management field as a whole
(Bartholomew & Smith, 2006). Likewise, a number of conceptual phenomena of interest
in fields such as strategic management often involve ‘‘unobservables,’’ making rigorous con-
struct validity efforts challenging; for example, the concept of opportunism from transaction
cost theory is inherently unobservable as are the unique resources that could lead to perfor-
mance advantages according to the resource-based view of the firm (Godfrey & Hill, 1995).
To remedy construct measurement problems germane to the management field, research-
ers have often relied on content analysis (Duriau, Reger, & Pfarrer, 2007; Morris, 1994), a
research method that uses a set of procedures to classify or otherwise categorize communi-
cation allowing inferences about context (Krippendorff, 2004; Weber, 1990). Typically
relying on archival narratives to extract criteria of interest to management scholars, content
analysis has been applied to assess elements such as CEO performance comparisons (e.g.,
Short & Palmer, 2003), aspects of organizational sensemaking (e.g., Gioia & Chittipeddi,
1991), and processes such as organizational sensegiving (e.g., Maitlis & Lawrence,
2007). Analysis is generally performed on organizationally produced texts such as CEO
shareholder letters, annual reports, and mission statements (Duriau et al., 2007).
This article is intended to fill a gap in the research methods literature by providing an
enhanced discussion of potential approaches to construct validation, when computer-aided
text analysis (CATA) is the primary tool used to capture theoretically based constructs of
interest for organizational researchers. Examples of software appropriate for CATA include
VBPro, CATPAC, Concordance, DICTION, General Inquirer, LIWC, NVivo, and MECA
(Krippendorff, 2004; Skalski, 2002). Integrating CATA to measure constructs of interest
provides numerous advantages to organizational researchers because CATA has higher
reliability than human coding with lower cost and greater speed (Neuendorf, 2002); yet,
a review of content analysis in organizational studies found that less than 25% of the articles
in major management journals in the past 25 years used CATA in their content analysis pro-
cesses (Duriau et al., 2007). The fact that incorporation of such techniques represents the
exception rather than the norm suggests that construct validation when using content anal-
ysis may be less than ideal and that scholars may not understand how to incorporate such
analyses into their empirical tests. To highlight the potential of CATA for future empirical
efforts, we highlight three contributions to organizational research methods with the goal of
providing a ‘‘user’s guide’’ to CATA, with the potential to offer greater validity than pre-
vious applications of content analysis.
First, we examine the degree to which methods for addressing validity are incorporated
in organizational studies using content analysis. We examine and build on the sample used
in the review of Duriau et al. (2007) organizational studies using content analysis and exam-
ine the content, external, discriminant, and predictive (nomological) validity of such work.
Content analysis experts have suggested that elements of validity such as nomological or
predictive validity are rare in studies using content analysis (McAdams & Zeldow, 1993;
Neuendorf, 2002; Weber, 1990), and such issues may also be problematic in organizational
Short et al. / Construct Validation Using CATA 321
studies using content analysis. Examination of other elements surrounding the use of con-
tent analysis in the organizational literature may also yield insight relevant to construct vali-
dation procedures.
Our second contribution, following in the tradition of other work appearing in Organiza-
tional Research Methods (e.g., Fendt & Sachs, 2008; Karren & Barringer, 2002; Lance,
Butts, & Michels, 2006; Rogelberg & Stanton, 2007), is to provide a ‘‘user’s guide’’ for
future empirical efforts consistent with the goals and values of organizational research. For
example, little guidance exists to aid researchers for developing specialized or custom dic-
tionaries that are needed when using content analysis with individual words as the unit of
analysis. This is somewhat surprising given that a number of predefined dictionary-based
programs have been accepted into the management literature. For example, the predefined
dictionaries in the DICTION software program have been used to examine the language of
charismatic leadership (e.g., Bligh, Kohles, & Meindl, 2004a, 2004b). Short and Palmer
(2008) argued that future efforts could advance management research by supplementing
these predefined dictionaries provided by content analysis software programs with
researcher-specified custom dictionaries. Content analysis experts such as Krippendorff
also noted that ‘‘Most content analyses would benefit from the construction of special-
purpose dictionaries, but developing a dictionary from scratch can be a formidable task.
It is not surprising, therefore, that content analysts usually try to build on available diction-
aries before they attempt to develop their own’’ (2004, p. 287). The suggestion to create
custom dictionaries needed to advance the management field is unlikely to become reality
if guidance is not provided for a thoughtful approach toward building such dictionaries.
At the same time, some advice offered in content analysis texts may not always be com-
pletely compatible with the goals of management researchers using content analytic meth-
ods, who generally incorporate both deductive and inductive approaches when using
content analysis (Doucet & Jehn, 1997; L. Doucet, personal communication, December
14, 2008; B. Kabanoff, personal communication, December 14, 2008; T. Pollock, personal
communication, December 14, 2008). Indeed, such approaches, common in organizational
applications, can be contrasted with content analysis texts such as Krippendorff (2004) who
suggested, ‘‘Deductive and inductive inferences are not central to content analysis’’ (p. 36).
Thus, the specific guidance offered in content analysis texts may result in inconsistent
applications for organizational scholars. For example, when presenting a flowchart for a
typical content analysis process, Neuendorf (2002) suggested:
You may use standard dictionaries (e.g., those in Hart’s program DICTION) or originally cre-
ated dictionaries. When creating original dictionaries, be sure to first generate a frequency list
from your text sample, and examine for key words and phrases. (p. 50)
The inductive suggestion to first examine a frequency list based on a particular text sam-
ple when creating new dictionaries seems contrary to the advice provided when standard
dictionaries, such as those found in Hart’s (2000) DICTION (which were deductively devel-
oped independent of any particular type of narrative text) are used. This advice may also be
problematic for management theorists who are interested in a deductive, rather than induc-
tive, analysis of narrative texts. As illustration, Wade, Porac, and Pollock (1997) examined
the justification of CEO pay, explaining their content analysis procedure by noting, ‘‘We
322 Organizational Research Methods
first created a starting membership array for a concept by listing words and phrases that a
priori seemed to fit the category’s definition’’ (p. 649). Given somewhat unclear or poten-
tially conflicting recommendations in the literature, our second goal is to provide sugges-
tions concerning the creation and use of word lists in a manner that is compatible with the
use of CATA for organizational research. Consequently, we highlight how content, discri-
minant, external, and predictive validity can be used in empirical efforts using CATA to
encourage more rigorous efforts when content analysis is applied in organizational research.
Our third contribution is to provide an illustration of our CATA suggestions using the
construct of entrepreneurial orientation. Although entrepreneurship scholars have called for
the incorporation of content analysis to measure entrepreneurial orientation (Lyon, Lump-
kin, & Dess, 2000), relatively few studies have used content analysis to assess elements of
this construct (e.g., Chen & Hambrick, 1995). Given that content analysis may be an espe-
cially useful tool for examining adolescent theories (theories that have not been developed
or tested adequately, Sonpar & Golden-Biddle, 2008), we believe that illustrating how
entrepreneurial orientation can be assessed through content analysis of shareholder letters
is an especially attractive approach that offers value to entrepreneurship scholars as well
as to the broader management field.
Following our three primary contributions, we first review methods for assessing con-
struct validity when using content analysis and assess the extent to which such techniques
have been used in research to date. Second, we suggest a number of possible procedures to
enhance validity when using CATA. Third, we provide an illustration using the construct of
entrepreneurial orientation, concluding with implications for future research efforts.
Examining Validity in Content Analysis
To examine validity in studies using content analysis, we systematically investigated the
extent to which different types of validity applicable to research using content analysis were
assessed. We based our analyses on the sample of 98 studies in the recent review of Duriau
et al. (2007) in Organizational Research Methods of content analysis usage in management
research. In evaluating these studies, we assessed four types of validity commonly
addressed in the management literature: content, external, discriminant, and predictive
validity. A brief description of each type of validity, a description of how we coded each
type of validity, and the results of our investigation follow. A summary of our coding
results from these studies is displayed in Table 1.
Content validity is the degree to which a measure encapsulates the full domain of a par-
ticular construct (Nunnally & Bernstein, 1994). We coded studies as addressing content
validity if authors evaluated content analysis procedures for their appropriateness of repre-
senting the full domain of their study’s constructs. For example, we found that 79 of 98
(81%) studies relied on deductive procedures, based on a priori theory, when creating their
coding schemes. Another example of this evaluative process is having content experts judge
the adequacy of words in a content analysis dictionary to represent a particular construct.
We found that 21 studies of the 98 (21%) created word lists or dictionaries to represent their
study’s constructs, 11 of 98 studies (11%) described the process whereby authors selected
words to represent the study’s constructs, but only 5 of the 98 (5%) studies addressed
Short et al. / Construct Validation Using CATA 323
content validity through evaluating the adequacy of chosen words to represent other con-
structs. These five studies described varying processes to demonstrate content validity. For
instance, three of the five studies used expert judges to ensure that the chosen words fit the
construct of interest (e.g., Doucet & Jehn, 1997; Ferrier, 2001; Mossholder, Settoon, Harris,
& Armenakis, 1995), while the other two studies analyzed the meaning of selected diction-
ary words in actual sentences from the study’s text to ensure chosen words reflected the
meaning of the construct (e.g., Abrahamson & Park, 1994; Wade et al., 1997). In sum, while
it seems that most studies relied on deductive procedures, little detail was provided concern-
ing how authors selected and validated coding schemes.
External validity is the ability to generalize findings across multiple settings (Cook &
Campbell, 1979). We coded studies that applied a content analysis coding scheme to more
than one sample to evaluate the study’s predicted relationships as demonstrating external
validity. Our examination of the 98 studies revealed that only 12 (12%) demonstrated exter-
nal validity by including multiple samples in their research design. Given that most content
analysis studies rely on archival data, it is perhaps surprising that so few studies examined
external validity.
Discriminant validity involves the extent to which a construct is distinct from other con-
structs (Campbell & Fiske, 1959). We coded studies as addressing discriminant validity if
they examined dimensionality, created word lists for multiple constructs with the hopes of
establishing dimensionality, or used exploratory or confirmatory factor analysis to potentially
distinguish constructs. We found that only 5 studies of the 98 (5%) assessed the dimension-
ality or provided any procedures to assess discriminant validity of their study’s constructs.
