Professional Documents
Culture Documents
Dijksterhuis NPD5reasons TIFN2016
Dijksterhuis NPD5reasons TIFN2016
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/233358244
CITATIONS READS
9 106
1 author:
Nina Michaelidou
Loughborough University
50 PUBLICATIONS 844 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Nina Michaelidou on 29 August 2015.
The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
A Typology of Consumers’ Variety-seeking Disposition based on Inherent Needs
Nina Michaelidou
Birmingham Business School
University of Birmingham
Edgbaston
Birmingham B15 2TT
United Kingdom
Tel: + 44 (0)121 414 8318
Fax: + 44 (0)121 414 7791
Email: N.Michaelidou@bham.ac.uk
1
Abstract
Consumers’ variety-seeking ‘has preoccupied many researchers in the last three decades and
many explanations and models have been produced. The concept is relevant to market
segmentation but is has not been fully explored. This study explores variety-seeking in order
to derive a consumer typology based on the strength of the disposition to seek for variety, and
the intrinsic needs that trigger this disposition. A quantitative methodology is adopted using a
sample comprising customers of a large UK-based clothing retailer. Factor and cluster
analyses are used to derive a variety-seeking disposition typology. Findings indicate four
and the build-in needs that drive variety-seeking disposition. Clusters also discriminate in
brand commitment, interest in fashion, age, gender, education and occupation. A number of
implications derive from this study in relation to targeting and marketing strategy.
2
Background
products, individuals make their brand selections on the basis of the variety they seek (Orth,
2005). Variety-seeking has preoccupied many researchers in the last three decades and many
explanations and models have been produced. Early research concentrated on modelling
experience) caused by satiation with product attributes (e.g. Farquhar & Rao, 1976; Hagerty,
1983; Huper & Reibstein, 1978; Jeuland, 1978; McAlister, 1982). Such models are expressed
as Markov processes using ‘transition probabilities’, each stating the probability of choosing
a particular brand B i directly following the selection of brand B j (Feinberg, Kahn &
McAlister, 1992, 1994; Givon, 1984; Kahn, Kalwani & Morrison, 1986; Lattin & McAlister,
1985; Trivedi 1999). Other models are ‘variable order’ and pattern consumption types in that
they allow for different types of behaviour such as pure inertia, pure variety-seeking and
The basic premise of these models is that repeat consumption of initially preferred products
or product attributes (e.g. flavour, colour, format, etc.) declines over time, leading to satiation
(McAlister, 1982). Thus, variety-seeking is the result of individuals feeling satiated with
consuming the same product attributes (e.g. physical and sensory) including format, flavour,
odour and colour (Inman, 2001; McAlister, 1982). Satiation with a product or its attributes is
thus one of the many underlying causes of variety-seeking which can be classified as being
either internal or external (Hoyer & Ridgway, 1984; Kahn, 1995; Van Trijp, Hoyer & Inman,
1996).
3
Other internal causes of variety-seeking highlighted in the literature include the inability of
consumers to make the right choice (Celsi & Olson, 1988; Huber & Reibstein, 1979), their
attempt to balance product attributes to maximize utility (Farquhar & Rao, 1976), multiple
purchases for future consumption (Kahn, 1995; Kahn & Lehman, 1991; Simonson, 1990),
consumers’ attempts to simplify the shopping task, and consumers’ optimal stimulation level
(Hoyer & Ridgway, 1984; Orth, 2005; Orth & Bourrain, 2005a; Raju, 1980; 1983; Van Trijp
& Steenkamp, 1992). On the other hand, external causes include out-of-stock situations and
special offers or promotions which are often used to stimulate varied purchases (Kahn &
Louie, 1990; Trivedi & Morgan, 2003; Van Trijp et al., 1996). However, the dominant view
in consumer research is that consumers tend to seek variety due to internal reasons such as
satiation with a brand’s attributes or the need for stimulation (McAlister & Pessemier, 1982).
behaviour (e.g. brand switching) focusing mostly on consumption histories and purchase
sequences (e.g. Trivedi, 1999). However, these approaches do not capture ‘true’ variety-
seeking’ (McAlister & Pessemier, 1982) since they focus on observed types of behaviour
& Baumgartner, 1992). They are therefore not ‘stable’ over time in terms of indicating
psychological causes of variety-seeking such as optimal stimulation level (e.g. Rohm &
risk taking (e.g. Givon, 1985; Orth, 2005; Raju, 1980; Trivedi, 1999), although research has
4
not fully explored variety-seeking as a segmentation base on its own. Yet, in terms of
promotional approaches. For example, high variety seekers would be easily induced to switch
to an alternative brand or to try a new product (Kumar & Trivedi, 2006; Trivedi & Morgan,
2003), whereas low variety seekers tend to be brand loyal and resistant to adopting new
products (e.g. Mittelstaedt, Grossbart, Curtis & Devere, 1976). Similarly, Van Trijp and
Steenkamp (1992) support the view that variety-seeking serves as a criterion for
segmentation, and assists marketers in adapting marketing strategies to consumer needs more
effectively.
