Determinants of consumer-generated-content usage for apparel shopping: The moderating effect of gender

Abstract

This study determined the influence of perceived ease of use, perceived enjoyment, trustworthiness, knowledge and competence as potential determinants of consumer-generated-content usage for apparel shopping in a sample of young adult consumers. The data was obtained from 455 young adult social media users using an anonymous questionnaire and the model was tested through structural equation modelling (SEM) approach. The eleven hypotheses were empirically tested. The findings confirmed all hypotheses in a significant way, with the exception of H2, H3, H5 and H7. Important to note on the study results is that while perceived usefulness has positively affected the attitude, its relationship with the intention to use consumer-generated content is not significant. However, knowledge and competence strongly influence attitudes. The indirect effect of perceived usefulness on intention via attitude, the mediating impact of perceived usefulness on the relationship between perceived ease of use and intention, the mediating effect of knowledge and competence on the relationship between perceived enjoyment and attitude have all been confirmed. This study stands to add new knowledge to the present body of consumer behaviour literature in Africa—a context that is often ignored by academics in developing countries.

Keywords:

• perceived ease of use
• perceived enjoyment
• trustworthiness
• knowledge
• consumer-generated-content usage
• apparel shopping

PUBLIC INTEREST STATEMENT

The need to understand consumers’ behaviour and decision-making when buying clothes is increasingly becoming important. Therefore, the current study results provide tentative support for the proposal that perceived usefulness, attitudes, and trustworthiness of consumer-generated content should be recognised as significant social media usage drivers by young adults in South Africa. The results of this empirical study are expected to have a positive effect on both practitioners and academics. For academics, the study offers current insights into consumer-generated content within social media communities. From the practitioners’ perspective, this study suggests that, for platform and marketing managers to leverage social media in retail and fashion marketing, they should pay attention to customer decision-making processes that affect the use of user-generated content (CGC) for online fashion shopping. It is also of the utmost importance to emphasise the consistency, reliability and expertise of the content thus produced.

1. Introduction

The advent of social media has witnessed a dramatic increase in online interaction and digital WOM communication (Nambisan et al., 2017). Marketers have also son the phenomenon and initiated brand communities on social media sites to engage consumers and facilitate and generate “buzz” to increase information sharing and, consequently,” drive sales. Social media’s growth and the proliferation of consumer-generated content have significantly changed consumer behaviour and online marketing strategies. Contemporary advances in social media are developing new information-sharing networks that are fundamentally transforming the way individuals construct shared realities. Most online fashion shopping sites have broad online communities of users who shop and scan, find, build, and share product details, personal styles, and content. User networks, such as Lyst, StyledOn, Polyvore ASOS Fashion Finder, Polyvore and general social networking sites, such as Instagram and Facebook, incorporate unique “apps’ to connect users to the fashion world and promote social interactions and fashion discussions (Olbrich & Holsing, 2011). The changing behaviour of customers about knowledge search and fashion search; and the introduction of social media into the marketing strategies of fashion merchandising sorganisations have influenced fashion marketing.

Internet-based networking platforms that encourage online community spaces and shared information sources have created a new way of coordinating efforts, opening up online communities for consumer-generated content (CGCs). The CGC discloses content created by individuals that involves content creation, origination and flow, and user use. In fashion e-tailing, fashion CGC means product display and presentation where shoppers can shop from exclusive fashion content produced by community members such as user-generated “looks” and “models” (Olbrich & Holsing, 2011).

Fashion retailing is one of the big areas most influenced by the online phenomenon. Social media offers a unique platform for fashion consumers to express their views and thoughts online in the form of text, photographs and videos through customer review sites, social networking sites, forums, and media sharing sites, among others (Wolny & Mueller, 2013). Several online social shopping sites and communities combine social networking with online shopping, enabling young fashion shoppers to shop online on sites such as Osoyou, ASOS fashion finder (Dennis et al., 2010). Previous research by (Halvorsen et al., 2013) and (Camacho-Otero et al., 2019) shows that consumer fashion reviews are helpful in fashion consumers as they direct prospective buyers on what to buy and where to buy. Also, some research indicates that CGC greatly affects fashion buying decisions (Zain et al., 2018). In fashion, CGC’s visual types have become the networking elements of the online fashion-shopping community. The essence of fashion revolves around esthetics and taste, meaning visual CGCs such as consumer-generated “looks” and “models” have become key social elements that promote social engagement, interaction and growth in online fashion consumption (Dennis et al., 2010; Olbrich & Holsing, 2011).

Nonetheless, given the growing prevalence of social media, research results indicate that a large proportion of Internet users still do not use CGC during the pre-purchase process (Wang et al., 2018). A recent survey conducted by We are Social (2018) reveals that 47% of South African Internet users use the Internet as a source of information during the pre-purchase process. It illustrates the need to consider the variables that affect CGC’s use of fashion preferences. Current studies have studied the determinants of CGC and the usage and adoption of online customer feedback using attitudes theories such as the Technology Acceptance Model [TAM] (Amoroso & Hunsinger, 2009; Lorenzo-Romero et al., 2014), the Planning Likelihood Model [ELM] (Wang & Lee, 2019); the Knowledge Adoption Model [IAM] (Erkan & Evans, 2016) and the Self-Determination Theory[(SDT] (Hasbullah, Sulaiman & Mas’od, 2020). Recent studies have successfully established the significant function of key determinants of online usage reviews such as perceived usefulness (PU) and perceived ease of use (PEOU) of the CGC as originally proposed in TAM, including trustworthiness and information derived from credibility theory. Given the recent increase in social media research, the determinants of CGC usage for the particular purpose of fashion shopping have hardly been the subject of any of these studies. To adapt and maximise the use of these new channels, retailers of clothing and apparel should also have a deep understanding of the determinants of CGC usage and adoption to improve customer interaction and gain a more comprehensive insight into consumers’ actual and real-time perceptions.