Predicative, or nomological, validity is the extent to which measures predict other con-
structs to which they are theoretically linked (Cronbach & Meehl, 1955). We coded studies
as addressing predictive validity if the relationship between content analysis constructs and
other theoretically related constructs were analyzed. We discovered that 41 of the studies in
the review of Duriau et al. (2007; 42%) displayed predictive validity.
Table 1
Examination of Validity in Content Analysis Research
Content Analysis Studies Used by
Duriau et al. (2007)
Entrepreneurial Orientation Studies
Using Content Analysis
Number of studies 98 9
Deductive approach 79 of 98 (81%) 9 of 9 (100%)
Evaluated content validity of
word dictionaries
5 of 98 (5%) 0 of 9 (0%)
Computer-aided text analysis
(CATA)
24 of 98 (25%) 0 of 9 (0%)
Assessment of multiple
samples
12 of 98 (12%) 0 of 9 (0%)
Assessment of construct
dimensionality
5 of 98 (5%) 0 of 9 (0%)
Assessment of predicative
Validity
41 of 98 (42%) 9 of 9 (100%)
Citation support 36 of 98 (37%) 9 of 9 (100%)
324 Organizational Research Methods
To complement our examination of content analytic studies highlighted by Duriau et al.
(2007), we identified nine additional studies from the entrepreneurship and strategic man-
agement literatures that used content analysis to examine dimensions of entrepreneurial
orientation (the construct we examine in our illustration of potential procedures to assess
construct validity when using content analysis). Similar to our findings from the Duriau
et al. (2007) studies, we found that various forms of validity were not consistently addressed
in the nine entrepreneurial orientation studies. Although 8 of 9 (89%) studies used key word
dictionaries, only two described how words were selected for the dictionary and neither of
these demonstrated content validity by providing an explanation of the process for word
selection to accurately represent the studies’ constructs. Additionally, no study demon-
strated external validity through the use of multiple samples. Although discriminant validity
was not always applicable (some studies examined only one dimension of entrepreneurial
orientation), it is perhaps disappointing that no study addressed dimensionality given that
entrepreneurial orientation has been theorized as a multidimensional construct (Covin &
Slevin, 1989; Lumpkin & Dess, 1996). We were encouraged to find that all nine studies
demonstrated predictive validity by analyzing the relationship between entrepreneurial
orientation and firm performance, given that such linkages have consistently been theorized
in the entrepreneurship literature (Lumpkin & Dess, 1996).
In addition to assessing the extent to which validity was addressed in content analytic
studies, we also wanted to gauge the extent to which organizational researchers relied on
established content analysis methodology texts or other relevant sources when applying
content analysis to a particular question of interest. Thus, we examined the reference
sections of the 98 content analysis articles reviewed in Duriau et al. (2007), as well as the
9 content analysis articles measuring elements of entrepreneurial orientation, to assess
whether scholars referred to other content analysis research studies (scholarly articles
on the topic of content analysis) or statistical texts (books) about content analysis. Our
analysis revealed that content analysis studies often describe their procedures but fre-
quently do not cite other content analysis research in developing these procedures. For
instance, 63% of the content analysis studies identified by Duriau et al. (2007) did not cite
any content analysis–related work such as content analysis textbooks or other content
analysis studies that used a particular content analysis approach. Of those studies from
the Duriau et al. (2007) sample that did include a content analysis citation, the most fre-
quent was to Holsti (1969) at 10% followed by Weber (1990) at 9% and Jauch, Osborn,
and Martin (1980) at 7%. There are two potential interpretations of our findings. The gen-
eral lack of citing established content analysis work might suggest that management
researchers are not incorporating the most rigorous standards when applying content anal-
ysis to management research questions. Alternatively, researchers may in fact be incor-
porating established work into their techniques but provide little detail to explain their
process or guide future efforts.
Overall, our analyses show that no single study used all of the potential methods to assess
validity that we reviewed. The general lack of content validity process descriptions, as well
as infrequently demonstrated external, discriminant, and predictive validity and citation
support in developing content analysis procedures, suggest the need to propose validity
guidelines for studies using content analysis. Accordingly, in the next section, we suggest
a number of possible procedures to address issues of validity when using CATA.
Short et al. / Construct Validation Using CATA 325
Construct Validation Using CATA: Recommended Procedures
In this section, we suggest a number of procedures for validating constructs using content
analysis. Our goals are twofold. First, we seek to provide content analysis procedures that
are consistent with the applied nature of organizational research. Second, our aim is to pro-
vide a number of possibilities for researchers seeking to enhance construct validity when
incorporating content analysis. We believe our procedures can be applied to any construct
(unidimensional or multidimensional), as long as content analysis can be used to convey
meaning of naturally occurring phenomena (Van Maanen, 1979). However, we realize that
differences in research goals, as well as practical considerations (e.g., journal page length
restrictions), will make it unlikely that all of these procedures will be incorporated into any
given study. We begin by discussing elements related to content validity, followed by exter-
nal validity, reliability, dimensionality, and predictive (nomological) validity. We do not
mean to suggest that researchers who do not follow each of these techniques risk conducting
a study that lacks validity. Rather, we build on the guidelines of others who suggest that
more than one approach is desirable when seeking to establish validity (e.g., Campbell &
Fiske, 1959; Scandura & Williams, 2000). Furthermore, we offer a number of possibilities
to provide clear guidance for organizational researchers interested in examining validity
when using content analysis in general, and specifically the CATA approach we outline and
illustrate below. A summary of our proposed procedures is provided in Table 2.
Content Validity
Construct measurement begins with a sound theoretical definition of the concept of inter-
est. Thus, content validity involves an assessment of measures by experts/judges or a pro-
cess capable of examining a match between theoretical definition and empirical
measurement, which is demonstrated when a measurement procedure properly samples the
theoretical content domain of a construct (Nunnally & Bernstein, 1994). We advocate an
approach to content analysis using CATA with single words as the unit of analysis. This
type of content analysis is well established in psychological research traditions and is based
on the assumption that the words people use provide valuable insight related to the thought
processes reflected in narrative texts (Pennebaker, Mehl, & Niederhoffer, 2003). In the fol-
lowing sections, we provide guidance for the use of both deductive and inductive
approaches to CATA.
Deductively derived word lists. We recommend CATA begin with a deductive approach
that has been used by a number of others in the management field, where word lists are first
developed (independent of organizational documents) and then applied to meaningful firm-
level narratives (e.g., Porac, Wade, & Pollock, 1999; Wade et al., 1997).1 A deductive pro-
cess requires theory to design the coding scheme (Potter & Levine-Donnerstein, 1999).
Despite the prevalence of this technique in the literature (e.g., Dowling & Kabanoff,
1996; Weber, 1990), little guidance exists concerning how this method could be applied
to a variety of constructs germane to management research when existing dictionaries do
not adequately capture the construct’s content domain. We agree with others who advocated
deriving words from prior theory when possible (Dowling & Kabanoff, 1996; Weber, 1990)
326 Organizational Research Methods
and agree with organizational scholars who argued that content analysis might be an espe-
cially useful tool for examining adolescent theories (Sonpar & Golden-Biddle, 2008). How-
ever, we believe the only necessary condition is that researchers can create plausible word
lists to measure the construct of interest in relevant organizational narratives.
We suggest a four-step process for this deductive approach to content analysis. First,
researchers develop a working definition of the construct, preferably derived from existing
literature. Second, an initial assessment of the dimensionality of the construct of interest
should be made. Third, a list of key words that correspond to the formal definition is devel-
oped to capture the construct of interest. If the construct is conceptualized as multidimen-
sional, multiple discrete word lists are created for each subdimension. When developing
word lists, examining previously validated scales can yield insight and word suggestions.
An exhaustive list of synonyms should also be considered for the referent construct (or mul-
tiple lists if the construct is multidimensional). For example, Rodale’s (1978) The Synonym
Finder has been used in previous semiotic analysis (e.g., Markel, 1998). Fourth, word lists
are validated and rater reliability assessed (cf. Deephouse, 1996, 2000). Ideally, word lists
are assessed by a judge or panel of experts who are knowledgeable in the specific topic. We
Table 2
Recommended Procedures to Enhance Construct Validity When Using CATA
Deductive content validity
1. Create working definition of construct of interest (use a priori theory when possible)
2. Initial assessment of construct dimensionality based on existing literature
3. Develop an exhaustive list of key words from the formal definition to capture the construct of interest. (If
the construct is hypothesized to be multidimensional, multiple discrete word lists should be created for each
subdimension)
4. Validate word lists using content experts and assess rater reliability
Inductive content analysis
1. Identify commonly used words from narrative text of interest using DICTION or other CATA software
2. Identify or create a working definition of the construct of interest to guide word selection
3. Identify words that match the construct of interest
4. Establish initial interrater reliability
5. Refine and finalize word lists
Assess external validity
1. Select appropriate samples and relevant narrative texts to examine construct of interest
2. Compare two relevant samples when possible
Ensure reliability
Assure reliability by analyzing texts using a computer-aided technique such as DICTION, TEXTPACK,
WordStat, NVivo, or another that computes word counts (see Neuendorf, 2002, for a review of computer-
aided text analysis and associated programs)
Assess dimensionality (for multidimensional constructs)
Assess construct dimensionality using visual inspection of the correlation matrix. If dimensions are uncor-
related, they might be assessing different constructs and dimensions might exhibit problems of convergent
validity. If dimensions are correlated over .5, the construct may not be multidimensional. If dimensions
exhibit too high a correlation, consider collapsing subdimensions to form a single measure (or less
subdimensions)
Assess predictive (nomological) validity
Examine ability to predict theoretically related variables not captured via content analysis using regression,
structural equation modeling, or other relevant statistical technique
Short et al. / Construct Validation Using CATA 327
recommend Holsti’s (1969) method for assessing interrater reliability using the following
formula (PAO¼ 2A/nAþ nB) where PAO is proportion agreement observed, A is the number
of agreements between the two raters, and nA and nB are the number of words coded by the
two raters. Although there is no generally accepted ‘‘rule of thumb’’ for interrater reliability
coefficients analogous to the .70 heuristic for coefficient a, Riffe, Lacy, and Fico (2005)
and Krippendorff (2004) suggested interpreting values greater than .80, and Ellis (1994)
indicated coefficients between .75 and .80 as indicative of high reliability.
Inductively derived word lists. Although we advocate a deductive approach as a starting
point for our CATA approach, we believe there is value in exploring an inductive approach
to content analysis where narrative texts of interest are examined for potential words not
identified through deductive procedures alone. This approach is consistent with previous
organizational scholarship that has used a combination of inductive and deductive methods
in developing word lists (i.e., Doucet & Jehn, 1997; Kabanoff, Waldersee, & Cohen, 1995).