Although previous research has profiled variety seekers (e.g. Rohm & Swaminathan, 2004),
causes of variety-seeking, specifically by considering optimal stimulation level (OSL) and its
inherent needs, in order to cluster consumers. Clustering individuals using OSL and its ‘built-
in’ needs provides a better and fuller psychological explanation and profile of variety-seeking
inherent needs, are likely to differ in attitudes and behaviour. The approach used in this study
stems from the field of environmental psychology (Mehrabian & Russell, 1974) which
individuals’ consumption experiences (Eliashberg & Sawhney, 1994; Van Trijp &
personality traits (e.g. dogmatism) which affect its scope and levels (Hoffman & Novak,
5
Psychographic segmentation has been explored in C2C and B2B contexts, and for both online
and offline retailing (e.g. Barry & Weinstein, 2009; McGoldrick, Keeling & Beatty, 2008;
variables such as personality traits, dispositions, values and lifestyle have been used to
segment consumers into clusters (Hassan, Craft & Kortam, 2003; Honkanen, Olsen &
Myrland, 2004; Wedel & Kamakura, 1999). These variables or ‘bases’ were developed and
are used for segmentation ‘…in response to the need for a more lifelike picture of consumers
and a better understanding of their motivations’ (Wedel & Kamakura, 1999, p.11). In
particular, Mitchell & Bates (1998) classified UK consumers into segments based on their
decision making styles. The authors identified and profiled four segments (‘Trend Setters’,
suggesting that marketers should adapt their strategies when targeting individuals with
different decision making styles (Michell & Bates, 1998). Kucukemiroglu, (1997) classified
consumers on the basis of ethnocentrism, while Roddy, Cowan & Hutchinson (1996) profiled
consumers based on their attitudes to organic food. More recently, Volkov, Harker & Harker
similar fashion, Rigopoulou et al. (2008) classified Greek consumers into two segments (i.e.
used involvement, values, and brand associations, to segment individuals (e.g. Lockshin,
Spawton & Macintosh, 1997; Orth, McDaniel, Shellhammer & Lopetcharat, 2004; Ross,
2007).
6
Psychological variables such as personality and predispositions influence consumption
patterns and explain perceptions (e.g. intentions, risk) or behaviours (e.g. switching, loyalty,
etc.) (Hoyer & Ridgway, 1984; Singh, 1990), which have implications for marketing and
advertising strategy (Gilbert & Warren, 1995). This is equally true for general personality
characteristics have a much stronger relationship with behavioural patterns across product
categories. [For a review of segmentation bases see Dibb, Simkin, Pride, & Ferrell, 2006;
Hassan et al., 2003]. On this basis, given OSL’s relatively stable nature and its potential
relationship with the adoption decision process (Mittelstaedt et al., 1976), promotional
strategy (Trivedi, 1999) and brand choice (Orth, 2005), individuals’ variety-seeking caused
by OSL, represents a useful basis for clustering consumers. The concept of OSL and its
relationship with variety-seeking are discussed in the following section. The methodology
and analysis sections follow, which then lead to the findings and discussion.
McAlister & Pessemier, 1982; Steenkamp & Baumgartner, 1992). Environmental psychology
suggests that individuals have a need for an optimal stimulation level (OSL) which is based
on different OSL needs (change, risk, etc.) that are linked to personality traits such as
dogmatism and intolerance (Raju, 1980). OSL posits that individuals seek to maintain an
optimal level of environmental stimulation, and that any deviation from optimality will lead
7
stimulation (Eliashberg & Swahney, 1994; Hebb, 1955; Leuba, 1955; Raju, 1981; Steenkamp
& Baumgartner, 1992; Wood & Swait, 2002). Previous research has shown that individuals
exhibit variety-seeking dispositions in order to maintain their OSL (Raju, 1980; Steenkamp &
Baumgartner, 1992). Previous research advocates that OSL leads to variety-seeking (e.g. Orth
& Bourrain, 2005b; Vazquez-Carrasco & Foxall, 2006). Thus, on this basis, OSL is viewed
‘source’ (e.g. inherent needs such as need for change, need for risk, need for sensation, etc.
Hoffman & Novak, 1996; Steenkamp & Baumgartner, 1992) of variety-seeking dispositions,
and to determine how variety-seeking is manifested (e.g. switching, use innovation, etc.) in
specific product situations (Price & Ridgway, 1982). For example, individuals with high
variety-seeking disposition driven largely by the need for change, will be more likely to
switch between familiar product alternatives, while individuals with high variety-seeking
disposition driven by the need for risk will be more willing to innovate (e.g. use innovation,
Van Trijp & Steenkamp, 1992). Similarly, individuals with a high need for sensation will
place more emphasis on emotional criteria such as touch, feel and taste when evaluating
alternatives.
Methodology
The data was acquired through the use of a URL embedded questionnaire (Wilson & Laskey,
2003) which was used for cost effectiveness and speed of transmission (Kent & Brandal,
8
2003). The questionnaire was sent to a sample1 of three thousand individuals. The
shopping behaviour for clothing. Clothing was selected as a context because it is pertinent to
both men & women of all ages, and it is a product category where variety-seeking
clothing brands and 2) browse for clothing in multiple shops or channels, which indicates a
variety-seeking disposition. The sample was randomly derived from the customer database of
a large UK clothing retailer across their retail, catalogue and Internet operations. Potential
respondents received a URL address in the form of a hypertext link included in an e-mail
message sent by the retailer. The e-mail invited recipients to visit the webpage to complete
the questionnaire. A pilot test preceded the main launch, and a completion incentive in the
form of vouchers was offered. Five hundred and seven usable questionnaires were returned.
