Applied Cognitive Psychology, Appl. Cognit. Psychol. 25: 605–614 (2011)Published online 22 June 2010 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/acp.1729
The Psychological Representation of Cor
porate ‘Personality’PHILIPP E. OTTO1*, NICK CHATER2 and HENRY STOTT3
1European University Viadrina, Frankfurt (Oder), Germany2University College London, London, UK3Decision Technology, London, UK
*CorrScharE-ma
Copy
Summary: As with any other object, people represent companies along a number of dimensions. But what are the key psychologicaldimensions that best describe companies, organizations, or brands? We apply research methods initially developed for studyingattitudes, including attitudes to other people, to look at how the public represents corporate ‘personality’. The major dimensionsthat psychologically differentiate companies resemble human factors of personality and can be labelled Honesty, Prestige,Innovation, and Power. These dimensions are confirmed after a time gap of 1 year, also capturing specific changes in the rating ofindividual companies. The proposed methodology not only has substantial commercial value in helping companies understand andtrack their public perception, but scales of this type can potentially guide and manage the decision-making of individuals or groupsinside and outside rated organizations, thus influencing their organizational culture. Copyright # 2010 John Wiley & Sons, Ltd.
In the continuous interaction with our environment, we need
not only to be able to quickly perceive new information but
also have to rely on existing information as a benchmark.
Behaviour always takes place under a specific frame or is
embedded in context, and evaluations are always in relation
to similar objects of the respective class of objects. But is
there a general mechanism which describes this formation
of differences between perceived objects? Do we apply
comparable processes in different domains? Kelly (1955)
proposed the theory of personal constructs for personality
evaluations, using methods that appear more generally
applicable. Osgood (1962) developed methods for finding
semantic ‘dimensions’ for objects of all kinds. More recently
computational corpus analysis has been applied to ‘dimen-
sionalizing’ semantic materials in a uniform way (Griffiths &
Steyvers, 2004; Landauer &Dumais, 1997). Moreover, in the
literature on categorization, it is typically assumed that
uniform principles guide the representation of diverse
categories, although certain fundamental distinctions (e.g.
the distinction between natural kind versus artefact concepts)
are sometimes viewed as representationally significant
(Gelman, 1988; Murphy, 2002; Sloman & Malt, 2003).
The focus in this work is how well existing research
methods developed for uncovering psychological dimen-
sions can be transferred to understanding how people
represent companies. Its public perception may be a
substantial factor within a company and for its outside
relations, guiding consumer purchasing decisions as well as
potentially influencing investors’ decision making. Thus, it
would be of considerable practical interest to have a
workable model of the dimensions of what we term corporate
‘personality’, which proposes a taxonomy of companies
based on their public perception. To tackle this question, we
derive an evaluation process for measuring generally
perceived company differences, which combines existing
methods to provide a new and unified approach.
espondence to: Philipp E. Otto, Europa-Universitat Viadrina, Grobernstrabe 59, D-15230 Frankfurt (Oder), Germany.il: [email protected]
right # 2010 John Wiley & Sons, Ltd.
PREVIOUS RESEARCH AND CONCEPTUAL
BACKGROUND
Research investigating how people represent complex objects
can be traced back to Osgood and colleagues, who postulated
the existence of general psychological dimensions for
evaluating objects (Osgood, 1962; Osgood, Tannenbaum, &
Suci, 1957). In their ‘semantic differential’ approach, lists of
adjectives were searched for which best capture meaningful
differences between items. The claim was made that a
restricted number of descriptive properties can be sufficient to
differentiate items within a wide range of categories of objects,
reaching from colours and shapes to stories and people. More
recent approaches have applied the semantic differential to a
range of specific domains, including product categories (Hsu,
Chuang, & Chuang, 2000; Katz, Aakhus, Kim, & Turner,
2002; Mondragon, Company, & Vergara, 2005), perceptual
categories (Ohnishi et al., 1996; Oyama, Yamada, & Iwasawa,
1998; Tessarolo, 1981) and names (Hartmann, 1985).
Another evaluation approach for complex objects is the
psychometric method. Here scales are developed for capturing
differences. Underlying factors are extracted which link
directly or indirectly to featural differences, but either way
explain variations in observed behaviour. This perspective is
common in personality research where different ‘personality
dimensions’ are used to explain individual differences
(compare Cattell, 1965; Eysenck & Eysenck, 1985; Goldberg,
1993). Prior research in brand perception has documented
interesting parallels between the perception of people and
the perception of brands (i.e. Epstein, 1977). Lievens and
Highhouse (2003) take a related approach in the evaluation of
organizational attractiveness. When describing an organiz-
ation, similar descriptors are used as when describing
categories like ‘friends’ or ‘strangers’. Thus, it appears that
people may represent organizations as having a corporate
‘personality,’ analogous to human personality. However, Aaker
(1997) who formalizes the specific dimensions for ‘brand
personality’ and Slaughter, Zicker, Highhouse, and Mohr
(2004) who did the same for ‘organization personality’ do
observe differences between human and company personality.