Such an approach is also consistent with the call for more engaged scholarship that contends
that knowledge is best produced by a cooperative arrangement between scholars who are
theory driven and practitioners who are closer to the phenomena; we believe that incorpora-
tion of deductive and inductive content analysis helps to bridge the knowledge transfer
problem characteristic of the divide between theory and practice (Van De Ven & Johnson,
2006).
We suggest a five-step process for an inductive approach to CATA content analysis.
First, a comprehensive list of commonly used words in the narrative of interest should be
identified. Second, scholars should identify or create a working definition of the construct
of interest. In step 3, the authors should examine words from the list of frequently used
words in the document of interest and assess whether they match the construct to be mea-
sured. In step 4, authors (or an additional rater) should independently examine the lists to
identify words associated with the construct of interest and then calculate an initial interra-
ter reliability (using the same Holsti [1969] method for assessing interrater reliability as
with the deductive analysis). Establishing reliability at this stage is important to assess gen-
eral agreement as to words that represent the construct of interest. In step 5, the list may be
refined by a discussion of potential words identified by the coders through an iterative pro-
cess until both raters agree that a particular word can be expected to meaningfully relate to
the construct of interest in the narrative. For multidimensional constructs, the inductive pro-
cedure could potentially yield additional words that relate to specific dimensions of a con-
struct. Conversely, the inductive analysis might also yield new dimensions not conceived
by previous efforts.
External Validity
External validity continues to be a key concern in management research methods (Savall,
Zardet, Bonnet, & Peron, 2008; Scandura & Williams, 2000), and techniques for establishing
external validity can be readily applied to research using content analysis (Neuendorf, 2002).
External validity refers to the confidence for which an inference regarding a causal relation-
ship can be generalized across different types of persons, settings, and times (Cook &
Campbell, 1979). For researchers using content analysis, two key decisions should be
328 Organizational Research Methods
considered carefully. The first decision involves the selection of narrative texts for analysis.
Selection of appropriate documents for content analysis is critical to ensure that the con-
struct can reasonably be expected to be detected in a given narrative. Thus, a fit between
research question and the type of document used for CATA should be considered prior
to the coding of documents. Examples of fit between document type and research questions
include mission statements used to examine organizational philosophy (e.g., Palmer &
Short, 2008; Pearce & David, 1987) and political speeches used to investigate elements
of espoused leadership (e.g., Bligh et al., 2004a, 2004b). The most commonly used narrative
text in the management literature, however, is the CEO’s letter to shareholders in annual
reports (Duriau et al., 2007), and these narratives have been analyzed to examine research
topics such as corporate strategy (Bowman, 1984) and managerial attention (D’Aveni &
MacMillan, 1990). Overall, we believe the key consideration is that the narrative text being
analyzed should provide an adequate setting with the construct that content analysis is being
used to detect (Krippendorff, 2004).
A second decision involves selecting an adequate sampling frame. Sampling decisions
are critical for establishing generalizability of study findings, but the decisions made by
organizational researchers concerning selection of a particular sampling frame have often
lacked adequate justification (Short, Ketchen, & Palmer, 2002). In some cases, practical
concerns may encourage reliance on certain samples such as the common usage of large
firms in the strategic management literature to allow pairing narrative documents with
archival firm performance data obtained from databases (e.g., COMPUSTAT). Regardless
of the sample used, similar to selection of the narrative text, researchers should be careful to
select a sample that is guided by theory and provides an adequate match with the research
question of interest (Short et al., 2002).
A comparison of different sampling frames provides additional insight into research
questions of interest. For example, use of the same measurement and analysis on a different
population constitutes an empirical generalization and a key form of replication (Tsang &
Kwan, 1999). Our review of content analysis studies suggests that demonstration of this
type of external validity is fairly uncommon in organizational research (12% of studies
examined by Duriau et al., 2007). This is problematic given the enthusiastic encouragement
for replication and triangulation in the management field (Hambrick, 2007; Scandura &
Williams, 2000; Singh, Ang, & Leong, 2003), and such efforts seem ideal for research using
content analysis.
Reliability
Reliability refers to the consistency achieved in construct measurement in terms of its
stability, dependability, and predictability (Kerlinger & Lee, 2000). To demonstrate ade-
quate reliability, we recommend the use of CATA as it has the advantages of greater relia-
bility and is less tedious than when content analysis is performed by human coders
(Dowling & Kabanoff, 1996). Additionally, the use of CATA exhibits high test–retest relia-
bility when examining transcripts of individuals over time (Schnurr, Rosenberg, Oxman, &
Tucker, 1986). Comparisons between word-count and human-scored content analysis have
found that CATA techniques are more accurate (Rosenberg, Schnurr, & Oxman, 1990). A
Short et al. / Construct Validation Using CATA 329
number of CATA programs are now available commercially such as DICTION, TEXT-
PACK, WordStat, or NVivo (see Neuendorf [2002] for a review of associated programs).
Dimensionality
Dimensionality refers to the association of measurement to a single construct; in the case
of multidimensional constructs, each dimension should be associated with a single factor to
ensure adequate discriminant validity (Nunnally & Bernstein, 1994). We recommend cre-
ating multiple word lists when constructs of interest are conceptualized as multidimen-
sional. When using content analysis, researchers can examine content analysis results
from multiple word lists and assess construct dimensionality simply by using visual inspec-
tion of the correlation matrix comparing multiple word lists (cf. Hair, Anderson, Tatham, &
Black, 1998). If dimensions are uncorrelated, researchers might consider whether they are
actually assessing different constructs. If results demonstrate (a) excessively high correla-
tions (more than .80), (b) no correlations, or (c) negative correlations (when none are
hypothesized), Hair et al. (1998) recommended conducting additional analyses to assess
dimensionality. For example, if dimensions exhibit too high of a correlation, consider col-
lapsing subdimensions to form a single measure, or conduct a factor analysis of word lists to
suggest fewer dimensions of a particular construct of interest.
Predictive Validity
Predictive validity (often dubbed nomological or criterion-related validity) refers to the
extent to which the operationalization predicts other constructs consistent with theoretically
derived a priori expectations (Cronbach & Meehl, 1955). It is demonstrated when con-
structs of interest are linked with others that are theoretically related. What sets this type
of validity apart from others is its emphasis on the role of theory and accompanying hypoth-
esis tests (Kerlinger & Lee, 2000). Unfortunately, predictive validity has often been under-
used in studies relying on content analysis (Neuendorf, 2002; Weber, 1990). To avoid issues
associated with common method variance, we encourage researchers to examine dependent
variables not captured via content analysis. In the strategic management literature, for
example, organizational performance operationalized with objective, archival data have
generally been the key dependent variable of interest (Meyer, 1991). Predictive validity can
be assessed using regression, structural equation modeling, or other accepted methodologi-
cal research methods associated within a particular research stream.
Validating CATA in Organizational Research: An Illustration Using
Entrepreneurial Orientation
To illustrate the potential application of our suggestions for construct validation when
using content analysis, we examine the construct of entrepreneurial orientation (Covin &
Slevin, 1988, 1989; Lumpkin & Dess, 1996). We believe this construct is attractive for a
number of reasons. First, although research in entrepreneurship in general has been charac-
terized by low paradigmatic development (Ireland, Reutzel, & Webb, 2005; Ireland, Webb,
330 Organizational Research Methods
& Coombs, 2005), entrepreneurial orientation has become a central construct in both the
strategic management and entrepreneurship literatures (Lyon et al., 2000; Tang, Tang, Mar-
ino, & Zhang, & Li, 2008). For example, according to scholar.google.com, Lumpkin and
Dess’s (1996) seminal article that appeared in Academy of Management Review has been
cited more than 1,000 times.
Despite the prevalence of the entrepreneurial orientation construct, numerous different
measures are used across this research stream. In many cases, scales are modified by drop-
ping (or adding) items without providing theoretical justification or demonstrations of psy-
chometric adequacy (e.g., Atuahene-Gima & Ko, 2001; George, Wiklund, & Zahra, 2005;
Slater & Narver, 2000). A meta-analysis of 51 entrepreneurial orientation studies by Rauch,
Wiklund, Frese, and Lumpkin (2004) reported that different variations (consisting of alter-
native or additional dimensions) of the foundational scales (e.g., Miller, 1983; Covin & Sle-
vin, 1989) were used in almost half of the studies, concluding that future research needs
more reliable and valid scales of the dimensions of entrepreneurial orientation (including
alternative approaches to measurement). Lyon et al. (2000) specifically advocated the use
of content analysis to measure entrepreneurial orientation.
Despite the espoused benefits of content analysis, its use in entrepreneurship studies has
been relatively sparse (Chandler & Lyon, 2001). However, content analysis has been
demonstrated as a useful approach to measuring aspects of entrepreneurial orientation
(e.g., Chen & Hambrick, 1995), and nine articles thus far have used content analysis to mea-
sure at least some elements of the construct. Yet, few studies have examined all five dimen-
sions outlined in Lumpkin and Dess’s (1996) Academy of Management Review article (see
Hughes & Morgan [2007] for a rare exception), and no study to date has used content anal-
ysis to link all five dimensions to firm performance in a single study. Thus, we believe
entrepreneurial orientation is an appropriate construct to demonstrate how content analysis
can make a methodological contribution to aid substantive research questions in the entre-
preneurship literature, especially when measurement has posed challenges in the past.
Furthermore, we believe that illustrating how entrepreneurial orientation can be assessed
with content analysis of shareholder letters is an especially attractive approach and offers
value to management researchers.
Content Validity
We began by identifying a formal definition of the entrepreneurial orientation construct
from the literature. We relied on the Lumpkin and Dess’ (1996) definition of entrepreneur-
ial orientation as the ‘‘processes, practices, and decision-making activities that lead to new
entry’’ (p. 136). Our second step involved an initial assessment of the content domain and
dimensionality of the entrepreneurial orientation construct based on an extensive review of
the literature. In the case of entrepreneurial orientation, considerable conceptual and
empirical work supports a five-dimension conceptualization of entrepreneurial orientation:
autonomy, competitive aggressiveness, innovativeness, proactiveness, and risk taking
(Lumpkin & Dess, 1996). In our third step, we developed an exhaustive list of words to cap-
ture each of the five entrepreneurial orientation dimensions. Had theory suggested entrepre-
neurial orientation as a unidimensional construct, we would have created a single word list;
because entrepreneurial orientation has been a priori theorized as multidimensional, we
Short et al. / Construct Validation Using CATA 331
created a discrete and exhaustive word list for each of its theoretically based dimensions
from The Synonym Finder (Rodale, 1978). As an example, the dictionary for autonomy
included variants of the word (autonomous), as well as synonyms (self-directing). It is
important that word lists used are exhaustive, mutually exclusive, and at the appropriate
level of measurement, so that each word can be associated with one and only one dimension
in the case of multidimensional constructs (Neuendorf, 2002).