Although the sample derived from the customer database of a single clothing retailer, all
respondents had internet access, and 95 per cent shopped online for multiple products and
brands, resulting in 252 different brands being mentioned including designer brands, high
street brands, department store-own brands and lower-end brands. In addition, respondents
indicated purchases of clothing using multiple channels (e.g. high street, catalogues
Measures
Variety-seeking
1
The sample was randomly selected and was representative (thus externally valid) of the specific population under
consideration because it fulfilled a number of conditions according to Schillewaert et al. (1998). First, a central register of
the research population existed (customer database). Second, all members of the population had the chance to be selected
since contact information was available. Third, no non-response bias occurred when comparing late versus early
respondents.
9
Variety-seeking disposition was captured using an optimal stimulation scale from
environmental psychology (e.g. Joachimsthaler & Lastovicka, 1984; Mehrabian & Russell,
1974). This scale was used in order to capture more needs (e.g. risk, sensation) that trigger
variety-seeking disposition, since other scales (e.g. ‘change seeker index’ (Steenkamp &
Baumgartner, 1992)) capture only the need for change or novelty or sensation. However,
according to Wahlers and Etzel (1990), this scale had problems in terms of item loadings,
therefore it was decided to re-examine this scale before its use in the study. In line with
Churchill (1979), the scale was reduced, resulting in 25 items being retained from the original
Discriminating Variables
validity checks which validate their cluster solutions. Ketchen and Shook (1996) report that
22 per cent of the studies they reviewed2 assessed criterion-related validity through the use of
1997; Orth et al., 2004; Rohm & Swaminathan, 2004; Ross, 2007; Singh, 1990). Variety-
seeking disposition has been theoretically linked to a number of variables including brand
Huber & Leeflang, 2007; Hoyer & Ridgway, 1984). Additionally, Raju (1980) found that
individuals with high OSL levels tended to have more exploratory tendencies compared to
individuals with lower OSL levels (Raju, 1980). Thus variety-seeking often manifests via
unfamiliar products (Price & Ridgway, 1982). In line with previous research, this study uses
2
24% used tests on clustering variables, 38% did not specify, 4% used hold-out samples and 7% used expert opinion
(Ketchen and Shook, 1996).
10
behavioural, attitudinal and demographic variables to validate variety-seeking clusters.
Specifically, the following variables were used: interest/attitude towards clothing, measured
by the overall score of a 3-item scale adapted from Van Trijp et al. (1996) [Clothing interests
exploration through shopping [I like to shop around; I like window shopping; I have an
evaluate clothing brands], knowledge [I understand the features to evaluate clothing brands]
and commitment [I would choose another brand if my preferred brand is not available] were
measured by single items adapted from Jensen, Carlson and Tripp (1989). All items were
The sample consisted of 14% males and 86% females spread across the following age groups:
5.9% in 16-24, 45.6% in 25-34, 39.7% in 35-44, 7.6% in 45-54 and 1.3% over 55 years of
age.
Scale Reduction
Based on the procedure outlined by Churchill (1979), and using a convenient sample, the
length of the scale was reduced by examining the item to total correlations, and eliminating
those items with values significantly lower than other items (or close to zero). Twenty-five
items were retained from the original 40 (see Finn & Kayande, 2004). Measures were taken
to ensure the reliability and content/face validity of the scale in view of the reduction.
Although this scale has not been tested extensively in the literature, the authors (Mehrabian &
Russell, 1974) advocate that the scale can be summed up to assess a single trait, but also it
11
reflects five individual factors (needs) for which the authors only give examples in terms of
which items corresponded to each factor (p.42). The reliability of the reduced scale was thus
assessed, and indicated an alpha value of above .80. Additionally, the content validity of the
scale was also ensured in order to include items reflecting needs underlining optimal
stimulation level. Thus, the reduced scale comprised of 25 items capturing all five inherent
needs (change, risk, activity, sensation, novelty.) in line with Mehrabian and Russell (1974).
Prior to cluster analysis, the scale was subjected to principal components exploratory factor
analysis (EFC) with orthogonal rotation (Varimax). EFA was performed prior to CFA for
three reasons: 1) The scale has been reduced to 25 items 2) The authors of the original scale
(Mehrabian & Russell, 1974) do not provide adequate information in terms of which items
correspond to the factors and 3) Problems reported by Wahler and Etzel (1990) in terms of
cross loadings of items. The solution derived from EFA indicates 7 distinct and interpretable
factors, explaining 55% of the variance (Table 1). Items with loadings below .35 were
excluded from the analysis to enable a clearer interpretation of the factors (Lockshin et al.,
1997; Tabachnick & Fidell, 2001). Because the scale is composed of inherent needs, and
given that it can be summed up with a single variable (Mehrabian & Russell, 1974) there is
was no cross loading on factors above .35 that would have potentially complicated the
interpretation of the factors. Reliability analysis showed that coefficient alpha values range
Confirmatory factor analysis using AMOS 16.01 was subsequently conducted to further
assess scale dimensionality in line with Gerbing and Anderson (1988). Maximum Likelihood
12
was used as the method of estimation. Chi-square was significant ( χ 2 (244)=489, p<0.00),
however relative chi-square (i.e. divided by degrees of freedom) was lower than 2 as
recommended by Carmines and McIver (1981). Findings indicate an acceptable fit for the
seven factor structure (GFI= .931, AGFI=.908, TLI=.871, CFI=.895, RMSEA= .044). AVE
was calculated to gauge discriminant validity (Fornell & Larker, 1981). Average variance
extracted for each factor is: 0.91 for F1, 0.94 for F2, 0.93 for F3, 0.94 for F4, 0.94 for F5,
0.86 for F6 and 0.95 for F7, exceeding the benchmark of 0.5 (Fornell & Larcker, 1981).