This paper presents the results of an empiric study that examines the relevant factors that attract consumers to the use of CGC for the purpose of the fashion purchase decision. The paper shall proceed as follows. The article briefly reviews the literature, provides a description of the conceptual model and the proposed hypotheses, followed by an explanation of the research methodology and discussion of the research results. The final section examines the implications of the study and the concluding remarks.

2. Model development

The rising relevance of CGC and social media in general, has not escaped the attention of researchers to retail fashion and apparel. Recent studies have investigated the role of CGC in information search and fashion shopping (Hasbullah, Sulaiman, & Mas’od, 2020); knowledge sharing and green fashion communities (Cervellon & Wernerfelt, 2012) and brand communities in luxury fashion brand equity (Brogi et al., 2013) among others. Potential consumers visit social media platforms when seeking knowledge to assist their consumer decision-making in fashion shopping and to shape their perceptions and images of styles (Nash, 2019). The primary concern of CGC platform developers and marketers may be focused on ease of use, the perception of future users of these platforms, and the behavioural outcomes arising from these perceptions. In the basis of the latter, an Extended Technology Acceptance Model offers an essential structure for understanding the determinants of consumer use of CGC for fashion shopping. Following Davis (1989) study, existing research on the extended TAM incorporated antecedents, mediating, and moderating components to explain technology adoption behavior (Al-Gahtani, 2011; Kamal et al., 2020).

2.1. Extended technology acceptance model

The relationship between perceived usefulness, attitudes towards CGCs, and adoption of CGCs can be explained in the context of adoption theories. According to the Technology Acceptance Model (Davis, 1993), perceived usefulness has a positive effect on attitudes, which positively affects behavioural intentions. The causal relationship between attitudes and intentions is also suggested in the Theory of Reasoned Behaviour (Ajzen & Fishbein, 2000). Several studies have validated the relevance of the external factors in the Technology Acceptance Model in determining technology adoption (e.g., Abdullah & Ward, 2016; Weerasinghe & Hindagolla, 2018). Considering CGC use must be approached from the perspectives of voluntary acceptance, perceived enjoyment, knowledge, and competence, CGC perceived trustworthiness plays a vital role in the favourable attitude toward usage. Using an extended technology acceptance model, this paper explores the effects of perceived usefulness, perceived ease of use, perceived enjoyment, knowledge and competence, perceived trustworthiness, and attitudes on the intention to use CGC for fashion shopping (ETAM).

Perceived usefulness is positively associated with attitudes, as consumers tend to develop a positive affect towards CGC if they expect to benefit from the use of CGC (Bhattacherjee & Sanford, 2006). Therefore, the current study posits that the perception of usefulness of CGC has an impact on consumer attitudes towards CGCs, as consumers tend to develop positive attitudes towards CGCs in cases where they believe it is functional.

Furthermore, attitudes towards CGCs and its perceived usefulness have positive effects on the adoption of CGCs because individuals tend to maintain beliefs, affect and behaviour consistent with significant others (Bhattacherjee & Sanford, 2006). Such causal relationships are empirically confirmed through a number of CGC adoption contexts (Kim & Johnson, 2016). For prospective fashion shoppers looking for information and style, the usefulness of CGC is linked to how much CGC helps them in their shopping experience and decision-making. Thus, perceived usefulness is presumed to affect fashion shoppers’ attitude and intent to use CGC for information and styles. It is therefore hypothesised that:

H1: Perceived usefulness has a positive effect on consumer attitude towards CGC usage for fashion shopping

Perceived usefulness and intention to use CGC for fashion shopping.

Besides, the model assumes a direct relationship between perceived usefulness and behavioural intention. The basis of such a relationship is that people develop their intentions towards behaviors that they consider useful, beyond any positive or negative feelings they may have towards behaviour. Many studies have verified this direct relationship between perceived usefulness and intention (e.g., Bashir & Madhavaiah, 2015). For example, the study conducted by Chen et al. (2019) supports the hypothesis that, in addition to the indirect relationship of attitudes, there is a direct relationship between perceived usefulness and community members’ intention to participate in an organisation hosted online communities. Based on the discussion, the following hypothesis is proposed:

H2: Perceived usefulness has a positive effect on intention to use CGC for fashion shopping.

Perceived ease of use and attitude towards CGC usage for fashion shopping.

The Technology Acceptance Model [TAM] (Davis, 1989) maintains that the perceived usefulness (PU) and the perceived ease of use (PEOU) of a specific service are fundamental determinants of the individual’s attitude and intention to use the service. In this paper’s specific context, PEOU refers to the degree to which a person believes that the use of the CGC platform would be effortless. For example, when selecting a style, a prospective shopper may have a number of information sources options. The consumer is likely to find the one that is cheaper and easier to use. Based on the above discussion, this study therefore assumes that the assessment of the amount of effort required by individual shoppers using the CGC platform will influence their attitudes as well as their perception of usefulness, indicating the likelihood or subjective probability of individual shoppers using online CGC for shopping information or details. Based on this discussion, it is hypothesised that:

H3: Perceived ease of use has a positive effect on attitude towards CGC usage for fashion shopping.

Consumer attitudes towards CGC and intention to use CGC for fashion apparel shopping

Ajzen (2011) described the attitude as an individual’s willingness to respond favourably or unfavourably to an object. In the perspective of the Technology Acceptance Model, Davis (1989) described the attitude towards use as the degree of evaluative impact that an individual associates with the use of the target technology in an activity. In this article, the attitude creates a reference to the use of CGC for shopping information. The attitude-behaviour relationship is apparent in customer behaviour dynamics (e.g., Tobias-Mamina & Maziriri, 2020) and has therefore been well established in traditional marketing literature, hence the hypothesis:

H4: Consumer attitudes towards CGC have a positive effect on intention to use CGC for fashion apparel shopping.

