Consumer Adoption of Fast-Fashion, Differences of Perceptions, and the Role of Motivations Across the Adoption Groups

The data were obtained through an online survey conducted from April to May, 2015 among a sample from the population of fast-fashion consumers in Turkey. The survey questionnaire consisted of four parts. In the first part, the sociodemographic characteristics and the purchasing habits of the participants were asked. In the second part, the DSI on a 5 point-Likert scale developed by Goldsmith and Hofacker [21] was used to categorize the participants. In the third part, the participants were required to rate their perceptions of fast-fashion including nine aspects by using a 5-point Likert scale. The aspects were modified from the literature [2,3,4, 26] pertaining to the fast-fashion products as being “in-style products”, “low-priced”, “having acceptable level of quality”, “frequent renewal of collections”, “opportunities towards socialization”, “association with brand names”, “imitating the luxury fashion products”, “expressing self-image”, and “allowing to act differently during purchasing”. In the fourth part, 16 items were used to explore the role of conformity, social or status image, and uniqueness on different consumer adoption groups with 4, 3, and 3 items adopted from studies of Eastman, Goldsmith and Flynn [27], Tian, Bearden and Hunter [28] and Goldsmith and Flynn [6], respectively.

In the present study, the responses of 244 participants were analyzed. The participants were first categorized based on the adoption groups proposed by Rogers [7] into “innovators”, “early adopters”, “early majority”, “late majority” and “laggards”. The questions were coded so that a high score reflected higher levels of innovativeness. The responses of the six questions were summed up and provided a DSI score. Accordingly, fast-fashion consumer groups were identified as “innovators”, “early adopters”, “early majority”, “late majority”, and “laggards” corresponding to the domain-specific scores of (30–26), (25–21), (20–16), (15–11), and (10–6), respectively.

The differences in the perceptions of fast-fashion among different consumer groups were investigated by using the nonparametric Kruskal-–Wallis test and Mann-Whitney U test by using IBM SPSS, v.23.0 statistical software at a 90% confidence interval.

To model the relationship between the consumer groups and influential motivational factors, the ordinal logistic regression method was used. Logistic regression for the ordinal response variable, which is also called as proportional odds model, was used to assess the influence of statistically significant explanatory variables on consumer groups using XLSTAT and IBM SPSS, v. ersion 23.0 software at a 95% confidence interval. The proportional odds model [29] can be defined [29] as represented in Eq. (1), as follows: ln(P(Y≤j|x)P(Y>j|x))=αj+(β1X1+β2X2+…βpXp)…j=1,………,J−1\ln \left( {{{P\left( {Y \le j|x} \right)} \over {P\left( {Y > j|x} \right)}}} \right) = {\alpha _j} + \left( {{\beta _1}{X_1} + {\beta _2}{X_2} + \ldots {\beta _p}{X_p}} \right) \ldots j = 1, \ldots \ldots \ldots ,J - 1

Within this model, j is the number of response levels of the response variable innovativeness, which corresponds to consumer groups, and p is the number of the explanatory variables, which corresponds to the motivational factors, “Conformity”, “Social or status image”, and “Uniqueness”. The probability of an equal or smaller response Y ≤ j is compared with the probability of a larger response, Y > j where j is the rank of the ordinal responses. P(Y ≤ j | x) is the cumulative probability for response level j when the explanatory variables take value x and a cumulative probability for Y is the probability that Y falls at or below a particular point. Consequently, j−1 linear predictors are obtained for comparing the response levels. The {a_j} are called cut point (intercept) parameters and they are non-decreasing in j [29].

Proportional odds mean that the effect of any explanatory variable for each response level must be consistent, or have parallel slopes, across all levels of the response. Therefore, as the first step in ordinal logistic regression, the test of parallel lines was done for evaluating the assumption that the relationships between the explanatory variables and the logits are the same for all linear prediction equations [30].

The significance of the explanatory variables was tested by using the likelihood ratio test, the Wald test, and the score test. In the likelihood ratio test, the model without any explanatory variable (baseline model) is compared against the model with all the explanatory variables (final model). The Wald test compares the maximum likelihood estimate of the slope parameter βi, to an estimate of its standard error [31]. The score test is a test of the null hypothesis in which the corresponding coefficients are located between two binary coefficients [32].

The relationship between the explanatory variables and the response levels was evaluated by calculating pseudo R² values. However, in the logistic regression analysis, pseudo R² is difficult to interpret. Instead, proportional odds ratios are more likely to be more useful [33,34,35]. Therefore, odds ratios, which represent the constant effect of an explanatory variable on the likelihood that one outcome will occur, were calculated. Finally, linear predictor equations, comparing the influences of the consumer groups on the adoption cycle that is based on innovativeness, were proposed.

