Moussa Touzani 2008

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/229566029

The Perceived Credibility of Quality Labels: A Scale Validation with


Refinement

Article in International IJC · August 2008


DOI: 10.1111/j.1470-6431.2008.00713.x

CITATIONS READS

95 2,975

2 authors:

Salim Moussa Mourad Touzani


University of Gafsa, Tunisia NEOMA Business School
46 PUBLICATIONS 535 CITATIONS 81 PUBLICATIONS 895 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Salim Moussa on 30 April 2021.

The user has requested enhancement of the downloaded file.


International Journal of Consumer Studies ISSN 1470-6423

The perceived credibility of quality labels: a scale validation


with refinement
Salim Moussa and Mourad Touzani
Higher Institute of Management, University of Tunis, Tunisia

Keywords Abstract
Quality labels, signal theory, credibility, scale
validation. In this paper, we present quality labels as signals that reduce problems that arise under
asymmetric information. We propose to closely scrutinize the concept of signal credibility,
Correspondence which is a key determinant of signalling effectiveness. In order to assess the perceived
Salim Moussa, Institut Supérieur de Gestion, credibility of a quality label, we offer a revisited version of a scale originally proposed by
41 Rue de la Liberté, Cité Bouchoucha, Le Larceneux. The data used in this paper involve three different labels and were collected
Bardo 2000, Tunisie. E-mail: using self-report surveys administered to 602 respondents. Based on findings from a
salimmoussa@yahoo.fr variety of reliability and validity tests, the scale demonstrates good psychometric proper-
ties. Both theoretical and managerial implications are discussed, along with limitations and
doi: 10.1111/j.1470-6431.2008.00713.x future research directions.

los and Krystallis, 2003; Larceneux, 2003b; McEachern and


Introduction Schröder, 2004; Tang et al., 2004; De Pelsmacker et al., 2005;
Over the last decade, the proliferation of quality labels, certifica- Getz and Shreck, 2006; D’Souza et al., 2007), and many questions
tion marks and seals of approval has become a worldwide phe- regarding their impact on consumer decision-making process and
nomenon. Fuelling this increase in quality labels are (a) the consumer behaviour remain unanswered.
growing consumer concerns about the safety of fresh meat and With the present study, we are trying to shed some light on the
food products in general as a result of numerous scandals and perceived credibility of quality labels (PCQL) among consumers
crises that have affected the food industry (such as bovine spongi- by answering the following two questions: (1) How credible are
form encephalopathy – so-called ‘mad cow’ disease – in the early quality labels? (2) How can we measure it? The purposes of this
1990s and the recent outbreak of ‘bird flu’ virus H5N1); (b) the paper are twofold: (a) to present quality labels as signals that
increased consumers’ knowledge about links between diet and reduce the information asymmetry between supplier and buyer
health; and (c) the latest significant development of green and about unobservable product quality; and (b) to refine and validate
ethical consumerism. Therefore, consumers have become more a scale originally proposed by Larceneux (2001) that assessed the
and more concerned not only about health and safety claims but perceived credibility of quality labels.
also about the ethical and ecological aspects of their consumption Towards these aims, this paper is organized around the follow-
(Fraj and Martinez, 2007; Llopis-Goig, 2007). ing lines. In the first section, the theoretical background is pre-
In response to consumers’ growing demand for healthy/safe, sented. A definition of the PCQL concept and a brief description of
organic, regional-specific, fair trade, reliable, power-saving and the PCQL scale are given in the next section. The third section
environmental-friendly products, manufacturers and retailers are focuses on scale validation issues (i.e. item selection and judge-
currently flagging products displaying a specific logo (or label) ment, questionnaire and data collection, factorial structure and
that ascertain one or more of these quality attributes (McEachern internal consistency, unidimensionality and convergent validity,
and Warnaby, 2004). However, not all of the quality labels cur- discriminant and predictive validities). In the fourth section, both
rently appearing on products in the marketplace are issued by theoretical and managerial implications are discussed, along with
third parties. Many of them are the result of testing or certifying limitations and future research directions.
schemes maintained by retailers and manufacturers who are
directly involved in the production and sale of these products.
Despite the ubiquity of quality labels that has brought a ‘new’
Literature review
attribute that is free and immediately present at the moment of
Information economics and signal theory
purchasing into the consumer choice process, quality labels have
received relatively modest but upward interest within the market- Following information economic models of search behaviour
ing literature (Parkinson, 1975; Laric and Sarel, 1981; Beltramini (Stigler, 1961), Nelson (1970) proposed in his seminal paper a
and Stafford, 1993; Miyazaki and Krishnamurthy, 2002; Fotopou- distinction between two types of goods: search and experience