606 P. E. Otto et al.
So far these different studies stand next to each other
without providing integrated insights into human decision
making. To generate a more general taxonomy of perceived
company differences, these studies are integrated into a
single framework. To combine these different approaches
into one unifying concept of corporate personality, we begin
with an exploratory study, to establish some of the natural
dimensions along which people differentiate companies. In
the light of this pilot study, our first study allows us to
systematically evaluate these and other adjective dimensions
for companies, provided by the literature, to get an
understanding of relative importance of the company
descriptors. Our second study then uses a subset of
descriptors, which provided to be important for measuring
company differences, to position diverse companies on these
adjective dimensions and to generate the factors of corporate
personality. Here, relations to different economic charac-
teristics are highlighted. In the third study, the stability of the
derived corporate personality factors is tested for a time gap
of 1 year, also measuring changes in the perception of
individual companies over this time span.
PILOT STUDY: EXPLORING COMPANY
PERCEPTION
The main aim is to explore, in a relatively open-ended method,
the natural dimensions which best describe the concept
‘company’. Later, we want people to rate companies on these
different adjective dimensions. But first we need a method to
generate further candidates, which might usefully discriminate
between companies, to not miss out fundamental descriptors of
company differences. For this we use an experimental
technique called Repetory Grid (RepGrid) which was
introduced by Kelly (1955). RepGrid was first used by Kelly
for the evaluation of individual personality differences. Tomake
sure the concept is derived by the participant and not induced by
the procedure, he introduced an iterative method which is
content neutral. This keeps the guidance by the actual questions
asked at a minimum and only builds on formerly given answers,
without needing any specific material and only providing a
content free frame. This general method is especially suitable
for the generation of dimensions people naturally use to
evaluate objects and can directly be applied for the analysis of
any concept. The one difference in the case here is that
the objects of analysis are companies instead of people. The
selected companies and the derived adjective dimensions of the
individual concepts are to be included in the later analysis.
Method
Six postgraduate students or university staff (three male;
three female; average age 27) took part in the study and were
paid £6 ($11) each. The individual RepGrid session lasted
approximately 60minutes and took place as a one-to-one
interview. The material consisted of a card for each elicited
company (element) and an initially blank table intowhich the
adjectives would be written. The same table was later used
for the rating of the elicited companies on the different
derived adjective dimensions.
Copyright # 2010 John Wiley & Sons, Ltd.
First, each participant had to name nine different well-
known companies, which were written on cards. This was
followed by a procedure to derive the adjective dimensions
for the corresponding objects, which Kelly (1955) called
‘triadic elicitation’, in which randomly three cards as triples
of companies were selected by the experimenter. Two of the
company cards were contrasted here with a third one and the
participant’s task was to produce ‘bipolar’ pairs of adjectives
that differentiated the third company from the other two.
These descriptions were always depicted in bipolar dimen-
sions, where one adjective was describing the two companies
on the one side and another adjective was describing the
single company on the other side, presenting opposing
adjectives like ‘international’ versus ‘national’, ‘large’
versus ‘small’, ‘friendly’ versus ‘unfriendly’, etc. These
comparisons were done repeatedly, so that each company
was selected once as the ‘single’ company, resulting in nine
descriptions for every participant. In a last step all nine
companies were rated on a scale from 1 to 5 on these derived
bipolar adjectives, all in line with Kelly’s (1955) standard
RepGrid method. The results were analysed according to
concept homogeneity and inter-individual variability.
Results
All participants found it easy to generate bipolar adjectives
to separate their selected companies. Also the similarity
between the companies, based on the rating for their elicited
adjectives, were supported by the participants, as the
grouping of the clustering results was ad hoc confirmed
by the participants. An example of a derived solution is given
in Figure 1. The named companies, the generated dimen-
sions, and the ratings are shown for this participant. As in the
case of the other participants, the adjective dimensions group
into higher level dimensions, shown by the hierarchical
clustering result which is based on the individual ratings. For
all different RepGrid solutions see Appendix A.
Named companies mainly represent large retailers or
famous brands. They cover supermarkets and banks as well
as current or potential employers and favourite product
producers. Participants are similar in what companies they
select, with the same company picked once by four out of the
six participants. The frequency with which the selected
companies co-occur within the sample is ‘Tesco’ four times,
‘Sainsbury’ and ‘Costcutters’ three times, and ‘H&M’, ‘BT’
and ‘Barclays’ two times. For comparing companies people
use similar adjectives, as is apparent from inspection of
Table 1, where the same adjectives reoccur within a small
sample of six participants. These qualitative overlaps in
choice of adjectives used to distinguish companies, broadly
support the assumption of a naturally agreed concept of
company differences. Common themes appear to be quality,
price, general appearance and contact experiences.
Discussion
The elicited dimensions for company evaluations have
overlaps within the sample but also with company
descriptors developed in previous studies (Table 2). First
they show commonalities with the concept of brand
personality (Aaker, 1997) and organization personality
Appl. Cognit. Psychol. 25: 605–614 (2011)
Figure 1. RepGrid example solution
Table 1. Company differentiation dimensions elicited for the self-selected companies
Person 1 Person 2 Person 3 Person 4 Person 5 Person 6
Common Affordable Dominant Attractive Abstract AdversarialEnjoyable Close Freedom of action Cheapa Cheapa BigEssential Durable Identity Competitive Educated CompetentHidden importance Formal Internationala Distant Helpfula ConcernedNeeded Luxurious Powerful Feminine Influential ExploitativeNice Qualityb Qualityb Helpfula Physical Internationala
Prestigious Rare pos. experience Spacious Modern Professional Qualityb
Secondary Relaxeda High status Regular Trustworthy Socially responsibleSpecific Rigid Typical Relaxeda Useful Well priced
Here only the one side of the dimension is shown. The other pole is its opposite (i.e. big–small) or its negation (i.e. common–uncommon).aOccurs twice.bOccurs three times.