Table 1 shows the sociodemographic profile and purchasing habits of the participants in relation to consumer adoption groups. The proportion of the participants in each consumer group was found to be consistent with the innovation/adoption curve of Rogers [7]; in this item in the literature, it has been demonstrated that the fashion is accepted and diffuses from innovators to early adopters, and then moves into a peak stage where a larger number of consumers begin to adopt the fashion. Subsequently, the diffusion process moves to a decline stage, and the number of adopters decreases.

Figure 1. demonstrates the adoption cycle of Rogers’ model and the fast-fashion consumers grouped in the present study. In Rogers’ model, innovators covered 2.5% of the population; on the other hand, early adopters, early majority, late majority, and laggards covered 13.5%, 34.0%, 34.0%, and 16.0% of the population, respectively [7]. In the present study, however, the ratios of innovators, early adopters, early majority, late majority, and laggards corresponded to 5.3%, 13.9%, 38.1%, 36.1%, and 6.6% of the participants, respectively. The results revealed that the first three fast-fashion consumer groups (i.e., innovators, early adopters, and early majority) covered more of the population when compared to the percentages of the population covered by the same categories in Rogers’ adoption curve. This can be attributed to the fact that the features of fast-fashion products satisfy the demands of the innovators and early adopters [24]. Moreover, the percentage of the innovators was greater, whereas that of the laggards was lower for the fast-fashion consumers when compared to Rogers’ model. It was also observed that the peak category for the fashion consumers belonged to the early majority with 38.1%, whereas, in Rogers’ model, the peak percentage again belonged to the early majority, though it was lower compared to the fast-fashion consumers and was equal to the percentage of the late majority. The higher rate of adoption of fast-fashion consumers is compatible with the short life cycle of fast-fashion products.

Figure 1

With regard to gender, it can be noticed that almost three-quarters of the innovators were female, whereas around 70% of the laggards were male (Table 1). As the fashion innovativeness of the participants increased, the number of females also increased. Another remarkable point is that the ratio of the participants with a “moderate to high” and “high” level of income was 23% for innovators, whereas other consumer groups had a lower ratio in terms of those income levels. Likewise, 92% of the innovators purchased clothes more than once a month, which was the highest percentage among all consumer groups. Concerning the purchase expense ratio, the ratio of the participants who allocated above 30% of their income increased as the fashion innovativeness increased. It can be concluded that female consumers are more innovative in the fast-fashion market. Moreover, as consumers’ level of innovativeness increases, there are more consumers at the postgraduate level, the level of income increases, the frequency of clothing for shopping rises, and the budget allocated for shopping for clothing increases.

According to the Kruskal-–Wallis test results, the perceptions of fast-fashion products including “being in-style products”, “expressing the self-image”, “imitating the luxury fashion products”, and “frequent renewal of collections”, differed significantly according to the consumer groups (p < 0.10). These results are consistent with those in previous studies. Fashion innovators were defined as consumers who are more interested in fashion and shopping and use clothing to express their individuality [14, 15, 18, 23, 36]. In addition to the primary motivation of feeling different, expressing their self-image through usage of innovative apparel was a concern for the fashion innovators. Further, fast-fashion is identified as “allowing dreams of luxury to come true” [4]. The differences among the perception of the fast-fashion products “imitating the luxury fashion” among the consumer groups can be attributed to the fact that high fashion innovators have a willingness to possess unique luxury products, and fashion innovators are the key luxury fashion consumer group [37]. Finally and importantly, high fashion innovators are the earliest consumers for innovations [38].

When all consumer groups were compared in pairs by using the Mann-–Whitney U-test in terms of consumers’ perceptions, it was observed that the fashion innovators’ perception of fast-fashion products in regard to “imitating the luxury fashion products” differed from the perceptions corresponding to all other consumer groups. The perception of the laggards towards fast-fashion products in terms of “frequent renewal of collections” was found to be significantly different from the other consumer groups. In other words, frequent renewal of collections was of no concern to laggards. Moreover, both the consumer groups of the “late majority” and the “laggards” did not perceive fast-fashion as a tool for “expressing the self-image”. No significant difference was observed between other paired comparisons of consumer groups.

The three explanatory variables (“conformity”, “social or status image”, and “uniqueness”) were used as explanatory variables in the ordinal logistic regression analysis. These three explanatory variables were used to obtain linear prediction equations that compared the categorized response variable, which, in the present study, was the influence of the consumer groups on the adoption cycle that is based on innovativeness.