526 International Journal of Consumer Studies 32 (2008) 526–533 © The Authors


Journal compilation © 2008 Blackwell Publishing Ltd
S. Moussa and M. Touzani Perceived credibility of quality labels

goods. Search goods were defined as products whose quality the warranties and money-back guaranties). Each of these categories
consumer can verify before purchasing (e.g. clothing). Experience can be further subdivided on the basis of whether the expenditure
goods were defined as products whose quality can be ascertained depends on an actual sale. Sale-independent default-independent
only after buying and experiencing the product (e.g. foods). signals, such as investments in advertising and quality certifica-
Darby and Karni (1973) took up Nelson’s theory and broadened tion, involve expenditure that occurs whether or not a sale takes
it in two perspectives. First, they turned from a product differen- place. In contrast, sale-dependent signals, such as a low introduc-
tiation to a distinction between product attributes because a tory price and slotting allowances, involve expenditures only in the
product can incorporate diverse attributes. Second, they intro- presence of a sales transaction.
duced credence attributes in addition to search and experience
ones. Credence attributes are product characteristics that cannot be
Quality labels as quality signals
evaluated by search prior to purchase or experience after purchase
and are very costly to evaluate at all (Darby and Karni, 1973, Certification marks and quality labels can be defined as ‘marks
pp. 68–69). For many products, taste, safety, health, ethical and used upon or in connection with the products and services of one
ecological qualities are very often experience or credence or more persons to certify regional or other origin, material, mode
attributes (Grunert, 2005; Jahn et al., 2005), and when the object of manufacture, quality, accuracy, or other characteristics of such
of exchange is such a product, sellers usually possess more infor- goods or services’ (Phelps, 1949). As sale-independent default-
mation about these qualities than do buyers. This information independent signals,2 quality labels are designed to provide the
differential is called information asymmetry (Arrow, 1963; consumer with product quality information, hence reducing the
Akerlof, 1970; Kirmani and Rao, 2000; Karstens and Belz, 2006). information asymmetry between suppliers and buyers (Fotopoulos
Therefore, in the predominance of experience and credence and Krystallis, 2003; Larceneux, 2003b). In other words, a quality
attributes, many market dysfunctions, such as adverse selection label assists the ill-informed consumer by transforming credence
(Akerlof, 1970), may arise. With adverse selection, George or experience attributes into ‘quasi-search attributes’ that the con-
Akerlof showed that when uncertain about the sellers’ quality, sumer can search for (Caswell, 1998; Jahn et al., 2005; Karstens
consumers will tend to look for bargains that minimize the risk of and Belz, 2006). Take for example a golden organic apple, which
being tricked. However, this bargain search has the adverse effect has a number of different attributes: (a) the price and the colour
of driving out higher quality sellers from the market because they can be searched prior to the purchase; (b) the taste of the apple can
are indisposed to accept lower prices, leaving behind only the be experienced after the purchase; whereas (c) the kind of farming/
lemons (i.e. lower quality sellers; Akerlof, 1970, pp. 489–492). production can hardly be verified by the consumer – neither before
A possible answer to the lemons problem could be the seller’s nor after the purchase of the product. Only a credible organic label
transfer of information to the customer in order to indicate its can convert this credence attribute into a search one.
quality. This type of information forwarding is referred to as However, quality labels are also a blurry category that covers
signalling (Spence, 1973). Nobel prize-winning economist many different things. They can be awarded by manufacturers,
Michael Spence defines signals as attributes or activities that groups of manufacturers, retailers, government bodies, certifiers
convey information about the characteristics of an economic agent and independent organizations (e.g. consumer associations, maga-
(e.g. job applicants, firms, consumers) and that are subject to his zines). The criteria for awarding the labels can either be very strict
manipulation (Spence, 1973, p. 357). A more elaborated definition or non-existent at all. Some quality labels refer to very specific
considers signals ‘as a marketer-controlled, easy-to-acquire infor- qualities, such as labels indicating organic production, whereas
mational cue, extrinsic to the product itself that consumers use to others are intended as general quality labels (Grunert, 2005).
form inferences about the quality or the value of the product’ Therefore, just giving consumers more information will not reduce
(Bloom and Reve, 1990, p. 59). asymmetry when consumers do not feel confident about using the
Commonly studied signals in the marketing literature include information. For this reason, the credibility of the signal represents
brand name (Erdem et al., 2006), price (Rao, 2005), advertising the main aspect behind an equilibrium in which high- and low-
spending (Moorthy and Hawkins, 2005), warranties (Sobermann, quality sellers can be separated (Kirmani and Rao, 2000).
2003), money-back guarantee (Moorthy and Srinivasan, 1995),
retailer reputation (Purohit and Srivastava, 2001), certification
The perceived credibility of a quality
(Mishra, 2006) and quality labels (Fotopoulos and Krystallis,
label and the PCQL scale
2003; Larceneux, 2003a,b; Jahn et al., 2005; Karstens and Belz,
2006). Signalling theory suggests that signal credibility determines
According to Kirmani and Rao (2000), these signals can be whether a signal conveys information effectively. But what is
classified1 as either default-independent or default-contingent ‘credibility’? According to The American Heritage Dictionary of
signals. Default-independent signals are signals that involve the English Language, in its fourth edition, credibility is the
up-front money expenditure and in which monetary loss occurs quality, capacity or power to elicit belief. Eisend (2002) states that
independently of whether the firm default on its claim (advertising ‘credibility describes a person’s perception of the truth of a piece
expenditures, brand investments and certification costs belong to of information’. Therefore, credibility is a perceived quality; it
this group). Conversely, default-contingent signals are signals in does not reside in an object, a person or a piece of information.
which the monetary loss occurs only when the firm does not
adhere to its claim (the main representatives of this group are 2
For several quality labels, up front expenditures are essential to meet
certain norms or standards of production/sale, and this money spent occurs
1
Other classifications do exist (see Bhattacharya, 1980). whether or not a sale happens.