Corporate personality 607
(Slaughter et al., 2004), sharing one and two company
adjectives respectively; second with Osgood’s semantic
differentials (Heise, 1970; Osgood et al., 1957) sharing four
company adjectives. In the former a large spectrum is used to
find the most useful dimensions for companies. The latter
assumes more general dimensions which apply for different
Table 2. Co-occurrence of named adjectives from different sources
RepGrid Brand
Number in totala 48 42Brand 1 —Organization 2 9Semantic differential 4 1
aAll named 54 adjective dimensions resulting from the RepGrid technique were co1997, p. 354) and Organization (Slaughter et al., 2004, p. 89) personality adjectiveOsgood’s Semantic Differentials represent his three dimensions Evaluation, Potencyis 118 different company adjectives.
Copyright # 2010 John Wiley & Sons, Ltd.
sorts of objects and which have been labelled Evaluation,
Potency, and Activity. By letting people derive the
dimensions for describing differences between companies
here, a more open-ended yet direct way is chosen to generate
the important dimensions for differentiating between
companies. The RepGrid method can be seen as a more
Organization Semantic differential
23 239 1— 11 —
nsidered of which six were redundant due to double naming. Brand (Aaker,s were taken from the respective studies and their derived descriptive factor., and Activity (Heise, 1970, p. 237). The total number for all sources together
Appl. Cognit. Psychol. 25: 605–614 (2011)
608 P. E. Otto et al.
direct approach for finding useful adjectives on which to
evaluate differences between companies, being positioned
somewhere in between these two separate approaches.
The elicited dimensions might prove useful for future
evaluations of companies. They are complementary to
adjectives generated in the existing literature, thus poten-
tially adding formerly neglected areas of systematic
company differences. But, of course, only a larger data set
will allow for reliable interpretation. Therefore, this pilot
study only prepares for the following studies. The results
derived here are used to produce a more systematic study. To
further evaluate the dimensions differentiating between
companies, in Study 1, we compare the newly derived
company concepts with the company descriptors used in
previous studies. Thereby, the number of potentially useful
adjectives is pruned to facilitate later evaluations.
Table 3. Most stable differentiating company adjectives
AdjectiveTotalSTD
STD means(over
companies)
STD meansdivided bytotal STD Source
Technical 1.18 0.92 0.78 BPLuxurious 1.22 0.90 0.74 RepGridInternational 1.38 0.98 0.71 RepGridUpper class 1.20 0.82 0.68 BPCool 1.20 0.78 0.65 BPQuiet 1.07 0.68 0.63 SDFormal 1.26 0.79 0.63 RepGridExploitative 0.97 0.59 0.61 RepGridLeader 1.19 0.72 0.60 BPOriginal 1.15 0.67 0.58 OP and BPPopular 1.07 0.62 0.58 OPNoisy 1.22 0.70 0.57 SDStatus 1.13 0.65 0.57 RepGridCheap 1.22 0.69 0.57 RepGridLow class 1.22 0.69 0.57 OPYoung 1.07 0.60 0.56 BP and SDQuality 1.20 0.67 0.56 RepGridOld 1.19 0.65 0.55 SDGood looking 1.21 0.66 0.54 BPWell priced 1.06 0.57 0.54 RepGridGlamorous 1.30 0.70 0.54 BPCreative 1.09 0.59 0.54 OPCompetitive 1.01 0.54 0.54 RepGridPowerful 1.20 0.64 0.54 RepGrid
and SDEducated 1.15 0.61 0.53 RepGridIntelligent 1.08 0.57 0.53 BPBusy 1.03 0.54 0.52 OPBig 1.06 0.55 0.52 RepGrid
and SDSecure 1.12 0.58 0.51 BPPrestigious 1.14 0.58 0.50 RepGridSecondary 1.10 0.55 0.50 RepGrid
Thirty-one adjectives where the proportion of variance explained by com-pany mean differences is above the cutoff of 50%. The sources are asfollows: RepGrid, the newly derived dimensions; SD, semantic differential;BP, brand personality; OP, organization personality.
STUDY 1: COMPANY EVALUATION DIMENSIONS
A simple rating method is used to evaluate the different
proposed descriptive adjectives. This is done to system-
atically compare the different sources of company descrip-
tors. Existing company personality adjectives, the semantic
differentials, and the RepGrid company adjectives generated
in our pilot study are used to measure a restricted number of
companies to figure out their descriptive values. Besides
finding the most useful dimensions, redundancies are
captured and the adjectives are put into relation. It also
enables us to determine a set of adjectives which best
describes company differences. This allows a reduction in
the number of adjectives required to then scale-up the
resulting methods in Studies 2 and 3.
