تحلیل عاملی تأییدی موجودی سبک مشتری برای محصولات ورزشی (PSISP)
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|20803||2013||11 صفحه PDF||سفارش دهید||7260 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Sport Management Review, Available online 19 September 2013
Consumers are bombarded every day by numerous promotion messages, and their decision making in purchasing sport goods or services is frequently confused by these advertised information (Lysonski, Durvasula, & Zotos, 1996). For this reason, research in consumer decision-making styles has become increasingly popular. In 2009, Bae, Lam, and Jackson developed the Purchaser Style Inventory for Sport Products (PSISP) to identify consumers’ shopping behaviors. However, the PSISP was exploratory in nature. The purpose of this study was to validate the PSISP using confirmatory factor analysis. Participants (N = 455) were college students in the southern region of the United States. Fit indices (e.g., CFI = .92, SRMR = .068, RMSEA = .065: 90% CI = .062; .068) indicated the model provided reasonable fit to the data. After model respecification, the 37-item PSISP-II model significantly (p < .001) improved and included nine latent factors: Quality, Brand, Fashion, Recreation, Price, Impulse, Confusion, Habit, and Endorsement. It was concluded that the PSISP-II was a reliable scale in measuring consumer decision-making styles in purchasing sport products.
In the United States, sport is a $422-billion industry (Plunkett, 2011). The sales of sporting and fitness equipment, sports apparel, licensed merchandize, and athletic footwear in 2010 were over $74 billion (Sporting Goods Manufacturers Association, 2011). This was a 3.5% increase over 2009, the largest one-year growth in the sporting goods industry in nearly 20 years; and for the first time, the sporting goods industry outperformed gross domestic product (GDP) since 2007. To compete for the market share, companies offering similar products or services with their competitors try all possible means to advertise and promote their merchandises to the consumers. Such advertisements and promotions are not limited to the “traditional” forms of delivery methods (e.g., television commercials, direct mailing, newspapers, and magazines), but are also extended to electronic formats (e.g., websites and e-mails). Recently, marketers are more creative in how they reach the consumers – the use of mass media. Twitter, Facebook, Texting, Shutterfly, RueLaLa, and Groupon are just some of the ways in which marketing and advertising messages are being transmitted and connected to the potential consumers. On the other hand, consumers are bombarded every day by numerous promotion messages, and their decision making in purchasing sport goods or services is frequently confused by these advertised information (Lysonski, Durvasula, & Zotos, 1996). Since consumers’ decision-making styles in shopping play an important role in consumer behavior studies, individual shopping styles have been determined and applied in developing marketing segmentation (Tai, 2005). In the last decade, research in consumer decision-making styles, in particular consumer shopping behaviors for sports products, has become increasingly popular (e.g., Bae et al., 2009 and Kwon and Armstrong, 2002). Nevertheless, very few, if any, studies have examined the psychometric properties of the scales that measure consumer decision-making styles. Without a sound and valid scale, the results and implications of the research studies cannot be warranted.