2.2. Motivational principle

Many studies have applied motivational theory to different contexts. Among these, in the marketing field, Holbrook (1994) applies a motivational theory that describes experiential value interpretation based on experiences involving direct use or remote enjoyment of goods and services. The Motivational Model (Sheth et al., 1991) assigns value consumption to actual, conditional, social, emotional and epistemological meaning. Experience value expectations focus on experiences involving either direct use or remote enjoyment of products and services providing extrinsic and intrinsic benefits (Nigam, 2012; Sullivan et al., 2012). Extrinsic benefit interpretation is usually derived from shopping encounters that are utilitarian (Singh, 2019). In contrast, understanding the inherent benefit is the product of pleasure and playfulness (experience) rather than the accomplishment of a task. Fashion consumers may use CGC sites as a means of achieving an end other than the act of surfing itself, such as online shopping, social networking among others (extrinsic benefit) or as an end to itself (intrinsic benefit). Intrinsic value, therefore, is the perceived pleasure associated with the use of a specific technology used to access fashion knowledge rather than the output of using that technology.

Prospective shoppers may want to enjoy the act of browsing CGC pages, viewing customer fashion photographs and videos, and reading their comments and reviews of styles and looks. When it comes to fashion apparel knowledge quest, Castaneda et al. (2007) found that customers might be less involved in the outcome of the information search process (extrinsic motivation) but more emotionally motivated. The “perceived enjoyment” concept is used to capture the inner motivation. Lately, perceived enjoyment has gained growing attention in marketing literature.

The construct was analysed in relation to TAM as one of the predictive variables for the general use of computers, internet use, instant messaging devices, e-learning, and online shopping. In various study contexts, perceived enjoyment was measured as a weaker indicator of consumer acceptance relative to perceived usefulness and ease of use (Mathieson et al., 2001; Tobias-Mamina & Maziriri, 2020). Nevertheless, a significant influence of perceived enjoyment has been identified in hedonic environments such as sports, home or leisure environments, and in some online settings (Van der Heijden, 2004). CGC is assumed to be a hedonistic device (Allam et al., 2019). The hedonic system’s importance is the degree to which users enjoy pleasure while using the program (Allam et al., 2019). Nonetheless, as this study concerns the functional or instrumental value of CGM for e-shopping, it is not clear if the hedonic value of “perceived enjoyment” may have an effect. Therefore, this study measures the effect of perceived enjoyment on customer attitudes and behavioral intent to use CGC for apparels shopping. The viewpoint of extrinsic motivation in the proposed model is addressed by perceived usefulness. Previous studies in psychology indicate that higher rates of intrinsic motivation typically contribute to a desire to spend more time on a given task and, as a result, form the sense of ease of use (Carsrud et al., 2017; Venkatesh, 2000). More intrinsically driven consumers are more likely to underestimate the complexity of using technology. In addition, perceived enjoyment has been shown to affect online shopping attitudes (Ha & Stoel, 2009). In light of the above, the following hypotheses are proposed:

H5: Perceived enjoyment has a positive effect on consumer attitude towards using CGC for fashion apparel shopping.

3. Trustworthiness and competence

A variety of studies have applied the principle of source credibility to different contexts. Notable among these in marketing communication is Hovland and Mandell’ (1952) application of the source credibility principle, which describes how the presumed legitimacy of the source of communication is decided in part by the persuasiveness of communication (e.g., Babu et al., 2018; Johnson et al., 2011). The source credibility model (Kerstetter & Cho, 2004) indicates that trustworthiness and competence are the key determinants of online acceptance of information. Credibility includes the degree to which the information source is considered authentic, competent and trustworthy by the information recipient (Wang et al., 2018).

The importance of credibility in the adoption of CGC and in the specific context of fashion and fashion products is underscored by the discreet nature of CGC and online fashion reviews, as users are confronted with the assessment of the opinions of complete strangers (Dennison & Montecchi, 2017). A multiplicity of source credibility dimensions (e.g., dynamism, elegance, authority, personality) has been proposed. While some of these dimensions have been questioned, there tends to be a consensus on the dimensions of trustworthiness and competence (Henderson et al., 2017). These two dimensions also appear to be the most relevant to the unique context of CGC. Thus, this study sconceptualises credibility as a two-dimensional construct, with competence and trustworthiness as distinct dimensions. Consumers planning to use CGC for shopping may therefore be concerned with the reliability and competence or skills of individuals posting content. On the contrary, some scholars argue that, since non-commercial information is perceived as more unbiased and credible, consumers tend to treat data from their peers (in this case, CGC contributors) as trustworthy (Kulmala et al., 2013). Simply explained, expertise refers to competence and intelligence, whereas trustworthiness is the obvious honesty and integrity of the source. Fashion products are considered high-involvement products linked to the consumer’s personal identity, so consumers continue to seek information from different sources before making a purchase decision. Therefore, it is expected that the perception of the trustworthiness of CGC sources by prospective shoppers will have a positive impact on their attitude towards the use of CGC for fashion apparel shopping. Hence it is posited that:

H6: Perceived trustworthiness has a positive effect on attitude towards CGC usage for fashion apparel shopping.

H7: Perceived trustworthiness has a positive effect on consumer intention to use CGC for fashion apparel shopping.

H8: Perceived knowledge and competence positively influence attitude towards using CGC for fashion apparel shopping.