Method
Participants rated a list of adjectives in relation to a set of
companies. As illustrated in the pilot study, the different
sources show overlaps in the adjectives used. Here we
included all proposed descriptors (Aaker, 1997; Slaughter
et al.; Heise, 1970) and our newly derived dimensions. Only
redundant adjectives, which described the same dimensions,
were left out.
Twenty students (10 male; 10 female; average age 26) took
part in the study who were paid £12 ($22) each. The computer
based rating lasted approximately 120minutes and took place
in separate rooms for each individual. Participants had to
rate 20 companies on all the 118 adjectives resulting from the
different studies in Table 2. A Likert Scale from 1 to 5 was
used for each adjective always taking both ends of the
dimension as a single evaluation, so that i.e. ‘good’ and ‘bad’
were rated separately with the complementary anchor of the
scale being its own negation (‘not good’ and ‘not bad’). The
companies were taken from the pilot study. The six companies
which were named more than once and in addition 14
representatives for different industries were used. The
companies were displayed together for each adjective to
keep the context in which the comparisons are made stable.
But the company order was randomized in each set and the
adjective order was randomized over participants.
Copyright # 2010 John Wiley & Sons, Ltd.
Results
The company ratings were analysed according to the relation
of inter-company variability to inter-individual variability.
For this we introduce the measure of Company Differentia-
bility, which describes how strong a specific adjective
differentiates between companies, while yielding consistent
ratings for a single company across the different raters.
The Company Differentiability score equals the variance of
the company means divided by the total variance. This
measure of Company Differentiability relates the adjective’s
descriptive power of differentiating between companies to
their inter-individual variation. The more stable the adjective
is over participants and the better it distinguishes between
companies, the higher is its value. The highest adjectives in
Company Differentiability, correspondingly being the most
stable and strongest between-company differentiating adjec-
tives, are shown in Table 3. These adjectives are selected by a
criterion of stable Company Differentiability, meaning that
their Company Differentiability score is equal to or above .5.
Thus, for those 31 adjectives, at least half of the total
variance is explained by the differences between company
mean values. Adjectives below this cutoff of 50% were left
Appl. Cognit. Psychol. 25: 605–614 (2011)
Corporate personality 609
aside due to their relative high inter-individual variability.
With this criterion of stable Company Differentiability,
RepGrid adjectives were included the most with 14 out of 31
representing our newly elicited dimensions. For the other
sources there are ten, six, and five cases in the highly
differentiating group for the Brand Personality, Semantic
Differential, and Organization Personality adjectives
respectively. The number of 31 adjectives is in line with
the hierarchical clustering results, where a strong decrease in
the fit of the model, measured by R square (RSq), is observed
somewhere around this number (Figure 2).
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
110 100 90 80 70 60
RSq
Number of C
Figure 2. RSq for number of cluster
Figure 3. Hierarchical clustering tree for
Copyright # 2010 John Wiley & Sons, Ltd.
The clustering results for the 31 highest scoring adjectives
in Company Differentiability are shown in Figure 3. These
clustering results are based on the average company values
over participants. The different clustering steps describe
different aggregation levels for company evaluation. The
adjectives group together into higher levels, forming
different aspects of company characteristics. They nicely
separate into four to six clusters which describe company
characteristics on a higher level. Representative classes as
higher order clusters are ‘quality/prestige’, ‘power’, and
‘price’, with the adjectives ‘technical’ and ‘quiet’ as outliers.
50 40 30 20 10 0lusters
s in the Pearson cluster history
the highly differentiating adjectives
Appl. Cognit. Psychol. 25: 605–614 (2011)
610 P. E. Otto et al.
These classes form different groups of adjectives for the
evaluation of companies.
Discussion
When evaluating a larger number of adjectives from different
sources, which all describe companies’ characteristics, the in
the pilot study newly elicited adjectives prove highly useful.
This illustrates the possible improvements we can achieve for
the evaluation of companies. By comparing the adjectives on
their Company Differentiability score, the more important
adjectives are isolated. These can be seen as the more general
descriptors for companies.
Interesting is not only how single adjectives describe
companies but also how these aggregate into factors. A first
step into this direction is done in this part by grouping the
adjectives into clusters. Note that these adjectives are broadly
in line with Osgood’s three general Semantic Differentials:
Evaluation (prestige/quality), Potency (power), and Activity
(price). However, a larger body of companies is needed for
generating the fundamental dimensions which guide
company evaluations. Thus, our next study applies a smaller
number of adjectives to a much wider range of companies.
STUDY 2: COMPANY POSITIONING
What are the fundamental factors for company perception?
To estimate the relative importance of the different company
dimensions and to learn more about their similarities, we ran
a further study to distill the most important factors for the
evaluation of companies. Study 2 uses the adjective
evaluation results of Study 1 and expands the analysis to
a large number of companies. This is done to derive a general
taxonomy of perceived company features which then can be
related to for example other company characteristics.
