Corporate Reputation Review Volume 18 Number 1

Stakeholder Tracking and Analysis: The

RepTrak® System for Measuring Corporate
Reputation

Charles J. Fombrun
Reputation Institute, New York, NY, USA

Leonard J. Ponzi
Reputation Institute, New York, NY, USA

William Newburry
Florida International University, Miami, Fl, USA

ABSTRACT reports tests conducted to validate the measures

‘Reputation’ is increasingly recognized for its empirically across five stakeholder groups in six
influence in creating stakeholder support and countries. Multivariate analyses confirm the exis-
engagement with companies. Both researchers and tence and stability of the seven underlying dimen-
practitioners would therefore benefit from having a sions in the factor structure, each of which is
rigorous instrument to measure reputations and constructed from a set of 3–4 underlying attributes.
the ability to develop predictive modeling of repu- The modeling therefore validates the RepTrak®
tation’s impact on stakeholder outcomes. The scorecard as a tool for measuring and tracking multi-
RepTrak® System evolved from studies conducted stakeholder perceptions of companies.
by Reputation Institute since 2000 to provide a Corporate Reputation Review (2015) 18, 3–24.
systematic tool for tracking and analyzing stake- doi:10.1057/crr.2014.21
holder perceptions that could help companies better
manage their reputation and its effects on stake- KEYWORDS: corporate reputation; cross-cultural;
holder behaviors. Prior research has demonstrated reputation measurement; scale development; stakeholder
the validity of the RepTrak® Pulse as a short management
form measure of ‘corporate reputation’. This
study reports empirical tests developed to validate
the seven dimensions that the RepTrak® System
uses to predict corporate reputation and stake- INTRODUCTION
holder support. Although these seven dimensions ‘Reputation’ is increasingly recognized for
have been verified internally by Reputation its influence on stakeholder support and
Institute, this methodology and its validation engagement with companies (Fombrun,
have not been reported publicly, limiting 1996, 2012). Both researchers and practi-
researchers’ and practitioners’ abilities to use the tioners would therefore benefit from having Corporate Reputation Review,
seven dimensions in their respective efforts. To a rigorous instrument to measure reputations Vol. 18, No. 1, pp. 3–24
© 2015 Macmillan Publishers Ltd.,
shed light on the model’s structure, this paper and the ability to develop predictive models 1363-3589

www.palgrave-journals.com/crr/
 Stakeholder Tracking Using RepTrak®

of reputation’s impact on stakeholder out- The RepTrak® System recognizes the fact
comes. Recognizing a growing need by both that a company’s overall reputation is rooted
practitioners and academics for a better con- in the perceptions of its stakeholders
ceptual and empirical tool for assessing and (Newburry, 2010), each of which responds
managing reputation – and the lack of vali- to different signals or informational inputs
dated instruments for doing so – Reputation (Spence, 1973; Prabhu and Stewart, 2001;
Institute launched a global project in 1998 to Basdeo et al., 2006). By examining the kinds
understand and measure the diverse factors of informational inputs that influence stake-
associated with corporate reputation. The first holder perceptions of a company, we can
measurement instrument that resulted from better predict the dimensions that are likely
our initial exploration was the Reputation to trigger stakeholders’ emotional reactions
Quotient (RQ), a six-dimension scale con- of admiration, liking and trust toward a firm
structed from 20 attributes (Fombrun et al., – its reputation.
2000). The four-attribute RepTrak® Pulse Stakeholder management is an important
measure was pulled out of the RQ in 2005 component of corporate strategy in general
and used to create a separate measure of a (eg, Freeman, 1984; Donaldson and Preston,
person’s emotional attachment to a company 1995) and the study of corporate reputation
(eg, Christian, 1959). Ponzi et al. (2011) in particular (Fombrun, 1996; Fombrun,
demonstrated the reliability and validity of 2012). Social-constructionists view reputa-
the RepTrak® Pulse scale as a measure of tion as a composite of different types of sta-
reputation, and since 2005 it has been exten- keholder perceptions of a firm (Rindova
sively tested and shown to have high face and and Martins, 2012). They note that percep-
content validity (Sarstedt et al., 2013). tions of firms come from many sources,
The full RepTrak® System was created in many of which would not be considered
2005–2006 to provide executives with an ‘valid signals’ by traditional economic the-
analytical instrument that could be used, ory (Rindova and Martins, 2012). Signaling
not only to track and assess stakeholder theory (Spence, 1973) relies on information
perceptions of companies, but that would economics to discuss the behavior of inter-
also enable a more comprehensive under- acting actors under conditions of informa-
standing of the underlying informational tion asymmetry and uncertainty. In a
drivers of reputation that elicit emotional marketplace, sellers send signals to buyers
attachment. The system is based on measur- through strategic actions such as prices,
ing a company’s overall reputation using the warrantees or return policies to demonstrate
RepTrak® Pulse and decomposing that the quality of their products and other firm
emotional attachment into an underlying set competencies (Basdeo et al., 2006). More
of dimensions and attributes, and predicting broadly, companies send signals to their
their effects on stakeholder support. various stakeholder groups in order to
The rigorous methodological under- influence how they are perceived (Prabhu
pinnings of the model and its validation have and Stewart, 2001; van Riel, 2012). How-
not been reported publicly to date, limiting ever, these stakeholders also receive signals
the ability researchers and practitioners have from other sources, such as formal media
to use the dimensions in their work. The (eg, newspapers, TV; Van Den Bogaerd and
purpose of this paper is to report empirical Aerts, 2014; Mason, 2014), social media
tests done to validate the seven dimensions of (eg, blogs; Fan et al., 2013), friends and
the RepTrak® System and predict corporate industry competitors, all of which influence
reputation and stakeholder support across the perceptions individuals have of firms,
five stakeholder groups in six countries. and in turn, their reputations.