International Journal of Consumer Studies 32 (2008) 526–533 © The Authors 527


Journal compilation © 2008 Blackwell Publishing Ltd
Perceived credibility of quality labels S. Moussa and M. Touzani

Consequently, in discussing the credibility of a quality label, we


Items selection and judgement
are always discussing its perceived credibility.
The scholarly examination of the concept of credibility is as Based on a literature review (Larceneux, 2001, 2003b), a set of 17
old as the discussion of rhetoric itself, having originated with the items was developed. Driven by face validity considerations
ancient Greeks (e.g. Aristotle, Cicerone). Previous works have (Hardesty and Bearden, 2004), these items were presented to four
detected many different dimensions of credibility (for reviews, marketing academics to determine whether each item should be
see Ohanian, 1990; Eisend, 2002). In a recent study, Eisend retained for further analyses. The four ‘expert judges’ were given
(2002) found that there are two main dimensions of source cred- the definition of the PCQL concept and were instructed to rate
ibility in marketing communication, namely competence and each of the 17 items in the following scale: (a) clearly represen-
trustworthiness. Source credibility is a term commonly used to tative; (b) somewhat representative; and (c) not at all representa-
imply a communicator’s positive characteristics that affect the tive of the construct under scrutiny. Items that were not rated as
receiver’s acceptance of the message.3 Eisend also concluded representative were dropped, resulting in nine items solution.
that these dimensions are the same as those identified by A pretest sample of 32 students was then conducted. Face-to-
Hovland et al. (1953, p. 21) in their seminal work on source face interviews were adopted. The two chief objectives of this
credibility and labelled expertness and trustworthiness. The pretest were to eliminate ambiguous and redundant items and to
expertness dimension captures the perceived knowledge, experi- determine whether there is any problem with the original French
ence, skill and competence of the source. Trustworthiness refers version of the items. From respondent comments, some of the
to an audience’s belief that the source provides information in statements were rephrased. In addition, two items were dropped
an honest manner, without a motivation for manipulation or because they appeared to be extremely redundant. The remaining
deception. pool of seven items was used for the main study data collection.
Turning to PCQL, the literature review revealed that the concept
was evoked and investigated in a small number of works (e.g.
Questionnaire and data collection
Parkinson, 1975). Expertise, trustworthiness and impartiality were
depicted as the principal characteristics of a credible seal or label Three grocery products (i.e. cookies, yogurt and shampoo) dis-
and were compared with other sources of information (i.e. adver- playing three different quality labels (i.e. an ISO 9001:2000 cer-
tisements, salespersons and friends). However, neither a definition tification mark, élu produit de l’année or voted product of the year
of the concept nor its measure was reported. Recently, the PCQL 2006, and Victoire de la beauté 2006) were selected. These prod-
concept was further studied by Fabrice Larceneux. In his works, ucts reflect experience goods (Nelson, 1970) relatively inexpen-
Larceneux (2001, 2003a,b) concludes that quality-label credibility sive and frequently purchased. The retained quality labels fit with
likely depends on the awarder’s credibility. To be credible, a label the two label categories identified by Larceneux (2003a), namely
must come from a third-party organization, foreign to the technical labels (ISO 9001:2000 certification mark) and experien-
manufacturer/seller, competent and not at all interested in the sale tial labels (voted product of the year 2006 and Victoire de la beauté
of the product bearing the label. Note that this definition stresses 2006). Thus, three versions of the questionnaire were developed.
on the independence of the awarding organization, its expertise All were administrated in French with quality-label logo and
and the honesty of its intentions. Based on the above definition, product pictures.
Larceneux (2001) proposed a scale containing six items measured For each quality label/product pair, students were intercepted on
on a seven-point Likert-type scale (i.e. 1 = strongly disagree, campus and were asked: (a) to indicate their agreement on a series
7 = strongly agree). But given that the study carried out by Larce- of seven items about the PCQL; and (b) to evaluate products
neux (2001) was exploratory in spirit, data analysis was restricted bearing each quality label on a series of seven statements, covering
to principal component analysis and internal consistency assess- product perceived quality (PPQ; four items) and product purchase
ment through Cronbach’s coefficient alpha computation intention (PPI; three items). For the PCQL items as well as for
(a = 0.846). The data analysed were collected from a convenience other constructs, a seven-point Likert-type scale was used (i.e.
sample of only 60 respondents (Larceneux, 2001, pp. 12–14). 1 = strongly disagree, 7 = strongly agree).
Consequently, this scale should be scrutinized further. We turn to A total of 627 self-administrated questionnaires were collected
this issue just next. with convenience sampling method. Twenty-five surveys contain-
ing missing data about the theoretical concept of interest were
discarded, resulting in an effective sample size of 602 responses.
The PCQL scale refinement Sample socio-demographical characteristics appear in Table 1.
and validation The sample of 602 observations was split randomly into an explor-
atory analysis sample (N1 = 301) and a validation sample
The process used to refine and validate the PCQL scale followed
(N2 = 301).
the procedures recommended by Churchill (1979) and Gerbing
and Anderson (1988), among others. The scale evolved from items
selection and judgement to construct validity tests. Factorial structure and internal consistency
Scores on PCQL seven items for N1 = 301 observations were
factor analysed by using the spss version 12 (SPSS Inc., Chicago,
3
The persuasiveness of source credibility is well documented. Interested IL, USA) principal axis analysis. Principal axis analysis was
readers are directed to Pornpitakpan’s (2004) recent literature review for retained as the method of extraction as principal components
more details. analysis is not always appropriate when developing scales