Method
Sixty-four large UK companies and globally operating
international companies (UK: 48 companies; international:
16 companies) were evaluated on 41 adjectives in an online
survey. These 41 adjectives represent the dimensions
Figure 4. Initial fa
Copyright # 2010 John Wiley & Sons, Ltd.
captured with the 31 adjectives high on Company
Differentiability and in addition keeping social adjectives,
like friendly and helpful, and trust related adjectives, like
pleasant and personal, in the pool. The total number of
people participating in this study was 1282 (40% female,
38.5 average age). Participants were recruited and paid via
the Ipoints web-service, which is a UK nation-wide panel
system for online data collection, primarily used in
commercial market research. This study was run in
September 2005.
Each evaluation lasted approximately 10 minutes. Every
participant evaluated all 64 companies on four randomly
allocated adjectives. The companies were displayed together,
but were randomized for each adjective. Each adjective
was rated by at least 100 participants (average 125 ratings)
on a 5-point Likert Scale, as in Study 1. These ratings
were averaged for each company adjective. The resulting
company mean values on each adjective were then used in a
factor analysis to derive the underlying dimensions of
company evaluations.
Results
The usage of the adjectives on a broader range of companies
enables a grouping of the different adjectives. These groups
or factors illustrate the underlying categorical differences
and nicely link to economic measures for company
performance.
In the factor analysis of this data, the eigenvalues of the
principal factors flatten out after the fourth factor which is
illustrated in Figure 4. Although Factor 5 is, with an
eigenvalue of 1.08 slightly above one, due to the drop before
we only consider a 4-factor solution in the further analysis, in
line with Cattell (1966). The data gathered in the year 2006,
which also supports this selection, is the topic of the next
section. The complete Equamax rotated factor solution is
given in Appendix B. The rotated solution nicely separates
into four factors that can be labelled ‘Honesty’, ‘Prestige’,
‘Innovation’ and ‘Power’ (see top of Figure 5). The first
factor which we named ‘Honesty’ captures fairness and
trustworthiness of a company. ‘Prestige’ is a dimension of
how valued a company is. With ‘Innovation’ the vividness
ctor solution
Appl. Cognit. Psychol. 25: 605–614 (2011)
Figure 5. Equamax rotated factor solution 2005 and 2006
Corporate personality 611
and flexibility of a company is described. ‘Power’ as the
fourth factor captures the importance or dominance of a
company.
These factors are then put in relation to economic
company descriptors taken from Datastream, which is a data
provider for historical company information. Here the factor
values for the British companies where correlated with
economic measures of size, evaluation, growth, and profit.
Table 4 shows the Spearman correlation for the four factors
based on the 25 companies listed on the London Stock
Exchange. The derived factors show direct relations to
objective company characteristics. The Prestige factor
strongly correlates with the measure for company size and
company profit; the Innovation factor correlates with the
measure for company growth.
Discussion
These factors can be of potential use for the understanding
of company perception and the development of alternative
evaluation criteria for companies. The factor correlation also
indicates possible relations with economic measures, which
Table 4. Factor Spearman correlation for company performancemeasures
Factor 1 Factor 2 Factor 3 Factor 4Honesty Prestige Innovation Power
Size �0.02 0.56a �0.23 0.06Evaluation 0.13 0.24 �0.27 0.33Growth �0.14 �0.32 0.52a �0.14Profit �0.14 0.75a �0.31 0.27
Company measures are taken from Datastream, where Size stands for thetotal assets employed, Evaluation for the market to book value, Growth forthe 3-year growth in sales, Profit for the pre-tax profit.aSignificant on the p< 0.01 level.
Copyright # 2010 John Wiley & Sons, Ltd.
might proof useful as performance predictors. Though,
evaluations over a longer time horizon and on a larger body
of companies are necessary to prove the strength and the
directionality of these relations.
Also interesting relations can be drawn from the derived
factors to Osgood’s concept of Semantic Differentials
(Osgood et al., 1957). The ‘Innovation’ factor fits with
their Activity dimension, ‘Prestige’ goes together with the
Evaluation dimension, and ‘Power’ with the Potency
dimension. Only ‘Honesty’ comes in as an additional factor
for the evaluation of companies, which seems not to be part
of Osgood’s original schema. The broader implications of
these results are discussed in more detail later. First, we test
the stability of the factor solution in a further study, to
evaluate how well these corporate personality factors capture
general company differences.
STUDY 3: COMPANY REEVALUATION
How stable are the corporate personality factors over time?
For estimating the stability of the factor solution, we ran
a re-evaluation of the companies on the same adjective
dimensions after a time gap of 1 year. This does not
only test the stability of the proposed measure but also
enables us to track changes in the perception of individual
companies on these factors. Study 3 reruns Study 2 after
exactly 1 year. The same method is used, but the number
of companies is further increased. Thereby, the stability
of the measure is tested in two ways. First, it is checked if
the measure generalizes to a larger set of companies.
Second, possible factor changes over time are evaluated.
For the subset of companies which are included in both
studies, also individual variations on this measure are
captured.
Appl. Cognit. Psychol. 25: 605–614 (2011)
612 P. E. Otto et al.
Method
Ninety-two large UK companies and globally operating
international companies (UK: 65 companies; international:
27 companies) were evaluated on 41 adjectives using the
same method as in Study 2. Fifty-nine companies from the
previous study were included and 32 new companies from
different sectors were added. Five from the 64 companies had
to be left out because they did not want to be included in a
further evaluation.