4 Corporate Reputation Review Vol. 18, 1, 3–24 © 2015 Macmillan Publishers Ltd. 1363-3589
 Fombrun, Ponzi and Newburry

The strategic objective of corporate com- Second, we empirically demonstrate the

munication is to align stakeholders with the soundness of the model as an instrument that
goals of the organization (van Riel, 2012) in can be applied across stakeholders, industries
order to ensure that stakeholders develop a and countries. Third, we suggest that because
sense of trust in a company (Srivastava and the RepTrak® System is a tool that was rig-
Chakravarti, 2009; Van Der Merwe and orously developed and validated, it has prac-
Puth, 2014). However, stakeholders respond tical relevance for tracking and analyzing the
to different signals, making it critical for reputations of companies globally.
practitioners to understand and communicate
across the multiple facets of companies to
which stakeholders are exposed, hence the REPUTATION AND ITS DIMENSIONS
need for a multidimensional framework for Prior literature suggests that a distinction can
measuring reputation internationally. be made between stakeholder assessments
In the next section, we examine seven based on a generalized view of reputation and
types of information signals that past literature assessments based on specific dimensions.
suggests as influencers on the formation of Lange et al. (2011) conducted a literature
corporate reputation, and which make up the review from which they identified three
dimensions of the RepTrak® System. Since major reputation conceptualizations: being
reputation assessments have been shown to known, being known for something and
vary across stakeholder groups (van Riel and generalized favorability. ‘Being known’ and
Fombrun, 2007; van Riel, 2012) and across ‘generalized favorability’ are broad percep-
national and other environmental contexts tions of a company. By contrast, ‘being
(eg, Michaelis et al., 2008; Deephouse et al., known for something’ suggests a dimensional
2009; Brammer and Jackson, 2012), we assess basis for reputation measurement. Consistent
the reliability and validity of the seven with this interpretation, Lange et al. (2011:
dimensions by examining the stability of the 166–167) noted that the critical distinction
model when tested across five stakeholder between ‘being known for something’ and
groups in multiple industries, in six countries, the ‘generalized favorability’ dimension ‘is that
specifically: key opinion leaders (in Brazil), the the latter reflects the perceiver’s approach –
general public (in the United States), investors avoidance reactions to the generalized global
(in Spain), physicians (in Switzerland) and perceptions of the firm, while the former
customers of the insurance industry (in reflects perceiver expectations for particular
Denmark and Sweden). While it would be desired or undesired organizational attributes
difficult and costly to test the universal validity or outcomes’. The distinction between a
of the model across all stakeholder groups and company’s overall reputation and specific
research settings, by demonstrating the validity reputation dimensions is also analogous to the
of the RepTrak® System across a diverse set of distinction between a company’s overall brand
respondents, industries and countries, we sug- and its product brands (eg, Smith et al., 2010),
gest that the system has the potential for and between general and specific brand ima-
generalizability. ges (Sonnier and Ainslie, 2011).
Overall, the research we report in this Building on the international qualitative
paper makes the following contributions. work already completed by Reputation
First, we validate the seven-dimensional Institute to develop the RQ instrument
RepTrak® framework derived from prior (Gardberg and Fombrun, 2002; Gardberg,
literature that addresses the need to manage 2006), Reputation Institute extended the
stakeholders in general and on each of these research by conducting a wide range of
seven reputation dimensions in particular. interviews with reputation managers, senior

© 2015 Macmillan Publishers Ltd. 1363-3589 Vol. 18, 1, 3–24 Corporate Reputation Review 5
 Stakeholder Tracking Using RepTrak®

communications managers and functional therefore assesses perceptions of a company’s

heads (eg, human resources, marketing and offerings based on whether they are thought
finance) of global companies headquartered to be high in quality, in value and service, and
around the world. These executive inter- in their ability to meet customers’ needs.
views were supplemented by consumer focus
groups, a number of which were run in the
United States, Europe and Asia, as well as in Innovation
nine countries of Latin America (Carreras As an important firm asset (Fang et al., 2011),
et al., 2013). This broad range of stakeholder innovation inherently relates to doing some-
interviews and focus groups conducted thing new or differently, and so readily gen-
between 1999 and 2006 were used to iden- erates an emotional reaction of respect and
tify the seven dimension structure of corpo- admiration for the innovator, and therefore
rate reputation. We review these dimensions reputation. Research confirms that there is a
below in terms of their conceptual roots. We relationship between innovation and reputa-
then empirically validate the RepTrak® tion, and recognizes that positive regard is
model for analyzing corporate reputation often dependent upon effective commu-
based on these seven dimensions. nication about an innovation (Courtright
and Smudde, 2009). Companies that adapt
quickly to change, launch new products and
Products/Services develop new ideas are more likely to earn
Most stakeholders know of a company from respect and admiration – and many publica-
its product and service offerings in the mar- tions such as Forbes, Bloomberg and Business
ketplace, and its reputation is likely to be Week compile and publish rankings of inno-
influenced by perceptions of its product vative companies, thereby conveying infor-
brands (Rao et al., 1999; Smith et al., 2010). mation that adds visibility to innovators. These
Some are more familiar, others less so, publications signal to all observers about a
depending on the particular characteristics of company’s innovativeness – thereby adding to
the company’s touch points with specific sta- their reputation. RepTrak®’s ‘innovation’
keholders. All stakeholders, and customers dimension assesses perceptions of a company as
more so than others, can be expected to innovative and adaptive.
develop perceptions of a company based on its
products and services – the quality of its
offering, the price at which it sells, its per- Workplace
ceived value, the customer support provided Our qualitative research suggests that most
and the belief in the company’s willingness to stakeholders like and respect companies that
stand behind its products and services (Dawar maintain good workplaces. Research asserts
and Parker, 1994; Lange et al., 2011). Signals that satisfied employees are more likely to
from the marketplace can also color the commit to long-term involvement, less likely
impressions that non-customer stakeholders to turn over and so more likely to act as
have of any company – and so the degree to ambassadors of the company and give a good
which those stakeholders will experience the employer a favorable rating. In turn, a firm’s
company as admirable, likeable, trustworthy reputation as a good workplace is critical in
and well regarded. Game theory models posit recruiting a high-quality workforce (Alniacik
that reputations, in fact, are built mainly from et al., 2012; Nolan et al., 2013). Various
investments companies make to increase pro- publications highlight how companies treat
duct quality (Milgrom and Roberts, 1986). their employees. Fortune regularly releases
RepTrak®’s ‘products/services’ dimension ‘The 100 Best Companies to Work For’ and