The chi-square value obtained as the result of the test of parallel lines, which evaluated the assumption that the relationships between the explanatory variables and the logits were the same for all the logits, was 11.281 (df = 9) with a p-value of 0.257 (>0.05). This means that the categories of the response variable, which was the influence of the consumer groups on the adoption cycle that is based on innovativeness, were parallel with each other. In other words, the coefficients belonging to the explanatory variables (Estimate (b) values in Table 3) were equal for each consumer category.

The model-fitting information regarding the likelihood ratio test, the Wald's test, and the score test are given in Table 2. The chi-square values of the Llikelihood ratio test, the Wald's test, and the score test were obtained as 55.577 (df = 3), 50.996 (df = 3), and 55.345 (df = 3), respectively, with the p-values of 0.000. Thus, the explanatory variables, as a group, contributed significantly to the prediction of the response variable innovativeness.

In the next step, the effect and the significance of the explanatory variables were evaluated. According to the results of the final ordinal logistic regression model employed with three explanatory variables, the p-values of 0.000, 0.000, and 0.437 for the explanatory variable 1 (social or status image), the explanatory variable 2 (uniqueness), and the explanatory variable 3 (conformity), respectively, were obtained. This indicated that two explanatory variables, “social or status image” and “uniqueness”, were significant on the influence of the consumer groups on the adoption cycle that is based on innovativeness, confirming the earlier findings in the literature [8]. The significant relationship between “social or status image” and fashion innovativeness can be attributed to the fact that consumers who are willing to express their social or status image through their clothing are adapting faster to new trends, styles, or fashion products. Watson and Yan stated that the unique style characteristics of fast-fashion should be the complement of the self-image [39]. They also indicated that fast-fashion consumers enjoy purchasing trendy products at affordable prices; this finding was confirmed by Bagwell and Bernheim, who highlighted emphasized that those consumers having the willingness to seem wealthy purchase a large number of products at low prices [40]. The other significant motivation underlying fashion innovativeness was “uniqueness”. Previous studies have shown that to look unique, fast-fashion consumers prefer purchasing scarce products that are produced in low quantities [4]. Moreover, Jordaan and Simpson concluded that 25.8% of fashion innovativeness can be explained by the variation in the need for fashion uniqueness [20]. The insignificance of “conformity” was attributed to the fact that the individuals who prefer fast-fashion products did not face a need to display conformity with others [8]. Innovator consumers are likely to throw away their clothing to avoid conformity [41]. As can be seen in the fast-fashion adoption curve, fast-fashion consumers are more innovative, and differentiation motivations (such as the need for uniqueness and expression of their social or status image) were influential rather than socialization forces (such as being in conformity with their social circle) [19].

To assess the model for possible numerical errors, 95% of the confidence intervals were checked. For the significant explanatory variables, smaller confidence intervals suggest greater precision of the variable. Since the 95% confidence intervals for two significant variables were found to be 0.507 and 0.491, which are lower than 1, a lack of significance was not expected [31, 42]. The parameter estimates are shown in Table 3. As there were five consumer groups, four intercepts were obtained.

The proportional odds models, i.e., linear predictor equations, comparing the influences of the consumer groups on the adoption cycle that is based on innovativeness, are given below. Since there were five consumer groups, four linear prediction equations (Eqs (2)–(5)) were obtained.

The coefficient estimates of the explanatory variable “social or status image”, with a value of 0.790, showed that it had the most significant relationship with the consumer category, followed by “uniqueness” with the coefficient estimate value of 0.550.

Three different pseudo R² summary statistics that can be used to assess model fit were obtained as follows: Cox and Snell, 0.203; Nagelkerke, 0.218; and McFadden, 0.084. The pseudo R² summary statistics in the present study would be considered somewhat weak. Instead, proportional odds ratios were calculated to understand the effect of the variables on the categories of the response variable innovativeness. The Exp (β) is the exponentiated beta, or the proportional odds ratio, which assesses the effect of the variable for any of the comparison categories, are the same regardless of where the intercept is made [43]. In ordinal logistic regression, an odds ratio >1 leads to an increase in the level or category of the response variable, whereas an odds ratio <1 leads to a decrease in the level or category of the response variable [44]. The odds ratio of the “social or status image” was 2.203. This means that a one-unit increase in the “social or status image” variable increased the level of fast-fashion consumer adoption 2.203 times. Similarly, a one-unit increase in the “uniqueness” variable increased the level of fast-fashion consumer adoption 1.733 times. As a result, the innovation level of consumer adoption groups, which is a categorical variable, can be strongly explained by these two motivations.