The single-serve wine option in South African restaurants: product involvement, risk perception and information-related behavioural effects

ABSTRACT

Wine is a highly-valued-added end product of an important agricultural value chain. In this product category, the single-serve wine by-the-glass (WBG) option in restaurants presents a largely underutilized business opportunity. Academic research examining consumer behavioural psychology-based constructs in the situational consumption context of restaurants, has also not kept pace with this market reality. Hence, this study establishes how product involvement, risk perception, and information processing, affect consumption of WBG by consumers in the situational milieu of restaurants. Following scale validation by conducting confirmatory factor analysis and thereafter fitting a structural equation model, the relationships between constructs are examined by utilising a sample of 1,038 South African consumers representative of dining across all restaurant categories. The findings contribute to the literature by showing that distinct motivational relationships exist between the involvement and perception of risk constructs and information-related behaviour of consumers engaging with the WBG option in restaurants. The stable nature of consumers' enduring involvement with wine products evokes motivational processing by triggering situational involvement upon purchasing WBG and perception of risk arousal of both the cognitively-evaluated (psychological, social and functional) and non-cognitively-evaluated (financial, physical and time) risk types.

KEYWORDS:

• Product involvement
• risk perception
• motivational process
• information-related behaviour
• restaurants
• Wine by-the-glass
• WBG
• South Africa

1. Introduction

Wine is a highly-valued-added end product of what is an important agricultural value chain. The wine market splits into two main categories: off-premise (take-home retail sales) and on-premise (on-site consumption, e.g., at restaurants). Of specific relevance to the on-premise category, is the fact that the growth trend in single-serve portions of food and beverage products is expected to continue into the next decade until 2030 (Mintel 2019). Although the marketing opportunity offered by the single-serve WBG (wine-by-the-glass) option is significant, its utilisation has been below the growth potential it represents (Frost 2015; Yoon and Stacy 2015). Durham et al. (2004) found that offering WBG is an important factor in realising at least some wine sales. Among the other opportunities associated with the WBG option, expensive wines could be sold faster if more WBG options were made available in restaurants (Frost 2015). More specifically, providing WBG options in restaurants articulates to a five percent increase in the bottle price for similar wines offered in bottles, and a twelve percent increase in the bottle margin (Dearden et al. 2021). However, utilising the WBG opportunity will need a shift in mindset by many restaurants that view WBG largely as a way to offer consumers “something” instead of “nothing” in the form of single-serve wine options.

Despite the business opportunity that WBG sales in restaurants represents, academic research in this area of marketing and consumer behavioural responses is still at a juvenile stage. This is quite surprising in view of the fact that the hospitality industry, and particularly, the foodservice component thereof, present many situations that are conducive to the formation of attitudes (Lacey et al. 2009). Consumers’ interpretation of a specific situation is instrumental in their attitude formation that takes place (Carvalho et al. 2018) and, in turn, can affect their behaviour, for example, ordering wine in a restaurant (Lacey et al. 2009), and thus could have an impact on strategies of marketing to them. Despite this gap in the knowledge base, most of the research focus has been on the wine taste preferences of consumers (Giacalone et al. 2014; Rinaldo et al. 2014). Attitude formation from a psychological viewpoint has received less attention (Thomé et al. 2016; Carvalho et al. 2018), while even fewer (Bruwer et al. 2017; Bruwer and Cohen 2019) focused on the motivational perspectives of consumers’ relationship with WBG. That is, despite attempts in the form of exploratory studies to develop risk perception scales (Bruwer et al. 2017; Bruwer and Cohen 2019) and work by Terrier and Jaquinet (2016) on the pairing of food and wine as a risk-reduction strategy (RRS). These studies however, treated perceived risk as a unidimensional construct and did not conduct the conventional reliability and validity testing on the risk measurement scale.

Consumer behavioural psychology (Saad 2021) applies a motivational perspective to product involvement and risk perception as constructs, and their effect on the subsequent behavioural responses of seeking and dissemination of product information (Klerck and Sweeney 2007). Hence, there is a need for a theory-rooted study, viewed through the behavioural psychology lens (Miranda and Duarte 2022), to determine how the individual WBG risk perception dimensions interrelate with product involvement and information-processing behaviour. Such a study will benefit from a strong theoretical grounding using a framework such as a motivational process model provides (Dholakia 2001; Sun et al. 2010). Theory-building research on WBG consumption in restaurants specifically focused on these pivotal constructs, will not only lay a theoretical foundation in the hospitality management domain, but open an avenue for academic discourse. Hence our study’s objective is to confirm a theory-based model of product involvement and perception of WBG risk that includes information-related behaviours in the motivational process of consumers when exposed to the situation of WBG ordering in restaurants. It achieves this by successfully treating WBG risk perception as a multi-dimensional construct in the restaurant context.

In the agricultural value chain, the wine category has been a prominent and highly value-added final product over many decades, even hundreds of years. Not surprisingly, from a generic viewpoint, its end-use has been extensively examined in the literature from, what has often been, a behavioural perspective. Despite the importance of the on-premise sector to the wine product category, there is however, a paucity among previous research to focus on consumption in the on-premise sector, notably in restaurants. With single-serve consumption in the on-premise sector rapidly on the increase (Mintel 2019), WBG consumption in restaurants is an area on which minimal research attention has been focused (Bruwer and Cohen 2019). Not only does our study contribute to the theory base by providing nuanced insights from the perspective of holistic and consumer-centric behavioural psychology (Saad 2021; Miranda and Duarte 2022), but it does this in South Africa, a country where such research has to date been absent. More specifically, we develop a theory-based motivational process model and use structural equation modelling (SEM) to test the relationships between focal behavioural psychology constructs for this specific consumption situation.

2. Literature review and hypotheses

Consumer behaviour, and per se marketing, is a reasoned action and consumption-relevant as a result of cognition and its conscious thinking processes (Grunert and Grunert 1995). Therefore, the focal motivational consumer behaviour constructs and psychological processes of product involvement, perceived risk, and information processing as these relate to the restaurant environment and the WBG option, can be theoretically explicated by means of a motivational process model (Dholakia 2001). Via this avenue, the current study contributes by providing theoretically grounded insights of various related consumer behavioural psychology constructs in the restaurants’ situational setting.

2.1 Product involvement

Involvement is “a motivational state that affects the extent and focus of consumers’ attention and comprehension processes” (Celsi and Olson 1988, p. 210), “as well as overt behaviours such as shopping and consumption activities” (Tarkiainen and Sundqvist 2009, p. 847). The definition of consumer involvement is consistent with our study’s motivational perspective.

Involvement has been a subject of interest in the marketing literature largely because of its motivational influence on behavioural outcomes, for instance, consumers’ seeking, dissemination and processing of information and their influence on buying behaviour (e.g., Warrington and Shim 2000; Bruwer et al. 2017). Since the early days of studying involvement, researchers distinguished between two types: enduring and situational involvement (e.g., Bloch and Richins 1983; Celsi and Olson 1988). The main differentiation between enduring and situational involvement is temporal duration and thus persistence, with enduring involvement less temporal and more persistent than situational involvement (Richins and Bloch 1986, Warrington and Shim 2000). Situational involvement is characterised by its fluctuating nature caused by the temporary arousal of the consumer’s interest inherently linked to the timeframe of the purchase decision (Park and Moon 2003), and/or the use of a product in a specific situational context (Warrington and Shim 2000), such as ordering the WBG option from the wine menu in a restaurant.