6 Corporate Reputation Review Vol. 18, 1, 3–24 © 2015 Macmillan Publishers Ltd. 1363-3589
 Fombrun, Ponzi and Newburry

Forbes publishes a list of ‘The 25 Best Places agencies. The more a company is perceived as
to Work’. Both broadcast workplace signals ethical and transparent, the more likely it is to
into the reputational marketplace that add generate admiration and trust in the minds of
visibility to employers who treat their most stakeholders – and hence to build repu-
employees well. Special interest stakeholders tation because ‘… the corporation tends to be
are regularly influenced by segment-specific viewed less as property and more as a public
publications publishing lists and information entity with a broad range of responsibilities to
about the employment practices of compa- creditors, workers, the public, and others’
nies, such as Working Mother Magazine’s ‘100 (Soleimani et al., 2014: 4). Companies them-
Best Companies for Working Moms’ and the selves often become signatories to institutional
Human Rights Campaign’s ‘Corporate codes of conduct to signal to stakeholders
Equality Index’ that assesses the diversity their principles and commitments, generate
policies of employers. Newburry et al. (2014) confidence that their internal practices are
used the workplace dimension of reputation sound – and thereby build reputation.
to examine issues related to foreignness and RepTrak®’s ‘governance’ dimension assesses
internationalization as they impact the stakeholder perceptions of a company as
attractiveness of employers in Latin America. ethical, fair and transparent.
Similarly, Martin et al. (2011) examined the
relationship between employer branding and
reputation. Signals that convey information Citizenship
about how fairly a company treats employees Qualitative inputs suggest that stakeholders
are likely to generate trust and respect among tend to respect and admire a company for
most stakeholders – and so contribute to their good deeds (eg, Orlitzky and Swanson,
building favorable reputations for those 2012). Moreover, empirical evidence suggests
companies. RepTrak®’s ‘workplace’ dimen- that corporate citizenship is a legitimacy
sion assesses perceptions of a company’s building strategic asset (Sridhar, 2012) that
practices in maintaining an environment that leads to various forms of company support
shows concern for employees, and for treat- (Aaron et al., 2012), and can even provide a
ing and rewarding them fairly and equitably. buffer that protects firms in times of crisis (Mio
and Fasan, 2012). As such, companies com-
monly expend significant dollar amounts and
Governance marketing efforts to promote these activities
Davis (2005: 143) defined corporate govern- and thereby build up a company’s image
ance as the ‘structures, processes and institu- (Gottschalk, 2013; Morris et al., 2013; Vlachos
tions within and around organizations that et al., 2013). Empirically, corporate social
allocate power and resource control among performance has been one of the main corre-
participants’. Given the growing complexities lates of corporate reputation (Lange et al.,
of multinational firms, governance is increas- 2011). By acting responsibly and commu-
ingly recognized as a key issue for firms nicating about it, companies signal that they are
(eg, Kim et al., 2011; Ghosh and John, 2009). good citizens, deserving of praise, and thereby
Having adequate governance structures in build trust and reputation. Good citizenship is
place to manage corporate reputation is itself a multidimensional construct, one com-
recognized as a key component of reputation monly understood to encompass notions of
management (Casado et al., 2014). Stake- environmental sustainability and responsible
holders are regularly exposed to information behavior (Tichy et al., 1997). Past research
about a company’s governance whether from suggests that corporate citizenship can cut
media, from auditors or from government both ways (eg, Koschate-Fischer et al., 2012).

© 2015 Macmillan Publishers Ltd. 1363-3589 Vol. 18, 1, 3–24 Corporate Reputation Review 7
 Stakeholder Tracking Using RepTrak®

Some stakeholders credit companies for act- Walsh et al., 2003). Past and current profit-
ing like good citizens and view it as a form of ability are important signals to investors about
relationship marketing, while others see it as the company’s operating success. It also signals
a distraction and an unnecessary drain on the likelihood of continuing profitability –
corporate resources. RepTrak®’s ‘citizenship’ indicating a company with strong future pro-
dimension assesses stakeholder perceptions of spects for growth. Expectations of future
a company as environmentally friendly, a profitability are important to all valuation
supporter of good causes and a positive con- models – and therefore are a powerful signal
tributor to society. about the strength of a company’s business
model. Profitability and growth prospects
have been shown to influence ratings of the
Leadership ‘world’s most-admired’ companies (Fombrun
CEOs can be important catalysts for generat- and Shanley, 1990), and have been consistent
ing admiration and trust with stakeholders correlates of reputation in other academic
(Flatt et al., 2013; Halff, 2013). Studies of studies (Lange et al., 2011). RepTrak®’s ‘per-
celebrity CEOs (Treadway et al., 2009), star formance’ dimension is therefore based on a
CEOs (Wade et al., 2008) and CEO Brands set of attributes that assess stakeholder percep-
(Bendisch et al., 2013) defend the importance tions of a company’s overall financial perfor-
of leadership in conveying a company’s suc- mance, profitability and growth prospects.
cess and performance to the financial com- Table 1 describes the seven-dimension
munity and other stakeholders. Moreover, model and 23 attributes of the RepTrak®
research confirms that managers do differ in scorecard. In the next section, we examine
their strategic abilities (Goldfarb and Yang, the validity of this model across a varied set of
2009). Appealing leaders attract favorable stakeholders, industries and countries.
media coverage and investor endorsements,
thereby signaling to all stakeholders the cred-
ibility of the company’s activities, increasing METHODOLOGICAL APPROACH
confidence and trust in the company, and To explore the cross-stakeholder validity of
thereby building corporate reputation. Pub- the RepTrak® System, we selected five
lished rankings of CEOs induce favorable studies conducted in 2010–2011. Each
perceptions of a company’s leaders, and can study was selected from past studies con-
build an appealing halo for the company ducted by Reputation Institute that used a
itself (Gaines-Ross, 2002; Westphal and common survey instrument with sufficient
Deephouse, 2011). RepTrak®’s ‘leadership’ sample size across different stakeholder
dimension is intended to assess perceptions of groups and countries. Although no finite
leaders as excellent and visionary managers, number of samples could demonstrate uni-
and strong endorsers of their companies. versal validity of the dimensions across all
stakeholders and countries, by selecting a
diverse set of stakeholders and countries, we
Performance seek here to demonstrate that the seven-
A common signal that influences how stake- dimension structure has reasonable validity
holders assess companies is ‘financial perfor- across a broad range of respondents and
mance’. Although stakeholders place different geographical settings. In each study, struc-
expectations on organizations, strong financial tural equation modeling (SEM) was then
performance is in part the consequence of applied to examine the reliability and
satisfying these diverse objectives (Donaldson validity of the multidimensional frame-
and Preston, 1995; Waddock and Graves, 1997; work. SEM has several advantages over