528 International Journal of Consumer Studies 32 (2008) 526–533 © The Authors


Journal compilation © 2008 Blackwell Publishing Ltd
S. Moussa and M. Touzani Perceived credibility of quality labels

Table 1 Sample socio-demographic characteristics (N = 602) Table 2 Summary of the principal axis factor analysis results (N1 = 301)

Characteristics Number in sample Items ma sb bic ad

Monthly household income PCQL1 4.73 1.698 0.704


-325 TDa 185 PCQL2 4.64 1.6 0.698
Between 325 and 650 TD 258 PCQL3 4.53 1.765 0.695 a = 0.847
Between 651 and 975 TD 106 PCQL4 4.91 1.53 0.735 (0.818–0.872)
Between 976 and 1300 TD 19 PCQL5 4.95 1.636 0.649
Between 1301and 1625 TD 14 PCQL6 4.78 1.724 0.679
+1626 TD 20 a
Item mean.
Gender b
Standard deviation.
Female 404 c
Factor loadings.
Male 198 d
Cronbach’s alpha reliability index (and its 95% confidence interval).
Age PCQL, perceived credibility of quality labels.
Mean 23.773
Median 23
Standard deviation 3.838
Urbanization
a confirmatory factor analysis for the remaining six items as speci-
Urban 543
fied by a multiple-indicator reflexive measurement model (Jarvis
Rural 59
et al., 2003) and estimated the model through the AMOS 4.01
a
$US1 = 1.209 Tunisian Dinar (31 January 2008 exchange rate). (Smallwaters Corp., Chicago, IL, USA) maximum likelihood
method (Arbuckle, 1999; Byrne, 2001).4 Following the recent
recommendations of Bentler (2007, p. 826), the overall model fit
(Fabrigar et al., 1999; Iacobucci, 2001). With principal axis fac- was evaluated by inspecting the comparative fit index (CFI),
toring, we seek to maximally account for the co-variance among Tucker–Lewis index (TLI), standardized root mean square
the original variables with a small number of common factors. residual (SRMSR) and root mean square error of approximation
The factor analysis was implemented by using several conven- (RMSEA) indices. These different indices point out an excellent
tional criteria (Worthington and Whittaker, 2006). First, the level of fit for the measurement model (Table 3).
Bartlett test of sphericity (the correlation matrix is identity) and Convergent validity was determined by inspecting the param-
Kaiser–Meyer–Olkin measure (KMO of sampling adequacy) were eter estimates. Specifically, large (>0.4), positive and statistically
inspected to verify whether the items shared a common core. significant estimates (t > 1.96) indicated that all loadings were not
Second, parallel analysis (Horn, 1965; Patil et al., 2008) was per- trivially different from zero. The composite reliability index (r,
formed to determine the number of factors to retain. Third, com- Jöreskog’s rho) and the average variance extracted (AVE or rVC)
munalities and factor loadings were inspected. Only items that showed values above the recommended thresholds of 0.70 and
exhibit a communality above 0.40 and a loading higher than 0.50 0.50, respectively (Fornell and Larker, 1981; Ping, 2004), which
were retained. Here, only one item did not show a high commu- provides additional evidence of convergent validity.
nality and a significant loading with the extracted factor. Before
deletion, the item was inspected to ensure that the original
meaning of the construct remained unchanged. A second factor Discriminant and predictive validity
analysis was conducted on the six remaining items (see Appen- In order to evaluate discriminant and predictive validities, a theo-
dix). The KMO yielded a value of 0.877 and the Bartlett test of retical model was proposed. According to signal theory, the model
sphericity obtained an approximated c2 value of 627 452 with an hypothesizes that the PCQL is positively related to the PPQ, which
associated significance of P < 0.001, signifying that the data set is is a prerequisite to the PPI (see Fig. 1). The model also posits a
adequate to factor analysis. Parallel analysis indicated a one-factor direct relationship between PCQL and PPI.
solution with a total extracted variance of 56.777%. The six items PPQ refers to the consumer’s judgement of the overall excel-
loaded heavily on the extracted factor. The loadings ranged from lence, esteem or superiority of a product relative to alternative
0.649 to 0.735 (see Table 2). Cronbach’s alpha index of internal product(s). PPQ is at a high level of abstraction than any specific
consistency was computed and obtained a value of a = 0.847 attribute of a product and differs from objective quality, as PPQ is
(95% confidence interval: 0.818–0.872; Duhachek et al., 2005). more akin to an attitudinal assessment of a brand (Zeithaml, 1988).
Four items were used to measure this construct based on scales
used by Dean (1999) and Erdem and Swait (2004). PPI refers to a
Unidimensionality and convergent validity
decision plan to buy a particular product (or brand) created
However, as written by Gerbing and Anderson (1988) and Ping
(2004), the exploratory factor analysis does not provide an explicit
test of unidimensionality. This important propriety refers to the 4
Prior to perform maximum likelihood estimation, values for skewness,
existence of a single trait or construct underlying a set of mea- kurtosis and Mardia’s test of multivariate normal distribution were
sures. Thus, a subsequent confirmatory analysis is needed for the inspected. Together they ascertain the non-deviation from multinormality
assessment of unidimensionality. Towards this end, we conducted (Lei and Lomax, 2005).