3.1. Moderating role of gender

Gender disparities in decision-making processes have been investigated in psychology, and it appears that a gender schema serves as a cognitive filter through which one categorizes personal attributes into masculine and feminine categories (e.g., He & Freeman, 2010). Gender-typed individuals, for instance, encode and process information using various socially constructed cognitive frameworks, which in turn assist determine and direct an individual’s perceptions (Davis & Wilson, 2016). This means that gender schemas may be viewed as a normative reference that leads to socialized behaviour that is congruent with the schema (Smith, 2013). As a result, the purpose of this study is to investigate how various genders influence the link between the primary elements of the online brand community and the adoption of CGC. This study suggests that the influence of gender on the link between trustworthiness, attitude, and behavioural intent is stronger for female users, whereas PU is stronger for male users.

Therefore, the study proposes the following general hypothesis:

H9-H11: Gender moderates the relationship between perceived usefulness (PU), trustworthiness (T), attitude and CGC usage intention.

4. Hypothesised model

Consumers’ perceptions of usefulness, ease of use, enjoyment, trustworthiness, expertise and knowledge are assumed to have an impact on their attitude and behavioural intention to use CGC for fashion shopping (see Figure 1). Behavioural intention describes the subjective probability of a person exhibiting a certain behaviour (Ajzen, 2011). According to Ajzen (2011), behavioural intention reflects the degree to which people are willing to try and how much effort they plan to make to fulfil the desired behaviour. Table 1 summarises all measures corresponding to the construct described, namely perceived ease of use, perceived usefulness, trustworthiness and usage intention.

Figure 1. Proposed conceptual model

Table 1. Demographic profile of respondents

	Frequency	%
Age
18–24	330	72.5
25–31	112	24.6
32–40	13	2.9
Monthly Expense
<1000	125	27.5
1001–2000	212	46.6
>2000	118	25.9
Year of online shopping
< 1 year	46	10.1
1–2 year	304	66.8
>3 years	105	23.1
Internet Frequency
Once a day	47	10.3
Several times a day	365	80.2
A few times a week	43	9.5

5. Methodology

5.1. Research design and sample selection

Following the post-positivist paradigm, this study applied a quantitative approach to data collection and interpretation. The technique adopted is similar to those used in other examples such as Tobias-Mamina and Maziriri (2020) and De Villiers, Chinomona and Chuchu (2018) in predicting behavioural intention. Theoretically, this study’s population is made up of millennial Internet users who have made an online purchase experience over the last six months. Since there is no such list of Internet shoppers, our sampling frame could not be standardised. Survey participants are young male and female consumers of fashion and clothing websites aged 18 to 40, a core online fashion customer category. Millennials are reported to be key social media users and display strong interest in fashion (Lyu et al., 2018). Purposive and snowball sampling methods are used to recruit participants. Purposive sampling is the process of finding one participant through another. It enables researchers to recruit experienced individuals in the phenomenon (Teddlie & Yu, 2007), such as community members of social e-shopping sites including Polyvore. They regularly use the site and consume CGCs like styles, tags, collages and reviews during their online shopping activities. Respondents were then recruited from public facilities (university campus, local mall). Between May and August 2019, 455 valid questionnaires were collected. The rule of thumb for PLS-SEM is that the total sample should be equal to the greater of: (1) 10 times the largest number of metrics used to evaluate the most complex structure; or (2) 10 times the largest number of structural paths for the specific latent structure of the structural model (Hair, Ringle & Starstedt, 2011). Thus, despite the widely suggested “ten-time rule,” a sufficient sample of 455 valid responses is considered to be more than adequate for the study. To eliminate differences in response patterns due to various reference points, all respondents were asked to respond to the questionnaire on fashion apparel.

5.2. Measures

There are two sections of the questionnaire. The first segment of the questionnaire includes information on the respondents’ socio-demographic characteristics, such as age, gender, educational level, and income level. The second segment consists of scale metrics chosen to assess research constructs based on current marketing literature measures. Proper modifications have been made to suit the current research context and purpose. “Perceived usefulness” and “Perceived ease of use” used a five-point scale measure adapted from Pavlou (2003); “Attitude” and “perceived enjoyment” both used five-point scale measures adapted from J. Kim and Forsythe (2009). “Trustworthiness” used a five-item scale measure adapted from Escobar-Rodríguez and Carvajal-Trujillo, while “Knowledge and competence” measures were adapted from Ohanian (1991). Finally, measures for the use of CGC were adapted from Venkatesh et al. (2012). All measures for this analysis are of a reflective nature. Trustworthiness and attitude were measured using five-point semantic differential scale. Measurement items for the remaining concepts were measured using a five-point Likert scale with anchors ranging from strongly disagree (1) to strongly agree (5) to reflect the level of agreement. The questionnaire was piloted by a group of 20 marketing experts and scholars to ensure accuracy and comprehension prior to the actual analysis. To assist participants, the photo-elicitation technique is included in the questionnaire to help participants make sense of the fashion apparel phenomenon and stimulate emotions and memory (McCormick & Livett, 2012) and to examine the perceptions and experiences of the visual CGC by shoppers.

5.3. Ethical consideration

An ethical clearance certificate (Protocol number: (BUSE/1243) to conduct the study was obtained from the University of the Witwatersrand Research Ethics Committee. Also, this research study proceeded in compliance with the ethical principles of academic research, including the security of respondents’ identity and interests and the confidentiality of the information provided by respondents. Respondents gave informed consent and the purpose of the study was also conveyed to the respondents.

In addition, this research study acted following the ethical standards of academic research, including protecting respondents’ identities and interest and assuring confidentiality of information provided by respondents. Respondents provided informed consent and nature of research was communicated to respondents.