8 Corporate Reputation Review Vol. 18, 1, 3–24 © 2015 Macmillan Publishers Ltd. 1363-3589
 Fombrun, Ponzi and Newburry

Table 1: The RepTrak® System: Dimensions and Attributes of Reputation

Dimension Attribute

Products & Services Offers high quality products and services

Products & Services Offers products and services that are a good value for the money
Products & Services Stands behind its products and services
Products & Services Meets customer needs
Innovation Is an innovative company
Innovation Is generally the first company to go to market
with new products and services
Innovation Adapts quickly to change
Workplace Rewards its employees fairly
Workplace Demonstrates concern for the health and well-being of its employees
Workplace Offers equal opportunities in the workplace
Governance Is open and transparent about the way the company operates
Governance Behaves ethically
Governance Is fair in the way it does business
Citizenship Acts responsibly to protect the environment
Citizenship Supports good causes
Citizenship Has a positive influence on society
Leadership Has a strong and appealing leader
Leadership Has a clear vision for its future
Leadership Is a well-organized company
Leadership Has excellent managers
Performance Is a profitable company
Performance Delivers financial results that are better than expected
Performance Shows strong prospects for future growth

more traditional techniques, particularly beyond the scope of this paper. However,
when the model to be evaluated is not our data-led methodological approach poin-
directly observable. ted us to a ‘reflective’ specified model, which
The hypothesized model we are testing refers to the case that observable variables (or
postulates that corporation reputation can be attributes) are ‘reflective’ or representative of
measured using 23 observed variables that a defined reputation construct – in our case,
load into a first-order structure that consists the RepTrak® Pulse. As reflective, the over-
of seven latent variables labeled as Products/ all model should be unidimensional and the
Services, Innovation, Workplace, Govern- items correlated. The model is graphically
ance, Citizenship, Leadership and Perfor- illustrated in Figure 1.
mance. These seven dimensions describe a The next section describes the sample and
single second-order factor structure measur- data cleaning procedures we applied. We
ing ‘Reputation’. then present the results of the first and
A review of the theoretical SEM-related second-order confirmatory factor analyses
literature on corporate reputation has conducted using AMOS v22.0 software.
revealed a number of different constructed A maximum likelihood estimation model
measures with a mix of formative and reflec- was adopted because it provided the most
tive models. A discussion of this literature is stable results for our sample sizes.

© 2015 Macmillan Publishers Ltd. 1363-3589 Vol. 18, 1, 3–24 Corporate Reputation Review 9
 Stakeholder Tracking Using RepTrak®

e1 High Quality e24

e2 Good Value

Products
e3 Stands Behind

e4 Meets Customer Needs

λ1
e25
e5 Innovative

e6 First to Market
Innovation
e7 Adapts to Change

e26 λ2
e8 Rewards Employees Fairly

e9 Concern for Employees’ Well -Being

Workplace
e10 Equal Opportunities in Workplace
λ3
e27
e11 Open and Transparent
λ4 Corporate
e12 Behaves Ethically
Governance Reputation
e13 Fair in Doing Business

e28
e14 Protects Environment λ5

e15 Supports Good Causes

Citizenship
e16 Positive Influence on Society λ6

e17 Strong and Appealing Leader e29

e18 Clear Vision of Future

Leadership
e19 Well Organized

e20 Excellent Managers λ7

e30

e21 Profitable

e22 Good Financial Results Performance

e23 Strong Growth Prospects

Figure 1: Hypothetical model of corporate reputation

SAMPLE AND DATA ● Study 2: A sample of key opinion leaders

Five stakeholder data sets were selected for in Brazil.
their ability to represent a cross-section of ● Study 3: A sample of investors and custo-
commonly measured stakeholders, industries mers assessing banks in Spain.
and countries (see Table 2): ● Study 4: A sample of doctors assessing a
pharmaceutical company in Switzerland.
● Study 1: A representative sample of the ● Study 5: A sample of customers in Denmark
general public in the United States. and Sweden assessing insurance companies.

10 Corporate Reputation Review Vol. 18, 1, 3–24 © 2015 Macmillan Publishers Ltd. 1363-3589
 Fombrun, Ponzi and Newburry

Table 2: Summary of Sample and Data Collection Methodologies

Stakeholder groups Study 1 Study 2 Study 3 Study 4 Study 5

General public Key Bank Physicians Insurance buyers

opinion customers
leaders & investors

Markets The United Brazil Spain Switzerland Denmark and

States Sweden
Year of study 2011 2010 2007 2007 2011
DC method Online CATI CATI Online Online
Company list 150 10 10 10 9
Sample source Panel Panel Client List Panel Panel
Initial sample size 4,652 1,291 2,328 585 3,300
Sample size after listwise 1,835 564 580 585a 942
deletion
Sample size after outliers 1,813 538 532 557 942
Final sample size 1,813 538 532 300b 500b
a
Includes EM imputed values
b
Random sample

Respondents in the five stakeholder samples when underlying normality assumptions do

were randomly selected from a larger sample not hold or missing data are not completely
set in each country, except for Study 3 in random (Little, 1992). With a minimum of
which a study sponsor provided a customer 20 records per variable, the five data sets were
list. Respondents were screened for their sufficiently large to achieve statistical power
familiarity with each of the companies and (McQuitty, 2004; Schreiber et al., 2006).
then asked to assess familiar companies on
each of the 23 attributes of the RepTrak®
scorecard, scored on 7-point Likert scales Treatment of Outliers
ranging from ‘Does not describe well’ to To legitimate multivariate analyses, all vari-
‘Describes very well’. Respondents also had ables in the model were presumed to follow
the option to reply ‘not sure’ to questions normality assumptions. Each data set was
posed. therefore examined for violations of normal-
ity. Outliers often contribute significantly to
departures from normality and to distortions
Treatment of Missing Values of the covariance matrix. Deleting outliers
Table 2 also shows the sample sizes and lowers multivariate skew and kurtosis. Kline
missing values in the initial data sets. Listwise (2005) recommends removing outliers if they
deletion was applied to records with missing reduce absolute values of skew to less than
values in Studies 1, 2, 3 and 5. In Study 4, to 3.0 and kurtosis to less than 10. These viola-
deal with the smaller sample size, missing tions of multivariate normality can also be
values were replaced with imputed values evaluated in one step by examining Mardia’s
using EM (Dempster et al., 1977), an algo- multivariate kurtosis. The analysis identified
rithm that produces acceptable results even outliers by calculating Mahalanobis distances