By its nature, enduring involvement is a phenomenon that has more stability in its representation of the personal interest of the consumer in the product over an extended period of time (Park et al. 2007). Moreover, a direct association between a person’s self-concept and enduring product involvement was found (Roe and Bruwer 2017), meaning that a potential relationship exists between enduring involvement and psychological risk. In the case of wine in a restaurant to accompany a meal, consumers who feel that the WBG options resonate strongly with their self-concept, will be more motivated to make a “satisfactory purchase”.

Product and/or environmental factors induce states of arousal and interest in both enduring and situational involvement (Guthrie and Kim 2009). These occur because both have similar behavioural responses in the form of attention to product-related information and motivation-induced search (Bruwer and Cohen 2019) for information and dissemination (Richins and Bloch 1986; Celsi and Olson 1988; Ogbeide and Bruwer 2013). Environmental aspects, inherent in the risk perceived when dining in a restaurant, induce the interest and arousal associated with situational involvement. Park and Moon (2003) found that situational involvement’s fluctuating nature is inherently linked to the timeframe of a purchase decision, which is typically present in the ordering of a WBG serving from the wine menu during a dining occasion at a restaurant.

The product involvement construct has long been associated with wine (Zaichkowsky 1985), and several later studies have utilised this premise and further developed the situational context of wine and consumer involvement (Charters and Pettigrew 2006; Ogbeide and Bruwer 2013; Bruwer et al. 2019). There is also evidence that information seeking and dissemination have associations with involvement (Ogbeide and Bruwer 2013). Information seeking furthermore, has an important role in wine purchasing in its use in response to risk perception by acting as a RRS (Johnson and Bruwer 2004).

2.2 Perceived risk

Perceived risk as a concept has a history dating back to the 1920s when it was first used in the economics and eventually in the marketing field (Bauer 1960). Its development has led to the belief that a better understanding thereof could be an important initial step in developing marketing strategies (Phillips and Hallman 2013). Since then, risk perception has been conceptualised in various ways but the one most commonly used by researchers is that of the consumer being faced by feelings of uncertainty when unable to foresee, in particular, negative outcome(s) of a buying decision (Mitchell 1999). It was found that the consequences, magnitude and likelihood that adverse consequences may occur if the product is purchased, are mapped into the perceived risk construct (Dowling and Staelin 1994).

The contemporary consumer perceived risk construct is multi-dimensional (Mitchell 1999; Blais and Weber 2006) and incorporates the various dimensions of risk (Bruwer et al. 2017; Bruwer and Cohen 2019). The construct currently comprises six risk dimensions: physical, functional, financial, time, social and psychological risk (Johnson and Bruwer 2004). These comprehensively account for and include the uncertainty(ies) a consumer may perceive in the WBG consumption situation in restaurants (Bruwer et al. 2017). The functional risk dimension measures the concern (wine) consumers have about the taste (performance) of the product, while the financial risk dimension measures the degree of possible financial loss. Physical risk is represented by the intensity of the feeling of discomfort and/or enduring detrimental health effects upon consuming the product. Psychological risk denotes the potential loss of self-esteem and negative affective states such as anxiety, guilt, etc. if the product is not consistent with persons’ perception of self-identity (Blais and Weber 2006). Social risk perception occurs when significant others will not approve of the product, in other words, another person is the source of this uncertainty (Bohnet et al. 2008). Time risk is the potential loss of time associated with making a (bad) purchase decision in terms of the time spent researching and purchasing the product (Bruwer et al. 2017). As a counterintuitive response, consumers adopt various RRS to help them cope with the situation, in the process exhibiting behaviours that are dependent upon the degree of importance of a particular type of risk (Blais and Weber 2006).

There is evidence that a consumer’s perception of risk is influenced by his/her involvement with the purchase decision (Dowling and Staelin 1994; Dholakia 2001; Bruwer and Cohen 2019). Product involvement and perceived risk both act in motivating consumer behaviours from a motivational perspective. The extant literature is largely in agreement about the distinctiveness of the involvement and risk perception constructs (e.g., Dowling and Staelin 1994), although some authors suggest that perceived risk is either an antecedent (Bloch 1981), or a dimension of product involvement, or a consequence thereof (Laurent and Kapferer 1985).

The process of consumers purchasing wine products has generally been associated with, at times, high complexity and resultingly the perception of risk (Johnson and Bruwer 2004). This association with risk perception also transcends to the hospitality situational context and is a major factor affecting consumer decision-making (Bruwer and Rawbone-Viljoen 2013). Often risk perception depends on consumers’ interpretation of a specific consumption situation (Bruwer et al. 2017), e.g., when faced with the decision of ordering WBG in a restaurant setting. Lacey et al. (2009) found that, in the Australian restaurant sector, tasting a small wine sample before ordering a larger quantity, reduces the majority of perceived risk evoked by the taste of the product. The situational aspect in which WBG is ordered and consumed is therefore helpful towards developing deeper insights into the behavioural responses that consumers have to perceived risk. Their responses are thus dependent upon their interpretation of this situation.

Despite it being referred to in the discussion of risk perception occurring in the on-premise trade (Bruwer and Rawbone-Viljoen 2013), a limited number of research studies (Terrier and Jaquinet 2016; Bruwer et al. 2017; Bruwer and Cohen 2019; Acuti et al. 2020) have had as their focus WBG as RRS in restaurants. Although the WBG strategy is RRS-specific to the consumption situation in restaurants, its relevance to each type of risk encountered in this situational context, warrants further investigation.

2.3 Search for and dissemination of information

Because much choice and buying behaviour is preceded by information search behaviour in some form, such search behaviour has been a focal topic of research (Peterson and Merino 2003). Information search activities have the role of being important RRS (Lacey et al. 2009; Bruwer et al. 2013). Decisions to purchase vary in complexity, requiring even greater effort when searching for product information for the purpose of RRS (Lacey et al. 2009) and, especially so in a high-risk perception environment such as WBG ordering in a restaurant with a meal.

The extant literature provides evidence of differences that exist between hedonic and utilitarian products when consumers process information pertaining to these product types (Laurent and Kapferer 1985; Palazon and Delgado-Ballester 2013). Alike craft beer (Bruwer and Cohen 2022), WBG fits the context of a hedonic product which provides a sensory experience of fun and/or pleasure, especially for consumers with an experiential orientation (Bűttner et al. 2013). The situational context is an inherent aspect in the restaurant environment where WBG consumption occurs and adds to the hedonic perspective our study provides.

Previous research found that the type of involvement, whether enduring or situational, can be a driver of consumer information processing method and intensity (Celsi and Olson 1988; Park and Moon 2003). The motivation to process information is associated with the consumer’s degree of involvement with the various informational stimuli.