International Journal of Consumer Studies 32 (2008) 526–533 © The Authors 529


Journal compilation © 2008 Blackwell Publishing Ltd
Perceived credibility of quality labels S. Moussa and M. Touzani

Table 3 Measurement model parameter estimates (N2 = 301) Table 4 Parameter estimates for the measurement models (N2 = 301)

Construct Path Estimatea tb rc r VCd Construct Path Estimatea tb rc r VCd

Perceived credibility of lX11 0.631 –e 0.861 0.509 Perceived credibility of lX11 0.648 –e 0.861 0.510
quality labels (x1) lX21 0.706 10.006 quality labels (x1) lX21 0.699 10.260
lX31 0.691 9.841 lX31 0.686 10.099
lX41 0.841 11.248 lX41 0.832 11.679
lX51 0.707 10.017 lX51 0.707 10.354
lX61 0.691 9.845 lX61 0.702 10.290
Perceived product lY11 0.818 –e 0.876 0.639
Overall goodness of fit indices
quality (h1) lY21 0.771 14.426
c2 = 24.868, d.f. = 9, P = 0.003, c2/d.f. = 2.763 lY31 0.821 15.595
Tucker–Lewis index = 0.963 lY41 0.787 14.806
Comparative fit index = 0.978 Product purchase lY12 0.740 –e 0.823 0.609
Root mean square error of approximation = 0.077 intention (h2) lY22 0.769 12.164
Standardized root mean square residual = 0.029 lY32 0.829 12.742
a
Standardized factor loadings computed via AMOS 4.01 maximum Overall goodness of fit indices
likelihood method.
b
c 2 = 137.420, d.f. = 62, P = 0.000, c 2/d.f. = 2.216
Student’s t-test of the null hypothesis that the parameter estimate is
Tucker–Lewis index = 0.950
zero.
c
Comparative fit index = 0.961
Jöreskog’s rho index of composite reliability.
d
Root mean square error of approximation = 0.064
Fornell and Larker’s average variance extracted.
e
Standardized root mean square residual = 0.044
This path was fixed to one for model identification.
a
Standardized factor loadings computed via AMOS 4.01 maximum like-
lihood method.
b
Student’s t-test of the null hypothesis that the parameter estimate is
zero.
c
Jöreskog’s rho index of composite reliability.
d
Fornell and Larker’s average variance extracted.
e
This path was fixed to one for model identification.

Figure 1 Theoretical model.


rVC(h2)], which suggests that the constructs each have more error-
free (extracted) variance than variance shared with other con-
structs (g112 and g212). As can be seen in Table 6, the results support
through a choice/decision process. Three items were used for discriminant validity.
this construct based on the studies of Erevelles et al. (2001) and Turning to the statistical estimates of the structural paths (see
Netemeyer et al. (2004).5 Table 5), we note that all the proposed relationships in the model
Figure 2 shows the structural model that was estimated through were statistically significant at the P < 0.01 level. The prediction
the AMOS 4.01 maximum likelihood method. The parameter esti- of a positive relationship between PCQL and PPQ was supported
mates corresponding to the empirical model appear in Tables 4 (g11 = 0.601, t = 8.116, P < 0.01). The relationship between PCQL
and 5. The relatively low value (<0.08) of the RMSEA and the low and PPI was also significant (g11 = 0.311, t = 3.994, P < 0.01). This
value (<0.05) of the SRMSR, together with the high values (>0.95) provides an explicit evidence of the predictive validity of the
of TLI and CFI, suggest that the hypothesized structural model PCQL scale. Furthermore, PPQ was also significantly related to
provides a good fit to the data (Bentler, 2007). PPI (b21 = 0.451, t = 5.721, P < 0.01).
As Table 4 demonstrates, all the loadings ranged from 0.648 to
0.832. The smallest t-value of the standard estimates was 10.099,
which indicates a high significance of the loadings. In addition, all Discussion
multi-item scales had composite reliability values ranging from
0.823 for PPI to 0.876 for PPQ. AVE rVC showed values ranging Summary and implications
from 0.510 for PCQL to 0.639 for PPQ. According to Fornell and The primary objective of this research was to refine and validate
Larker (1981, p. 46), rVC can also be used to evaluate discriminant a measure of the PCQL. Signal theory was used to delineate
validity. The requirements for discriminant validity are fully sat- the PCQL construct domain. Several advocated psychometric
isfied if the squared correlation between constructs (g112 and g212) is procedures were used, including item judging, exploratory
less than either of their individual AVEs [rVC(x1), rVC(h1) and and confirmatory factor analyses, internal consistency and
unidimensionality assessment, and convergent, discriminant and
5
A bilingual expert fluent in both English and French translated these predictive validity testing. Across pretests and studies that cover
scales. The verbal equivalence between the French and the English ver- over 600 respondents and examine three different quality labels
sions was checked through back-translation with the help of a second displayed by three products, consistent evidence of the validity of
bilinguist. This process was repeated until agreement was reached. the revisited PCQL measure was found.