© 2015 Macmillan Publishers Ltd. 1363-3589 Vol. 18, 1, 3–24 Corporate Reputation Review 11
 Stakeholder Tracking Using RepTrak®

Table 3: Assessment of Multivariate Normality

Study All cases Final sample after removal of outliers

Skew Critical ratio of Kurtosis Critical ratio of Skew Critical ratio of Kurtosis Critical ratio of
skew kurtosis skew kurtosis

1 −0.77 −21.43 0.55 7.69 −0.73 −12.73 0.32 2.79

2 −1.58 −23.22 2.76 20.24 −1.63 −15.45 3.19 15.11
3 −1.00 −19.61 1.10 10.79 −0.89 −8.34 1.00 4.70
4 −1.00 −19.61 1.10 10.79 −0.89 −8.34 1.00 4.70
5 −0.11 −2.64 0.05 0.64 −0.16 −1.49 0.08 0.38
Overall improvement 15% 46% 26% 48%

for each record in the data set. The larger the data sets using exploratory factor analysis.
the distance, the more likely the record Second, we carry out a first-order con-
was to violate multivariate normality. We firmatory factor analysis on each of the
deleted observations with distance greater studies to test for the validity of the hypo-
than 100. Table 3 depicts the results of thesized seven-dimension factor structure of
the assessment of the multivariate normality corporate reputation. Third, we conduct a
of each stakeholder sample and overall second-order confirmatory factor analysis to
improvement from removing outliers. Since justify the dimensional hierarchy of corporate
any remaining non-normality could lead to reputation.
overestimation of χ2 fit statistics, potentially
leading to false rejection of any model tes-
ted, we also report additional fit statistics to Establishing Reliability
assess each of the multivariate models tested An exploratory factor analysis using princi-
using SEM. pal components was run on the 23 attri-
The final samples analyzed in this paper butes. The researchers selected an equamax
consisted of 1,813 random respondents non-orthogonal rotation because of its
drawn from the US general public assessing properties to equally distribute explained
the 150 largest US companies; 538 key opi- variance and clarify the underlying struc-
nion leaders in Brazil assessing 10 of Brazil’s ture of unexplored data sets (Hair et al.,
largest companies; 532 bank clients and 2006). Tables 4 and 5 depict diagnostic
investors in Spain assessing Spain’s 10 largest measures of each study, which included
banks; 300 physicians in Switzerland assessing KMO, Bartlett’s test and Communalities.
the world’s 10 largest pharmaceutical com- The measures were examined and found to
panies; and 500 insurance clients in Denmark be acceptable.
and Sweden assessing 9 insurance providers. An examination of the factor-rotated
structures in each study revealed a pattern
matrix that matched the hypothesized
DATA ANALYSES AND RESULTS RepTrak® framework, and attributes asso-
The results are presented in three parts. First, ciated with Innovation, Workplace and Pro-
we examine the reliability and validity of ducts/Services loaded consistently on the
each of the hypothesized dimensions across hypothesized dimensions. Two pairs of

12 Corporate Reputation Review Vol. 18, 1, 3–24 © 2015 Macmillan Publishers Ltd. 1363-3589
 Fombrun, Ponzi and Newburry

Table 4: Exploratory Factor Analysis Diagnostics

EFA Diagnostic

Study 1 Study 2 Study 3 Study 4 Study 5

Kaiser–Meyer–Olkin measure 0.990 0.988 0.972 0.960 0.983

of sampling adequacy (KMO)
Bartlett’s test of sphericity Approx. 62,862.029 16,557.721 12,987.239 7,576.367 29,537.866
χ2
d.f. 253 253 231 253 253
Sig. 0.000 0.000 0.000 0.000 0.000

Table 5: Communalities

Communalities – Extraction

Study 1 Study 2 Study 3 Study 4 Study 5

Adapts quickly to change 0.851 0.838 0.890 0.871 0.884

Behaves ethically 0.902 0.849 0.897 0.794 0.845
Clear vision for its future 0.900 0.854 0.858 0.834 0.895
Employee well-being 0.879 0.890 0.867 0.848 0.900
Environmentally responsible 0.895 0.973 0.869 0.824 0.844
Excellent management 0.888 0.893 0.805 0.896 0.892
Fair in the way it does business 0.893 0.850 0.786 0.889 0.919
First to market 0.954 0.840 0.888 0.827 0.858
High quality 0.896 0.875 0.844 0.898 0.899
High-performing 0.946 0.863 0.800 0.869 0.880
innovative 0.880 0.834 0.849 0.880 0.888
Meets customer needs 0.892 0.879 0.854 0.790 0.895
Offers equal opportunities 0.906 0.903 0.822 0.793 0.892
Open and transparent 0.891 0.928 0.887 0.817 0.821
Positive influence on society 0.887 0.863 0.876 0.818 0.892
Profitable 0.971 0.977 0.938 0.848 0.963
Rewards employees fairly 0.909 0.889 0.868 0.846 0.881
Shows growth prospects 0.878 0.875 0.834 0.882 0.896
Stands behind 0.897 0.859 0.888 0.836 0.887
Strong and appealing leader 0.878 0.828 0.839 0.890 0.894
Supports good causes 0.871 0.893 0.894 0.867 0.944
Value for money 0.887 0.861 0.885 0.812 0.909
Well organized 0.894 0.873 0.831 0.893 0.889
Extraction method: Principal component analysis.

dimensions cross-loaded within individual these dimensions were well-defined when

studies, namely the Performance and Lea- the five data sets were pooled across stake-
dership attributes, and the Citizenship and holders. The total explained variance ranged
Governance attributes. However, all four of from 83 to 90 percent across the five studies.