5.4. Data analysis

Structural Equation Using Partial Least Squares (PLS)

The Smart PLS toolkit was used to estimate the theoretical model using a modeling strategy with bootstrapping and 5000 re-samples (Ringle, Wende & Becker, 2015). Unlike the initial regression approaches, SEM uses the same technique to investigate the loading of the measured items on their expected latent variables (measurement model) as well as the hypothesized link between a series of dependent and independent constructs (structural model) (Sarstedt et al., 2017). This method allows for the measurement errors of the observable variables to be analyzed as part of the computation.

While eliminating biases in the parameter estimation characteristic of regression analysis, it is clear that the PLS methodology is best suited for probing probable causation and allows for reflective and formative frameworks (Sarstedt et al., 2017). The PLS method puts a minimal demand on measurement scales, sample sizes, and residual distributions, making it excellent for complicated models and predictive-oriented investigations with endogeneity (Hair et al., 2017). As this research is a preliminary attempt to explore the role of perceived enjoyment, knowledge and competence and the TAM factors in predicting the use of CGC by consumers for the specific function of apparel shopping, PLS was deemed most fitting(Hair et al., 2017). In this study, data analysis was carried out using the Smart PLS-3 software program, with 1000 re-samples performing bootstrapping.

6. Results

6.1. Sample characteristics

Table 1 shows the profile of the respondents to the sample. Of the 455 respondents, 72.6 percent were between 18 and 24 years of age, 24.6 percent were between 25 and 13 years of age, and 2.9 percent were between 32 and 40 years of age. The majority of respondents had a monthly budget of more than R1000. Nearly 67 % of respondents had less than one year of online shopping experience and 33% had more than one year of experience. The majority of respondents were mainly Web users who use the Web many times a day (80.2 per cent) or once a day (10.3 per cent), with a few using the Internet a few days a week (9.5 per cent).

Respondents were overwhelmingly familiar with CGC networks, with the majority using social networking sites such as Twitter, Myspace fashion, style-hive and shop-flick. Many respondents have used social sharing platforms such as YouTube and Instagram and user-generated reviews of third-party app stores (such as Mango, ASOS, Boohoo) and service providers’ websites such as Pinterest.

6.2. Measurement model assessment

Upon completing the data collection, a two-step analytical method proposed by Sarstedt et al. (2017) was used first to test the measurement model and validate the structural model. The study model is reflective with seven reflective constructs. First, we validated the outer model by checking the reliability, convergence and discriminatory validity of the different constructs (PU, PEOU, Trustworthiness, Knowledge & Competence, Perceived Enjoyment and Use Intention); second, we tested the inner (structural) model to determine the assumed relationship between the constructs and to assess the overall prediction of the proposed model. Composite Reliability (CR) value and average extracted variance (AVE) value of each construct must be evaluated for convergent validity. Factor loading loaded on their a-priori variables above 0.70 at 95 percent confidence level, with AVE values varying from 0.558 to 0.768 above the suggested threshold of 0.5 (Hair et al., 2017) as shown in Table 2. AVE values of 0.5 and higher mean that the latent variable explains more than half of its indicator variance. The Cronbach alpha and Dillon-Goldstein rho values for all seven constructs were robust and well above the lower limit of 0.6 (p < 0.01) (see Table 2), suggesting high reliability and further supporting the proper loading of indicators within their respective (a priori) constructs (Nunnally & Bernstein, 1994). The CR values for all variables ranged from 0.785 to 0.908, exceeding the recommended threshold of 0.7 (Hair et al., 2012), indicating high construct reliability. The seven constructs’ discriminant validity was measured using the HTMT ratio (Franke & Sarstedt, 2019; Henseler & Sarstedt, 2013). As shown in Table 3, each pair of reflective constructs has an HTMT value below 0.9, which indicates that a sufficient discriminant validity is defined. The Fornell and Larker criterion further confirmed discriminant validity. According to Fornell and Larker, the latent construct would have to share more variance with its assigned indicators than with any other latent element. The square root of the AVE of each latent construct would have to be greater than the maximum correlation of the construct with any other latent construct.

Table 2. Reliability and convergent validity

Construct	Items	Loadings	Std Beta	Cronbach’s α	AVE	CR	rho_A
Perceived Ease of use	PEOU1	0.809	0.807	0.807	0.634	0.874	0.809
	PEOU2	0.844
	PEOU3	0.775
	PEOU4	0.775
Perceived usefulness	PU1	0.715	0.744	0.748	0.655	0.849	0.840
	PU2	0.892
	PU3	0.811
Perceived Trustworthiness	PT1	0.647	0.746	0.750	0.573	0.841	0.794
	PT2	0.662
	PT3	0.828
	PT4	0.866
Knowledge & Competence	KC3	0.839	0.850	0.850	0.768	0.908	0.906
	KC4	0.847
	KC5	0.939
Perceived Enjoyment	PE1	0.691	0.819	0.820	0.626	0.829	0.940
	PE2	0.990
	PE3	0.649
Attitude towards the use of CGC	ATT1	0.832	0.699	0.701	0.558	0.785	0.665
	ATT2	0.832
	ATT3	0.537
Intention to Use CGC for apparel shopping	INT1	0.842	0.706	0.709	0.632	0.835	0.771
	INT2	0.631
	INT3	0.888

Table 3. Constructs’ discriminant validity

	ATT	INT	KC	PEOU	PE	PT	PU
Attitude	(0.747)
Intention	0.528	(0.795)
Knowledge & Competence	0.321	0.129	(0.876)
Perceived ease of use	0.379	0.189	0.133	(0.796)
Perceived enjoyment	0.103	0.055	0.031	0.212	(0.791)
Perceived trustworthiness	1.028	0.341	0.196	0.403	0.136	(0.757)
Perceived Usefulness	0.426	0.159	0.104	0.769	0.769	0.378	(0.809)

Note. Square roots of AVE shown on the diagonal.