Discriminant validity was established by comparing the AVE with the squared correlations
between the factors (Fornell & Larcker, 1981). Table 2 shows the factor loadings from the
CFA. All items have significant loadings on their corresponding factor, albeit some loadings
are <.60i.
The findings of the EFC and CFA show 7 distinct and interpretable factors and thus
differentiate from the 5-stucture of Mehrabian and Russell (1974), since they identify 2 new
needs. The factors represent inherent needs that underlie variety-seeking disposition, and are
likely to determine how variety-seeking disposition will be manifested (Mehrabian & Russell,
1974; Price & Ridgway, 1982). They are labelled as: need for novelty (1), need for
unpredictability (2), need for excitement (3), need for activity (4), need for risk (5), need for
13
Table 1 Rotated component matrix
Component
1 2 3 4 5 6 7
Osl1:I actively seek new ideas and experiences .671
Osl2:People wearing strange or weird clothing make me uncomfortable .794
Osl3:I prefer familiar people and places .616
Osl4:When things get boring I look for new and unfamiliar experiences .759
Osl5:I like to touch and feel a sculpture .805
Osl6:People view me as quite an unpredictable person .512
Osl7:I like to run through heaps of fallen leaves .791
Osl8:I sometimes like to do things that are a little frightening .702
Osl9:I prefer friends who are unpredictable .733
Osl10:I prefer an unpredictable life full of change to a more routine one .624
Osl11:I like surprises .539
Osl12:My ideal home would be peaceful and quiet .657
Osl13:I eat different kind of food most of the time .531
Osl14:As a child, I often imagined exploring the world .407
Osl15:I like novelty and change in my daily routine .584
Osl16:I am happiest when I feel safe and secure .615
Osl17:I do not pay much attention to y surroundings .694
Osl18:I like to go somewhere different nearly every day .426
Osl19:I do not like to have lots of activity around me .560
Oasl20:I am interested only in what I need to know .569
Osl21:I like meeting people who give me new ideas .615
Osl22:I would be content to live in the same house the rest of my life .437
Osl23:I like a job that offers variety and travel even if it involves some
.473
danger
Osl24:I avoid busy, noisy places .741
Osl25:I would enjoy risky sports such as mountain climbing or sky
.738
diving
14
Table 2 Standardized regression weights
Estimate P
osl15 <--- f1 .667 ***
osl7 <--- f6 .794 ***
osl21 <--- f1 1.082 ***
osl18 <--- f1 .511 ***
osl14 <--- f1 .505 ***
osl6 <--- f1 .412 ***
osl4 <--- f1 .640 ***
osl1 <--- f1 .546 ***
***
osl16 <--- f2 .957
***
osl10 <--- f2 .622 ***
osl9 <--- f2 .487 ***
osl20 <--- f3 .726 ***
osl17 <--- f3 .477 ***
osl13 <--- f3 .574 ***
osl11 <--- f3 .496 ***
osl24 <--- f4 .753 ***
osl19 <--- f4 .654 ***
osl12 <--- f4 .456 ***
osl25 <--- f5 .648 ***
osl23 <--- f5 .535 ***
osl8 <--- f5 .562 ***
osl3 <--- f7 .924 ***
osl2 <--- f7 .467 ***
osl5 <--- f6 1.070 .003
Cluster Analysis
Factor overall scores were then calculated to be used in the cluster analysis. The cluster
procedure involved 2 stages. The first stage concerned the internal validation. The data was
randomly divided into 2 subsets (e.g. Lockshin et al., 1997). The first subset was used to
generate the possible alternative cluster solutions using hierarchical cluster analysis with
Ward’s method (Dibb, 1998; Lockshin et al., 1997; Rohm & Swaminathan, 2004; Singh,
1990). The optimal number of clusters was determined by analysing and finding the largest
differences among clustering coefficients (Hair, Anderson, Tatham & Black, 1998). The
second stage involved using the second subset to conduct K-means cluster analysis using
15
three, four, and five cluster solutions, as indicated by the hierarchical cluster analysis in the
first subset. The initial centroids provided by the hierarchical analysis on the first subset were
used to parameterize the K-means analysis in the second subset (Punj & Steward, 1983).
Memberships from K-means analysis on the second subset were then compared with
memberships of a hierarchical cluster analysis on the second subset. The degree of agreement
between the K-means assignments [of the second subset] and the results of the hierarchical
analysis provides an indication of the stability of the solution (Punj & Steward, 1983). The
‘four’ cluster solution was chosen as the most appropriate solution in terms of stability and
reproducibility. The datasets were then combined and a final K-means cluster analysis was
conducted using the following procedure (Everitt, Landau & Leese, 2001). The final cluster
Clusters
1 2 3 4 ANOVA (F) p
Need for Novelty 2.95 2.29 3.02 2.70 48.096 .000
Need for Unpredictability 3.87 3.33 4.31 3.47 63.053 .000
Need for Excitement 2.35 1.81 2.69 2.73 57.701 .000
Need for Activity 2.72 2.46 3.52 2.81 52.190 .000
Need for Risk 3.73 2.46 3.88 2.46 157.943 .000
Need for Change 2.61 2.00 3.41 3.28 123.507 .000
Need for Sensation 3.68 2.06 2.42 2.92 103.621 .000
3.13 2.34 3.32 2.91
Overall VSD level
% 22.5 30.6 24.7 22.3
Cluster descriptors are based on overall scores. Scores range from 1 to 5 (high-low level).