Word lists were validated in step 4 using a multistep process with raters. In our example,
two raters examined the words generated by The Synonym Finder and compared them with
the theoretical definition of entrepreneurial orientation, deleting words that did not match the
specified dimension and adding words that could be associated with the construct. In our case,
both raters had published in the entrepreneurship literature and one scholar was blind to the
study’s purpose and not part of the author team. Of the 717 words generated by The Synonym
Finder and the 43 words added by our two raters, 244 words were selected by both raters as
representative of entrepreneurial orientation and were retained for subsequent analyses
(autonomy¼ 36 words, innovativeness¼ 86, proactiveness¼ 27, competitive aggressiveness
¼ 58, risk taking ¼ 37, respectively; the final word list is presented in Table 3).
To demonstrate interrater reliability of our nominal coding scheme (agree/disagree
between the raters), we used Holsti’s (1969) method discussed previously. Our observed
reliabilities ranged from .75 to .88 (reliabilities for each of the dimensions are presented
in Table 6), demonstrating consistency between our raters.
We supplemented our deductive analysis with an inductive procedure based on the word
choices evident in CEO letters to shareholders (the document of interest in our illustrative
example). First, we identified a comprehensive list of commonly used words from our sam-
ple of shareholder letters. Similar to others who have applied an inductive approach to dic-
tionary development (e.g., Abrahamson & Park, 1994; Doucet & Jehn, 1997), we generated
an exhaustive list of frequently used words within all texts across both of our samples. DIC-
TION was helpful in this regard as it calculates an ‘‘insistence score,’’ which assesses the
degree to which a text is dependent on often-repeated words. To calculate this score, DIC-
TION identifies all words mentioned three or more times within a particular text. We
extracted these words and created an exhaustive list of frequently used words across all
texts from our two samples. This analysis resulted in 3,331 frequently used words found
in our shareholder letter samples. Second, we looked for a general definition/explanation
of entrepreneurial orientation that could guide our inductive coding. Such a definition could
be generated by the researchers interested in conducting an inductive analysis or culled
from previous work. Given the prevalence of research in entrepreneurial orientation, we
used The Blackwell Encyclopedia of Management: Entrepreneurship (Hitt & Ireland,
2005) to guide us toward a general understanding of the nature of the entrepreneurial orien-
tation construct. The section of the encyclopedia included a general discussion of the entre-
preneurial orientation construct as well as specific definitions. For example, the authors of
this section (Lumpkin & Dess, 2005) noted (p. 104):
An early definition of entrepreneurial orientation was provided by Miller, who stated an entre-
preneurial firm is one that ‘‘engages in product market innovation, undertakes somewhat risky
ventures and is first to come up with ‘proactive’ innovations, beating competitors to the
punch.’’ (1983, p. 770)
332 Organizational Research Methods
Table 3
Word Lists for Entrepreneurial Orientation Dimensions
Entrepreneurial Orientation
Dimension Content Analysis Words With Expert Validation
Autonomy At-liberty, authority, authorization, autonomic, autonomous, autonomy, decon-
trol, deregulation, distinct, do-it-yourself, emancipation, free, freedom, free-
thinking, independence, independent, liberty, license, on-one’s-own, preroga-
tive, self-directed, self-directing, self-direction, self-rule, self-ruling, separate,
sovereign, sovereignty, unaffiliated, unattached, unconfined, unconnected,
unfettered, unforced, ungoverned, unregulated
Innovativeness Ad-lib, adroit, adroitness, bright-idea, change, clever, cleverness, conceive,
concoct, concoction, concoctive, conjure-up, create, creation, creative, crea-
tivity, creator, discover, discoverer, discovery, dream, dream-up, envisage,
envision, expert, form, formulation, frame, framer, freethinker, genesis, genius,
gifted, hit-upon, imagination, imaginative, imagine, improvise, ingenious,
ingenuity, initiative, initiator, innovate, innovation, inspiration, inspired,
invent, invented, invention, inventive, inventiveness, inventor, make-up,
mastermind, master-stroke, metamorphose, metamorphosis, neoteric, neoter-
ism, neoterize, new, new-wrinkle, innovation, novel, novelty, original, ori-
ginality, originate, origination, originative, originator, patent, radical, recast,
recasting, resourceful, resourcefulness, restyle, restyling, revolutionize, see-
things, think-up, trademark, vision, visionary, visualize
Proactiveness Anticipate, envision, expect, exploration, exploratory, explore, forecast, fore-
glimpse, foreknow, foresee, foretell, forward-looking, inquire, inquiry, inves-
tigate, investigation, look-into, opportunity-seeking, proactive, probe,
prospect, research, scrutinization, scrutiny, search, study, survey
Competitive aggressiveness Achievement, aggressive, ambitious, antagonist, antagonistic, aspirant, battle,
battler, capitalize, challenge, challenger, combat, combative, compete, comp-
eter, competing, competition, competitive, competitor, competitory, conflict-
ing, contend, contender, contentious, contest, contestant, cutthroat, defend,
dog-eat-dog, enemy, engage, entrant, exploit, fierce, fight, fighter, foe, intense,
intensified, intensive, jockey-for-position, joust, jouster, lock-horns, opponent,
oppose, opposing, opposition, play-against, ready-to-fight, rival, spar, strive,
striving, struggle, tussle, vying, wrestle
Risk taking Adventuresome, adventurous, audacious, bet, bold, bold-spirited, brash, brave,
chance, chancy, courageous, danger, dangerous, dare, daredevil, daring,
dauntless, dicey, enterprising, fearless, gamble, gutsy, headlong, incautious,
intrepid, plunge, precarious, rash, reckless, risk, risky, stake, temerity, uncer-
tain, venture, venturesome, wager
Additional inductively
derived words
Advanced, advantage, commercialization, customer-centric, customized,
develop, developed, developing, development, developments, emerging,
enterprise, enterprises, entrepreneurial, exposure, exposures, feature, features,
founding, high-value, initiated, initiatives, innovations, innovative, introduc-
tions, launch, launched, leading, opportunities, opportunity, originated, out-
doing, outthinking, patents, proprietary, prospects, prototyping, pursuing, risks,
unique, ventures
Source: Deductive word lists were developed with the aid of Rodale’s (1978) The Synonym Finder. Of the 717
words generated by The Synonym Finder and the 43 words added by our two raters, 244 words were selected by
both raters as representative of entrepreneurial orientation and were retained for subsequent analyses.
Short et al. / Construct Validation Using CATA 333
In step 3, two authors independently examined the lists to identify words associated with
entrepreneurial orientation and then calculated an interrater reliability (using the same Hol-
sti [1969] method for assessing interrater reliability as with the deductive analysis). Estab-
lishing reliability at this stage is important to assess general agreement as to words that
represent the construct of interest. Our initial interrater reliability for our inductive analysis
was .97. In step 4, the list was refined by a discussion of potential words identified by one
but not both authors through an iterative process until both authors agreed that a particular
word could be expected to meaningfully relate to the construct of entrepreneurial orienta-
tion in shareholder letters.
The identification of additional words using an inductive approach suggests the value of
both deductive and inductive methods when using CATA. Indeed, our inductive analysis
led to the identification of 41 additional words not generated by the deductive procedures.
To highlight the additional words from the inductive analysis, we present them separately
from the deductively defined lists in Table 3. The incorporation of an inductive approach
could also be used to identify the presence of additional entrepreneurial orientation dimen-
sions or the need for greater refinement of existing dimensions. For instance, one induc-
tively derived word, ‘‘commercialization’’ seems to capture unique elements of
entrepreneurship not fully developed in current conceptualizations of entrepreneurial orien-
tation. Commercialization is a process of converting knowledge into marketable products or
services and is an important aspect of entrepreneurship that is vital to new venture growth
(e.g., Kazanjian, 1988; Shane, 2004). Although this process is likely related to other entre-
preneurial orientation dimensions (e.g., proactiveness, risk taking, etc.), a firm’s propensity
to commercialize independent of being proactive or taking risks may represent a unique
dimension of its entrepreneurial orientation. Although we did not separate terms associated
with commercialization from other words identified through our inductive analysis, sub-
stantive efforts may benefit from an iterative process, where inductively derived words are
either incorporated into deductive lists or used to reveal new dimensions.
External Validity
We relied on shareholder letters as the organizational narrative that was particularly
applicable to the measurement of entrepreneurial orientation for a number of reasons. First,
letters to shareholders are excellent sources of managerial cognitions as they provide a
means for reconstructing perceptions and beliefs of authors (D’Aveni & MacMillan,
1990). Shareholder letters in particular signal the major topics and themes to which man-
agers attend (Barr, Stimpert, & Huff, 1992). Thus, annual report texts are useful for captur-
ing elements of top management’s values, beliefs, and ideologies (which include
entrepreneurial orientations). Second, the letter is the most widely read section of the annual
report (Courtis, 1982) and provides a forum for the CEO to voice thoughts on important
issues affecting the organization (Goodman, 1980). Indeed, empirical evidence suggests
that the CEO actively participates in the writing process of the letter (cf. Amernic, Craig,
& Tourish, 2007). Furthermore, the rhetoric in these texts has been related to organizational
actions and outcomes (Bowman, 1984; Michalisin, 2001).
For example, assertions of innovativeness (one dimension of entrepreneurial orientation)
in the letter to shareholders were related to firm reputation for innovation and the number of
334 Organizational Research Methods
trademarks applied for by the firm (Michalisin, 2001). Such results lend support to the
notion that communication in the shareholder letters reflects actual firm behaviors and
makes them particularly useful for examinations of entrepreneurial orientation.