A number of effects have been identified by research examining the relationship(s) between information-related behaviour with product involvement and risk perception (Phillips and Hallman 2013). Different types of word-of-mouth activities, such as opinion leadership and seeking, are influenced by enduring and situational involvement (Dholakia 2001) and by psychological and social risk perception (Lin and Fang 2006). Product- and/or situation-related information give rise to various types of risk perception aroused by the composite mental activity of processing such information (Dholakia 2001). Moreover, information search behaviour to reduce the perception of risk occurs at two levels, with consumers routinely conducting information search in order to acquire general product information and, at times, to reduce a specific type of risk associated with a situation in which consumption occurs (Dowling and Staelin 1994). The need for cognitive closure also plays a role in consumers’ information search behaviour (Choi et al. 2008), and more specifically, in situations in which the importance of action in the form of decision is high, which is a scenario classically present with the WBG decision in restaurants.

Wine is a product type requiring much information to completely describe its characteristics in order to be able to make an informed purchase and/or consumption decision (Dodd et al. 2005; Lesschaeve et al. 2012; Ellis and Caruana 2018). Investigating information-related processing and behavioural roles in the motivational processes leading to the actioning of the WBG buying decisions, thus has theoretical value.

2.4 Conceptual model of constructs’ relationships

The psychological processes and motivational consumer behaviour constructs can be theoretically explained by a conceptual model of product involvement, perceived risk, and information processing (Dholakia 2001). This should also hold for how these relate to the restaurant environment, and specifically the WBG option (Bruwer and Cohen 2019). Our study develops a motivational process model that offers a complete and integrated view of how the latent constructs relate to and influence one another (see Figure 1). A premise of this model is that all the latent constructs have a distinct nature and influence one another. The execution of the model is instrumental in contributing to the aforementioned gap in the literature.

Figure 1. Conceptual model of wine by-the-glass (WBG), product involvement, risk perception and information processing.

Based on the theoretical linkages between the discussed latent constructs, in the first instance, we propose that:

H1: Enduring involvement (E_I) positively influences a) situational involvement (S_I) with the wine product class, and b) psychological risk (Psych_R) perception of the WBG option.

When dining out in a restaurant, the feeling of situational involvement with wine is almost unavoidable when the decision is made to have, or not have, wine with the food. Despite the temporal nature of this emotional state, the subsequent motivational force created by situational involvement (Richins and Bloch 1986), is likely to motivate consumers to spend more effort on considering the elements of risk (psychological, social, functional, physical, financial, and time). When situational involvement is driven by enduring involvement, this further increases the consumer’s feelings of anxiety in what additionally is a product-related situation (wine with food in a restaurant) (Roe and Bruwer 2017). Based on this argument, we contend that:

H2: Situational involvement (S_I) positively influences the perception of a) psychological risk (Psych_R), b) physical risk (Phys_R), c) financial risk (Fin_R), d) social risk (Social_R), e) time risk (Time_R), and f) functional risk (Funct_R) aroused by the WBG option.

The risk first experienced mainly at an affective level (physical, financial and time), is likely to be next subjected to evaluation of the cognitive kind and resulting in the subsequent experiencing of specific cognitive risks (Flynn and Goldsmith 1993). As a result of the motivated state aroused by situational involvement in the restaurant environment, the consumer is likely to be stimulated to put extra effort into evaluating the cognitive risk elements (psychological, social, and functional) (Watson and Spence 2007). Psychological risk is likely to result because of what underlies this anxiety (Dholakia 2001). Although psychological risk perception is an antecedent of social and functional risk (Flynn and Goldsmith 1993; Dholakia 2001), we expand on this by positing that, in the WBG situation in restaurants, psychological risk is also an antecedent of the non-cognitive financial, physical and time risks, and we therefore posit that:

H3: The arousal of psychological risk (Psych_R) perception by the WBG option, positively influences the perception of a) social risk (Soc_R), b) functional risk (Funct_R), c) physical risk (Phys_R), d) financial risk (Fin_R), and e) time risk (Time_R).

The final component of the motivational process model is represented by behavioural responses which are the results of the state(s) of motivation arising from the consumer’s product involvement and the experience of risk. The consumption situation of WBG with a meal in a restaurant “creates’ situational involvement that may directly influence information seeking without the process of cognitive risk evaluation occurring (Dholakia 2001). The information-related behaviours of actively seeking product-related information and giving product advice to others are therefore, directly influenced by consumers’ involvement levels (Richins et al. 1992). Regarding wine as a product, there is broad consensus that consumers use information seeking activities primarily as a RRS (Johnson and Bruwer 2004; Lacey et al. 2009; Bruwer et al. 2013). The motivational process model employed in our study (Figure 1), shows the pathways along which the emotions resulting from the consumer’s situational and enduring involvement with wine as a product operate (Bruwer and Cohen 2022). These pathways are provided by the cognitive risk elements of social, psychological and functional risk perception evoked in the consumption situation (Bruwer et al. 2017). We therefore propose that:

H4: Seeking product-related information (I_S) before purchasing WBG is positively influenced by a) enduring involvement (E_I) and b) situational involvement (S_I) with the wine product class; and the perception of c) psychological risk (Psych_R), d) social risk (Soc_R), e) functional risk (Funct_R), f) physical risk (Phys_R), g) financial risk (Fin_R) and h) time risk (Time_R) aroused by the WBG option.

Evidence suggests consumers disseminate information about a product in the form of opinions and/or advice mainly because of the extent to which they perceive risk (Ogbeide and Bruwer 2013, and their enduring involvement therewith (Shoham and Ruvio 2008). They also engage in information dissemination because of the motivation power of the situational involvement force (Bloch and Richins 1983; Park and Moon 2003; Park et al. 2007), and their perception of psychological and social risk (Lin and Fang 2006). Social risk perception is particularly evident in a public consumption-purchase decision situation, which WBG consumption in a restaurant, often within a multi-person dining group, embodies (Mazzarol et al. 2007). The provision of opinions and advice occur when consumers use their involvement with the consumption situation and the product for the purposes of self-affirmation to receive reassurance from significant others, thus invoking an influence exerted by psychological risk perception (Shoham and Ruvio 2008). We therefore derive:

H5: Providing product information (I_P) before purchasing WBG is positively impacted by a) enduring involvement (E_I) and b) situational involvement (S_I) with the wine product class; and also, with c) psychological risk (Psych_R) perception aroused by the WBG option.

Figure 1 shows the conceptual model of the focal constructs, the proposed influences between constructs and on consumers’ information-related behavioural responses.

3. Methodology

3.1 Study context

We chose the South African restaurant industry as the focus of our study. South Africa is the largest economy on the African continent and a significant entity in the world wine market. The 2020 revenue of South Africa’s on-premise foodservice industry amounted to ZAR 81.6 billion ($US 5.68 billion) and its domestic wine market revenue to ZAR 47.2 billion ($US 3.3 billion equivalent). In the large markets in Europe (e.g., the UK), and also in the USA, the on-premise sector typically has a volume share of around 20%, while by contrast in South Africa the comparative figure is 32% with this volume contributing 38% of wine’s sales value in South Africa (Euromonitor International 2021). Considering that the country’s consumer food-service industry is valued at ZAR 130.3 billion ($US 9.0 billion) with drinks contributing 29.6% of that (Euromonitor International 2022), wine is therefore a key contributor to the South African restaurant industry, accounting for a significant portion of restaurant revenue.