Clusters
The strength of variety-seeking dispositions and the inherent needs are used to describe and
16
Cluster 1: This cluster includes 22.5% of the respondents. They have a medium level of
variety-seeking disposition which stems from various needs (Table 3). This group scores
second highest on need for excitement and need for change. Relative to the other clusters they
have a low score on need for risk and need for sensation. This cluster is named change
seekers.
Cluster 2: This cluster comprises the highest number of sample respondents (30.6%) and has
the highest variety-seeking disposition with the highest scores on all needs relative to the
other clusters. This cluster has the highest scores on all needs with a particularly high score
on need for excitement. Their high variety-seeking disposition is likely to be evident in their
Cluster 3: This cluster is the second larger cluster with 24.7% of respondents. Relative to the
other clusters, they have significantly the lowest level of variety-seeking disposition, in that
they score the lowest on five of the seven needs. It is interesting to note that this cluster
scores very low on the need for unpredictability, which indicates that respondents in this
cluster prefer to feel safe and secure and attempt to avoid unpredictability. They also score
low on need for risk, relative to the other clusters, thus they are labelled as risk and
unpredictability avoiders. In contrast, respondents in this cluster score second highest on the
Cluster 4: This cluster includes 22.3% of respondents and has the second highest variety-
seeking disposition level relative to the other clusters. Respondents in this cluster score
second highest on the need for risk and need for novelty. Irrespective of the other clusters,
and in terms of the strength of needs that drive variety-seeking disposition, this cluster scores
17
highest on need for risk and novelty seeking but lowest on need for unpredictability. They are
To establish the external validity of the cluster solution, criterion-related validity was
assessed using attitudinal, behavioural and demographic variables not included in the
shopping and exploratory tendencies. Reliability was established for attitude towards clothes
at a.74 and the discriminating variables were also assessed for discriminant validity.
Correlation values indicate low correlations ranging from .137 to .274 (p<.05) with the
exception of the value between attitude towards clothes and exploratory tendency 3 - ‘I have
Additionally, behavioural (frequency of shopping and amount spent) and demographic (age,
gender, education, occupation) variables were used to profile the clusters. Clusters were
validated via ANOVA and chi-square tests in line with previous research (Ketchen & Shook
1996; Lockshin et al., 1997; Orth et al., 2004; Roddy et al., 1996) ).
Findings of χ 2 tests for demographic variables show significant differences across clusters,
the sample is biased towards females, clusters differ in terms of the number of males
included, with clusters 2 and 4 having more males than clusters 1 and 3. Findings also show
that relative to the other clusters, clusters 2 and 4 are made up of more respondents between
18
the ages of 15 and 34. Cluster 3 has more respondents between the ages of 35 and 54. In
terms of education, clusters 2 and 4 have more respondents with university degrees compared
to clusters 1 and 3. Furthermore, relative to the other clusters, cluster 2 has more respondents
in education, medical services and middle management, while clusters 1 and 3 have more
Furthermore, ANOVA findings indicate that information search and exploratory tendencies (I
like to shop around; I like window shopping) do not discriminate the clusters. However,
significant differences were observed for exploratory tendency ‘I have an interest in fashion’
(F= 4.741, p<.05), attitude to clothes (F= 3.133, p<.05), frequency of clothes shopping
(F=4.480, p<.05) and brand commitment (F=2.928, p<.05). Findings also indicate that
knowledge (F=2.599, p<.10) and amount spent on clothes (F=2.400, p<.10) distinguish
clusters, but at 90% confidence intervals. Multiple comparisons were also calculated. Table 4
19
Table 4 Multiple comparisons
Brand commitment
1-3 -.250* .094 .040
Table 4 shows significant (p<.05) pairwise differences between clusters in terms of their
interest in fashion, brand commitment, frequency of clothes shopping, and attitude to clothes.
Clusters 1 (M 1.70, sd .959) and 4 (M 2.05, sd 1.12), and 2 (M 1.62, sd .962) and 4, differ in
terms of their interest in fashion, with clusters 1 and 2 having a significantly stronger interest
in fashion. Similarly cluster 1 (M 2.0, sd .95) has a significantly stronger attitude to clothes
compared to cluster 4 (M 2.37, sd .924). Significant differences were also observed in terms
.955), and 2 (M 4.21, sd .855) and 3, with clusters 1 and 2 tending to shop more frequently
than cluster 3. Clusters 1 (M 2.34, sd) and 3 (M 2.59, sd .767) also differ in terms of brand
brand. This is an interesting finding which shows that although both clusters 1 and 3 have
preferred brand. This may be explained by the fact that cluster 1 has a significantly stronger
20
General Discussion
This study uses the notion of optimal stimulation level to capture variety-seeking dispositions
(VSD) which is grounded on various intrinsic needs that determine how variety-seeking
specific product contexts (Hoyer & Ridgway, 1984). Variety-seeking has been linked to
market segmentation, although research in this domain is rather limited. Rohm and
Swaminathan (2004) found that overall, variety seekers are motivated by convenience in their
shopping and are interested in alternative shopping channels and brands. The authors also
report that variety seekers exhibit tendencies to plan their shopping trips. The authors’
findings are in line with previous research which highlights that variety-seeking is linked to
exploratory behaviours, as well as other constructs such as risk and behavioural intentions
Our findings complement this stream of research by arguing that a number of needs underlie
variety-seeking disposition via OSL that can be used to explain and profile variety seekers
fuller and better. We thus identify different variety-seeking segments with different levels of
inherent needs, indicating that individuals can be segmented and profiled, not only in terms of
the strength (or level) of their variety-seeking disposition or tendency (e.g. variety seekers vs.