The study was run on 1374 people (44% female, 36.5
average age). These new participants were again recruited
and paid via the Ipoints web-service. The study was run in
September 2006. Each evaluation lasted approximately 10
minutes. Every participant evaluated all 92 companies on
three randomly allocated adjectives. The companies were
displayed together, but were randomized for each adjective.
Each adjective was rated by at least 80 participants (average
100 ratings). The resulting factor solution is compared to
the factor results of Study 2 and changes for companies,
which were included in both evaluations, are highlighted.
Results
The re-evaluation of the companies tests the developed
method of company differences. On the one hand, it is
interesting if the generated factors are similar to the ones in
the previous study. On the other hand, it is important to check
the stability of the factors.
The factor analysis of this in the year 2006 gathered data
nicely compares to the 2005 data. The eigenvalues of the
principal factors again flatten out after the fourth factor
(Figure 4). This supports our initial assumption of a 4-factor
solution for perceived company differences. The differences
of the Equamax rotated factor solution between the years
2005 (Study 2) and 2006 (Study 3) are shown in Figure 5.
Here the five highest positive and the two highest negative
loading adjectives are shown for each factor. The order of the
most important adjectives varies, but only one to two
important adjectives change in the factors. Also the replaced
adjectives appear to describe similar dimensions and the
rotated solution again nicely separates into the four factors
we labelled ‘Honesty’, ‘Prestige’, ‘Innovation’ and ‘Power’
Figure 6. Ten highest and lowest companies on ‘Innovation
Copyright # 2010 John Wiley & Sons, Ltd.
(see Appendix C for the complete Equamax rotated factor
solution).When applying confirmatory factor analysis for the
2006 data based on the three highest loading adjectives of
each factor resulting from Study 2, the factor solution is
confirmed (see Appendix D), although independence of the
factors cannot be claimed (RMSEA¼ 0.17).
The stability of the factors is further supported by the high
correlation of the derived factor values for the companies
included in both evaluations. When based on the same factor
loadings, the Pearson correlation of the company factor scores
between 2005 and 2006 is .94 for ‘Honesty’, .87 for
‘Innovation’, .98 for ‘Prestige’ and .88 for ‘Power’. For
descriptive purposes the changes in the rank for the top and the
lowest 10 companies are shown for the least stable factor
‘Innovation’ in Figure 6. Between 2005 and 2006 mainly
German car producers gain in their perceived innovativeness.
The boost in innovativeness for the newspaper ‘The Sun’ could
possibly be attributed to its ongoing online marketing campaign.
Discussion
The re-evaluation with an extended number of companies
stresses the value of the corporate personality measure.
The factor solution holds after a time gap of 1 year and the
measure proofs to be robust. The high correlation between
the companies’ factor scores further supports the stability of
the factors. Besides, the re-evaluation tracks the overall
performance on these factors and captures changes of single
companies. Although isolated company activities can have
strong impacts on their public perception, long-term effects
for single actions close to the evaluation can be questioned and
corresponding influences are expected to be of short impact
for their corporate personality. These influences do not
undermine the stability of the derived factors, as the corporate
personality taxonomy captures generally perceived company
differences, which can vary over different time spans.
GENERAL DISCUSSION
The derived factors can potentially be applied in diverse
areas. This potential highly depends on the factors’
’ in 2006 with their changes in rank compared to 2005
Appl. Cognit. Psychol. 25: 605–614 (2011)
Corporate personality 613
universality and how stable they link to other company
characteristics of interest. These two questions guide the
general discussion.
How universal are the factors of corporate
personality?
A first, very positive, observation in relation to the likely
generalizability of factors of corporate personality is that
they connect well with Osgood et al.’s (1957) classic attempt
to find universal semantic differentials. This suggests that
there are general principles guiding the evaluations of objects
that are structuring how people judge companies. No doubt,
of course, the evaluation of companies has its idiosyncrasies
and possibly other objects might have their own peculiarities
too. Osgood et al.’s (1957) proposed general dimensions are
Activity, Evaluation and Potency. It is interesting to consider
how these original dimensions, which have a clear sense
when applied to, for example, living things, translate into
the related factors concerning companies. The translation of
the Activity dimensions, in the case of companies, into an
Innovation factor appears straight forward as innovation can
be seen as resulting from concerted activity within groups.
Activity is somehow equal to the rate of change at which an
entity updates itself. Our Prestige factor stands in close
relation to Osgood’s Evaluation dimension. The prestige of a
company is described by evaluative qualitative criteria.
Potency is a measure of strength and freedom of action, or
the ability to influence and create one’s environment.
The power of a company directly reflects this ability. The
additionally derived factor Honesty could somewhat be
understood as part of the Evaluation dimension, although
the results here show that a company’s honesty is somehow
distinct from the more general quality evaluations and should
be treated as a separate dimension. In passing, it is worth
noting that Honesty relates directly to questions of trust, the
public perception of which is presently a central concern in
many areas of commerce.
These conclusions are further supported by the study of
Slaughter et al. (2004) who also report an honesty dimension
for Organization Personality. Their first factor ‘Boy Scout’ is
described by adjectives like ‘friendly’, ‘family-oriented’,
‘pleasant’, ‘personal’, ‘attentive to people’, ‘helpful’,
‘honest’ and ‘cooperative’, which is quite similar to our
Honesty factor. Factor two is Innovativeness and factor three
is Dominance, which are directly in line with our results.