We chose the S&P 500 for our sampling frame. There is recognition in the strategic
management literature that large established firms may benefit or even be forced to adopt
an ‘‘entrepreneurial mindset’’ in highly competitive landscapes (Bettis & Hitt, 1995) and
thus should undertake entrepreneurial behaviors to grow and survive (Zahra, 1991). With
growing interest in entrepreneurship within large firms, it is not surprising that entrepre-
neurial orientation has not been limited to the study of small or nascent firms, but rather,
sampling frames in previous studies have ranged from small private firms to very large
public firms drawn from the Fortune 500 (e.g., Hult, Snow, & Kandemir, 2003; Zahra,
1996) as well as public firms with more than US$25 million in sales (e.g., Sarkar, Echambadi,
& Harrison, 2001). Because the S&P 500 is comprised of publicly traded firms, the use of
this sample allows for the collection of additional variables (i.e., measures of organiza-
tional size and firm performance) through a secondary source without the threat of intro-
ducing common method variance. These measures, obtained from the COMPUSTAT
database, were used to demonstrate nomological validity of our measure of entrepreneur-
ial orientation. Our criterion for the use of a particular letter is that the firm was consis-
tently listed in the S&P 500 for 2001–2005, resulting in letters and data from 450 firms in
2005 (our year of data collection).
To allow for an assessment of external validity, we collected an additional sample com-
prised of small, high-growth firms to allow for comparisons with our sample of S&P 500
firms. We selected firms from the Russell 20001 stock index for this comparison group.
The 2,000 firms in this index are drawn from a larger Russell 30001 stock index that rep-
resents approximately 98% of the total value of the U.S. equity market (Russell Invest-
ments, 2008a). In contrast, the Russell 20001 firms represent only 10% of the total
value of the Russell 30001 stock index and as such provide a large sample of small, pub-
licly traded firms (Russell Investments, 2008b). Moreover, the Russell 20001 is considered
an important benchmark index for small valued companies in investment portfolios (Cheng,
Liu, & Qian, 2006; Reilly & Wright, 2002). Specifically, we selected the 500 highest
growth firms in this index. Following the method used by Inc. magazine to select its top
500 private entrepreneurial firms, we calculated the average annual revenue growth for
each firm in the Russell 20001 index for more than a five-year period (2001–2005) and
selected the top 500 high growth firms for our comparison sample. Average annual revenue
growth for our Russell 20001 sample was much higher than the S&P 500 sample (i.e.,
148% vs. 18%), suggesting that our Russell 20001 sample involved firms who collectively
pursue entrepreneurial activities that spur such revenue growth (and thus, a more entrepre-
neurial sample to compare with our S&P 500 sample). Similar to our S&P 500 sample, we
collected shareholder letters for 2005 and content analyzed the resulting 205 shareholder
letters with DICTION using our entrepreneurial orientation custom dictionaries.
For both samples, we conducted one sample t tests (compared to a test statistic of zero)
for each entrepreneurial orientation dimension to evaluate the presence of language consis-
tent with an entrepreneurial orientation in shareholder letters (a zero result indicates that
language consistent with an entrepreneurial orientation is not present in our sample of
shareholder letters). As shown in Table 4, all dimensions were significant suggesting that
Short et al. / Construct Validation Using CATA 335
language consistent with an entrepreneurial orientation was communicated in the share-
holder letters across both samples.
We also directly examined external validity with the two samples. Because of the differ-
ences in available shareholder letters between the two samples, we took a random set of 205
firms from our S&P 500 sample to ensure that differences between samples were not influ-
enced by sample size differences. To evaluate mean differences between the samples, we
conducted a one-way analysis of variance (ANOVA). Results of this analysis revealed
no significant differences in mean values in entrepreneurial orientation dimensions with the
exception of proactiveness, where higher mean levels were detected in the shareholder let-
ters of Russell 20001 firms (See Table 5).
Reliability
To enhance reliability, we relied on CATA as it minimizes the error from human coders
(such as rater fatigue or inadequate coder training). Similar to human coding schemes,
CATA generally assesses content via word usage (Morris, 1994). Relying on text to study
cognition assumes that the insights about the authors’ mental models can be detected
through the presence of, absence of, and frequency with which certain concepts are used
in text (Carley, 1997). CATA is advantageous in that multiple texts can be analyzed in min-
utes with near perfect reliability and without coder bias (Stevenson, 2001).
A number of software programs are available for content analysis. In our analyses, we
used DICTION (Hart, 2000) because it has been used in previous research examining man-
agement constructs such as charismatic leadership (Bligh et al., 2004a, 2004b). Further-
more, researchers have advocated DICTION as content analysis software with the
potential to measure a number of theoretically based constructs of interest to strategic man-
agement and entrepreneurship research (Short & Palmer, 2008). Using CATA, we analyzed
1,512 single-spaced pages of text in the S&P 500 and Russell 20001 samples.
Table 4
Evidence of Language Representing Entrepreneurial Orientation Dimensions in
Shareholder Letters of Firms in S&P 500 and Russell 20001
S&P 500 Sample Russell 20001 Sample
N Mean SD t Test N Mean SD t Test
Autonomy 453 1.01 2.11 10.23* 205 .65 1.07 8.70*
Competitive aggressiveness 453 3.13 2.99 22.25* 205 2.04 2.17 13.46*
Innovativeness 453 12.76 8.94 30.38* 205 9.13 7.75 16.86*
Proactiveness 453 2.95 3.20 19.60* 205 3.07 3.62 12.14*
Risk taking 453 1.11 2.05 11.50* 205 .78 2.02 5.49*
Inductively derived entrepreneurial orientation words 453 18.94 13.46 29.85* 205 15.22 10.85 20.03*
Note: The results of this table were based on computer-aided text analysis using the word lists for entrepre-
neurial orientation presented in Table 3.*p < .01.
336 Organizational Research Methods
Dimensionality
To assess the multidimensionality of entrepreneurial orientation, we conducted two addi-
tional steps. An underlying assumption in construct measurement is that measures should be
associated with a single concept for unidimensional measures; for constructs that are pro-
posed to be multidimensional, each dimension should be distinct from but related to the others
(Edwards, 2000). When using content analysis to assess construct validity, we propose exam-
ination of the correlation matrix of the DICTION scores to reveal significant but not perfect
correlations between dimensions for a multidimensional construct; high correlations would
provide evidence of a unidimensional construct. Table 6 displays the correlations for each
dimension of entrepreneurial orientation in our two samples, based on the bivariate correla-
tions of the word lists used to measure entrepreneurial orientation dimensions. These results
provide evidence that entrepreneurial orientation is a multidimensional construct. For the
S&P 500 firms, all dimensions were correlated at the p < .05 or greater level, with the excep-
tion of risk taking and innovativeness. Similar to the S&P 500 sample, the correlations among
dimensions in the Russell 20001 sample were all positive except for the correlation between
autonomy and risk taking. Moreover, although not all dimensions in the Russell 20001 sam-
ple displayed a significant relationship with the other dimensions, each dimension was signif-
icantly related (at least p < .05) to at least one other dimension except risk taking. In both
samples, no correlations were higher than .40. Thus, as all measures were correlated less than
.5 with any other measure, evidence from our sample is consistent with Lumpkin and Dess’s
(1996) multidimensional conceptualization of entrepreneurial orientation.
Predictive Validity
In entrepreneurial orientation research, the most examined causal relationship of interest
is between entrepreneurial orientation and firm performance, where most studies assume a
Table 5
ANOVA Comparisons of S&P 500 to Russell 20001 Firms on Entrepreneurial
Orientation Dimensions
Entrepreneurial Orientation Dimension
S&P 500 Sample
(N ¼ 205)
Russell 20001 Sample
(N ¼ 205) F Test
Autonomy .61 .57 .16
Competitive aggressiveness 1.87 1.64 2.43
Innovativeness 7.31 7.15 .15
Proactiveness 1.67 2.61 14.68**
Risk taking .76 .53 4.00*
Inductively derived entrepreneurial
orientation words
11.42 14.06 9.16**
Total entrepreneurial orientation 20.54 21.90 3.05
Note: ANOVA ¼ analysis of variance. The results of this table were based on computer-aided text analysis
using the word lists for entrepreneurial orientation presented in Table 3. Entrepreneurial orientation dimensions
are standardized by the number of words in the shareholder letter.*p < .05.**p < .01.
Short et al. / Construct Validation Using CATA 337
direct and positive relationship (Rauch et al., 2004). To measure firm performance we relied
on Tobin’s Q, a measure that assesses the degree to which the stock market values a firm
relative to its replacement cost (Mehran, 1995) and is commonly used in the entrepreneur-
ship literature (e.g., Andrews & Welbourne, 2000; Dyer & Whetten, 2006). To allow for a
potential lag effect, we used a 2-year performance measure based on the average Tobin’s Q
for the years 2005 and 2006. To assess the relationship of the entrepreneurial orientation
dimensions with firm performance, we first controlled for organizational size using the
logarithm of full-time employees (Konrad & Mangel, 2000).
During the process of examining our shareholder letters, we noticed that there was con-
siderable variance in the length of these documents. The mean number of words used in our
sample of CEO letters to shareholders was 1,566 words with the shortest letter using only
126 words and the longest using 6,513 words. To control for such a wide discrepancy in text
length across our sample, we standardized our detected entrepreneurial orientation con-
struct words by dividing them by the total number of words found in the text for our tests
of predictive validity. This control method is similar to others used in content analytic stud-
ies and is appropriate when texts vary considerably in length (i.e., Doucet & Jehn, 1997;
Emrich, Brower, Feldman, & Garland, 2001).
Using stepwise regression, we found that the five entrepreneurial orientation dimensions
as a set significantly predicted performance for the S&P 500 and Russell 20001 samples.
As shown in Table 7, the magnitude of each dimension’s predictive power varied among
performance measures. For instance, in both samples, competitive aggressiveness and risk
Table 6
Intercorrelations of Entrepreneurial Orientation Dimensions to Assess
Dimensionality
Entrepreneurial Orientation Dimension 1 2 3 4 5 6
S&P 500 Sample
1. Autonomy .80
2. Competitive Aggressiveness .15** .75
3. Innovativeness .17** .38** .88
4. Proactiveness .10* .20** .30** .85
5. Risk Taking .11* .21** .06 .09* .83
6. Inductive Entrepreneurial Orientation (unique words) .18** .44** .57** .42** .13** .97
Russell 20001 Sample
1. Autonomy .80
2. Competitive aggressiveness .16** .75
3. Innovativeness .17** .39** .88
4. Proactiveness .01 .11** .19** .85
5. Risk Taking �.03 .13y .10 .07 .83
6. Inductive entrepreneurial orientation (unique words) .14* .32** .62** .36** .17* .97
Note: Holsti’s (1969) observed agreement proportions are shown in italics in the diagonal. The results of this
table were based on computer-aided text analysis using the word lists for entrepreneurial orientation presented in
Table 3.*p < .05.**p < .01.yp < .10.
