Decision-Making Styles of Young Malay , Chinese and Indian Consumers in Malaysia

Although consumer decision-making style represents a relatively consistent pattern of cognitive and affective responses, culture has been proven to impact significantly on individual values and attitudes, thus, culture is expected to have a significant influence on consumer decision-making styles. This paper investigates the decision-making styles of young Malay, Chinese and Indian consumers in Malaysia using Consumer Style Inventory (CSI) developed by Sproles and Kendall (1986). An examination of the psychometric properties of CSI shows some similarities and differences in consumer decision-making styles among the three ethnic groups, suggesting that ethnicity affects shopping behavior. Identification of new traits exclusive to each ethnic group suggests that CSI cannot be applied without considering the cultural factors.


Introduction
This study presents the findings from an exploratory study of consumer decision-making styles in Malaysia using the Consumer Styles Inventory (CSI) developed by Sproles and Kendall (1986).A consumer's decision-making style has been defined as "a patterned, mental, cognitive orientation towards shopping and purchasing, which constantly dominates the consumer's choices.[... ] these traits are ever-present, predictable, central driving forces in decision-making" (Sproles, 1985, p. 79).Sproles (1985) proposed that consumers adopt a "shopping personality" that is relatively enduring and predictable in much the same way as psychologists view personality in its broadest sense.
Based on his review of previous literature, Sproles (1985) identified 50 items related to consumers' cognitive and affective orientation towards shopping and buying.Employing a factor analysis technique, Sproles found that six out of nine traits were confirmed to be present.Sproles and Kendall (1986) refined this inventory and accordingly developed a more parsimonious scale consisting of 40 items.The Consumer Style Inventory (CSI) that they developed consists of eight mental characteristics of decision-making styles, as depicted in Table 1.
Whereas many studies investigated decision-making styles across different countries, to date, no research has been conducted to determine consumer decision-making styles across different ethnic groups within a national boundary.Radder et al.'s (2006) study is the only exception that examines decision-making styles of Chinese, Motswana and Caucasian students in South Africa.In Malaysia, two published studies on decision-making styles are identified, namely Mokhlis (2009) and Wan Omar et al. (2009), both of which sampled the student market.While these two studies provided important insights into consumer behavior within the specific context of Malaysia, the studies did not looked at ethnic variation in consumer decision-making styles.It is believed that ethnic groups in Malaysia may also have certain distinctive characteristics in terms of decision-making styles that could be of equal interest to both researchers and marketers.
The purpose of this study is to examine consumer decision-making styles of the three main ethnic groups in Malaysia namely Malay, Chinese and Indian.The aim is to demonstrate that consumer decision-making styles differs according to consumers' ethnicity and that consumer behavior can be predicted from an understanding of the cultural personality of consumers.

The Instrument
The questionnaire consisted of Sproles and Kendall's (1986) 40-item Likert scaled Consumer Style Inventory (CSI).The instrument researchers used was Consumer Style Inventory (CSI) developed by Sproles and Kendall (1986).Each item was answered by a 5-point Likert scale; ranging from 1 (strongly disagree) to 5 (strongly agree).The reliabilities of the CSI scale, according to Sproles and Kendall (1986), ranged from 0.48 to 0.76.The items were randomly ordered in a self-administered CSI instrument to counterbalance possible order effects.In addition, some demographic questions were included in the questionnaire.

The Subjects
The study was carried out in Malaysia.Among examples of plural societies, Malaysia shows an unusually balanced ethnic structure of two dominant groups, the Malays who make up 53.4 per cent and the Chinese who make up 26 per cent.As well there are 7.7 per cent Malaysians of Indian ethnic origin (Department of Statistics Malaysia, 2005).Due to the cultural differences that exist in the origins of different communities, there is a noticeable absence of homogeneity in the behavior of consumers in Malaysia where the nature of its domestic market is highly characterized by the "ethnically segmented consumer markets" (Mohd. Salleh, Teo & Pecotich, 1998, p. 481).Such unique characteristic provides a particularly appropriate context for this study from which a sample representing diverse ethnic consumer groups can be drawn from its population.
A convenience sample of 600 undergraduate students was selected from a public university with an enrollment of over 6,000 students in the state of Terengganu, Malaysia.Respondents were given a copy of the questionnaire to complete outside class hours, together with assurances regarding the anonymity and confidentiality of their data.They were informed that their participation was voluntary and could be withdrawn at any stage.Completed questionnaires were returned to the authors the following week.Of 560 students responded to the survey, 73 questionnaires were rejected.The final sample consisted of 487 undergraduate students.The ethnic breakdown is as follows: Malay, n = 260 (53.4 per cent); Chinese, n = 115 (23.6 per cent) and Indian, n = 112 (23 per cent).
Although student samples are not representative of all cross-sections of the population, they are considered appropriate for cross-cultural theory testing.Using a relatively more homogeneous group such as undergraduate students is particularly helpful to minimize random error that might occur by using a heterogeneous sample such as the general public (Calder, Philips & Tybout, 1981).This is because the likelihood of error within the measurement model being inflated by situational factors inherent in diverse samples (e.g.education, age, income and social class) is reduced when respondents are homogeneous across demographic and behavioral characteristics, thus resulting in less "extraneous variation" (Peterson, 2001).Sherman et al. (1999) also support the use of students as subjects within consumer research and confirm its acceptance by stating that 86 per cent of the articles published in the Personality and Social Psychology Bulletin used students as their human subjects.