3.2 Measures

Data collection to test the study’s hypotheses was achieved by executing an online survey among dining patrons of restaurants in South Africa on a nationally geographically representative sample of restaurant dining patrons. The collection of the data was completed in late 2019, just prior to the outbreak of the covid-19 pandemic, meaning that the pandemic and actions such as “lockdowns’ had no effect on the results of our study. The survey instrument consisted of three sections: seven-point Likert measurement scale items for product involvement, WBG risk perception, and information seeking-dissemination, as well as associated consumption metrics and, also socio-demographics.

The product involvement (enduring and situational) construct was measured using scale items from Bruwer and Cohen (2019), following their adjustment to the contexts of the wine product category and the situational aspect in restaurants. After adjustment to the wine and restaurant contexts, we used items from Bruwer and Cohen (2022) to measure information seeking and dissemination. The risk perception construct and its elements were measured using 24 adapted scale items from Bruwer and Cohen (2022) and Dholakia (2001).

We used procedural (ex ante) remedies to control for common method variance (CMV) as a countering action to the self-reported nature of the data (Mackenzie and Podsakoff 2012). In the process, we adopted counterbalancing questions order and different response formats. For example, we asked respondents about their product involvement and risk perception before asking about their information processing to avoid priming effects. Respondents had to answer all questions thus there was no missing data and no non-response bias issues.

We used a decompositional approach as in Taylor and Todd (1995) to measuring the respective constructs. Such an approach has been previously used in studies of involvement (Richins and Bloch 1986; Ogbeide and Bruwer 2013) and risk perception (Dholakia 2001; Bruwer et al. 2017).

3.3 Sample and procedure

Data collection was by means of an online survey obtained from a geographically representative sample of restaurant dining patrons in South Africa. The participants were recruited from the national data base of a professional third-party consumer panel provider. The sample frame consisted of people from their South African “general population” database who complied with certain qualification criteria: wine consumers, 18 years and older (in accordance with the country’s alcohol consumption regulations), dined out at a venue fitting the restaurant classification (Euromonitor International 2022) at least once in the past month, where they bought and consumed WBG. The nomenclature of the seven restaurant categories is the result of adapting the USA categorisation (Canziani et al. 2016) and combining that with the South African consumer food-service industry categorisation (Euromonitor International 2022), which will aid future studies.

We collected information from respondents during a 7-day time frame ensuring coverage of all week and weekend days. An online survey was programmed in Qualtrics using the panel provider’s online survey platform. A series of decision rules and attention check questions were implemented to obtain quality data. Additional to the advantage an online survey has of eliminating invading the privacy of diners when collecting information in restaurants in situ (Bruwer et al. 2017), our survey successfully sampled respondents throughout South Africa and across all of its restaurant categories. We contacted 4,267 people with 2,347 initially agreeing and qualifying to participate in the research study. A total of 1,250 persons completed it from which 1,038 surveys could be utilised for final analysis (44% response rate). Data analysis was executed by means of SPSS 26 and the R package Lavaan software.

The socio-demographics and consumption metrics of the respondents are exhibited in Table 1. The gender split is 55% female and 45% male and the average age of the sample is 43.9 years, which is close to the mid-point of the Generation-X age cohort. More specifically, the age generation cohort distribution (Pew Research Center 2015) is 32% Millennials (born after 1980), 39% Generation-X (born 1965–80), 23% Baby Boomers (born 1946–1964), and 2% Silent Generation (born 1928–45). Table 1 also shows that 68% are married or living with a significant other in a household in which 3.4 persons live together (usually 2 adults and 1 child); education wise 70% possess a university or post-secondary qualification; and have median monthly earnings of ZAR 51,979 (US$ 1.00 = ± ZAR 18.00) in the household.

Table 1. Socio-demographics, restaurant category patronage and WBG metrics of respondents.

Socio-demographic variable	Males	Females	Total (n = 1,038)
Gender	45.3%	54.7%	100.0%
Age generation
18–37 years old Millennials	48.6%	27.9%	35.3%
38–53 years old Generation-Xers	29.9%	44.6%	39.4%
54–72 years old Baby Boomers	18.8%	25.7%	23.2%
73–90 years old Silent Generation	2.7%	1.8%	2.1%
Age mean (years)	40.9 years	45.6 years	43.9 years
Gross monthly household income*
Household income (median)	ZAR 45,377	ZAR 55,996	ZAR 51,979
Educational status (highest level)
No post-secondary qualification	22.3%	23.2%	22.9%
Undergraduate post-secondary qualification	56.2%	53.1%	54.2%
Postgraduate post-secondary qualification	21.5%	23.7%	22.9%
Marital status
Married or co-habiting with de facto partner	67.1%	69.0%	68.3%
Single	26.6%	16.9%	20.3%
Divorced, separated or widowed	6.3%	14.1%	11.4%
Persons in household	3.52 persons	3.28 persons	3.37 persons
Restaurant category#
Full-service family restaurant	28.0%	39.0%	35.1%
Specialist steakhouse (e.g., Spur Group)	34.8%	17.0%	23.3%
Fine-dining upscale restaurant	15.8%	18.7%	17.6%
Casual ethnic food-service (e.g., Italian)	9.5%	18.1%	15.0%
Bar/pub/tavern	7.9%	3.6%	5.1%
Licensed café (e.g., Supermarket in-store)	2.4%	3.0%	2.8%
Quick-service/limited-service restaurant	1.6%	0.6%	1.1%
Wine consumption on the dining occasion
Dining group size (persons)	3.50	3.79	3.69
No. of WBG servings consumed by respondent	2.14	1.82	1.95
No. of 750 ml bottles also consumed by group	0.88	0.91	0.89
No. of WBG servings equiv. consumed by group	6.52	6.36	6.42
Amount spent^a on wine on the dining occasion	ZAR 201.90	ZAR 185.73	ZAR 191.84

^a ZAR Rands # Categorisation as per Canziani et al. (2016) and Euromonitor International (2022).

As far as the respondents’ last dining experience before participating in the study is concerned, Table 1 reveals that the most frequented restaurant category was full-service family restaurants (35%), with specialist steakhouses (23%), fine-dining upscale (18%), casual ethnic food-service (15%), and all other restaurant categories (9%) following. On this dining occasion the WBG consumption metrics reveal that the consumers spent an average of ZAR 192 on wine (WBG servings and full bottle) and also that the dining groups were multi-person in nature (an average of 3.7 persons per group). A mere 2% of respondents wasn’t part of a multi-person group, electing to dine out alone. The multi-person nature of the dining groups could, at least in part, explain why respondents ordered 1.95 WBG servings (150 ml. size each) on average. Moreover, the multi-person nature of the dining group could explain why, for the table, an additional 0.9 standard-size 750 ml bottle(s) was/were also ordered by the diners who perhaps preferred either a different wine type to the available WBG options available, or felt that WBG was too expensive relative to the quantity of wine offered, etc. Males consumed volumetrically 18% more and spent 9% more money on WBG than females in the dining situation. This is put in some perspective by the fact that the overall monthly wine consumption of females is 11% less than that of males.