نتیجه گیری انگلیسی
The purpose of this study was to validate the PSISP scale using confirmatory factor analysis. The original PSISP has 42 items. After model respecification, five items were removed and the final PSISP-II model included 37 items under nine dimensions: Quality (4 items), Brand (4 items), Fashion (5 items), Recreation (5 items), Price (3 items), Impulse (3 items), Confusion (4 items), Habit (3 items), and Endorsement (6 items). Overall, the inter-factor correlation coefficients (see Table 3) indicated there was no strong relationship among the nine factors. For this reason, a second-order model was not proposed or tested during the model respecification process. The CR of all nine factors was above the .70 standard. This indicated that all items collectively contributed a good overall reliability of the factor. Likewise, the VE showed that reasonable variances were extracted by each factor. In fact, the VE indicates the proportion of variance that is explained by an underlying factor in relation to that due to measurement error. For instance, the VE of the Quality factor was 0.78, meaning that 78% of the variance is explained by the Quality factor, while only 22% is due to measurement error. Nevertheless, the VE is a rather conservative estimate, it is sometimes acceptable even if the value is below .50 (Hatcher, 1994). Based on the goodness-of-fit indices of the CFA, the 37-item modified PSISP-II (see Appendix A) provides reasonable fit to the data. In spite of this, the model is still not perfect; and there is room for improvement. Previous researchers indicated that the fit of a model is affected by, among other things, its complexity and specification (Bollen and Long, 1993, Gerbing and Anderson, 1993 and Kaplan, 2000). Fan, Thompson, and Wang (1999) classified their four-latent-variable model (with three to four indicators per latent variable) as “moderate complexity” (p. 63). In fact, most researchers using structural equation modeling involved two to six latent variables, with about two to six indicators for each latent variable (Gerbing & Anderson, 1993). Based on this standard, the nine-factor PSISP model can be considered as high complexity, which may hamper its model fit. On the other hand, using too few indicators per latent variable is inappropriate. In their Monte Carlo study, Anderson and Gerbing (1984) found a greater chance of nonconvergence and improper solutions with two indicators per factor, especially with small sample sizes (e.g., N < 150). MacCallum (1995) pointed out that models with low numbers of parameters relative to the number of measured variable variances/covariances were highly disconfirmable, and that “for such models, bad fit to observed data is entirely possible” (p. 30). On the other hand, the structural parameters were unbiased when the models have three or more indicators per factor ( Gerbing & Anderson, 1985). Viewing this, the PSISP-II maintains at least three indicators per latent variable during the entire scale development process ( Loehlin, 1998). This study also examined gender differences among the nine dimensions of the PSISP-II. The results showed that females had significantly higher mean scores in Fashion, Price, and Habit than the male. This means female participants are paying more attention to the fashion trend and styles, more price conscious, and have a higher level of brand loyalty than their counterparts. One interesting finding is the gender differences on the Impulse dimension. Though the factor was not significant at the .05 level, it was significant at the .10 level (see Table 5). So, it is still safe to say that women shoppers are more likely than men to make their purchase by impulse. The results of this study are consistent with previous research in comparing shopping behaviors between female and male consumers. For example, Sondhi and Singhvi (2006) found that women tended to be relying more on the appearance (e.g., look and feel) of the products than men when shopping for clothes. According to Birol and Nuri (2007), price sensitivity was more common in females than in males. In fact, Underhill (1999) pointed out 86% of women looked at the price tags when they shopped, but only 72% of men did. On the other hand, Chen, Green, and Miller (2008) suggested that “very good after sales services” could increase brand loyalty of female consumers; while Rocereto and Mosca (2012) study concluded that pleasurable emotions toward hedonic products had a stronger driving force to brand loyalty behaviors for women than for men. In light of previous research and the findings of this current study, different marketing strategies are necessary when targeting female shoppers in order to maximise profitability. Bear in mind that women are more involved in fashion and are interested in shopping and shop more often for clothing in general than men (Browne and Kaldenberg, 1997, Chiger, 2001 and Flynn et al., 2000). Women also like to spend more time shopping and demand more spacious and pleasant shopping surroundings than men do (Underhill, 1999). Retail store managers should be knowledgeable and understand gender differences so that they can tailor their marketing practices accordingly. For instance, an open concept with a relaxing shopping environment is important for female shoppers. Managers targeting women consumers should also provide stylish up-to-date trend clothing with competitive price, and pay attention to their sales services. In conclusion, the PSISP has sound psychometric properties and can be used to assess consumers’ purchase style across different samples. According to Jöreskog and Sörbom (1993), a fit model does not necessarily mean a correct or best model because there may be many equivalent models as determined by the fit indexes. This is true for the PSISP model because it is only in its initial stage. Future researchers should reexamine the PSISP with samples from different countries to further examine its factor structure and invariance across gender, race, etc. In addition, further examination of the psychometric properties, such as the convergent and divergent validity of the PSISP is required. For example, the PSISP could be compared to other similar scales to see whether they were developed with the same degree of emphasis on scale construction and specificity for the consumer behavior setting (i.e., convergent validity).