The rapidly growing black middle class, referred to as middle-of-the-pyramid (MOP) consumers, has been one of the most noticeable phenomena in post-apartheid society, and has played an important role in reshaping the South African society (Lappeman & Neethling 2017). They tend to attract researchers as a result of possible new theoretical and practical contributions in relation to international trends in the buying behaviour of this segment (African Development Bank 2011). Middle-of-the-pyramid consumers have grown to become a major market in countries such as Brazil, Russia, India, China and South Africa (BRICS) (Asian Development Bank 2010). Growth has also been noted in countries in sub-Saharan Africa, South Asia, East Asia and Latin America (African Development Bank 2011). The size of the MOP market differs between countries and regions, and is responsible for greater demand for, and consumption of, consumer products (Kravets & Sandikci 2014). The movement of South African marketers to focus on the more prosperous segments is reflected in similar trends in emerging markets in developing countries, but the growing MOP is more composite because of the diversity of the market and its ‘cultural paradox’ and heterogeneous nature (Potgieter, Wiese & Strasheim 2013). This market segment is still underestimated and sometimes overlooked by marketers, especially in leisure behaviour and consumption (Kruger & Saayman 2014). Similarly, research on the factors that influence MOP consumers’ buying behaviour in sub-Saharan Africa is still limited (Jackson 2016).

The purpose of this exploratory research was therefore to explore the limited knowledge in the current South African MOP dining experience literature by determining the expectations of MOP consumers when dining out at full-service restaurants. This information is invaluable for the hospitality sector in that it will allow restaurateurs to target this local emerging market segment in an improved manner.

The study design comprised both qualitative (to develop and refine the measuring instrument) and quantitative methods (to collect the data from the MOP consumers) to enrich the quality of the research, and to assess consumer expectations of the dining experience (Cresswell & Clark 2011).

Table 2 reports the factor loading of all the items and indicates those items with values lower than 0.40 that were not interpretable for the study (Burns & Bush 2014). Of the dining experience items, two (Q1 ‘meal is exactly what you ordered’, Q2 ‘food tastes good to you’) did not load sufficiently on any of the components. These items were excluded from the dining experience dimension factors and 48 items were retained. Only those items with a greater than one variance explained were retained. Composite factors were identified using factor loadings and factor scores for items loading on each factor. These composite factors served as the independent variables in subsequent analysis for the study and should be interpreted taking into consideration the item content of each factor.

The 48 items resulted in a 10-component solution based on Eigenvalues greater than 1.00, which explained 66.293% of the cumulative variance in the data. The initial confirmatory factor analysis of the 10-component solution with 48 items from the exploratory principal component analysis fitted the data reasonably well. Factor 10 had only two items. Previous researchers (Antun & Gustafsson 2005) confirm that a parsimonious model with fewer items would be more useful in surveying a real restaurant situation. Least factors also tend to improve the factor matrix. It was decided to force a solution with only eight factors, as shown in Table 2, which resulted in a solution explaining 61.739% of the variance in the data and a six-component solution with a variance of 55.592%. Owing to the larger number of overlapping items in the six-component solution, the eight-component solution was used as the final solution. Some of the items did not fit in content-wise with the factor theme content in which they were grouped, while others overlapped with items under the same factor. The items with cross-loads were evaluated in the context of theory. Even though the loading was lower on the other factor, and the factor loadings were both greater than 0.4 (Burns & Bush 2014), the factor was categorised according to where interpretability was best. Items Q14 ‘Waiter is professional in service’ and Q15 ‘Timeliness of service acceptable’, which fell under the food factor, were thus moved to the service factor. Item Q13 ‘Food is freshly served’, which fell under the restaurant health factor, was moved to the food factor because of its applicability to that category. A similar procedure was performed in the study by Kruger, Saayman and Slabbert (2015), where some factors were retained as part of the alternative possibility. The resulting internal consistency remained acceptable for all the factors, as shown in Table 2.

At this point, a conclusion can be reached regarding the validity of the dining experience dimensions of the measuring instrument. The commonality between variables was relatively high, ranging from 0.377 to 0.768. The 48 items yielded eight factors that were interpretable, namely food, service, atmosphere, layout, social connectedness, food health, restaurant health and accessibility. The renamed factors differ from the categories used in the questionnaire because of the grouping and characteristics of the items in the factors. They also differ from the original five factors proposed by Antun et al. (2010) and the items added after the interviews and used in the questionnaire, namely food (11 items), service (11 items), atmosphere (17 items), social (9 items) and health (2 items) (Table 1). Even though the factors seem to be different, it can be concluded that the atmosphere factor could be divided into subsections related to general atmosphere, layout and accessibility, as noted by Ryu and Jang (2008a). The health factor could likewise be divided into food and restaurant health. The overall patterns identified in the EFA were similar to those in the study of Antun et al. (2010), except the separation of layout and accessibility from atmosphere factor, and the split of health and food factor in the restaurant (Barber, Goodman & Goh 2011).