3.4 Structural equation model (SEM) fitting

To characterise our dataset’s variability, we conducted skewness and kurtosis tests. Skewness ranged from−0.894 to−1.799, thus below the threshold of 1.0. The kurtosis for all items ranged between−1.224 and 1.634 and conform with the threshold level of between−2.00 and 2.00 (George and Mallery 2010). The tests for skewness and kurtosis thus confirm that the distributions of the collected data are acceptable for conducting SEM analysis. SEM is a particularly suitable analysis technique when several dependent and independent variables are involved in complex models, to test the theories involved (Bagozzi and Yi 2012; Nunkoo et al. 2013). We executed CB-SEM analysis to examine the five constructs in our model using the two-step procedure due to its greater incidence of use by researchers (Nunkoo et al. 2013; Miranda and Duarte 2022).

4. Results and discussion

4.1 Structural model evaluation

Given the fact that previous attempts in the form of exploratory studies in the restaurant domain treated perceived risk as a unidimensional construct (Terrier and Jaquinet 2016; Bruwer et al. 2017) and did not comprehensively test the scale(s)” discriminant validity and reliability, these were deemed necessary requirements in our study. Hence, we first tested the psychometric properties of the WBG risk scale’s reliability, discriminant, nomological and convergent validity. We next conducted confirmatory factor analysis (CFA) with the Lavaan R software package structured in accordance with the risk perception framework (Klerck and Sweeney 2007) to obtain each item’s loading on the dimensional variables of WBG risk perception. Cronbach alphas for the constructs’ scales to determine their reliability were also computed. We used the following inclusion criteria to identify the final set of items for the WBG risk perception scale:

• – Standardised loadings on each latent construct’s selected items are >0.5

• – Items selected for each latent construct, combine for a composite AVE >0.50 and alpha >0.70

The CFA’s outputs indicate satisfactory coefficient loading at > .50 of 24 WBG risk scale items on their dimensional factors and all goodness-of-fit (GoF) statistics above the minimum thresholds (see Table 2). These procedures resulted in a final 24 item-set being selected, following the removal of one and the reversal of two scale items (see Appendices A and B).

Table 2. WBG risk perception constructs confirmatory factor analysis (CFA).

Factors	Mean	SD	Loading	AVE	Construct reliability	Z-stat	p > [z]
F1: Functional risk				0.533	0.733	22.303	<0.001
WBG_R1 (reversed)	5.55	1.408	0.113			2.413	0.016
WBG_R2	4.31	1.735	0.742			30.173	<0.001
WBG_R3	3.61	1.705	0744			28.075	<0.001
WBG_R4	4.38	1.707	0.753			28.550	<0.001
F2: Physical risk				0.586	0.735	28.334	<0.001
WBG_R5	3.49	1.766	0.553			19.019	<0.001
WBG_R6	4.94	1.725	0.781			34.747	<0.001
WBG_R7	4.32	1.721	0.741			31.235	<0.001
F3: Financial risk				0.519	0.723	22.263	<0.001
WBG_R8	5.05	1.560	0.512			24.614	<0.001
WBG_R9	4.09	1.864	0.604			27.094	<0.001
WBG_R10 (reversed)	3.74	1.692	0.134			22.081	0.037
F4: Social risk				0.646	0.865	29.598	<0.001
WBG_R11	2.70	1.689	0.544			14.178	<0.001
WBG_R12	2.29	1.515	0.732			30.065	<0.001
WBG_R13	2.01	1.332	0.823			39.947	<0.001
WBG_R14	2.04	1.302	0.691			22.568	<0.001
WBG_R15	2.11	1.390	0.833			42.814	<0.001
WBG_R16	1.93	1.235	0.772			28.013	<0.001
F5: Psychological risk				0.686	0.885	30.119	<0.001
WBG_R17* (deleted)	5.18	1.569	-0.098			-	-
WBG_R18	1.95	1.236	0.744			25.984	<0.001
WBG_R19	1.91	1.193	0.573			14.938	<0.001
WBG_R20	2.03	1.295	0.851			42.916	<0.001
WBG_R21	2.04	1.307	0.824			36.090	<0.001
WBG_R22	2.16	1.461	0.793			30.665	<0.001
F6: Time risk				0.513	0.711	22.440	<0.001
WBG_R23	4.60	1.833	0.573			18.583	<0.001
WBG_R24	3.41	1.671	0.721			26.130	<0.001
WBG_R25	3.80	1.816	0.674			22.608	<0.001

Measurement using 7-point Likert Scale: 1 =  Strongly Disagree …  … ..7 = Strongly Agree.

Goodness-of-fit stats: Chi-square: 1319.94 (452) RMSEA = 0.059 GFI = 0.916 AGFI = 0.907 p = 0.000.

^a Scale item deleted from the final WBG risk perception scales used in SEM model fit.

For the six perceived risk dimensions (factors), all alpha reliability coefficients and AVEs were satisfactory (Table 2). These values range from alpha = .885 and AVE = .686 (psychological risk) to alpha = .711 and AVE = .513 (time risk), thus comfortably satisfying the Cronbach alpha criterion of .700 or greater and AVE 0.50 or greater (Nunnally and Bernstein 1994). When considering that social risk (alpha = .865 and AVE = 0.646) had the second highest values after psychological risk, the fact that these two cognitively evaluated risk dimensions produced robust results, is a positive outcome in that it confirms the strong impact of psychological and social risk on the product involvement construct (Dholakia 2001). The WBG risk scale yielded a satisfactory average inter-item covariance of 0.374 and composite alpha of 0.833. While achieving the desired Cronbach alpha threshold, the physical, financial, functional, and time risk dimensions in our model had acceptable, although relatively low AVEs values, which resulted in the reverse coding of two items (functional and financial risk) and the deletion of one (psychological risk). There is support for the retention of the constructs with lower AVEs due to one or two weaker latent variables in the SEM measurement model (Hair et al. 2011). The CFA results nevertheless confirm the convergent and discriminant validity of the WBG risk measurement scale with both the factor loadings ≥0.5 and AVE values ≥ the squared correlations (SCs) of the constructs (functional and financial risk excepted, before item reversal) (Sideridis et al. 2014).

In the case of the enduring involvement, situational involvement, information seeking and information provision scales, all alpha reliability coefficients and AVEs were satisfactory. Table 3 shows that these values range from alpha = .798 and AVE = .601 (enduring involvement) to alpha = .912 and AVE = .838 (information seeking).

Table 3. Reliability analysis of involvement and information processing constructs scale items.

Latent construct	Inter-item covariances	AVE	Alpha	Chi-Square	p
Enduring Involvement (E_I)	1.404	0.601	0.798	409.360	0.000
Situational Involvement (S_I)	1.683	0.664	0.808	57.301	0.000
Information Seeking (I_S)	2.305	0.838	0.912	8.857	0.000
Information Provide (I_P)	2.008	0.692	0.818	105.109	0.000

The results in Table 2 and Table 3 indicate the empirical distinctiveness of the latent constructs, thus facilitating fitting the measurement model of the motivational process.

4.2 Measurement model fit

The model was estimated using the R package of Lavaan’s SEM module. The model’s GoF statistics and other parameters are presented in Table 4 and Figure 2 and show the significant and non-significant structural paths. Confirmation of the satisfactory overall fit of our model is provided by the GoFs in Table 4. The .057 RMSEA of our model complies with the recommended RMSEA threshold of .08 or less (Sideridis et al. 2014). Our model also complies with all other SEM model GoF thresholds (Nunkoo et al. 2013) in that GFI = .958 (recommended GFI is ≥ .900), AGFI = .917 (recommended AGFI is ≥ .900), and SRMR = .051 (recommended SRMR is ≤ .08).