530 International Journal of Consumer Studies 32 (2008) 526–533 © The Authors


Journal compilation © 2008 Blackwell Publishing Ltd
S. Moussa and M. Touzani Perceived credibility of quality labels

Figure 2 Empirical model.

Table 5 Structural path parameter estimates (N2 = 301) (or cues) akin to the total food quality model (Grunert, 2005) can
provide more understanding about effects of quality labels on
Independent Dependent
consumer behaviour before and after purchase.
variable variable Path Estimate t R 2a
Given the brevity of this measure, managers may find it useful
x1b h1c g11 0.601 8.116 0.361 to determine whether the quality label that their products are
x1 h2d g21 0.311 3.994 0.469 bearing or will bear is perceived as highly credible by consumers.
h1 h2 b21 0.451 5.721 This scale can also be used by retailers, magazines and third-party
a
Coefficient of determination for the system of equations. organizations to evaluate the perceived credibility of their own
b
Perceived credibility of quality label. labels against other quality labels present in the marketplace.
c
Product perceived quality.
d
Product purchase intention. Limitations and future research directions
Like most studies, ours is not without limitations. However, these
limitations raise questions that could be the subject of further
Table 6 Assessment of discriminant validity (N2 = 301)
research. First, our sample was convenience-based and was mainly
Construct x1 h1 h2 composed by students. Whether our results will generalize to all
a
the population is not known. We suggest that future replications
x1 0.510
could retest and calibrate the scale on random samples of consum-
h1 0.361b 0.639a
h2 0.096c 0.203d 0.609a
ers. Second, although we looked at three different quality labels,
all retained products were frequently purchased grocery goods.
a
Average variance extracted (r VC). Whether this measure would hold as well for quality labels asso-
b
Square of g11. ciated to durable goods, services or web sites has yet to be tested.
c
Square of g21. This scale could also be translated to other languages and extended
d
Square of b21. to other countries to see how the PCQL varies across cultures.
However, it is necessary to note that quality labels seldom
operate in isolation as manufacturers and sellers regularly use a
The above results have both theoretical and managerial impli- combination of signals to convey quality information. Understand-
cations. Theoretically, to our knowledge, this is the first study that ing the interactions between these signals can prove to be highly
offers a valid multi-item measure for the PCQL construct. As there useful (Miyazaki et al., 2005). Thus, future research is needed to
had been no published studies that validated such a measure with determine how quality labels operate in the presence of other
rigorous psychometric procedures, the results reported here should effective signals like brands (Erdem et al., 2006), warranties
be considered as part of a larger effort of trying to develop more (Sobermann, 2003) or prices (Rao, 2005).
dependable measures and empirical research about quality labels.
With this improved scale, relationships between PCQL and other
theoretically related constructs can henceforth be further studied.
Acknowledgements
In this spirit, a more comprehensive conceptual framework that The authors would like to acknowledge the support, comments and
deals with how consumers infer quality from a variety of signals suggestions of the guest editor and the two IJCS anonymous

International Journal of Consumer Studies 32 (2008) 526–533 © The Authors 531


Journal compilation © 2008 Blackwell Publishing Ltd
Perceived credibility of quality labels S. Moussa and M. Touzani