© 2015 Macmillan Publishers Ltd. 1363-3589 Vol. 18, 1, 3–24 Corporate Reputation Review 13
 Stakeholder Tracking Using RepTrak®

Table 6: Factor Loadings and Correlations for Each Study

Factor loadings & correlations

Study 1 Study 2 Study 3 Study 4 Study 5

Corporation Reputation & RepTrak® Pulse 0.848 0.848 0.873 0.864 0.871
Products & Services 0.978 0.976 0.993 0.948 0.960
Innovation 0.989 0.989 0.994 0.906 0.941
Workplace 0.960 0.962 0.949 0.845 0.926
Governance 0.970 0.969 0.989 0.915 0.993
Citizenship 0.977 0.983 0.977 0.881 0.968
Leadership 0.997 0.997 0.975 0.946 0.966
Performance 0.982 0.975 0.935 0.909 0.932

In each study, we examined the relia- Confirming Internal Validity: The First-
bility of the seven dimensions of the Order Factor Model
hypothesized model using Cronbach’s α A first-order confirmatory factor analysis was
(Nunnally and Bernstein, 1994). Seven run to examine the variance shared by
Cronbach’s α coefficients were calculated observed variables with the latent or unob-
for each study, and a total of 35 tests exam- served variables that were hypothesized to
ined. α coefficients ranged from a low of explain ‘Corporate Reputation’. The aim of
0.84 (Physicians-Innovation dimension) to these analyses was to demonstrate that the
a high of 0.96 (Insurance-Product dimen- dimensions of each first-order model con-
sion). Cronbach’s α’s on each test were all verge. We carried out a first-order con-
above 0.7, indicating strong scale reliability firmatory factor analysis with maximum
(Nunnally and Bernstein, 1994). In addi- likelihood on the seven-factor measurement
tion, we tested the effect of deleting an model illustrated in Figure 2.
attribute on each dimension scale. All The first-order confirmatory factor analy-
dimension alpha coefficients were higher sis produced an acceptable fit across all five
with all of the hypothesized dimension studies. While the literature is specific with
attributes included than with any of the regards to the fit indices, to assess how well
attributes excluded, further supporting scale the models represented the data, we followed
reliability. These analyses provide support Hair et al.’s (2010) suggestion regarding a
for the internal consistency of the dimen- mix of fit indices, and examined χ2, one
sional structure of the RepTrak® System. incremental fit test (ie, Comparative Fit
Finally, the Exploratory Factor Analysis Index, CFI), one goodness-of-fit index
structure and loadings were examined. (ie, Trucker-Lewis Index, TLI), and one
Table 6 depicts the standard estimates for badness-of-fit index (ie, Root Mean Square
each study factor loadings and correlations. Error of Approximation, RMSEA). These
The results illuminate that the overall struc- indices have been shown in past research to
ture is unidimensional and the items are cor- demonstrate very little random variation due
related. Each study structure and attribute to sample size, number of parameters, model
factor loadings on each dimension provided misspecification or method of estimation
adequate support for the convergent and (Fan et al., 1999). The χ2 test statistic is sig-
discriminant validity of the model. nificant and is discussed below. A CFI of 0.90

14 Corporate Reputation Review Vol. 18, 1, 3–24 © 2015 Macmillan Publishers Ltd. 1363-3589
 Fombrun, Ponzi and Newburry

High Quality e1

Good Value e2

Products
Stands Behind e3

Meets Customer Needs e4

Innovative e5

First to Market e6
Innovation
Adapts to Change e7

Rewards Employees Fairly e8

Concern for Employees’ Well-Being e9

Workplace
Equal Opportunities in Workplace e10

Open and Transparent e11

Behaves Ethically e12

Governance
Fair in Doing Business e13

Protects Environment e14

Supports Good Causes e15

Citizenship
Positive Influence on Society e16

Strong and Appealing Leader e17

Clear Vision of Future e18

Leadership
Well Organized e19

Excellent Managers e20

Profitable e21

Performance Good Financial Results e22

Strong Growth Prospects e23

Figure 2: Validating dimensionality using first-order confirmatory factor analysis

and above testifies to strong scale uni- However, as mentioned earlier, there are
dimensionality. A CFI and TLI above 0.90 several problems with solely relying on this
indicates convergent validity. RMSEA values index. The χ2 test statistic is problematic
between 0.05 and 0.08 are considered when used with data that are not multi-
acceptable. Table 7 presents the fit statistics variate normal; it is extremely sensitive to
we examined across the five data sets and sample size, and also affected by the num-
demonstrates a robust model fit. ber of parameters in the model
The most commonly cited fit index in (Schermelleh-Engel et al., 2003; Satorra
the literature is the χ2 test statistic. and Bentler, 2001). In a large sample, a χ2

© 2015 Macmillan Publishers Ltd. 1363-3589 Vol. 18, 1, 3–24 Corporate Reputation Review 15
 Stakeholder Tracking Using RepTrak®

Table 7: Goodness-of-Fit Indices of First- Larcker (1981) argue that lack of dis-
Order Factor Models criminant validity creates uncertainty about
whether the results confirm a hypothesized
Study χ2 d.f. CFI TLI RMSEA VE structure or whether the results demon-
strate statistical inconsistencies (Farrell,
1 1574.0*** 209 0.98 0.97 0.06 0.83 2010). To operationalize discriminant
2 416.8*** 209 0.99 0.99 0.04 0.80
validity, we compared AVE with the
3 758.7*** 209 0.96 0.95 0.07 0.78
4 808.8*** 209 0.92 0.90 0.10 0.74
amount of shared variance (Farrell, 2010).
5 586.7*** 209 0.98 0.97 0.06 0.82 Shared variance is the amount of variance
that the latent variable is able to explain in
***P-value < 0.000 another latent variable, and is calculated as
Note: χ2 Chi-Square, d.f. Degrees of Freedom, CFI the square of the correlation estimates or
Comparative Fit Index, TLI Trucker–Lewis, RMSEA Root factor loadings.
mean Error of Approx., VE Variance Extracted An appropriate test for discriminant valid-
ity involves demonstrating that the AVE of
test almost always leads to rejection of the observed attributes of a dimension is sig-
model. To this end, the p-values associated nificantly larger than the shared variance of
with the computed χ2 did not exceed 0.05 the dimensions in the model (Fornell and
for each of the five stakeholder groups. Larcker, 1981). Table 8 shows that the AVE
However, it is often noted that ‘incon- of each dimension was indeed greater than its
sistency among indices is common, and shared variance with any other dimension
having the chi-square as the outlier is par- across all studies, with one exception out of
ticularly common’ (Eagle et al., 2001: 13). 420 tests (Study 1, Innovation < – > Perfor-
As such, we concluded that our overall set mance), thereby supporting the discriminant
of analyses demonstrate strong model fit. validity of the model.