Figure 2. Estimated model of wine-by-the glass (wbg) product involvement, risk perception and information processing.

Table 4. Final SEM outer model loadings outputs.

Latent variable	Item	Loading	Standard error	z-stat	P
Enduring Involvement (E_I)	E_I1	0.704	0.025	27.641	<0.001
Enduring Involvement (E_I)	E_I2	0.683	0.026	26.727	<0.001
Enduring Involvement (E_I)	E_I3	0.671	0.028	23.939	<0.001
Situational Involvement (S_I)	S_I1	0.771	0.022	35.250	<0.001
Situational Involvement (S_I)	S_I2	0.822	0.025	32.579	<0.001
Functional Risk (Funct_R)	WBG_1_r	0.104	0.044	2.196	0.028
Functional Risk (Funct_R)	WBG_2	0.742	0.025	29.939	<0.001
Functional Risk (Funct_R)	WBG_3	0.741	0.027	27.893	<0.001
Functional Risk (Funct_R)	WBG_4	0.764	0.026	29.022	<0.001
Physical Risk (Phys_R)	WBG_5	0.533	0.030	17.610	<0.001
Physical Risk (Phys_R)	WBG_6	0.760	0.033	23.163	<0.001
Physical Risk (Phys_R)	WBG_7	0.773	0.032	24.254	<0.001
Financial Risk (Fin_R)	WBG_8	0.941	0.285	3.289	<0.001
Financial Risk (Fin_R)	WBG_9	0.333	0.099	3.093	0.002
Financial Risk (Fin_R)	WBG_10_r	0.104	0.044	2.196	0.028
Social Risk (Social_R)	WBG_11	0.442	0.031	13.888	<0.001
Social Risk (Social_R)	WBG_12	0.732	0.025	29.090	<0.001
Social Risk (Social_R)	WBG_13	0.832	0.021	39.460	<0.001
Social Risk (Social_R)	WBG_14	0.694	0.030	22.778	<0.001
Social Risk (Social_R)	WBG_15	0.834	0.019	42.470	<0.001
Social Risk (Social_R)	WBG_16	0.771	0.028	27.183	<0.001
Psychological Risk (Psych_R)	WBG_18	0.754	0.028	26.474	<0.001
Psychological Risk (Psych_R)	WBG_19	0.582	0.039	14.970	<0.001
Psychological Risk (Psych_R)	WBG_20	0.853	0.020	42.913	<0.001
Psychological Risk (Psych_R)	WBG_21	0.811	0.024	34.415	<0.001
Psychological Risk (Psych_R)	WBG_22	0.794	0.027	29.141	<0.001
Time Risk (Time_R)	WBG_23	0.572	0.034	16.930	<0.001
Time Risk (Time_R)	WBG_24	0.753	0.036	20.526	<0.001
Time Risk (Time_R)	WBG_25	0.634	0.036	17.537	<0.001
Information Provide (I_P)	I_P1	0.814	0.022	38.491	<0.001
Information Provide (I_P)	I_P2	0.852	0.021	38.853	<0.001
Information Seeking (I_S)	I_S1	0.923	0.016	56.964	<0.001
Information Seeking (I_S)	I_S2	0.913	0.015	60.309	<0.001

Goodness-of-fit stats: Chi-square = 1230.831 (439) RMSEA = 0.057 GFI = 0.958 AGFI = 0.917.

SRMR = 0.051; p = 0.000.

4.3 Structural parameters for hypothesised paths

Table 5 displays the path estimates and structural parameters in relation to the study’s hypotheses, while the final measurement model (Figure 2) shows the significant hypothesised paths.

Table 5. Parameter estimates for hypothesised paths.

Path between constructs	β (Beta)	Stan. Error	z-stat	p > [z]	Result
H1a: END_Involv→SIT_Involv	0.514	0.042	12.294	<0.001***	Supported
H1b: END_Involv→Psycho_Risk	0.268	0.054	4.986	<0.001***	Supported
H2a: SIT_Involv→Psycho_Risk	0.104	0.047	2.119	0.034**	Supported
H2b: SIT_Involv→Phys_Risk	0.133	0.060	2.229	0.026**	Supported
H2c: SIT_Involv→Fin_Risk	0.217	0.097	2.242	0.025**	Supported
H2d: SIT_Involv→Social_Risk	0.073	0.035	2.006	0.045**	Supported
H2e: SIT_Involv→Time_Risk	0.195	0.059	3.309	<0.001***	Supported
H2f: SIT_Involv→Funct_Risk	0.129	0.041	3.124	0.002***	Supported
H3a: Psych_Risk→Social_Risk	0.688	0.039	17.603	<0.001***	Supported
H3b: Psych_Risk→Funct_Risk	0.372	0.036	10.233	<0.001***	Supported
H3c: Psych_Risk→Phys_Risk	-0.031	0.054	-0.579	0.563	Not supported
H3d: Psych_Risk→Fin_Risk	-0.066	0.043	-1.541	0.123	Not supported
H3e: Psych_Risk→Time_Risk	0.206	0.055	3.743	<0.001***	Supported
H4a: END_Involv→INFO_Seek	0.155	0.046	3.385	<0.001***	Supported
H4b: SIT_Involv→INFO_Seek	0.443	0.044	9.986	<0.001***	Supported
H4c: Psych_Risk→INFO_Seek	0.235	0.049	2.913	<0.001***	Supported
H4d: Social_Risk→INFO_Seek	0.077	0.045	2.013	0.042**	Supported
H4e: Funct_Risk→INFO_Seek	0.072	0.036	1.985	0.047**	Supported
H4f: Phys_Risk→INFO_Seek	0.039	0.039	1.005	0.315	Not supported
H4g: Fin_Risk→INFO_Seek	0.072	0.038	1.989	0.048**	Supported
H4h: Time_Risk→INFO_Seek	-0.019	0.040	-0.486	0.627	Not supported
H5a: END_Involv→INFO_Prov	0.408	0.047	8.720	<0.001***	Supported
H5b: SIT_Involv→INFO_Prov	0.308	0.048	6.489	<0.001***	Supported
H5c: Psych_Risk→INFO_Prov	0.277	0.035	3.913	<0.001***	Supported

***z < 2.58 **z < 1.96 *z < 1.645.

The analysis confirms that consumers’ level of enduring involvement (E_V) with wine as a product positively affects both consumers’ situational involvement (S_V) with the WBG options in restaurants (H_1a: β = 0.514; z = 12.294***), while at the same time, activating their psychological risk perception (H_1b: β = 0.268; z = 4.986***). The results also show that consumers’ situational involvement with WBG in restaurants significantly affects all the risk types evoked by the WBG option: psychological (H_2a: β = 0.104; z = 2.119**), physical (H_2b: β = 0.133; z = 2.229**), financial (H_2c: β = 0.217; z = 2.242**), social (H_2d: β = 0.073; z = 2.006**), time (H_2e: β = 0.195; z = 3.309***), and functional risk (H_2f: β = 0.129; z = 3.124***). This deviates from the finding of Bruwer and Cohen (2022) who found that, in the case of craft beer in restaurants, situational involvement only significantly affects psychological, but not social and functional risk perception. This finding is both insightful and a contribution to the theory base in that it confirms, across the full spectrum of risk types, that the WBG consumption situation in restaurants activates these emotions and that the perception of risk varies in intensity. The adoption of risk-reduction strategies (RRS) by consumers can mitigate the negative consequences on their behaviour as later further described in the implications sub-section of this document.