HTMT values shown below the diagonal

The variance-inflation-factor coefficient is measured in order to detect possible multi-collinearity. The validated model was then tested on a gender basis using 1) multi-group t-tests and 2) PLS-SEM software permutation tests (Vinzi & Russolillo, 2013).

6.3. Structural model and hypotheses results

After the validation test, PLS SEM is used to assess the explanatory power and predictive validity of the proposed model, as well as the magnitude of the path coefficients and the importance of the shypothesised relationship. Since the PLS is predictive, the R² value is a crucial criterion for assessing the structural model, which introduces the amount of variance explained by the endogenous latent variable. While there are no generalities as to whether R² is high or low, Henseler and Sarstedt (2013) describes R² levels of 0.67, 0.33 and 0.19 as strong, moderate and small. As shown in Figure 2 and Table 4, the model describes 57.6 percent of the variance in attitude indicating a significant level and 15.8 percent of the variance in usage intention indicating weak level.

Figure 2. PLS results of the structural model

Table 4. R² (explained variance) and Q² (predictive relevance test)

Endogenous construct	R^2	Q^2
Intention to use	0.577	0.065
Attitude	0.158	0.304

6.3.1. Model predictive relevance (Q²⁾

The additional predictive value of exogenous latent variables was measured using the Stone-Geisser Q Test. The blindfolding technique was used to determine the cross-validated redundancy test Q². Values of Q² > 0 mean that exogenous constructs are of predictive relevance for the endogenous construct under consideration, whereas values below zero imply a lack of predictive relevance (Hair et al., 2012). All Q² values vary considerably above zero, suggesting a high predictive capacity of the exogenous constructs.

Model Fit: The sStandardised Root Mean Square Residual (SRMR)

The SRMR is a model fit estimation measure. Table 5 shows that the study model SRMR = 0.071 is lower than the recommended threshold of < 0.08 (Hu & Bentler, 1998) and χ² = 813.066, NFI = 0.713 indicating good model fit.

Table 5. Model fit summary

	Estimated model
SRMR d_ULS d_G NFI	0.071 1.387 0.455 0.713

6.3.2. Path coefficient (β) and T-statistics estimation

To test whether the path coefficients vary significantly from zero, bootstrapping was used to calculate t-values. The non-parametric bootstrapping protocol with 455 cases, 1,000 subsamples, was used (Hair et al., 2012). The analysis found that four of the eight hypothesised relationships in the inner model were significant. The hypothesised relationships are shown in Table 6.

Table 6. Results of the hypothesised relationship

	Regression Path	Path Coefficient(β)	t-Value	Supported
H1	Perceived Usefulness->Attitude	0.129	2.694*	Supported
H2	Perceived Usefulness->Intention	−0.075	1.119	Rejected
H3	Perceived Ease of Use->Attitude	−0.028	0.542	Rejected
H4	Attitude->Intention	0.283	2.897*	Supported
H5	Perceived Enjoyment-> Attitude	−0.082	1.309	Rejected
H6	Perceived Trustworthiness->Attitude	0.708	20.630**	Supported
H7	Perceived Trustworthiness->Intention	0.085	0.800	Rejected
H8	Knowledge & Competence->Attitude	0.103	2.313*	Supported
H9	Perceived Trustworthiness*Gender*Intention	0.335	1.249	Rejected
H10	Perceived Usefulness*Gender* Intention	0.584	2.417*	Supported
H11	Attitude*Gender* Intention	0.295	1.039	Rejected

Note: *Significant p < 0.05; p < 0.01**

H3 has no relationship because of factors such as low t- statistics (β = −0.028; t = 0.542).The perceived usefulness was found to positively impact the attitude (β = 0.129, t = 2.694); while its direct effect on the intention (β = −0.075; t = 1.119) was insignificant. Knowledge and competence were found to have a positive impact on the attitude (β = 0.103; t = 2.313). Consistent with behavioural theories, attitudes provide a positive and significant impact on the decision to use CGC for apparel shopping (β = 0.283; t = 2.897). While the respondents’ assessment of the content’s reliability has a small significant impact on the decision to use the CGC (β = 0.800), the t-statistic is lower than the required 1.96. Perception of trustworthiness has the strongest relationship with attitude (β = 0.708; 20.630) and is found to have no relationship with intention because the finding is insignificant (β = −0.075; t = 1.119). Respondents’ perception of enjoyment did not affect their disposition towards the use of CGC for fashion apparel (β = −0.82; t = 1.309). The greater the beta coefficient, the stronger the latent exogenous construct’s impact has on the latent endogenous construct. Figure 2 shows that, compared to other β values in the model, the perceived trustworthiness factor had the highest path coefficient β = 0.708, which showed that it had a higher variance value and a high effect on the attitude towards the use of CGC when shopping online. Figure 2 shows the graphic representation of all model path coefficients. Thus, the results endorsed H1, H4, H6 and H8; they did not support H2, H3, H5 and H7. In order to investigate hypotheses that were not supported (H2, H3, H5 and H7), we tested the mediation effect using bootstrapping.

In order to examine the hypothesis that were not supported (i.e., H2, H3, and H7) mediation effect was tested. Thus, the Sobel Test confirmed the significance of the indirect influence of perceived usefulness through attitude (β = 0.189, z = 7.567; p < 0.01). The mediating impact of perceived usefulness on the relationship between perceived ease of use and usage intention has also been established (β = 0.173, z = 6432; p < 0.01). Likewise, the mediating influence of Knowledge and Competence on the relationship between Perceived Enjoyment and Attitude has been established (β = 0.096, z = 4.438; p < 0.01).