Convergent Validity
To assess convergent validity, we examined Establishing Construct Validity: The
the loadings of each factor and the average Second-Order Factor Model
variance extracted (AVE) of the scale. The seven endogenous constructs (the repu-
Across studies, each observed variable had a tation dimensions) are viewed as mediating
significant loading on its respective latent variables into a single second-order latent
construct ( P < 0.001) with values ranging variable. Second-order factor models were
from 0.78 to 0.94. AVE represents the var- therefore developed from the underlying
iance in the independent variables accoun- first-order models. The purpose of these
ted for by each latent variable. Across the second-order models is to provide external
five studies, AVE was greater than 0.7, construct validity by confirming that the
which exceeds the recommended 0.5 RepTrak® framework maintains the hypo-
benchmark value (Fornell and Larcker, thesized second-order factor structure by
1981), and confirms the measure’s con- converging on the latent reputation construct
vergent validity. variable (Spector, 1992). We used the pre-
viously validated RepTrak® Pulse measure of
corporate reputation as the external depen-
Discriminant Validity dent variable (Ponzi et al., 2011). Figure 3
Establishing discriminant validity is required describes the model specification whose
for latent variables analysis. Fornell and construct validity we sought to establish.

16 Corporate Reputation Review Vol. 18, 1, 3–24 © 2015 Macmillan Publishers Ltd. 1363-3589
 Fombrun, Ponzi and Newburry

e1 High Quality e24

e2 Good Value
Products
e3 Stands Behind

e4 Meets Customer Needs

λ1
e25
e5 Innovative

e6 First to Market
Innovation
e7 Adapts to Change

λ2
e26
e8 Rewards Employees Fairly

e9 Concern Employee Well-Being Workplace r

e10 Equal Opportunity Workplace λ3
e27 Good Overall Reputation e31
e11 Open and Transparent
λ4 Corporate RepTrak
ξ ξ2
Good Feeling About e32
e12 Behaves Ethically
Governance Reputation 1 Pulse Trust e33
e13 Fair in Doing Business
Admire and Respect e34

e28 λ5
e14 Protects Environment

e15 Supports Good Causes Citizenship

e16 Positive Influence on Society
λ6

e17 Strong and Appealing Leader e29

e18 Clear Vision of Future

e19 Well Organized Leadership

e20 Excellent Managers λ7
e30
e21 Profitable

e22 Good Financial Results Performance

e23 Strong Growth Prospects

Figure 3: Establishing construct validity using second-order confirmatory factor analysis

Table 9 summarizes the goodness-of-fit Table 9: Goodness-of-fit Indices of Second

indices for the five second-order factor Order Factor Models
models that were tested. Each model pro-
vides support for the hypothesis that the Study χ2 d.f. CFI TLI RMSEA
seven dimensions of the RepTrak® System
are components of a second-order con- 1 2218.7*** 313 0.98 0.97 0.06
2 777.6*** 316 0.98 0.98 0.05
struct that can be described as ‘Corporate
3 1241.3*** 314 0.95 0.94 0.08
Reputation’. 4 964.7*** 311 0.93 0.92 0.08
In sum, these analyses confirm the under- 5 845.9*** 312 0.97 0.97 0.06
lying dimensional structure of corporate
reputation hypothesized in the RepTrak® ***p < 0.000
System. The results are persuasive because Note: χs2 Chi-square, d.f. Degrees of Freedom, CFI
they verified the dimensionality of the cor- Comparative Fit Index, TLI Trucker–Lewis, RMSEA Root
porate reputation measure across five stake- mean Error of Approximation

holder groups, in multiple industries and

across six geographies. The results therefore
provide robust empirical support for the the confidence they need to apply the model
reliability and validity of the RepTrak® Sys- for measuring corporate reputations across
tem and provide researchers and practitioners stakeholders, industries and geographies.

© 2015 Macmillan Publishers Ltd. 1363-3589 Vol. 18, 1, 3–24 Corporate Reputation Review 19
 Stakeholder Tracking Using RepTrak®

DISCUSSION AND CONCLUSIONS specific impact that the seven dimensions of

Researchers and practitioners have struggled reputation have on outcomes of interest,
for years to measure rigorously the reputa- including stakeholders’ supportive intent (eg,
tions of companies (Dowling and Gardberg, Newburry, 2010) and its manifestation in a
2012). The need for a rigorous comparative company’s sales, risk profile and financial
measure has been made abundantly clear in performance (eg, Smith et al., 2010).
conferences and journals for decades. Despite From a practitioner standpoint, a validated
extensive conversations and debate, there management tool with predictive ability is
remains a lack of consensus about the paramount to tracking and managing stake-
dimensionality and structure of the construct holder perceptions and relationships. The
‘corporate reputation’ (see, eg, Barnett and RepTrak® System provides practitioners a
Pollock, 2012). toolbox of complementary measures for both
The analyses reported in this paper took a tracking and analysis of reputations and for
point of departure in various studies that have linking to outcomes. Of particular value to
relied on Reputation Institute’s RepTrak® managers is the demonstrated validity of the
System for measuring corporate reputations. RepTrak® System to measure cross-industry,
By applying multivariate analyses to five cross-stakeholder and cross-country percep-
representative data sets, we provide robust tions of stakeholders. Multinational compa-
evidence of the reliability, internal validity nies need this capability to address the
and external validity of the RepTrak® Sys- diversity of the global environment in which
tem’s seven dimensions and 23 attributes. they operate. Consistent tracking of stake-
A first important theoretical implication of holder perceptions is one thing. The ability
this study is the explicit attention given to to use the RepTrak® System to predict
validating a framework that incorporates profitable pathways for improving business
both the seven dimensions of reputation and outcomes, however, is another. With a vali-
its generalized representation (the RepTrak® dated tool in hand, practitioners are in a bet-
Pulse). Doing so should prove invaluable for ter position now to build predictive models
future academic research by enabling a more to explain where best to allocate their scarce
thorough understanding of corporate repu- resources in order to optimize outcomes.
tation and its influences. It suggests that the This study is not without its limitations, of
emotional appeal of a company – its reputa- course. First, our validation relied on five
tion – can be partly explained by the infor- studies of stakeholders in six countries. Given
mational content of the signals to which the increasing recognition of the importance
stakeholders respond. of national and other environmental contexts
By demonstrating the effectiveness of the (eg, Prabhu and Stewart, 2001; Michaelis
signaling process (eg, Spence, 1973; Prabhu et al., 2008; Berger and Fitzsimons, 2008),
and Stewart, 2001), the model invites future studies should validate the measure
researchers to develop more fine-grained with additional stakeholder groups in other
theories to explain how specific dimensions cultural contexts. Given known differences
of reputation influence outcomes. While a between Western and Eastern cultures
few such studies have been conducted (eg, Chen et al., 2005), validation of the
(D’Souza et al., 2013; Newburry et al., 2014; framework with respondents in Asia is partic-
Vidaver-Cohen and Brønn, 2013), we see ularly needed. Second, future research should
considerable opportunity to complement the also examine links between RepTrak®’s
vast majority of past research that has focused dimensions and their direct and indirect
narrowly on the generalized reputation con- impacts on supportive behaviors and business
struct. Future research should identify the outcomes (Newburry, 2010), consistent with