Table 5 shows mixed results for our supposition that aroused psychological risk is not only an antecedent of social and functional risk, but also of physical, financial, and time risk in the WBG situation in restaurants. Psychological risk positively affects social (H_3a: β = 0.688; z = 17.603***), functional (H_3b: β = 0.372; z = 10.233***) and time risk (H_3e: β = 0.206; z = 3.743***), but not physical (H_3c: β = -0.031; z = -0.579) and financial risk (H_3d: β = -0.066; z = -1.541). This contributes to theory by reaffirming the cognitively-evaluated risk relationship between psychological and social and functional risk, and adds time risk thereto. Further work on exploring the effects of psychological risk on financial and physical risk is needed.

Consumers’ inclination to obtain product-specific information (I_S) prior to purchasing the WBG option was strongly positively affected by both enduring involvement (E_I) with the wine product (H_4a: β = 0.155; z = 3.385***) and situational involvement (S_I) with the WBG option (H_4b: β = 0.443; z = 9.986***). Among the six risk types, four positively affected inclination to seek product-related information: psychological (H_4c: β = -0.235; z = 2.913***), social (H_4d: β = 0.077; z = 2.013**), functional (H_4e: β = 0.072; z = 1.985**) and financial (H_4g: β = 0.072; z = 1.989**), while the two other risk types did not support the hypotheses: physical (H_4f: β = 0.039; z = 1.005) and time (H_4h: β = -0.019; z = -0.486). This finding underlines the importance of sound product information (functional) and perception of value (financial) and confirms the pivotal role of the cognitively-evaluated risk types (psychological social and functional) in the purchase process in the situational milieu of restaurants.

Finally, the proposed impacts of enduring involvement (E_I) (H_5a: β = 0.408; z = 8.720***) and situational involvement (S_I) (H_5b: β = 0.308; z = 6.489***) on the consumer’s propensity to distribute information (I_P) related to the product before purchasing the CB option were also supported, as was the impact of aroused psychological risk (H_5c: β = 0.277; z = 3.913***). Consumers with a high level of enduring involvement are likely to act as opinion leaders on the occasion of a dining group having a meal at a restaurant, based on the finding that the level of enduring involvement positively affects the need for consumers to disseminate information about wine to others.

Figure 2 shows the structural pathways of the hypotheses that were supported (solid line) and those not supported (dotted line). The results of the concepts validated in the SEM model, confirm the scales’ nomological validity and provide enrichment of their theoretical network.

5. Conclusions, implications, limitations, and future research recommendations

The current study makes a substantive contribution to the theory base by providing nuanced insights from the perspective of holistic and consumer-centric behavioural psychology. First, by creating, confirming and further developing consumer behavioural psychology insights pertaining to the focal consumer-centric constructs. Second, by developing a unique motivational process model, building further on the works of Bruwer et al (2017) and Bruwer and Cohen (2019), that explicates the processes that cause product involvement, risk perception and information-related responses and by which they impact one another. Third, by theoretically explaining the psychological processes in a cogent and uncomplicated way through this model. Finally, by successfully treating a theory-based motivational process model of risk perception as a multi-dimensional construct that includes involvement and information-related behavioural responses in the situational context of restaurants where consumers have exposure to a WBG ordering situation.

The research confirms the existence of relationships of a distinctive nature between the closely-related constructs of product involvement, perception of risk. and information-processing behaviour of consumers engaging with the WBG option when having meals in restaurants. Also confirmed are the situational involvement construct’s unstable nature, the stable nature of enduring involvement, and the pivotal role of psychological risk on social, functional and time risk as well, as on information processing. The enduring involvement state with wine brings about increased motivation from the situational aspects of WBG purchasing in the restaurant setting, by triggering situational involvement and the arousal of WBG risk perception of not only cognitively-evaluated risks (psychological, social and functional), but also non-cognitively-evaluated risks (financial, physical and time). The findings also confirm that WBG risk perception is a distinct multi-dimensional construct thus strengthening the knowledge base.

The results further contribute to theory by adding the time risk dimension as a non-cognitively-evaluated risk type to the already established cognitively-evaluated risk relationships that psychological risk have with social and functional risk (Dholakia 2001). The existence of the relationships between these cognitively-evaluated risk constructs is also reconfirmed. We further confirm that, for the wine product class, a positive relationship exists between information seeking behaviour and risk perception by consumers in restaurants. We also find that both enduring involvement and situational involvement, the latter in contrast to Dholakia (2001), influence both product-related information seeking and dissemination.

Although few studies have examined the WBG consumption aspect in restaurants with meals, these span three countries: Australia (Bruwer et al. 2017), the USA (Bruwer and Cohen 2019), and South Africa (current study). A comparison of the headline metrics shows the largest WBG consumption (2.76 serves per person; 51% female weighted, and dining group size of 4.61 persons) in the USA market (Bruwer and Cohen 2019), with Australia slightly “behind” (2.38 serves per person, 58% female weighted, and dining group size of 3.25 persons) (Bruwer et al. 2017). In South Africa, by comparison, there were 1.95 WBG serves per person, a 55% female component in a dining groups size of 3.69 persons in the current study. With only these indicators available to compare the maturity stage of the WBG market in South Africa with that of the other countries, it seems safe to say that it is most likely in the intermediate stage, which gives further credence to its potential to become more prominent in future.

A number of practical implications emanate from our study’s findings for developing sound marketing strategies. Because of their involvement with wine, consumers are motivated to try various types of wine, which, all things equal, should have an encouraging effect on them to buy WBG in a restaurant setting. For restaurant managers, making the WBG option available is a strategy capable of increasing consumers’ learning about, and interest in wine. For example, in partnership with restaurants they supply their wines to, wine tasting experiences could be organised to make consumers aware and enable them to taste various WBG options.

A good understanding of consumers’ involvement with both the wine product and the WBG situation in a restaurant is central for restaurateurs in their quest to reduce risk perception and its adverse effects among consumers. Because restaurants do not have control over consumers’ enduring involvement with the wine product, their strategies should aim at ways in which they can reduce risk perception evoked by consumers’ situational involvement for each risk type pertaining to the WBG option. These actions can be focused on, but are not limited to, sommeliers’ recommended WBG selections, user-friendly wine menus, suggestions for food and wine pairing, WBG special price offers, notice boards containing WBG information, responsible alcohol intake information, etc.

Facilitating the employment of RRS-specific to reducing pivotal psychological and social risks will go some way towards yielding more revenue for restaurants via the WBG option. Because of the dynamics between the members of dining groups, restaurants should strive to establish who the group’s decision-maker for ordering wine is and focus their efforts more on this person. Knowing who the consumer is remains key to the success of restaurants and information about the market segment(s) patronising a restaurant and, can also be put to good use to match the wine ordering situation with the type of consumer dining there. Consumers with high levels of enduring involvement are likely to act as opinion leaders based on our finding that the level of enduring involvement strongly affects the need for consumers to disseminate information about wine to others. Therefore, restaurants’ digital marketing strategies could be aligned with this. Training of restaurant staff should align with the abovementioned practical implications.