reviewers. The first author wants to thank Hamida Skandrani, Fraj, E. & Martinez, E. (2007) Ecological consumer behaviour: an
Karima Ghzaeil, Nadia Gouta, Hayet Ben Ayed and Hounaida empirical analysis. International Journal of Consumer Studies, 31,
Dhaou for their helpful comments during the first stages of this 26–33.
research. Gerbing, D.W. & Anderson, J.C. (1988) An updated paradigm for scale
development incorporating unidimensionality and its assessment.
Journal of Marketing Research, 25, 186–192.
Getz, C. & Shreck, A. (2006) What organic and fair trade labels do not
References
tell us: towards a place-based understanding of certification. Interna-
Akerlof, G.A. (1970) The market for ‘lemons’: quality uncertainty and tional Journal of Consumer Studies, 30, 490–501.
the market mechanism. Quarterly Journal of Economics, 84, 488– Grunert, K.G. (2005) Food quality and safety: consumer perception and
500. demand. European Review of Agricultural Economics, 32, 369–391.
Arbuckle, J.L. (1999) AMOS 4.01 [Computer software]. SmallWaters Hardesty, D.M. & Bearden, W.O. (2004) The use of expert judges in
Corp., Chicago, IL. scale development: implications for improving face validity of mea-
Arrow, K.J. (1963) Uncertainty and the welfare economics of medical sures of unobservable constructs. Journal of Business Research, 57,
care. American Economic Review, 53, 941–973. 98–107.
Beltramini, R.F. & Stafford, E.R. (1993) Comprehension and perceived Horn, J.L. (1965) A rational and test for the number of factors in factor
believability of seals of approval information in advertising. Journal analysis. Psychometrika, 30, 179–185.
of Advertising, 22, 3–13. Hovland, C.I., Janis, I.L. & Kelley, H.H. (1953) Communication and
Bentler, P.M. (2007) On tests and indices for evaluating structural Persuasion. Yale University Press, New Haven, CT.
models. Personality and Individual Differences, 42, 825–829. Iacobucci, D. (2001) Factor analysis. Journal of Consumer Psychology,
Bhattacharya, S. (1980) Nondissipative signaling structures and dividend 10, 75–76.
policy. Quarterly Journal of Economics, 95, 1–24. Jahn, G., Schramm, M. & Spiller, A. (2005) The reliability of certifica-
Bloom, P.N. & Reve, T. (1990) Transmitting signals to consumers for tion: quality labels as a consumer policy tool. Journal of Consumer
competitive advantage. Business Horizons, 33, 58–66. Policy, 28, 53–73.
Byrne, B.M. (2001) Structural Equation Modeling with AMOS: Basic Jarvis, C.B., Mackenzie, S.B. & Podsakoff, P.M. (2003) A critical
Concepts, Applications, and Programming. Lawrence Erlbaum, review of construct indicators and measurement model misspecifica-
Mahwah, NJ. tion in marketing and consumer research. Journal of Consumer
Caswell, J.A. (1998) How labelling of safety and process attributes Research, 30, 199–218.
affects markets for food. Agricultural and Resource Economics Karstens, B. & Belz, F.-M. (2006) Information asymmetries, labels and
Review, 27, 151–158. trust in the German food market. International Journal of Advertising,
Churchill, G.A. (1979) A paradigm for developing better measures 25, 189–211.
of marketing constructs. Journal of Marketing Research, 21, Kirmani, A. & Rao, A.R. (2000) No pain, no gain: a critical review of
64–73. the literature on signalling unobservable product quality. Journal of
Darby, M.R. & Karni, E. (1973) Free competition and the optimal Marketing, 64, 66–79.
amount of fraud. Journal of Law and Economics, 16, 67–88. Larceneux, F. (2001) Proposition d’une échelle de mesure de la
Dean, D.H. (1999) Brand endorsement, popularity and event sponsor- crédibilité d’un signe de qualité. DMSP Working paper n°289.
ship as advertising cues affecting pre-purchase attitudes. Journal of Larceneux, F. (2003a) Segmentation des signes de qualité: labels expéri-
Advertising, 28, 1–12. entiels et labels techniques. Décisions Marketing, 29, 35–46.
De Pelsmacker, P., Janssens, W., Sterckx, E. & Mielants, C. (2005) Larceneux, F. (2003b) Stratégie de la signalisation de la qualité:
Consumer preferences for the marketing of ethically labelled coffee. l’impact des labels sur le processus de décision des consommateurs.
International Marketing Review, 22, 512–530. Doctoral dissertation, University of Paris IX Dauphine, France.
Duhachek, A., Coughlan, A.T. & Iacobucci, D. (2005) Results on the Laric, M.V. & Sarel, D. (1981) Consumer (mis)perceptions and usage of
standard error of the coefficient alpha index of reliability. Marketing third party certification marks, 1972 and 1980: did public policy have
Science, 24, 294–301. an impact? Journal of Marketing, 45, 135–142.
D’Souza, C., Taghian, M., Lamb, P. & Peretiatko, R. (2007) Green deci- Lei, M. & Lomax, R.G. (2005) The effect of varying degrees of nonnor-
sion: demographics and consumer understanding of environmental mality in structural equation modeling. Structural Equation Modeling,
labels. International Journal of Consumer Studies, 31, 371–376. 12, 1–27.
Eisend, M. (2002) Dimensions of credibility in marketing communica- Llopis-Goig, R. (2007) Fair trade and global cognitive orientation: a
tion. Asia Pacific Advances in Consumer Research, 5, 366–373. focus on Spanish fair trade consumers. International Journal of
Erdem, T. & Swait, J. (2004) Brand credibility, brand consideration, and Consumer Studies, 31, 468–477.
choice. Journal of Consumer Research, 31, 191–198. McEachern, M.G. & Schröder, M.J. (2004) Integrating the voice of the
Erdem, T., Swait, J. & Valenzuela, A. (2006) Brands as signals: a cross- consumer within the value chain: a focus on value-based labelling
country validation study. Journal of Marketing, 70, 34–49. communications in the fresh-meat sector. Journal of Consumer
Erevelles, S., Roy, A. & Yip, L.C.S. (2001) The universality of the Marketing, 21, 497–509.
signal theory for products and services. Journal of Business Research, McEachern, M.G. & Warnaby, G. (2004) Retail ‘quality assurance’
52, 175–187. labels as a strategic marketing communication mechanism for fresh
Fabrigar, L.R., Wegner, D.T., MacCallum, R.C. & Strahan, E.J. (1999) meat. International Review of Retail, Distribution and Consumer
Evaluating the use of exploratory factor analysis in psychological Research, 14, 255–271.
research. Psychological Methods, 4, 272–299. Mishra, D.P. (2006) The role of certification in service relationships:
Fornell, C. & Larker, D.F. (1981) Evaluating structural equation models theory and empirical evidence. Journal of Retailing and Consumer
with unobservable variables and measurement error. Journal of Mar- Services, 13, 81–96.
keting Research, 18, 39–50. Miyazaki, A.D. & Krishnamurthy, S. (2002) Internet seal of approval:
Fotopoulos, C. & Krystallis, A. (2003) Quality labels as a marketing effects on online privacy policies and consumer perceptions. Journal
advantage. European Journal of Marketing, 37, 1350–1374. of Consumer Affairs, 36, 28–49.