The reliability of the questionnaire was determined by using item analysis with coefficient alpha measurement. Reliability indicates the accuracy and consistency of the measuring instrument and is measured using a reliability coefficient. To aid in interpretability, varimax rotation was used to establish a clearer separation of the components (Hair et al. 2010). Excluding factor loadings of less than 0.4 resulted in a simple structure (Thurstone 1947), with each of the eight components showing a number of strong loadings (Table 3).

The Cronbach’s alpha coefficient was calculated for the dining experience dimensions of the questionnaire, yielding internal consistency reliability of at least 0.7 or higher, which is appropriate (Pallant 2010). The generally accepted lower limit for Cronbach’s alpha is 0.70, although it may decrease to 0.60 in exploratory research, as stated by MacCaullum et al. (2001). The Cronbach’s alpha coefficient was calculated for each of the eight factors as well as for the overall scale. The results per factor are given in Table 3.

Each of the extracted components demonstrates acceptable internal consistency (above 0.6), as illustrated by the Cronbach’s alpha coefficients listed in Table 3. The Cronbach’s alpha coefficients ranged from 0.693 to 0.921 for the eight dining experience factors, all of which were significant at the alpha level of 0.1 (Hair et al. 2010). Orthogonal rotation was done to minimise correlation between the factors, and thus results in better interpretation after rotation. The final eight factors were food, service, atmosphere, layout, social, food health, restaurant health and accessibility. The factors displayed reliability, and it can be concluded that the multiple measurement items are highly reliable in measuring each item. These factors support the researcher’s initial conceptualisation of the important dining experience dimensions. According to the mean scores, the dimensions ranged from the highest to the lowest in descending order of importance as follows: restaurant health, service, accessibility, food, atmosphere, layout, food health and social. There was more agreement among respondents regarding the importance of restaurant health (standard deviation [SD] = 0.719) than social (SD = 1.141) as dining experience dimensions. The highest level of agreement was related to food (SD = 0.651), and the lowest level of agreement was related to social (SD = 1.141) as dining experience dimensions. The health factor was extended to the restaurant, which is not entirely unexpected, as the importance of health issues in restaurants was supported by Sulek and Hensley (2004) and Barber and Scarcelli (2010). The layout of the interior of the restaurant was also added as a separate factor. As confirmed by Ryu and Jang (2008b), the spatial layout and seating comfort have an immediate effect on consumers’ perceptions of quality, pleasure and excitement as well as an unconscious effect on the intention of the consumer to return in future. It can be concluded that the reliability of the questionnaire was acceptable.

The eight underlying dimensions (service, social, food, layout, restaurant health, atmosphere, food health and accessibility) are independent variables that influence MOP consumers in their decision-making with respect to dining out, but each dining experience dimension has a particular level of importance. The subscales for the extracted components were obtained by calculating the mean of the item loading on each of the subscales, as shown in Table 3. The most important dining experience dimension was restaurant health, and social was the least important.

This study noted both similarities and differences in comparing the results with the dining experience dimensions of importance in the study of Antun et al. (2010) (DINEX), and with TANGSERV (Baloglu, Mao and Busser (2010) and several South African studies (Kruger & Saayman 2016; Mhlanga, Hattingh & Moolman 2015; Mhlanga & Tichaawa 2016). Restaurant health was found to be the most important dining experience dimension in this study, but it was accorded lesser importance in the original DINEX (Antun et al. 2010), and did not feature as important in other South African studies as well (Mhlanga et al. 2015; Mhlanga & Tichaawa 2016). Moreover, the South African studies (Kruger & Saayman 2016) did not consider the importance of the social dining experience dimension, which was noted as being least important in this study and in DINEX. The TANGSERV scale used by Baloglu et al. (2010) mentioned the importance of accessibility, but at a lower level of importance than in this study. The South African studies conducted at full-service restaurants, as in this study (Mhlanga 2015; Mhlanga, Moolman & Hattingh 2013; Mhlanga & Tichaawa 2016; Moolman 2011; Petzer & Mackay 2014), emphasise the importance of food quality, service and ambience. This study found these dining experience dimensions to be of moderate importance, similar to the findings of Antun et al. (2010) in DINEX. The reported differences of this study may be because of different characteristics in the selected sample and to the types of restaurant categorised as full-service restaurants.