338 Organizational Research Methods
taking displayed a significantly negative relationship with performance while innovative-
ness and proactiveness demonstrated a positive relationship to performance (although these
positive relationships were significant only in the S&P 500 sample). Autonomy was not sig-
nificantly related to performance in either sample. Table 8 shows the significant relation-
ships of the inductively defined entrepreneurial orientation words for both samples. In
Table 9, we examine both deductive and inductive lists and again find significance for both
samples. Overall, we found a fairly consistent pattern relating the entrepreneurial orienta-
tion measures to performance for both samples.
Our results with regard to mixed dimensionality and nonfindings vis-a-vis entrepreneur-
ial orientation dimensions and performance are not surprising and are consistent with find-
ings in this research stream (cf. Kreiser, Marino, & Weaver, 2002; Rauch et al., 2004). For
example, the relationship between proactiveness and performance has been equivocal. Sar-
kar et al. (2001) found a positive relationship between proactiveness and firm performance
(with proactiveness in smaller firms exhibiting the strongest relationship to performance).
Conversely, Morgan and Strong (2003) found a negative but nonsignificant relationship
between proactiveness and firm performance in their sample of medium-to-large high tech-
nology and industrial manufacturing firms. This pattern of results is also found in the
dimension of competitive aggressiveness. Ferrier (2001) reported a significant positive rela-
tionship with the number of competitive actions and firm performance, whereas Lumpkin
and Dess (2001) revealed that competitive aggressiveness was negatively but not signifi-
cantly related to firm performance. Additionally, in a rare study that included all five
dimensions of entrepreneurial orientation, mixed relationships were found with each
dimension and firm performance (Hughes & Morgan, 2007). In this analysis, proactiveness
and innovativeness were positively related to performance, while risk taking was negatively
Table 7
Hierarchical Regression Analyses for Deductively Defined Entrepreneurial
Orientation Measures Predicting Tobin’s Q: S&P 500 Firms and Russell 20001
Firms
S&P 500 Sample Russell 20001 Sample
1 2 1 2
Firm size (logarithm of employees) �.09 �.13** �.18** �.14*
Autonomy �.03 .02
Competitive aggressiveness �.08 �.11
Innovativeness .19** .15*
Proactiveness .12* .18*
Risk taking �.21** �.15*
R2 .01 .12 .03 .09
F for change in R2 3.46* 11.09** 6.05* 3.97**
Note: Table contains standardized regression coefficients. Measurement of entrepreneurial orientation dimen-
sions was based on computer-aided text analysis using the word lists for entrepreneurial orientation presented in
Table 3. Entrepreneurial orientation dimensions are standardized by the number of words in the shareholder
letter.*p < .05.**p < .01.
Short et al. / Construct Validation Using CATA 339
related and competitive aggressiveness and autonomy were positively but not significantly
related to firm performance. Our results incorporating deductive and inductive lists yielded
R2s of .14 and .13 for S&P 500 and Russell 20001 samples, respectively. These results rep-
resent some of the strongest relationships of entrepreneurial orientation to firm performance
to date (cf., Stam & Elfring, 2008; Tang et al., 2008; Walter, Auer, & Ritter, 2006).
Table 8
Hierarchical Regression Analyses for Inductively Based Entrepreneurial
Orientation Measure Predicting Tobin’s Q
S&P 500 Sample Russell 20001 Sample
1 2 1 2
Firm size (logarithm of employees) �.09 �.07 �.17* �.17*
Inductively derived entrepreneurial orientation words .21** .08
R2 .01 .05 .03 .04
F for change in R2 3.67 19.37** 5.93* 1.10
Note: Table contains standardized regression coefficients. Inductive entrepreneurial orientation is generated
from a computer-aided text analysis using a word list generated from an inductive analysis of CEO 2005
shareholder letters. This measure is standardized by the number of words in the shareholder letter. Tobin’s Q is
calculated using an average of performance for the years 2005 and 2006.*p < .05.**p < .01.
Table 9
Hierarchical Regression Analysis for Deductive and Inductive Entrepreneurial
Orientation Measures Predicting Tobin’s Q: S&P 500 Firms and Russell 20001
Firms
S&P 500 Sample Russell 20001 Sample
1 2 1 2
Firm size (logarithm of employees) �.09 �.10* �.18* �.14*
Autonomy �.03 .00
Competitive aggressiveness �.09 �.11
Innovativeness .15** .11
Proactiveness .10* .15*
Risk taking �.19** �.14*
Inductively derived entrepreneurial
orientation words (unique words)
.17** .20**
R2 .01 .14 .03 .13
F for change in R2 3.31 11.41** 6.05* 4.78**
Note: Table contains standardized regression coefficients. Measurement of entrepreneurial orientation dimen-
sions was based on computer-aided text analysis using the word lists for entrepreneurial orientation presented in
Table 3. Entrepreneurial orientation dimensions are standardized by the number of words in the shareholder
letter.*p < .05.**p < .01.
340 Organizational Research Methods
Discussion
We examine the use of methods to assess construct validity when using content analysis
and provide a number of useful procedures for future research efforts. By doing so,
researchers will be able to measure concepts of interest via content analysis for a variety
of subjects, where traditional survey or other measurement techniques are difficult to apply
while maintaining acceptable standards of validity. By implementing the checks on valid-
ity, we suggest, researchers can have more confidence in the inferences made from their
work. Failure to capitalize on such opportunities could result in inconsistent findings, lim-
iting our understanding of substantive topics of interest.
The contributions of our work should be viewed in light of its limitations. For example,
examinations of other types of validity would further strengthen and supplement the tests
we propose. Convergent validity (assessment of correlations between two measures of the
same construct) could potentially be assessed by comparing traditional survey measures of
entrepreneurial orientation with the content analytic measures we assessed. Given tradi-
tional challenges in response rates of CEOs, this could be a daunting task with sampling
frames such as the S&P 500. An alternative means to assess convergent validity would
be to compare CATA with results based on human coders. For example, research using
DICTION detected a high level of agreement between human coders and predefined DIC-
TION variables used to measure charismatic leadership (Bligh et al., 2004a, 2004b). Further
assessment of discriminant validity could be achieved when constructs such as entrepre-
neurial orientation are uncorrelated with dissimilar constructs such as consideration and
initiating structure dimensions of leadership (cf. Weber, 1990). In short, the limitations
of the current study represent fruitful avenues for future research.
The use of CATA may be accompanied by a number of trade-offs to human-coded anal-
ysis. For example, CATA is less context sensitive than human coders for detecting the mean-
ing of a word within a sentence. However, Rosenberg et al. (1990) found that computers
calculating word frequencies outperformed human coders in a project that assigned psychia-
tric patients to a diagnostic category based on a content analysis of their speech. Such a find-
ing suggests that the human coding advantage is less clear than some scholars might suspect.
Any advantage of human coding over CATA, however, must be weighed in light of the
advantages of CATA (i.e., perfectly reliable analysis and ability to analyze a large number
of texts quickly). Yet, our CATA approach may be less sensitive to examining the temporal
elements of word choice. For example, the unexpected finding that risk taking was signifi-
cantly and negatively related to firm performance in both of our samples could be a function
of word lists representing the risk-taking dimension of entrepreneurial orientation potentially
referring to past, present, and future risks all in the same shareholder letter. Such possibilities
suggest that complementing CATA with human coding could be helpful in discovering nuan-
ces concerning how a given construct is portrayed in a given sample of interest.
Our process is also useful for examining substantive research questions in a novel man-
ner that may allay previous concerns associated within some research streams. For example,
as mentioned in our illustration, tests linking entrepreneurial orientation to firm perfor-
mance have been equivocal (e.g., Smart & Conant, 1994; Zahra & Covin, 1995). Entrepre-
neurial orientation studies have often used perceptual measures of performance (e.g., Naldi,
Short et al. / Construct Validation Using CATA 341
Nordquist, Sjoberg, & Wiklund, 2007), but this practice could be problematic by introdu-
cing potential for common method variance. Our examination ameliorates such issues
because assessing entrepreneurial orientation through content analysis and performance via
COMPUSTAT provides an improvement over studies that have captured both constructs
through survey measures.
Extensions of our proposed procedures could yield further benefit. For example, in
exploring entrepreneurial orientation, we examined external validity by comparing two
samples using the same narrative text (i.e., CEO letters to shareholders). Our examination
of different samples using the same measures with two different populations constitutes an
empirical generalization, a key form of replication in the organizational sciences (Tsang &
Kwan, 1999). Analysis of different populations using an alternative data source would pro-
vide an additional form of empirical generalization. For example, entrepreneurship scholars
have called for content analysis of IPO prospectuses as a particularly useful data source
(Marino, Castaldi, & Dollinger, 1989). Previous content analyses found little link between
information found in CEO prospectus statements and organizational performance (e.g.,
Daily, Certo, & Dalton, 2005), but applying the dictionaries we created to prospectus state-
ments would provide an additional perspective. Other documents such as mission state-
ments have been advocated as important narratives that reflect organizational identity
(Palmer & Short, 2008), and examination of mission statements may also provide valuable
insight concerning articulation of entrepreneurial orientation when defining an organiza-
tion’s raison d’etre. Differences in firm performance have often been found based on indus-
try (e.g., Rumelt, 1991; McGahan & Porter, 1997) and strategic group membership (e.g.,
Short, Ketchen, Palmer, & Hult, 2007). Thus, comparisons between differences in specific
industry contexts (such as high-tech industries) and other industry contexts could provide an
additional extension for demonstrating external validity.
Conclusion
Measurement in the organizational literature is dynamic, and sophisticated analytic tech-
niques continue to emerge and evolve. Despite such advancements, little guidance exists in
terms of how to validate many abstract theoretical constructs that are assessed via content
analysis. It is our hope that our process will provide a useful blueprint toward that end and
begin a dialogue that will improve construct validation in the organizational sciences when
analyses use content analysis.