Analysis and Results
This study utilized exploratory factor analysis (EFA) with principal component method to extract a small number of latent variables (factors) from a large number of observed variables (40-items on the CSI).The varimax procedure of orthogonal rotation, generally regarded as the "best" and most commonly used (Hair et al. 1998) was applied on the principal component solutions.This procedure was favored for this study since it minimizes correlation across factors and maximizes within the factors.Varimax procedure gives a clearer separation of the factors and has proven very successful as an analytic approach to obtaining an orthogonal rotation of factors (Hair et al. 1998).
To assess the internal consistency of each factor group obtained, a reliability analysis was conducted by calculating the Cronbach's alpha for each factor.The assumption behind this approach was that the items of a measure work together as a set and should be capable of independently measuring the same construct.The items should be consistent in what they indicate about the concept being measured.For consistency, it was decided that reliability should not lower than 0.5, the minimum acceptable level suggested by Kerlinger and Lee (2000).

Validation of the Sproles and Kendall's Eight Factor CSI Model
The 40-item CSI inventory was factor analyzed using data from the three samples.For the purpose of comparing the factor solution with Sproles and Kendall's (1986) findings, a constrained eight-factor solution was extracted.The 40 items were attributed to the respective factors according to the eight factors.Table 2 presents the results of factor analysis for the three samples compared with the results of the original Sproles and Kendall model.The solution explained a total variance of 47.3 per cent in the Malay sample, 55.1 per cent in the Chinese sample and 51.1 per cent in the Indian sample.These figures were higher than that of Sproles and Kendall (1986) whose eight-factor solution explained only 46 per cent.
In general, the eight-factor structure of the CSI was not completely replicated in all three Malaysian samples.Of the 40-item inventory, 57.5 per cent of the items in the Malay sample, 55 per cent in the Chinese sample and 45 per cent in the Indian sample loaded on the Sproles and Kendall (1986) specified factors.
Table 2 also shows the reliability coefficients across the three samples for various factors, as measured by Cronbach's alpha.Coefficient alphas for the eight-factor model across the three samples were generally low, with only three out of eight factors in each case having good internal consistency reliabilities.The Recreational, Price Conscious and Brand Loyal factors had poor alpha coefficients (below 0.5) across the three samples, which suggest that items did not measure these factors effectively and that scale items need re-designing to improve measurement of the trait.