20 Corporate Reputation Review Vol. 18, 1, 3–24 © 2015 Macmillan Publishers Ltd. 1363-3589
 Fombrun, Ponzi and Newburry

similar studies on brand value (eg, Goldfarb Bendisch, F., Larsen, G. and Trueman, M. (2013)
et al., 2009). Finally, studies of additional ‘Fame and fortune: A conceptual model of
CEO brands’, European Journal of Marketing, 47(3/4),
industries and stakeholders could further 596–614.
confirm the generalizability of the mea- Berger, J. and Fitzsimons, G. (2008) ‘Dogs on the
surement system. street, pumas on your feet: How cues in the
Overall, this paper has put forward a environment influence product evaluation and
robust validation of the RepTrak® System choice’, Journal of Management Research, 45(1), 1–14.
Brammer, S. and Jackson, G. (2012) ‘How regulatory
for measuring reputations, one that practi-
institutions influence corporate reputations: A cross-
tioners can rely upon to track their reputa- country comparative approach’, in, M.L. Barnett and
tions on a continuous basis, to analyze the T.G. Pollock (eds.), The Oxford Handbook of Corporate
underlying dimensional drivers of their Reputation, Oxford University Press, Oxford,
reputations and to link to outcome measures pp. 297–319.
of interest to their businesses, such as sup- Carreras, E., Alloza, A. and Carreras, A. (2013)
Corporate Reputation, LID Publishing, London.
portive intent, sales, investor and employee
Casado, A.M., Peláez, J.I. and Cardona, J. (2014)
churn, and financial performance (Fombrun, ‘Managing corporate reputation: A perspective on
1996). Given the increasingly recognized the Spanish market’, Corporate Reputation Review,
importance of reputation to all organizations, 17(1), 46–63.
the availability of a robust measurement tool Chen, H., Ng, S. and Rao, A.R. (2005) ‘Cultural
such as RepTrak® can help improve how we differences in consumer impatience’, Journal of
Marketing Research, 42(3), 291–301.
manage, not only reputational issues, but the
Christian, R.C. (1959) ‘How important is corporate
intangible economic asset that a corporate image?’ Journal of Marketing, 24(2), 79–80.
reputation represents. Courtright, J.L. and Smudde, P.M. (2009) ‘Leveraging
organizational innovation for strategic reputation
management’, Corporate Reputation Review, 12(3),
245–269.
Acknowledgements Davis, G.F. (2005) ‘New directions in corporate
The authors thank Professor Naomi Gard- governance’, Annual Review of Sociology, 31,
berg, Sebastian Taciak, RI Manager – Global 143–152.
Research & Analytics, CRR Editor Cees van Dawar, N. and Parker, P. (1994) ‘Marketing universals:
Riel and three anonymous reviewers for Consumers’ use of brand name, price, physical
appearance, and retailer reputation as signals of
their helpful comments and suggestions on product quality’, Journal of Marketing, 58(2), 81–95.
previous drafts of this paper. Deephouse, D.L., Li, L. and Newburry, W. (2009)
‘Institutional and national culture effects on
corporate reputation’. Academy of Management Best
REFERENCES Paper Proceedings, August, Chicago, pp. 1–6.
Aaron, J.R., McMillan, A. and Cline, B.N. (2012) Dempster, A.P., Laird, N.M. and Rubin, D.B. (1977)
‘Investor reaction to firm environmental ‘Maximum likelihood from incomplete data via the
management reputation’, Corporate Reputation EM algorithm’, Journal of the Royal Statistical Society.
Review, 15(4), 304–318. Series B (Methodological), 39(1), 1–38.
Alniacik, E., Alniacik, U. and Erdogmus, N. (2012) Donaldson, T. and Preston, L.E. (1995) ‘The
‘How do the dimensions of corporate reputation stakeholder theory of the corporation: Concepts,
affect employment intentions?’ Corporate Reputation evidence, and implications’, Academy of Management
Review, 15(1), 3–19. Review, 20(1), 65–91.
Barnett, M.L. and Pollock, T.G. (eds.) (2012) The Dowling, G.R. and Gardberg, N.A. (2012) ‘Keeping
Oxford Handbook of Corporate Reputation, Oxford score: The challenges of measuring corporate
University Press, Oxford. reputation’, in, M.L. Barnett and T.G. Pollock
Basdeo, D.K., Smith, K.G., Grimm, C.M., Rindova, (eds.), The Oxford Handbook of Corporate Reputation,
V.P. and Derfus, P.J. (2006) ‘The impact of market Oxford University Press, Oxford, pp. 34–68.
actions on firm reputation’, Strategic Management D’Souza, C., Taghian, M. and Sullivan-Mort, G.
Journal, 27(12), 1205–1219. (2013) ‘Environmentally motivated actions

© 2015 Macmillan Publishers Ltd. 1363-3589 Vol. 18, 1, 3–24 Corporate Reputation Review 21