Note
1. We sought guidance from a number of statistical texts, as well as relevant management research, when
considering our proposal for content validity. We were also in contact with Roderick Hart, creator of one of the
most well-known computer-aided text analysis programs—DICTION. We were intrigued by Professor Hart’s
thoughts and inquired about his process when developing DICTION. Professor Hart noted the following:
Unlike many computerized language analysis programs that depend on users to create their
own dictionaries, DICTION is a theoretically derived program built around five key dimen-
sions of language behavior. Its search dictionaries were deductively derived, not inductively
342 Organizational Research Methods
derived, with the key factor being face validity—do the categories in fact measure what they
appear to measure? In content analysis, that is ultimately the most important kind of validity
because categories that look foolish upon inspection will not be used. Thus, before deploying
DICTION, users should inspect its dictionaries carefully. For example, if the words in its Praise
dictionary (astute, beautiful, and eminent) do not seem praiseworthy, the user should find
another program. All 10,000 of DICTION’s search words can be inspected in this way, impli-
citly demanding that users give a priori thought to the phenomena they are studying. DIC-
TION’s ability to tap exactly the same conceptually derived dimensions every time it is
used is therefore its greatest limitation and its greatest strength. (Hart, 2008, personal commu-
nication, June 11, 2008)
References
Abrahamson E., & Park C. (1994). Concealment of negative organizational outcomes: An agency theory per-
spective. Academy of Management Journal, 37, 1302-1334.
Amernic J., Craig R., & Tourish D. (2007). The transformational leader as pedagogue, physician, architect, com-
mander, and saint: Five root metaphors in Jack Welch’s letters to stockholders of General Electric. Human
Relations, 60, 1839-1872.
Andrews A. O., & Welbourne T. M. (2000). The people/performance balance in IPO firms: The effect of the
chief executive officer’s financial orientation. Entrepreneurship Theory and Practice, 25, 93-106.
Atuahene-Gima K., & Ko A. (2001). An empirical investigation of the effect of market orientation and entre-
preneurship orientation alignment on product innovation. Organization Science, 12, 54-74.
Bagozzi R. P., Yi Y., & Phillips L. W. (1991). Assessing construct validity in organizational research. Admin-
istrative Science Quarterly, 36, 421-458.
Barr P. S., Stimpert J. L., & Huff A. S. (1992). Cognitive change, strategic action, and organizational renewal.
Strategic Management Journal, 13, 15-36.
Bartholomew S., & Smith A. D. (2006). Improving survey response rates from chief executive officers in small
firms: The importance of social networks. Entrepreneurship Theory and Practice, 30, 83-96.
Baruch Y. (1999). Response rate in academic studies—A comparative analysis. Human Relations, 52, 421-438.
Bettis R. A., & Hitt M. A. (1995). The new competitive landscape. Strategic Management Journal, 16, 7-19.
Bligh M. C., Kohles J. C., & Meindl J. R. (2004a). Charisma under crisis: Presidential leadership, rhetoric, and
media responses before and after the September 11th terrorist attacks. The Leadership Quarterly, 15,
211-239.
Bligh M. C., Kohles J. C., & Meindl J. R. (2004b). Charting the language of leadership: A methodological inves-
tigation of President Bush and the crisis of 9/11. Journal of Applied Psychology, 89, 562-574.
Bowman E. H. (1984). Content analysis of annual reports for corporate strategy and risk. Interfaces, 14, 61-71.
Boyd B. K., Gove S., & Hitt M. A. (2005). Construct measurement in strategic management research: Illusion or
reality? Strategic Management Journal, 26, 239-257.
Campbell D. T., & Fiske D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod
matrix. Psychological Bulletin, 56, 81-105.
Carley K. M. (1997). Extracting team mental models through textual analysis. Journal of Organizational Beha-
vior, 18, 533-558.
Chandler G. N., & Lyon D. W. (2001). Issues of research design and construct measurement in entrepreneurship
research: The past decade. Entrepreneurship Theory and Practice, 25, 101-113.
Chen M. J., & Hambrick D. C. (1995). Speed, stealth and selective attack: How small firms differ from large
firms in competitive behavior. Academy of Management Journal, 38, 453-482.
Cheng Y., Liu M. H., & Qian J. (2006). Buy-side analysts, sell-side analysts, and investment decisions of money
managers. Journal of Financial and Quantitative Analysis, 4, 51-83.
Short et al. / Construct Validation Using CATA 343
Cook T. D., & Campbell D. T. (1979). Quasi-experimentation: Design and analysis issues for field settings.
Boston: Houghton-Mifflin.
Courtis J. K. (1982). Private shareholder response to corporate annual report. Accounting and Finance, 22, 53-72.
Covin J. G., & Slevin D. P. (1988). The influence of organization structure on the utility of an entrepreneurial
top management style. Journal of Management Studies, 25, 217-234.
Covin J. G., & Slevin D. P. (1989). Strategic management of small firms in hostile and benign environments.
Strategic Management Journal, 10, 75-87.
Cronbach L. J. (1971). Test validation. In R. L. Thorndike (Ed.), Educational measurement (pp. 443-507).
Washington, DC: American Council on Education.
Cronbach L. J., & Meehl P. E. (1955). Construct validity in psychological tests. Psychological Bulletin, 52, 281-302.
Daily C. M., Certo S. T., & Dalton D. R. (2005). Investment bankers and IPO pricing: Does prospectus infor-
mation matter? Journal of Business Venturing, 20, 93-111.
D’Aveni R., & MacMillan I. (1990). Crisis and the content of managerial communications: A study of the focus
of attention of top managers in surviving and failing firms. Administrative Science Quarterly, 35, 634-657.
Deephouse D. L. (1996). Does isomorphism legitimate? Academy of Management Journal, 39, 1024-1039.
Deephouse D. L. (2000). Media reputation as a strategic resource: An integration of mass communication and
resource-based theories. Journal of Management, 26, 1091-1112.
Doucet L., & Jehn K. A. (1997). Analyzing harsh words in a sensitive setting: American expatriates in commu-
nist China. Journal of Organizational Behavior, 18, 559-582.
Dowling G. R., & Kabanoff B. (1996). Computer-aided content analysis: What do 240 advertising slogans have
in common? Marketing Letters, 7, 63-75.
Duriau V. J., Reger R. K., & Pfarrer M. D. (2007). A content analysis of the content analysis literature in orga-
nizational studies: Research themes, data sources, and methodological refinements. Organizational
Research Methods, 10, 5-34.
Dyer W. G., & Whetten D. A. (2006). Family firms and social responsibility: Preliminary evidence from the
S&P 500. Entrepreneurship Theory and Practice, 30, 785-802.
Edwards J. R. (2000). Multidimensional constructs in organizational behavior research: An integrative analyti-
cal framework. Organizational Research Methods, 4, 144-192.
Ellis L. (1994). Research methods in the social sciences. Madison, WI: WCB Brown & Benchmark.
Emrich C. G., Brower H. H., Feldman J. M., & Garland H. (2001). Images in words: Presidential rhetoric, char-
isma, and greatness. Administrative Science Quarterly, 46, 527-557.
Fendt J., & Sachs W. (2008). Grounded theory method in management research. Organizational Research Meth-
ods, 11, 430-455.
Ferrier W. J. (2001). Navigating the competitive landscape: The drivers and consequences of competitive
aggressiveness. Academy of Management Journal, 44, 858-877.
George G., Wiklund J., & Zahra S. A. (2005). Ownership and the internationalization of small firms. Journal of
Management, 31, 210-233.
Gioia D. A., & Chittipeddi K. (1991). Sensemaking and sensegiving in strategic change initiation. Strategic
Management Journal, 12, 433-448.
Godfrey P. C., & Hill C. W. L. (1995). The problem of unobservables in strategic management research. Stra-
tegic Management Journal, 16, 519-533.
Goodman R. (1980). Annual reports serving a dual marketing function—Report as survey. Public Relations
Quarterly, 36, 21-24.
Hair J. F., Anderson R. E., Tatham R. L., & Black W. C. (1998). Multivariate data analysis. Upper Saddle River,
NJ: Prentice Hall.
Hambrick D. C. (2007). The field of management’s devotion to theory: Too much of a good thing? Academy of
Management Journal, 50, 1346-1352.
Hart R. P. (2000). DICTION 5.0: The text-analysis program. Thousand Oaks, CA: Sage.
Hitt M. A., & Ireland R. D. (2005). The Blackwell encyclopedia of management: Entrepreneurship (2nd ed.).
Oxford, UK: Blackwell.
Holsti O. R. (1969). Content analysis for the social sciences and humanities. Reading, MA: Addison-Wesley.
Hughes M., & Morgan R. E. (2007). Deconstructing the relationship between entrepreneurial orientation and busi-
ness performance at the embryonic stage of firm growth. Industrial Marketing Management, 36, 651-661.
344 Organizational Research Methods
Hult G. T. M., Snow C. C., & Kandemir D. (2003). The role of entrepreneurship in building cultural competi-
tiveness in different organizational types. Journal of Management, 29, 401-426.
Ireland R. D., Reutzel C. R., & Webb J. W. (2005). Entrepreneurship research in AMJ: What has been published,
and what might the future hold? Academy of Management Journal, 48, 556-564.
Ireland R. D., Webb J. W., & Coombs J. E. (2005). Theory and methodology in entrepreneurship research.
Research Methodology in Strategy and Management, 2, 111-141.
Jauch L. R., Osborn R. N., & Martin T. N. (1980). Structured content analysis of cases: Complementary method
for organizational research. Academy of Management Review, 5, 517-526.
Kabanoff B. Waldersee, R., & Cohen M. (1995). Espoused values and organizational change themes. Academy
of Management Journal, 38, 1075-1104.
Karren R. J., & Barringer M. W. (2002). A review and analysis of the policy-capturing methodology in orga-
nizational research: Guidelines for research and practice. Organizational Research Methods, 5, 337-361.
Kazanjian R. K. (1988). Relation of dominant problems to stages of growth in technology-based new ventures.
Academy of Management Review, 31, 257-279.
Kerlinger F. N., & Lee H. B. (2000). Foundations of behavioral research (4th ed.). Orlando, FL: Thomson
Learning.
Konrad A. M., & Mangel R. (2000). The impact of work-life programs on firm productivity. Strategic Manage-
ment Journal, 21, 1225-1237.
Kreiser P. M., Marino L. D., & Weaver K. M. (2002). Assessing the psychometric properties of the entrepre-
neurial orientation scale: A multi-country analysis. Entrepreneurship Theory and Practice, 26, 71-94.