variety avoiders), but also the inherent needs which drive variety-seeking dispositions. Four
distinct clusters were identified and profiled using additional demographic, behavioural and
attitudinal variables (e.g. age, gender, education, occupation, attitude to clothes, brand
commitment). Clusters were found to differ in terms of the level of their inherent OSL needs
for clothes. Additionally, findings show that demographic variables (e.g. age, gender,
21
education and occupation), as well as attitudinal and behavioural variables (e.g. attitude to
clusters. On the other hand, information search, knowledge and amount spent on clothes do
The findings of the cluster analysis show that the majority of respondents exhibit a medium
level of variety-seeking disposition (e.g. ranging from 2.34 for cluster 2 and 3.32 for cluster
3). This is in line with previous research (Berlyne, 1968), although it is interesting to
highlight the variation observed in terms of the strength of different needs in each cluster. For
example, cluster 2 has the highest need for excitement, whereas cluster 3 has the lowest level
of need or no need for unpredictability. Cluster 2 also has a high score on need for change,
whereas cluster 3 has the lowest score for change. Similarly, cluster 1 has the lowest score on
The variation of the level of needs observed is further validated by the demographic and
behavioural differences identified. For example, cluster 2 which has the highest level of
cluster 3 (35-54). Similarly, clusters 1 and 2 shop for clothes more frequently than cluster 3.
However, no significant differences were observed for exploratory purchase tendencies (I like
to shop around and I like window shopping) albeit clusters 1 and 2 significantly differ from
risk taking and vicarious exploration (Price & Ridgway 1982; Raju 1980)) and other
Additionally, research should also explore a more generalisable sample, and use additional
22
behavioural and demographic variables such the number of brands bought, decision criteria,
The study adds to the existing literature by exploring a consumer typology based on inherent
needs that trigger variety-seeking disposition (e.g. Mitchell & Bates 1998; Rohm &
Swaminathan, 2004). This typology reflects a perspective which offers a ‘fuller’ explanation
of variety-seeking dispositions based on OSL needs, which can be used to cluster individuals
and has implications for targeting and marketing strategy. Individuals whose variety-seeking
disposition is triggered by different strength of needs, for example the need for change, risk,
etc., are expected to exhibit significant differences in attitudes and behaviours as well as
exploratory behaviours. This study shows that clusters differ on attitude towards clothes and
and its ‘sources’ has implications for the development of marketing strategy, since VSD
offers additional information based on which marketers can devise strategies to target
different clusters of individuals. For example, our findings show that individuals with a
relatively high variety-seeking disposition compared to the other clusters (e.g., clusters 2 and
4) score high on need for novelty, activity and risk. Individuals in clusters 2 and 4 tend to
have positive responses towards experiences entailing novelty and risk, and are also likely to
have a higher level of risk tolerance compared to other respondents such as the respondents in
clusters 1 and 3. Individuals in clusters 2 and 4 therefore would be more eager to try new
products or services, since these individuals have a need for novelty, tend to be more
innovative and perceive less uncertainty and consequences in their purchases compared to
23
other consumers (Fiore, Lee & Kunz, 2004). In addition, individuals with a high need for
risk and novelty tend to switch brands or exhibit multiple loyalties, particularly in the context
of clothes shopping where individuals own a portfolio of brands (Michaelidou & Dibb,
2009).
In contrast, individuals in clusters 1 and 3 score low on the need for risk and unpredictability,
and hence likely are to be risk avoiders. Risk avoiders are likely to be later adopters of
innovations, as they tend to avoid situations with high uncertainty and unforeseeable
consequences. Compared to the other clusters, clusters 1 and 3 therefore would use larger
amounts of information in their decision making, and would have a higher propensity to be
brand loyal. This finding is useful to marketers in devising strategies to target such
individuals; marketers can acquire a better profile of consumers if they consider inherent
OSL needs such as risk, novelty and unpredictability as segmentation bases, since OSL needs
determine behaviours such as risk taking, brand switching or loyalty, and shape attitudes.
Hence, when targeting individuals with high need for novelty and risk, marketers should
stress the variety and novelty of their product assortment; they should direct any new
offerings to individuals with a high need for novelty and risk. In contrast, when targeting
individuals with a low need for risk and unpredictability, marketers should attempt to support
consumers’ decision making processes with large amounts of information, and explain the
Furthermore, cluster 1 scores very low on need for sensation, a finding which indicates that
these individuals do not place significant weight on decision criteria involving sensory input
such as taste, smell and touch. This is an interesting and useful finding for both companies
who want to promote their online shopping operation, but also for e-tailers. For example,
24
online supermarkets such as Ocado (www.ocado.com) would be interested in targeting
individuals who do not choose food on the basis of sensory appeal such as smell, taste and
texture. Individuals with low or no need for sensation are therefore ‘ideal candidates’ for
online shopping of experiential products including, but not limited to, groceries such as fresh
In contrast, individuals in clusters 2, 3 and 4 who score relatively high on need for sensation
will be more interested in sensory or physical criteria (e.g. taste, smell, texture) when buying.