Western developed markets are known to have ageing populations. However, the South African market includes a significant MOP market segment, which is a growing brand-conscious group that tends to develop among young working adults. Marketers should therefore reconsider the dining behaviour of the MOP consumers, characterised by unique consumption needs, and thus have a potential impact of the total MOP market. More targeted marketing campaigns should focus on this group of consumers to understand how they continually shape decision-making as the MOP market segment.

This study highlighted that the expectations of MOP consumers should be related to particular dining experience dimensions of full-service restaurants. In this way, restaurateurs could improve (and even exceed) the particular consumer dining expectations. Because of the restaurant density in the Gauteng area, formal full-service restaurants should aim to differentiate themselves from other restaurant categories.

The study is among the first to analyse full-service restaurants in different locations for the South African MOP market segment and to propose and confirm the implications for their expected dining experience dimensions. An in-depth evaluation of the literature (as shown in Table 1) regarding important dining experience dimensions was done and their influence on decision-making related to dining out was adapted so as to be specific to MOP restaurant users. It is important for the hospitality industry to understand that there are different dining experience dimensions according to MOP needs and expectations, and that these should be identified as clearly as possible.

Al Khattab and Aldehayyat (2011) confirmed that it is not feasible to measure the dining experience with a single assessment tool because the consumer is part of the experience. Therefore, the researcher used both interviews and questionnaires to determine MOP consumers’ needs and expectations during the dining experience, and thus enriched the quality of research. The interviews made it possible to understand consumers’ needs and expectations, whereas the questionnaire was able to clarify the multivariate effect of the dining experience dimensions of the selected group of consumers.

Apart from enriching the academic literature, this study also makes a contribution to science. The present study shows that particular quality dining experience dimensions that influence the MOP market segment in choosing a restaurant for a dining experience. The implications are thus not limited to restaurants of a particular origin. Restaurant health, service, accessibility and food are the most important dining experience dimensions, followed by the atmosphere, layout, food health and social dining experience dimensions as factors that might influence diners’ expectations of a quality dining experience.

In assessing the empirical study, the main limitation was that the sample frame could not be compiled on the basis of the web-based questionnaire alone. Further prospective respondents were therefore selected through a convenience sample approach among consumers at full-service restaurants. The sample used for data analysis in the present study (363 cases) exceeded the absolute minimum recommended sample size of between 200 and 400 (Van Voorhis & Morgan 2007). Considering the number of estimated parameters in this study (50 dining experience items), the sample was relatively small. A larger sample size is recommended in future studies for more appropriate estimates in the study model.

Data from this study were collected only in Gauteng Province, using a convenience sample approach. The outcomes of this study can therefore not necessarily be applied to restaurants in other geographical areas of South Africa. The validity of the present study with respect to MOP consumers could be increased through a larger sample in different geographical areas.

Another limitation was that, in considering their expectations of a dining experience, the respondents might have focused on particular restaurants, irrespective of whether these were their favourite restaurants or not.

The study discussed MOP consumers’ priorities with respect to food, service, atmosphere and social connectedness as predictors of a quality dining experience. The study shows that there is a positive relationship between the levels of tangible dimensions in the restaurant industry. Although the intangible dimensions are more important, the tangible dimensions also perform an important role. The study revealed a number of similarities with previous studies, as discussed in the previous section. It can be concluded that the restaurant industry needs to consider particular dining experience dimensions when offering a service to MOP consumers. If they would like to attract more of this segment, they need to focus on the aspects of importance under each dining experience dimension.

Despite the apparent homogeneity of the selected group of respondents in terms of the demographic factors of group, age, gender and geographical location, this study revealed that there were differences in their perceptions of the phenomena under investigation. The eight dining experience dimensions extracted from the factor analysis were valid and it was possible to use them as independent variables in subsequent analysis in the study. The most important dining experience dimension factors were restaurant health, followed by service, accessibility, food, atmosphere, layout, food health and social. Restaurants should not underestimate what consumers expect from a dining experience, and should thus concurrently monitor and compare their consumer dining experience expectations to strengthen consumer retention, as well as for long-term sustainability and the restaurant’s reputation.