Krippendorff K. (2004). Content analysis: An introduction to its methodology (2nd ed.). Thousand Oaks, CA:
Sage.
Lance C. E., Butts M. M., & Michels L. C. (2006). The sources of four commonly reported cutoff criteria: What
did they really say? Organizational Research Methods, 9, 202-220.
Lumpkin G. T., & Dess G. G. (1996). Clarifying the entrepreneurial orientation construct and linking it to per-
formance. Academy of Management Review, 21, 135-172.
Lumpkin G. T., & Dess G. G. (2001). Linking two dimensions of entrepreneurial orientation to firm perfor-
mance: The moderating role of environment and industry life cycle. Journal of Business Venturing, 16,
429-451.
Lumpkin G. T., & Dess G. G. (2005). Entrepreneurial orientation. In M. A. Hitt & R. D. Ireland (Eds.), The
Blackwell encyclopedia of management: Entrepreneurship (pp. 104-107). Oxford, UK: Blackwell.
Lyon D. W., Lumpkin G. T., & Dess G. G. (2000). Enhancing entrepreneurial orientation research: Operatio-
nalizing and measuring a key strategic decision making process. Journal of Management, 26, 1055-1085.
Maitlis S., & Lawrence T. B. (2007). Triggers and enablers of sensegiving in organizations. Academy of Man-
agement Journal, 50, 57-84.
Marino K. E., Castaldi R. M., & Dollinger M. J. (1989). Content analysis in entrepreneurship research: The case
of initial public offerings. Entrepreneurship Theory and Practice, 14, 51-66.
Markel N. (1998). A semiotic contribution to libel action law. International Journal for the Semiotics of Law,
11, 49-56.
McAdams D. P., & Zeldow P. B. (1993). Construct validity and content analysis. Journal of Personality Assess-
ment, 61, 243-245.
McGahan A. M., & Porter M. E. (1997). How much does industry matter, really? Strategic Management Jour-
nal, 18, 15-30.
Mehran H. (1995). Executive compensation structure, ownership, and firm performance. Journal of Financial
Economics, 38, 163-184.
Meyer A. D. (1991). What is strategy’ s distinctive competence? Journal of Management, 17, 821-833.
Michalisin M. D. (2001). Validity of annual report assertions about innovativeness: An empirical investigation.
Journal of Business Research, 53, 151-161.
Miller D. (1983). The correlates of entrepreneurship in three types of firms. Management Science, 29, 770-791.
Morgan R. E., & Strong C. A. (2003). Business performance and dimensions of strategic orientation. Journal of
Business Research, 56, 163-176.
Morris R. (1994). Computerized content analysis in management research: A demonstration of advantages &
limitations. Journal of Management, 20, 903-931.
Short et al. / Construct Validation Using CATA 345
Mossholder K. W., Settoon R. P., Harris S. G., & Armenakis A. A. (1995). Measuring emotion in open-ended
survey responses: An application of textual data analysis. Journal of Management, 21, 335-355.
Naldi L., Nordquist M., Sjoberg K., & Wiklund J. (2007). Entrepreneurial orientation, risk taking, and perfor-
mance in family firms. Family Business Review, 20, 33-58.
Neuendorf K. A. (2002). The content analysis guidebook. Thousand Oaks, CA: Sage.
Nunnally J. C., & Bernstein I. H. (1994). Psychometric theory (3rd ed.). New York: McGraw-Hill.
Palmer T. B., & Short J. C. (2008). Mission statements in U.S. colleges of business: An empirical examination of
their content with linkages to configurations and performance. Academy of Management Learning and Edu-
cation, 7, 454-470.
Pearce J. A., & David F. (1987). Corporate mission statements: The bottom line. Academy of Management Exec-
utive, 1, 109-116.
Pennebaker J. W., Mehl M. R., & Niederhoffer K. G. (2003). Psychological aspects of natural language use: Our
words, our selves. Annual Review of Psychology, 54, 547-577.
Porac J. F., Wade J. B., & Pollock T. G. (1999). Industry categories and the politics of the comparable firm in
CEO compensation. Administrative Science Quarterly, 44, 112-144.
Potter W. J., & Levine-Donnerstein D. (1999). Rethinking validity and reliability in content analysis. Journal of
Applied Communication Research, 27, 258-284.
Rauch A., Wiklund J., Frese M., & Lumpkin G. T. (2004). Entrepreneurial orientation and business perfor-
mance: Cumulative empirical evidence. In S. A. Zahra, C. G. Brush, P. Davidsson, J. Fiet, P. G. Greene,
R. T. Harrison, M. Lerner, C. Mason, G. D. Meyer, J. Sohl & A. Zacharakis (Eds.), Frontiers of Entrepre-
neurship Research (pp. 164-177). Wellesley, MA: Babson.
Reilly F. K., & Wright D. J. (2002). Alternative small-cap stock benchmarks. Journal of Portfolio Management,
28, 82-95.
Riffe D., & Lacy S. & Fico, F. G. (2005). Analyzing media messages: Using quantitative content analysis in
research (2nd ed.). Mahwah, NJ: Erlbaum.
Rodale J. I. (1978). The synonym finder. Emmaus, PA: Rodale Press.
Rogelberg S. G., & Stanton J. M. (2007). Understanding and dealing with organizational survey nonresponse.
Organizational Research Methods, 10, 195-209.
Rosenberg S. D., Schnurr P. P., & Oxman T. E. (1990). Content analysis: A comparison of manual and com-
puterized systems. Journal of Personality Assessment, 54, 298-310.
Rumelt R. (1991). How much does industry matter? Strategic Management Journal, 13, 119-134.
Russell Investments. (2008a). Russell 30001 Index Fact Sheet. Retrieved March 23, 2009, from http://www.rus-
sell.com/indexes/characteristics_fact_sheets/us/Russell_3000_Index.asp
Russell Investments. (2008b). Russell 20001 Index Fact Sheet. Retrieved May 20, 2008, from http://www.rus-
sell.com/indexes/characteristics_fact_sheets/US/Russell_2000_Index.asp
Sarkar M. B., Echambadi R. A., & Harrison J. S. (2001). Alliance entrepreneurship and firm market perfor-
mance. Strategic Management Journal, 22, 701-711.
Savall H., Zardet V., Bonnet M., & Peron M. (2008). The emergence of implicit criteria actually used by
reviewers of qualitative research articles: Case of a European journal. Organizational Research Methods,
11, 510-540.
Scandura T. A., & Williams E. A. (2000). Research methodology in management: Current practices, trends, and
implications for future research. Academy of Management Journal, 43, 1248-1264.
Schnurr P. P., Rosenberg S. D., Oxman T. E., & Tucker G. J. (1986). A methodological note on content analysis:
Estimates of reliability. Journal of Personality Assessment, 50, 601-609.
Shane S. (2004). Encouraging university entrepreneurship? The effect of the Bayh-Dole Act on university
patenting in the United States. Journal of Business Venturing, 19, 127-151.
Short J. C., Ketchen D. J., & Palmer T. B. (2002). The role of sampling in strategic management research on
performance: A two-study analysis. Journal of Management, 28, 363-385.
Short J. C., Ketchen D. J., Palmer T. B., & Hult T. M. (2007). Firm, strategic group, and industry influences on
performance. Strategic Management Journal, 28, 147-167.
Short J. C., & Palmer T. B. (2003). Organizational performance referents: An empirical examination of their
content and influences. Organizational Behavior and Human Decision Processes, 90, 209-224.
346 Organizational Research Methods
Short J. C., & Palmer T. B. (2008). The application of DICTION to content analysis research in strategic man-
agement. Organizational Research Methods, 11, 727-752.
Singh K., Ang S. H., & Leong S. M. (2003). Increasing replication for knowledge accumulation in strategy
research. Journal of Management, 29, 533-549.
Skalski, P. D. (2002). Computer content analysis software. In K. A. Neuendorf (author), The content analysis
guidebook. Thousand Oaks, CA: Sage.
Slater S. F., & Narver J. C. (2000). The positive effect of a market orientation on business profitability: A
balanced replication. Journal of Business Research, 48, 69-73.
Smart D. T., & Conant J. S. (1994). Entrepreneurial orientation, distinctive marketing competencies and orga-
nizational performance. Journal of Applied Business Research, 10, 28-38.
Sonpar K., & Golden-Biddle K. (2008). Using content analysis to elaborate adolescent theories of organization.
Organizational Research Methods, 11, 795-814.
Stam W., & Elfring T. (2008). Entrepreneurial orientation and new venture performance: The moderating role of
intra- and extraindustry social capital. Academy of Management Journal, 51, 97-111.
Stevenson R. L. (2001). In praise of dumb clerks: Computer-assisted content analysis. In M. D. West (Ed.), The-
ory, method and practice in computer content analysis (pp. 3-12). Portsmouth, NH: Greenwood.
Tang J., Tang Z., Marino L. D., Zhang Y., & Li Q. (2008). Exploring an inverted u-shape relationship between
entrepreneurial orientation and performance in Chinese ventures. Entrepreneurship Theory and Practice, 30,
219-239.
Tsang E. W. K., & Kwan K. (1999). Replication and theory development in organizational science: A critical
realist perspective. Academy of Management Review, 24, 759-780.
Van De Ven A. H., & Johnson P. E. (2006). Knowledge for theory and practice. Academy of Management
Review, 31, 802-821.
Van Maanen J. (1979). Reclaiming qualitative methods for organizational research: A preface. Administrative
Science Quarterly, 24, 520-526.
Wade J. B., Porac J. F., & Pollock T. G. (1997). Worth, words, and the justification of executive pay. Journal of
Organizational Behavior, 18, 641-664.
Walter A., Auer M., & Ritter T. (2006). The impact of network capabilities and entrepreneurial orientation on
university spin-off performance. Journal of Business Venturing, 21, 541-567.
Weber R. P. (1990). Basic content analysis (2nd ed.). Thousand Oaks, CA: Sage.
Zahra S. A. (1991). Predictors and financial outcomes of corporate entrepreneurship: An exploratory study.
Journal of Business Venturing, 6, 259-285.
Zahra S. A. (1996). Governance, ownership, and corporate entrepreneurship: The moderating impact of industry
technological opportunities. Academy of Management Journal, 39, 1713-1735.
Zahra S. A., & Covin J. G. (1995). Contextual influences on the corporate entrepreneurship performance rela-
tionship: A longitudinal analysis. Journal of Business Venturing, 10, 43-58.
Short et al. / Construct Validation Using CATA 347