Their factors four (Thrift) and five (Style) are comparable to
our Prestige factor or Osgood’s general Evaluation dimen-
sion. Also Aaker’s (1997) first factor, named Sincerity,
describes an honesty dimension as the most important
dimension for their Brand Personality with the loading
adjectives ‘down-to earth’, ‘honest’, ‘wholesome’ and
‘cheerful’. Although receiving somehow differing labels,
also the Prestige factor (Competence and Sophistication) and
the Power factor (Ruggedness) find their representations. An
interesting question for future work is how far this viewpoint
holds for future evaluations of the company concept;
currently, it seems that some clear common themes arise
from different methods for evaluating how people perceive
companies. So far, research appears to be showing a
Copyright # 2010 John Wiley & Sons, Ltd.
relatively consistent picture which represents the natural way
of evaluating companies.
The Semantic Differential has been used for attitudinal
research and the evaluation of diverse objects. Besides for
persons, it has not been applied for many living objects. But
when researching the affective characterization of cities,
Ward and Russell (1981) derive at similar results with
reporting as the main first two factors an evaluation
(‘angenehm’) and an activity (‘erregend’) dimension. In
the light of their work, and our own results on company
perception stressing similarities between individual and
organizational identities, it is natural to ask how far Osgood’s
dimensions apply to other types of objects, e.g. animals or
celebrities? Can we find similar regularities in these
domains? If so, there really might be cognitive simplification
processes involved which are similar across evaluations of
complex entities. If different classes of objects are mapped
according to similar dimensions, we might be able to derive
valuable information for the understanding of basic cognitive
processes. An important general question arises, such as
whether the underlying dimensions are separable or integral.
If these dimensions are separable (and hence can directly be
judged independently by participants); or if they are integral
(in which case, reconstruction of the dimensions will
necessarily be indirect, using methods such as that described
here).
Altogether, this study nicely combines previous research
and isolates the important adjectives for company percep-
tions. The corporate personality factors provide a general
measure of perceived company differences by integrating the
different approaches in this field into a universal taxonomy
for companies. An open question remains in which way
these adjective-based dimensions link to other models of
personality. For example the first four ‘Big Five’ personality
factors (Goldberg, 1993) could be represented on an
organizational level characterizing groups of people: with
Honesty describing ‘Agreeableness’, Prestige measuring
‘Extraversion’, Innovation capturing ‘Openness’ and Power
linking to ‘Consciousness’ (leaving ‘Neuroticism without an
organizational representation).
How stable and useful are dimensions of corporate
personality?
With the introduction of personality factors for companies, a
new way of describing companies is provided. This mainly
aims at an alternative description of companies, which
directly reflects the public understanding of companies. But
we have seen in Study 2 that the proposed corporate factors
also show links to economic variables and might in general
be useful as performance indicators. Thus, tracking measures
of corporate personality might add important dimensions to
economic measures of company performance and could be
used both in shaping marketing and brand strategy, and
potentially also in evaluating and predicting company
success. Here for example the first factor of corporate
personality, namely Honesty, might nicely link to marketing
concepts like trust building or corporate reputation. An
important restriction of the presented approach is its purely
outside perspective on organizations. These more generally
Appl. Cognit. Psychol. 25: 605–614 (2011)
614 P. E. Otto et al.
perceived differences from the customer perspective do not
necessarily represent the inside perspective, although over
the long run the outside perspective is assumed to represent
the internal perception or vice versa. Though, the linkage to
internal values or organizational practices, nicely described
as ‘Organizational Cultures’ by Hofstede et al. (1990),
remains to be established.
One important point for the application of the factors is
their variability. We assume that the values on the factors are
not stable over time and that changes can be expected over
longer time horizons. Study 3 provides some first examples
for this, which stresses the potential of the corporate factors
as indicators to track changes over time. The performance of
a company over time on these dimensions could be used, for
example, to evaluate the impact of a high-profile advertising
campaign. Also sources of variations, due to, for example,
regional differences, cannot be ruled out, but prior work on
the inter-cultural stability of factors supports the assumption
of stable dimensions for company evaluations. Geeroms,
Vermeir, Van Kenhove, and Hendrickx (2005) generated
preferred Brand Personality factors across 11 countries. The
four corporate factors find representations in their 8-factor
solution (i.e. ‘Belonging’, ‘Recognition’, ‘Vitality’, and
‘Power’). But perhaps more interesting is the link they built
between these factors and consumer motives, which stresses
the interactive component of the evaluation process and
illustrates that motives or goals can play a key role here.
Corporate Personality could have relationships to many
different areas of interest. It can not only represent company
structures in different regions but also influence the climate
within organizations (i.e. Argyris, 1958) and the behaviour of
a company in themarket (i.e. Podolny, 1993). If so, there could
be links to issues like consumer motives and demands,
attractiveness as an employer and employment confidentiality,
differences in short term and long-term performances, as well
as the general social acceptability of a company. Thus, this
improved measure of corporate personality provides the
cornerstone for a wide range of practical applications.