Similar to other studies, ours has some limitations. The WBG motivational process could also be influenced by other consumer behavioural psychology constructs, such as opinion leadership, objective and/or subjective product knowledge, and product experience. Although the social nature of restaurant dining is inherently associated with multi-person dining groups, the influence of dining companions was not measured in our study.

Our study also springboards some future research recommendations. Other psychological constructs such as opinion leadership, objective and/or subjective product knowledge, and product experience could be included and their relationship with the involvement and WBG risk perception constructs examined. “Informal” opinion leadership that occurs within a dining group in a restaurant when wine choices are made, is likely to yield important further insights. The influence of food, e.g., type of cuisine, quantity consumed, etc. could be included in further research on the WBG context in restaurants. Further work on all the effects of situational involvement in the dining situation is also needed. The proven importance via our study of situational involvement in the restaurant dining situation, warrants further work on all of the effects of situational involvement. The impact of specifically, psychological risk, on both information seeking and information dissemination, should also be further examined. The body of theoretical knowledge our study generated, can be further enhanced in future studies that examine interaction effects of both a mediating and moderating nature, following the outcomes of first testing the primary pathways. Future research could possibly explore post Covid-19 pandemic WBG and whether single-serve helps to act as a hygiene risk-reduction factor. Similar future studies conducted in South Africa, could examine if the race socio-demographic aspect has a strong influence on consumer behaviour, and the nature of such influence(s). From an expenditure viewpoint, all possible expenditure permutations could be unravelled by measuring the expenditure on food, non-alcoholic beverages, etc. and how these relate to WBG expenditure.

Ethics approval statement

The research team was judged to be acting in a collaborative role to the retail partners and not collecting any personal identifying information, and hence the Ethics Committee of the University indicated that this study did not require ethics approval.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by National Research Foundation: [Grant Number 109156].

Appendices

Appendix A: Measurement of the reliability of WBG perceived risk scale items

Table

Risk dimension	Scale item	Scale mean if item deleted	Scale variance if item deleted	Corrected inter-item total correlation	Cronbach’s alpha if item deleted
Functional	R1 Drinking WBG allows for buying smaller quantities instead of a full 750 ml bottle of wine (reversed)	12.29	18.663	0.069	0.839
	R2 I would be concerned the wine served to me in a wine-by-the-glass format would not taste good because of the bottle having been opened a few days ago already	10.43	11.605	0.581	0.521
	R3 I would be afraid that WBG would not provide me with the level of benefits that I expected it to ^#	11.13	11.559	0.605	0.503
	R4 When buying WBG, I would worry about how reliable the WBG would be in terms of my taste expectation ^#	10.36	11.954	0.560	0.541
Physical	R5 WBG because it reduces the feeling of discomfort caused by a hangover	9.27	9.430	0.456	0.741
	R6 WBG because it makes it easier to control alcohol consumption level	7.81	8.554	0.591	0.578
	R7 WBG because of recommended quantity alcohol intake for health benefits	8.43	8.567	0.592	0.577
Financial	R8 WBG makes it possible to try some of the more expensive wines	7.83	6.997	0.246	0.189
	R9 WBG costs less and gives me more control over how much money I spend in total on wine when dining out	8.80	5.663	0.262	0.129
	R10 When buying WBG I would be concerned that I may not get my money’s worth (reversed) ^#	9.14	7.577	0.110	0.440
Social	R11 WBG because I cannot pick a 750 ml bottle for the table	10.38	30.658	0.427	0.874
	R12 WBG because not picking the right wine at a reunion would make me feel a fool	10.79	27.935	0.704	0.813
	R13 My friends’ and colleagues’ opinions about me buying WBG would cause me to feel concern ^#	11.08	29.086	0.741	0.807
	R14 WBG buying makes me feel antisocial	11.04	31.057	0.603	0.832
	R15 When buying WBG, I would be concerned about what people whose opinions are of value to me, would think of me, if I made a bad choice ^#	10.97	28.549	0.742	0.806
	R16 My buying WBG would cause me concern about what my friends would think of me if I made a bad choice	11.15	30.795	0.670	0.822
Psycho-logical	R17* WBG makes it possible to try new brands because I can order a number of small servings instead of one bottle of a single brand (deleted)	–	–	–	0.867
	R18 The thought of buying WBG would fill me with anxiety ^#	8.14	18.676	0.697	0.838
	R19 WBG because buying wine makes me feel anxious	8.18	20.586	0.521	0.877
	R20 When I think about buying WBG, I would experience tension ^#	8.06	17.458	0.787	0.814
	R21 The thought of buying WBG would make me feel uncomfortable ^#	8.04	17.803	0.738	0.827
	R22 I would worry a lot when buying WBG ^#	7.92	16.909	0.717	0.834
Time	R23 WBG because it is quicker to drink with a quick meal when I don’t have much time	7.21	8.937	0.469	0.637
	R24 WBG because it is quicker to order as there are fewer choices on the wine list ^#	8.40	9.191	0.538	0.551
	R25 WBG because I am not delayed in ordering as my choice is independent of the wine choice(s) of my dining companion(s) ^#	8.01	8.758	0.501	0.595

Note: WBG = Wine by-the-glass; Measurement using 7-point Likert Scale: 1 =  Strongly Disagree …  … ..7 = Strongly Agree Scale’s overall Cronbach alpha = 0.8133.

# Adapted from Dholakia (2001) for the purposes of this study; All other items are from the WBG scale of Bruwer et al. (2017).

*Scale item deleted from the final overall WBG risk perception scale used in SEM model fit.

Appendix B: reliability of involvement and information processing constructs’ scale items

Table

Construct	Scale item	Mean	Standard deviation	Corrected inter-item total correlation	Cronbach’s alpha if item deleted
Enduring Involvement (END_Involv)	EI-1 Because of my personality, I would rate wine as being of the highest importance to me personally	3.39	1.765	0.576	0.610
	EI-2 I could make many connections or associations between important experiences in my life and wine	4.16	1.754	0.551	0.641
	EI-3 In general, my wine would help others to see me as I would ideally like them to see me	2.93	1.652	0.526	0.670
Situational Involvement (SIT_Involv)	SI-1 When purchasing the wine on this last dining occasion, I had a high level of interest in the purchase process	4.50	1.588	0.662	0.796
	SI-2. On this last dining occasion, I put a lot of effort into the purchase of the wine	4.18	1.601	0.662	0.796
Propensity to Seek Information (INFO_Seek)	IS-1 Before buying the wine, I would obtain substantial information about the different brands and vintages	4.05	1.662	0.838	0.912
	IS-2 I would acquire a great deal of information about the different brands and vintages before buying the wine	3.96	1.656	0.838	0.912
Propensity to Provide Information (INFO_Prov)	IP-1 I would give other people advice about buying a particular brand of wine and vintage	4.41	1.680	0.692	0.818
	IP-2 I would like to influence my friends in their choice of a particular brand and vintage of wine	3.96	1.727	0.692	0818

Measurement using 7-point Likert Scale: 1 =  Strongly Disagree …  … ..7 = Strongly Agree1

# Adapted from Dholakia (2001) and the WBG scale of Bruwer et al. (2017).