6.3.3. Interpreting the Interaction Results

Using PLS Algorithm result R² for the main effect model (0.122) with addition of interaction variable “Gender” changed to R² for endogenous construct Intention to 0.158, a 0.036 change indicating 3.6% (f² = 0.0373) additional variance. R² for endogenous construct attitude changed from o.576 to 0.577 a 0.1% (f ² = 0.001) change. The next step was to calculate the size of the effect (f²) using an effect size calculator. Based on Cohen (1988) guidelines, we conclude that both endogenous variable attitude and usage intention had a small effect size based on their f² (Akgül et al., 2019). Interaction effect is further tested using the bootstrapping procedure. as shown on Figure 3 below:

Figure 3. Gender*Perceive usefulness*Usage Intention

Multi-group analysis results showed that significant differences only existed within one path of the proposed model. Explicitly, the path coefficient from perceived usefulness to usage intention was significantly higher than the corresponding path coefficients in the structural model. As seen in Table 7, the interaction between Perceived usefulness*Gender is positive (β = 0.575; t = 2.350) while the other two interactions are negative (β = −0.219; t = 0.815; β0.193; t = 0.620 respectively). However, the size and nature of interaction cannot be based on coefficients only hence Dawson’s (2014) product indicator approach was used to further examine moderation analysis. Since the moderator for the study is a categorical measure (gender), this study uses interaction plots (2 way with binary moderator) to facilitate the interpretation of gender as moderator effect between the relationship perceived trustworthiness, perceived usefulness, attitude and usage intention at +1SD (Male 1) and −1SD (female 0). Only three constructs have the interaction results (Gender*Attitude; Gender*perceived Usefulness and Gender*perceived trustworthiness). However, results shown on Table 7 indicate that Attitude*Gender, as well as Perceived Trustworthiness*Gender relationships, are not significant, suggesting lack of interaction effect, and only perceived usefulness was found to be significantly related to shoppers’ intention to use CGC for e-shopping. Therefore, only Perceived Usefulness*Gender qualifies for interaction plot. There were no significant differences between males and females found in the rest of the model’s paths, indicating a common pattern of behaviour in both genders in general. According to the current study, neither females nor men consider perceived usefulness while deciding whether or not to utilize CGC.

Table 7. Moderating effect tests

Hypothesis	Relationship	Std Beta	Std Error	t-Value
H9	Perceived Trustworthiness*Gender->Intention	−0.219	0.269	0.815
H10	Perceived Usefulness*Gender->Intention	0.575	0.245	2.350**
H11	Attitude*Gender->Intention	0.193	0.311	0.620

NB: **p < 0.01

A recall of the hypothesis H9 to H11: Gender moderates the relationship between perceived usefulness (PU), trustworthiness (T), attitude and CGC usage intention.

Interaction plots are described by looking at the gradient and direction of the slope. As shown in Figure 3, the line labelled for females (women) has a steeper and positive gradient compared to males (less steep and negative gradients), suggesting that the positive relationship is generally greater for females. Based on the gradient of the interaction plot Gender*Perceived usefulness, the female-represented line has a steeper positive gradient compared to the male (less steep and negative gradient), suggesting that the positive relationship is more robust for the female group, thus supporting the hypothesis (H9). According to the plot, females’ desire to use CGC increases as their perception of usefulness increases. In contrast, when males’ perceptions of usefulness grow, their propensity to use the platform appears to fall. This conclusion contradicts the findings of Leong et al. (2013), who found that male customers’ inclination to utilise technology is more significantly impacted by their perception of its utility. This gender*Perceived Usefulness->Intent relationship, females are more likely than males to utilise CGC for online retail purchasing.

7. Discussion and findings

7.1. Key findings

This study examines the current research gap in technology acceptance in fashion shopping environments, emphasising the role of perceived enjoyment, knowledge and competence, and integrates a gender perspective into the TAM model. It builds on and extends the TAM and Motivation theory by examining the moderating role of gender in the relationship between TAM constructs (i.e., perceived usefulness, perceived ease of use, attitude towards usage) the motivation constructs (i.e., perceived trustworthiness, perceived pleasure, knowledge and competence) and the intention of the shopper to use consumer-Generated Content for their fashion shopping.

Comparing our results with previous TAM research, this analysis reveals some significant findings. The study found substantial support for some of the traditional TAM-related constructs, such as perception of usefulness, attitude, and motivational theory, such as perceived trustworthiness, knowledge, and competence. However, this would imply that individuals who generally avoid CGC may be willing to use CGC for fashion shopping if they realise it is useful, enjoyable and straightforward to use whenever they shop online. By measuring the moderating impact of gender, building TAM and motivation variables, this study promotes a deeper understanding of shoppers’ intent to use online ratings and styles in their decision-making. The findings obtained from the estimation of PLS-SEM and multi-group and permutation tests are intended to advance theoretical knowledge of the degree to which CGC drivers influence apparel shoppers to use CGC and provide practical implications for retail and marketing professionals.

Contrarily to previous research in other contexts, perceived enjoyment and perceived ease of use have been found to have no impact on the attitude towards the use of CGC. Interestingly, several studies have not identified a clear connection between perceived ease of use and attitudes, and have recognised that this connection is entirely mediated by perceived usefulness in some technological contexts (Kim et al., 2008). Although the research supports the direct relationship between usefulness and attitudes, the results further support the direct influence of perceived trustworthiness and attitude and usage intention. Both PU and KC have a major impact on consumer attitudes towards the use of CGC (β = 0.707 and 0.103 respectively). These results are consistent with the TAM literature findings that found perceived usefulness exerting significant effect on behavioural intentions (Tobias-Mamina et al., 2020).