ACKNOWLEDGEMENTS
The authors thank the ESRC for a research grant to the first
author and Decision Technology for providing additional
resources.
REFERENCES
Aaker, J. L. (1997). Dimensions of brand personality. Journal of Marketing
Research, 34, 347–356.
Argyris, C. (1958). Some problems in conceptualizing organizational climate:
A case study of a bank. Administrative Studies Quarterly, 2, 501–520.
Cattell, R. B. (1965). The scientific study of personality. Baltimore: Penguin
Books.
Cattell, R. B. (1966). The Scree Test for the number of factors.Multivariate
Behavioral Research, 1, 245–276.
Epstein, S. (1977). Traits are alive and well. In D. Magnusson, & N. S. Elder
(Eds.), Personality at the crossroad (pp. 83–98). Hilsdale, NJ: Lawrence
Erlbaum Associates.
Eysenck, H. J., & Eysenck, M. W. (1985). Personality and individual
differences: A natural sciences approach. New York: London.
Copyright # 2010 John Wiley & Sons, Ltd.
Geeroms, N., Vermeir, I., Kenhove, P. V., & Hendrickx, H. (2005). A
taxonomy of consumer motives through preferred brand personality:
Empirical findings for 11 countries. Ghent University Working Paper.
Gelman, S. A. (1988). The development of induction within natural kind
and artifact categories. Cognitive Psychology, 20, 65–95.
Goldberg, C. R. (1993). The structure of phenotypic personality traits.
American Psychologist, 48, 26–34.
Griffiths, T., & Steyvers, M. (2004). Finding scientific topics. Proceedings of
the National Academy of Sciences, 101, 5228–5235.
Hartmann, T. (1985). A concept-specific semantic differential for the rating
of German 1st and 2nd names. Journal of Social Psychology, 125, 787–
788.
Heise, D. R. (1970). The semantic differential and attitude research. In G. F.
Summers (Ed.), Attitude measurement (pp. 235–253). Chicago: Rand
McNally.
Hofstede, G., Neuijen, B., Ohayv, D. D., & Sanders, G. (1990). Measuring
organizational cultures: A qualitative and quantitative study across twenty
cases. Administrative Science Quarterly, 35, 286–316.
Hsu, S. H., Chuang, M. C., & Chuang, C. C. (2000). A semantic differential
study of designers’ users’ product form perception. International Journal
of Industrial Ergonomics, 25, 375–391.
Katz, J., Aakhus, M., Kim, H. D., & Turner, M. (2002). Young user’s
attitudes towards ICTs: A comparative semantic differential study of the
mobile telephone. Annals of Telecommunications, 57, 225–237.
Kelly, G. A. (1955). The psychology of personal constructs (Vol. 1). New
York: Norton.
Landauer, T. K., & Dumais, S. T. (1997). A solution to Plato’s problem:
The Latent Semanctic Analysis theory of the acquisition, induction,
and representation of knowledge. Psychological Review, 104, 140–211.
Lievens, F., & Highhouse, S. (2003). The relation of instrumental
and symbolic attributes to a company’s attractiveness as an employer.
Personnel Psychology, 56, 75–102.
Mondragon, S., Company, P., & Vergara, M. (2005). Semantic differential
applied to the evaluation of machine tool design. International Journal of
Industrial Ergonomics, 35, 1021–1029.
Murphy, G. (2002). The big book of concepts. Cambridge, MA: MIT
Press.
Ohnishi, S., Shinkawa, A., Katto, M., Hamanaka, F., Kurioka, Y., Kawasaki,
M., et al. (1996). Evaluation of odor quality using a semantic differential
method for five odrants of a T&T olfactometer: The case for using
concrete adjectives. Chemical Senses, 21, 34–34.
Osgood, C. E. (1962). Studies of the generality of affectivemeaning systems.
American Psychologist, 17, 10–28.
Osgood, C. E., Tannenbaum, P. H., & Suci, G. J. (1957). The measurement of
meaning. Urbana: University of Illinois Press.
Oyama, T., Yamada, H., & Iwasawa, H. (1998). Synthetic tendencies as the
basis of sensory symbolism: A review of series of experiments by means
of semantic differential. Psychologia, 41, 203–215.
Podolny, J. M. (1993). A status-based model of market competition. Amer-
ican Journal of Sociology, 98, 829–872.
Slaughter, J. E., Zickar, M. J., Highhouse, S., & Mohr, D. C. (2004).
Personality trait inferences about organizations: Development of a
measure and assessment of construct validity. Journal of Applied Psy-
chology, 89, 85–103.
Sloman, S. A., & Malt, B. C. (2003). Artifacts are not ascribed essences, nor
are they treated as belonging to kinds. Language and Cognitive Processes,
18, 563–582.
Tessarolo, M. (1981). The musical experience: The semantic differential as a
research instrument. International Review of Aesthetics and Sociology of
Music, 12, 153–153.
Ward, L. M., & Russell, J. A. (1981). The psychological representation of
molar physical environments. Journal of Experimental Psychology, 110,
121–152.
APPENDICES
Due to the journal’s space limitations the additional material
can only be retrieved online at: http://econ.euv-ffo.de/�peo/
CM.html.
Appl. Cognit. Psychol. 25: 605–614 (2011)