Generation of a Modified CSI Model
In a bid to identify a more appropriate model for the Malaysian situation, the data were factor analyzed without constraining the factor solution.Factor models consisting of six, seven, eight and nine factors each were scrutinized to find the optimum solution.Only items that contextually fitted the factor and which had a factor loading of 0.4 and higher were included in the subsequent analysis.
Tables 3, 4 and 5 reports the results of EFA using principal component analysis with varimax rotation for the Malay, Chinese and Indian samples respectively.All factors had eigenvalues greater than one, which is a rule often used in judging the adequacy of the factor solution.Based on the factor analysis (factor loadings), the findings only incorporate the use of 25 of the original items for the Malay sample, 29 items for the Chinese sample and only 18 items for the Indian sample.Variations in the reliability (Cronbach alphas) for constructs also exist for each of the factors across the three samples.
For the Malay, an eight-factor solution was extracted with eigenvalues ranging from 1.15 to 4.09, which accounted for 59.1 per cent of the variance.Seven of the eight original CSI traits plus one new Malay factor were found (see Table 3).Except for factor 8 (Confused by Overchoice), the reliability coefficients were all above 0.5.
Factor 2, entitled Value Conscious, was identified as a new factor among this sample group.Four items loaded positively onto this factor, two of which from the Impulsive, Careless, and another two from Price Conscious constructs.This trait describes people who concern with how much they spend and would shop carefully to find the best value for money.The reliability for this factor was 0.5.
In the case of Chinese sample, an eight-factor solution was identified which accounted for 62.7 per cent of the variance and had a range of eigenvalues of 1.11 to 5.24.Seven of the eight original CSI traits plus one new factor exclusive to Chinese were found (shown in Table 4).Except for factors 7 (Price Conscious) and 8 (Careless), the reliability coefficients were all above 0.5.
Factor 6, entitled Shopping Avoidance, was identified as a new factor exclusive to this sample group.The factor is a combination of one item from the High-Quality Conscious and Recreational, Hedonistic constructs developed by Sproles and Kendall (1986).The sign of loadings for these two items was negative, suggesting the existence of consumers who dislike shopping and thus shop around very little because they make shopping trips fast.This factor approximates to an opposite of the Recreational, Hedonistic consumer trait.The alpha coefficient for this factor was 0.66.
A six-factor model best fitted the Indian sample.The factors had a range of eigenvalues of 1.16 to 3.04 and together accounted for 61.8 per cent of the variance.Only five of the eight original CSI traits plus one new factor were found (see Table 5).Except for factor 6 (Careless), the reliability coefficients were all above 0.5.
Factor 5, entitled Satisfying, was identified as a new factor exclusive to Indian sample group.Three items loaded positively onto this factor; two of which from the High Quality Conscious and one from the Recreational, Hedonistic constructs.This factor is best described by one item: "A product doesn't have to be perfect or best to satisfy me".Those scoring high on this factor could be expected to be willing to sacrifice quality in order to avoid spending much time shopping.The alpha coefficient for this factor was 0.53.
When looking at inter-ethnic similarities, five decision-making traits appear to be common across the three samples namely Fashion Conscious, Quality Conscious, Careless, Recreational and Confused by Overchoice.The reliability coefficients for these factors across the three sample groups were all above 0.5 with the exception of Careless for both the Chinese and Indian samples, and Confused by Overchoice for the Malay sample.

Conclusion
This study was a first attempt at verifying the generalizability of Sproles and Kendall's CSI across three ethnic groups within a Malaysian retail environment.The results revealed some interesting patterns in the decision-making traits of young Malay, Chinese and Indian consumers.Eight meaningful factors resulted for the Malay and Chinese samples, and six for the Indian sample.Five common decision-making traits (Fashion Conscious, Quality Conscious, Careless, Recreational and Confused by Overchoice) were confirmed across all three sample groups, albeit with different item loadings.In addition, three new decision-making traits were identified; one for each sample group, namely Value Conscious, Shopping Avoidance and Satisfying.
A fairly good results were obtained for the Malay and Chinese sample groups despite some items did not have acceptable scores to fit with both the eight-factor and modified models.The inventory however did not seem to be applicable to the Indian dataset, with only 18 out of 40 items loaded onto six factors.It appears that the CSI in its original configuration could not be applied to different cultures without modifications.The identified traits were therefore generalizable to some extent across populations, but did vary between cultural groups which support the general view that consumer decision-making processes are culturally dependent.To International market researchers and marketers, this finding poses a warning that instruments validated in one county on limited samples are not immediately applicable to other countries.The dimensions and items included in the inventory need to be tested before being used in another country setting or a modified model of decision-making traits may be necessary to more adequately account for the consumer behavior and retail environment of another culture (Walsh et al. 2001).
On a final note, this research was clearly based on the student consumer cohort aged 19-25 and thus the results did not represent the Malaysian population in general.Other segments, such as non-students, who may have different decision-making orientations, should be investigated.Also, additional studies comparing decision-making styles of other ethnic consumers as well as from different regions of Malaysia might produce interesting findings.The well-known national brands are best for me .733 The higher the price of a product the better its quality .722 The more expensive brands are usually my choice .696 The most advertised brands are usually very good choices .450 Factor

Table 1 .
Consumer style characteristics

Table 2 .
Consumer style characteristics: Eight factor model

Table 3 .
Factor analysis of the CSI: Malay sample

Table 4 .
Factor analysis of the CSI: Chinese sample

Table 5 .
Factor analysis of the CSI: Indian sample