532 International Journal of Consumer Studies 32 (2008) 526–533 © The Authors


Journal compilation © 2008 Blackwell Publishing Ltd
S. Moussa and M. Touzani Perceived credibility of quality labels

Miyazaki, A.D., Grewal, D. & Goodstein, R.C. (2005) The effect of Pornpitakpan, C. (2004) The persuasiveness of source credibility: a criti-
multiple extrinsic cues on quality perceptions: a matter of consis- cal review of five decades’ evidence. Journal of Applied Social Psy-
tency. Journal of Consumer Research, 32, 146–153. chology, 34, 243–281.
Moorthy, S. & Hawkins, S.A. (2005) Advertising repetition and quality Purohit, D. & Srivastava, J. (2001) Effect of manufacturer reputation,
perception. Journal of Business Research, 58, 354–360. retailer reputation, and product warranty on consumer judgments of
Moorthy, S. & Srinivasan, K. (1995) Signaling quality with money-back product quality: a cue diagnosticity framework. Journal of Consumer
guarantee: the role of transaction costs. Marketing Science, 14, 442– Psychology, 10, 123–134.
466. Rao, A.R. (2005) The quality of price as a quality cue. Journal of
Nelson, P. (1970) Information and consumer behaviour. Journal of Marketing Research, 42, 351–357.
Political Economy, 78, 311–329. Sobermann, D.A. (2003) Simultaneous signalling and screening with
Netemeyer, R.G., Krishnan, B., Pullig, C., Wang, G., Yagci, M., Dean, warranties. Journal of Marketing Research, 40, 176–192.
D., Ricks, J. & Wirth, F. (2004) Developing and validating measures Spence, M. (1973) Job market signalling. Quarterly Journal of Econom-
of facets of customer-based brand equity. Journal of Business ics, 87, 355–374.
Research, 57, 209–224. Stigler, G.J. (1961) The economics of information. Journal of Political
Ohanian, R. (1990) Construction and validation of a scale to measure Economy, 69, 213–225.
celebrity endorsers’ perceived expertise, trustworthiness, and attrac- Tang, E., Fryxell, G.E. & Chow, C.F. (2004) Visual and verbal commu-
tiveness. Journal of Advertising, 19, 39–52. nication in the design of eco-label for green consumer products.
Parkinson, T.L. (1975) The role of seals and certifications of approval in Journal of International Consumer Marketing, 16, 85–105.
consumer decision-making. Journal of Consumer Affairs, 9, 1–14. Worthington, R.L. & Whittaker, T.A. (2006) Scale development
Patil, V.H., Singh, S.N., Mishra, S. & Donavan, D.T. (2008) Efficient research: a content analysis and recommendations for best practices.
theory development and factor retention criteria: abandon the ‘eigen- Counseling Psychologist, 34, 806–838.
value greater than one’ criterion. Journal of Business Research, 61, Zeithaml, V.A. (1988) Consumer perception of price, quality, and value:
162–170. a means-end model and synthesis of evidence. Journal of Marketing,
Phelps, D.M. (1949) Certification marks under the Lanham act. Journal 52, 2–22.
of Marketing, 13, 498–505.
Ping, R.A. Jr. (2004) On assuring valid measures for theoretical models
using survey data. Journal of Business Research, 57, 125–141.

Appendix The French final version of the PCQL scale (seven-point Likert scale,
1 = completely disagree to 7 = completely agree)
PCQL1 Je peux me fier à ce que dit ce signe.
PCQL2 Ce signe émane d’un organisme ou d’experts reconnus.
PCQL3 Ce signe est honnête.
PCQL4 L’organisme qui remet ce signe a de bonnes intentions.
PCQL5 L’organisme a fait passer des tests sérieux avant de délivrer ce signe.
PCQL6 Ce signe m’inspire confiance.

International Journal of Consumer Studies 32 (2008) 526–533 © The Authors 533


Journal compilation © 2008 Blackwell Publishing Ltd

View publication stats

You might also like