The findings from this study highlight the mediating role of attitude. For example, the effect of perceived usefulness on usage intention is fully mediated by attitude towards CGC. In addition, the attitude construct partially mediates the relationship between perceived trustworthiness and the intention to use CGC for shopping purposes. This finding implies that customers/consumers intend to perform behaviours (in this case, the use of CGC for online shopping) to which they have a positive affect.in this study can be explained by the fact that the nature of the use of CGC for shopping is voluntary and is determined by individual involved in the act.

Conversely, the availability of several online and offline e-shopping information sources makes it particularly important to choose any specific source of information. The findings also show the significant impact of trustworthiness, knowledge and competence (expertise) on shoppers’ intention to use user-generated style boards (β = 0.708; 0.103, respectively). This finding supports a number of digital and marketing studies that also found trustworthiness and expertise to be critical in predicting the use of CGC and online reviews (e.g., Hu et al., 2016).

Most importantly, the multi-group and permutation tests results further indicate that only perceived usefulness-usage intention relationship is moderated by gender. These findings contribute to the ongoing debate on the moderating role of gender on the relationship between various behavioural antecedents and digital technologies. Of particular interest is the confirmation that PU is a central and primary determinant of CGC usage for female shoppers, although it has not been significant for male consumers. However, these results contradict other researches that have not found any variations in the relationship between usefulness and usage intention in their respective models with gender as intervening variable (e.g., Lian et al., 2014). While online shopping can be equally useful and beneficial to both females and males, some other applications (e.g., fashion CGC/online reviews) are often used by females only if the user-generated “looks” and “models” prove to be of value to females who appear to be more interaction driven. This may explain the stronger influence of perceived usefulness of user-generated style boards on females CGC usage intention.

The study found that PEOU has an insignificant effect on the attitude towards the use of CGC, most likely due to the age group of the respondents, who are predominantly young adults. The ease of use of CGC does not seem to be of any importance to millennials. Ease of use of CGC does not seem to be a significant enabler of young consumers’ shopping expedition.

7.2. Implications for theory and practice

Although this study expands the spectrum of fashion marketing research to CGC’s application in the context of online fashion shopping, it validates the significant role of perceived usefulness, perceived ease of use, perceived enjoyment, knowledge and competence, perceived trustworthiness in predicting the attitude and intention to use CGC for shopping. The study introduces additional consumer research frameworks that reflect the complex context of CGC usage in online retail shopping. The research also draws attention to variations in predictors of the use of technology in voluntary settings. As mentioned in the previous section, contrary to traditional TAM literature, factors such as perceived trustworthiness, perceived usefulness and perceived competence are of greater significance in the use of CGC in online fashion shopping.

This empiric research supports the aptness of TAM’s attitude construct when examining individual use intention in a non-institutional situation. The research findings indicate that three of the five psychological variables examined directly affect attitudes that partially mediate the relationship between these variables and usage intention. Consequently, attitude fully mediates the relationship between perceived usefulness and usage intention. The study also found that the intention of the shoppers to use CGC as a style board differs slightly by gender.

As a result, a variety of organisational implications arise from this analysis. The model should help managers understand how shoppers judge CGC websites. The study provides insight into the psychological factors that determine online shoppers’ decision to use CGC for fashion shopping. In order to exploit social media in retail and fashion marketing, these aspects need to be addressed by platform managers. It is also important to emphasise the quality, reliability and expertise of the content thus generated.

The findings suggest that CGC fashion shopping attitudes’ strongest antecedent is trustworthiness (β = 0.708; t = 20.630). It may represent the evaluation by fashion consumers of the degree of credibility of consumers who produce the content. Project managers therefore need to stress the aspect of trust and credibility. Fashion retail service providers intending to incorporate digital media into their marketing campaigns are encouraged to concentrate on CGC or consumer-generated “looks” and “designs” that express credibility attributes.

7.3. Limitations and future research

Despite this research’s significance, there are limitations to the research that offer relevant scope for further investigation. Most significant, the research adopted an attitude-based approach measuring the usage intention rather than the actual usage of CGC. Therefore, future studies can measure actual usage tracking respondent behaviour over a longer period. Secondly, this study only considered gender as a moderator; future studies could consider other interaction variables. Extant literature has identified gender identity as an aspect influencing behaviour (Ye et al., 2017).

Further studies could go beyond the concept of consumer intention to use CGC for shopping and reflect on elements that influence consumers to contribute content on the platforms. Despite these limitations, the proposed model provides a theoretical framework, for the development of retail marketing, particularly in the virtual environment. While this study sheds light on the role of perceived trustworthiness in influencing attitudes, further investigation of these links will be particularly valuable, as a limited number of studies have explored these causal factors in a model.

Additional information

Funding

The authors received no direct funding for this research.

Notes on contributors

Rejoice Jealous Tobias-Mamina

Dr. Rejoice Jealous Tobias-Mamina is currently a Post-Doctoral Research Fellow in theDepartment of Life and Consumer Sciences at the University of South Africa (UNISA). Shehas a keen interest in consumer behaviour.

Eugine Tafadzwa Maziriri

Dr. Eugine Tafadzwa Maziriri is an NRF rated researcher and he is currently working as aSenior Lecturer in the Department of Business Management at the University of the FreeState in Bloemfontein, South Africa. He has a keen interest in consumer behaviour.

Elizabeth Kempen

Prof. Elizabeth Kempen is the head of the College Office for Research and Postgraduate Studies in the College of Agriculture and Environmental Sciences at UNISA. She graduated with a PhD in Philosophy from the University of Liverpool in the UK. Her main research interest is in consumer decision making.