Consider for example the purchase of a house deodorant. Individuals with a high need for
sensation will choose on the basis of the smell and aroma, while individuals with a low need
for sensation will be more interested in the functionality and effectiveness of the house
deodorant in terms of eliminating odours; they will therefore place less emphasis on the
aroma and smell. Thus, for individuals with a high need for sensation, sensation seeking is a
key part of their shopping experience, and they will place considerable emphasis on sensory
appeal when evaluating products or services. On this basis, individuals with a high need for
sensation are likely to choose to shop in retail stores, rather than other shopping formats such
as the Internet, or direct mail. This finding is useful to companies with a multi-channel
operation, as well as e-tailers in targeting individuals with online products and services, since
Moreover, consumers’ varied levels of OSL needs (e.g., sensation, risk, unpredictability,
excitement, novelty and change) assist marketers when devising promotional messages that
focus on specific product attributes (e.g., sensory or non-sensory) and formats (e.g., affective
or cognitive, emotional or informational appeals) and assist consumers to make decisions. For
25
example, high sensation seekers (such as clusters 2, 3 and 4) are likely to be more interested
in messages of high sensation appeal that trigger greater sensory and affective responses,
based messages. Messages that target high sensation seekers therefore should highlight the
sensory attributes of the product (e.g., taste, smell, texture, etc.) and focus on cues such as
music, colour, illustrations, and endorsements. On the other hand, when targeting individuals
with a low need for sensation (e.g. cluster 1), marketers should stress the functional
characteristics of products (e.g. price, durability etc.) rather than the sensory or emotional
attributes.
Similarly, individuals with a low need for risk, unpredictability, sensation and excitement
will be more interested in messages that contain functional information about the product
such as price, quality, warranty and availability. Marketers should therefore focus on using
rational appeals to target such individuals, including, for example, problems and solutions,
Overall, our findings have practical implications, particularly in terms of segmentation and
targeting, as well as for advertising strategy and the design of messages. Marketers can adjust
individuals, and particularly the OSL needs that drive their variety-seeking disposition.
26
REFERENCES
Barry, J., & Weinstein, A. (2009). Business psychographics revisited: From segmentation
340.
Baumgartner, H., & Steenkamp, J.-B., E.M. (1996). Exploratory consumer buying behaviour:
121-137.
Bawa, K. (1990). Modelling inertia and variety seeking tendencies in brand choice behaviour.
Berlyne, D.E. (1960), Conflict, arousal and curiosity. New York: McGraw-Hill.
Carmines, E.G., & McIver, J. P. (1981). Unobserved variables. In G.W. Bohrnstedt & E.F.
Borgatta (eds.) Social Measurement: Current Issues. Beverly Hills, CA: Sage.
Celsi, R.L., & Olson, J.C. (1988). The role of involvement in attention and comprehension
Dibb, S. (1998). Market segmentation: Strategies for success. Marketing Intelligence and
Dibb, S., Simkin, L., Pride, W.M., & Ferrell, O.C. (2006). Marketing: concepts and
Eliashberg, J., & Swahney, M.S. (1994). Modelling goes to Hollywood: predicting individual
Everitt, B. S., Landau, S. & Leese, M. (2001), Cluster analysis, (4th Edition), London:
Arnold.
Farquhar, P.H., & Rao, V.R. (1976). A balance model for evaluating subsets of multi-
27
Feinberg, F.M., Kahn, B.E. & McAlister, L. (1992). Market share response when consumers
Feinberg, F.M., Kahn, B.E., & McAlister, L. (1994). Implications and relative fit of several
Fiore, A. M., Lee S.-F., & Kunz G. (2004). Individual differences, motivations, and
willingness to use a mass customization option for fashion products. European Journal of
Fornell, C., & Larcker D.F. (1981). Evaluating structural equation models with unobserved
Gilbert, F.W., & Warren, W.E. (1995). Psychographic constructs and demographic segments.
Givon, M. (1984). Variety seeking through brand switching. Marketing Science, 3 (1), 1-22.
Hagerty, M.R. (1983). Variety seeking among songs which vary in similarity. In R. P.
Bogozzi & A.M. Tybout (Eds.), Advances in Consumer Research, 10, Ann Arbor MI:
Hair, J.F., Anderson, R.E. Tatham, R.L., & Black, W.C. (1998), Multivariate data analysis,
Hassan, S.S., Craft, S., & Kortam, W. (2003). Understanding the new bases for global
Hebb, D.O. (1955). Drives and the C.N.S. (Central Nervous System). Psychological Review,
62, 243-254.
28
Helmig, B., Huber, J.-A., & Leeflang, P. (2007). Explaining behavioural intentions towardco-
Honkanen, P., Olsen, S., & Myrland, O. (2004). Preference-based segmentation: A study of
meal preferences among Norwegian teenagers. Journal of Consumer Behaviour, 3 (3), 235-
250.
Hoyer, W.D., & Ridgway, N.M. (1984). Variety seeking as an explanation for exploratory
behaviour: A theoretical model. In T.C. Kinnear (Ed.), Advances in Consumer Research, 11,
Huber, J., & Reibstein, D. J. (1979). Attitude measures and choice frequency: Some pitfalls
to be avoided. In Attitude Research Plays for High Stakes, American Marketing Association
Proceedings Series.
Jensen, T.D., Carlson, L., & Tripp, C. (1989). The dimensionality of involvement: An
empirical test. In T. K. Sroll (Ed.) ,Advances in Consumer Research, 16, Provo, UT:
attribute space: Theory and evidence of hybrid behaviour. International Journal of Research
29
Joachimsthaler, E.A.. & Lastovicka, J.A. (1984). Optimal stimulation level-exploratory
Kahn, B.E. (1995). Consumer variety seeking among goods and services. Journal of
Kahn, B.E., & Lehmann, D.R. (1991). Modelling choice assortments. Journal of Retailing, 67
(3), 274-299.
Kahn, B.E., & Louie, T.A. (1990). Effects of retraction of price promotions on brand choice
behaviour for variety seeking and last purchase loyal consumers. Journal of Marketing
Kahn, B.E., Kalwani, M.U., & Morrison, D.G. (1986). Measuring variety seeking and
reinforcement behaviours using panel data. Journal of Marketing Research, 23, 89-100.
Ketchen, D.J., & Shook, C.L. (1996). The Application of cluster analysis in strategic
management Research: An analysis and critique. Strategic Management Journal, 17 (6), 441-
458.
Kent, R., & Brandal, H. (2003). Improving e-mail response in a permission marketing
Kumar, A., & Trivedi, M. (2006). Estimation of variety seeking for segmentation and
Lattin, J.M., & McAlister, L. (1985). Using a variety seeking model to identify substitute and
(3), 330-339.
30
Leuba, C. (1955). Toward some integration of learning theories: The concept of optimal
Lockshin, L. S., Spawton, A. L., & Macintosh, G. (1997). Using product, brand and
purchasing involvement for retail segmentation. Journal of Retailing and Consumer Services,
4 (3), 171-183.
McGoldrick, P.J., Keeling, K.A., & Beatty, S.F. (2008). A typology of roles for avatars in
Michaelidou N., & Dibb, S.C. (2009). Brand switching in clothing: The roles of variety
Mittelstaedt, R.A., Grossbart, S.L., Curtis, W.W., & Devere, S.P. (1976). Optimal stimulation
level and the adoption decision process. Journal of Consumer Research, 3 (2), 84-95.
Orth, U.R. (2005). Consumer personality and other factors in situational brand choice
Orth, U.R., & Bourrain, A. (2005a). Ambient scent and consumer exploratory behaviour: A
Orth, U.R., & Bourrain, B. (2005b). Optimum stimulation level theory and the differential
31
(Eds.), Advances in Consumer Research, 32, Valdosta, GA: Association for Consumer
Research, 613-619.
Orth, U.R., McDaniel, M., Shellhammer, T., & Lopetcharat, K. (2004). Promoting brand
benefits: The role of consumer psychographics and lifestyle. The Journal of Consumer
Price, L.L., & Ridgway, N.M. (1982). Use innovativeness, vicarious exploration and
purchase exploration: Three facets of consumer varied behaviour. In Walker et al., (Eds.),
Association, 56-60.
Punj, G., & Steward, D.W. (1983). Cluster analysis in marketing research: Review and
Raju, P.S. (1980). Optimal stimulation level: Its relationships to personality, demographics
Raju, P.S. (1981). Theories of exploratory behaviour: Review and consumer research
and implications. In R. P. Bogozzi & A.M. Tybout (Eds.), Advances in Consumer Research,
Rigopoulou, I.D., Tsiotsou, R.H., & Kehagias, J.D. (2008). Shopping orientation-defined
Roddy, G., Cowan, C.A., & Hutchinson, G. (1996). Consumer attitudes and behaviour to
Rohm, A.J., & Swaminathan, V. (2004). A typology of online shoppers based on shopping
32
Ross, S.D. (2007). Segmenting sports fans using brand associations: A cluster analysis,
Schillewaert, N., Langerak, F., & Duhamel, T. (1998). Non-probability sampling for www
Simonson, I. (1990). The effect of purchase quantity and timing on variety seeking
Singh, J. (1990). A typology of consumer response styles. Journal of Retailing, 66 (1), 57-99.
Steenkamp, J.-B. E.M., & Baumgartner, H. (1992). The role of optimum stimulation level in
Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics, (4th Edition), Boston:
Trivedi, M. (1999). Using variety seeking based segmentation to study promotional response.
Trivedi, M., & Morgan, N. S. (2003). Promotional evaluation and response among variety
Van Trijp, H.C.M., Hoyer, W.D., & Inman, J.J. (1996). Why switch? product category level
explanations for true variety seeking behaviour. Journal of Marketing Research, 33 (3), 281-
292.
Van Trijp, H.C.M., & Steenkamp, J.-B. E. M. (1992). Consumers’ variety seeking disposition
33
Vazquez-Carrasco, R., & Foxall, R. G. (2006). Positive vs. negative switching barriers: The
influence of service consumers’ need for variety. Journal of Consumer Behaviour, 5, 367-
375.
Volkov, M., Harker, M., & Harker D. (2006). People who complain about advertising: The
aficionados, guardians, activists and seekers. Journal of Marketing Management, 22, 379-405
Wahlers, R.G., Dunn, M.G., & Etzel, M.J. (1986). The congruence of alternative OSL
measures with consumer exploratory behaviour tendencies. In R.J. Lutz, (Ed.), Advances in
Consumer Research, 13, Provo, UT: Association for Consumer Research, 398-408.
Wahlers, R.G. & Etzel, M.J. (1990). A structural examination of two optimal stimulation
level measurement models. In M.E. Goldberg, G.Gorn, & R.W. Pollay (Eds.), Advances in
Consumer Research, 17, Provo, UT: Association for Consumer Research, 415-422.
Wedel, M., & Kamakura, W. (1999), Market segmentation: Conceptual and methodological
Wilson, A., & Laskey, N. (2003). Internet based marketing research: A serious alternative to
Wood, S.L., & Swait, J. (2002). Psychological indicators of innovation adoption: Cross-
classification based on need for cognition and need for change, Journal of Consumer
i
Additional analysis could have been performed to eliminate items, yet it was decided to proceed to cluster
analysis without losing further content from the scale since the fit indices are acceptable and the factor loadings
are all significant.
34