Product risk refers to the loss felt by consumers should the product fail to meet or exceed the consumers’ expectations (Tandon et al. 2018:68), which means that the product may not perform as expected (Ariff et al. 2014:1). It was defined by Tariq, Bashir and Shad (2016) as ‘chances of the failure of products to meet its user’s requirements’. Product risk is related to performance risk (Tariq et al. 2016:96). In an online context, product risk is caused by the consumers’ inability to physically examine the product before the purchase (Farhana et al. 2017:225). According to Tariq et al. (2016:98), product risk has no significance on consumers’ attitudes towards online shopping behaviour. However, Aghekyan-Simonian et al. (2012:329) found that product risk presents a significant barrier to consumers who intend on purchasing online. In a study investigating consumers’ attitude towards online purchasing, Hong and Yi (2012:1304) found product risk to have a negative influence on consumers’ online purchase behaviour. In addition, product risk was found to have a negative influence on consumers’ attitudes towards online shopping (Ariff et al. 2014:1; Bhatti et al. 2018:7). Therefore, the following hypothesis was formulated:

Consumers who purchase products online usually have concerns about the delivery of the product. According to Tanadi et al. (2015:227), delivery risk refers to the potential loss consumers incur from delayed delivery, non-delivery and the delivery of damaged products. Non-delivery risk was found to have a negative influence on consumers’ attitudes towards online shopping (Ariff et al. 2014:1; Tariq et al. 2016:96). Furthermore, Tanadi et al. (2015:227) and Tariq et al. (2016:98) found delivery risk to have a significantly negative influence on consumers’ attitudes towards purchasing online. This was supported by Nawi et al. (2019:8) who found delivery risk to have no significant impact on consumer attitudes towards online shopping. Therefore, the following hypothesis was formulated. Contrary to the given findings, Hong et al. (2016:18) found a significant effect of delivery risk towards customer attitude in online shopping, which implies that the impact that delivery risk has on attitude towards online shopping differs across consumer groups:

According to Dai and Chen (2015:46), consumers’ personal and financial information can be put in a great danger because of security breaches, which result in consumers suffering identity theft and financial loss. Consumers who shop online develop a certain level of vulnerability from using online payment systems and authenticating the reliability of privacy policies (Tanadi et al. 2015:228). Security and privacy risk have been found to be the most prominent factors that influence consumers’ online purchasing decision (Hong & Yi 2012:1305) and that consumers with a stronger privacy and security concerns would not shop online (Schlosser, White & Llyod 2006:137). The risk is associated with consumers’ fear of the misuse of personal and financial information (Thakur & Srivastava 2015:153). Privacy risk in an online context refers to the potential loss of confidential information and exposure to identity theft (Featherman & Wells 2010:114), whereas security risk refers to concerns about monetary loss through using online payment systems (Thakur & Srivastava 2015:153). Privacy risk has been found as one of the factors that deters consumers from successfully completing online purchases that only require online payment methods (Hong & Yi 2012:1305). Moreover, Orubu (2016:17) found privacy risk to have significantly negative influence on consumers’ attitudes towards online shopping behaviour.

Security still remains a barrier for consumers to shop online (Rahman et al. 2018:2). Consumers’ issues are centred around concerns of secure payment systems, as most of them believe that online payment systems are not safe (Yadav, Sharma & Tarhini 2016:2, 18). This is consistent with Dai and Chen’s (2015:41) findings that security concern negatively influences consumers’ purchase intentions on online shopping platforms. Therefore, the following hypothesis was formulated:

Demographics tend to have a significant effect on consumers’ perceived risk towards online platforms. According to Marriott and Williams (2018:20), financial risk significantly influences male and female consumers’ intention to purchase online. Pascual-Miguel, Agudo-Peregrina and Chaparro-Peláez (2015:1554) found that women perceive more risk when they have the intention to shop online. However, Lin et al. (2018:1195) found that perceived risk has a higher negative influence on male consumers’ attitude towards online shopping, especially performance risk (Marriott & Williams 2018:26). Moreover, female consumers were found to have a higher positive attitude towards the financial security of online shopping platforms (Wu, Quyen & Rivas 2016:706). Pascual-Miguel et al. (2015:1554) stated the gender difference is narrowing in the adoption of online shopping behaviour. It is thus important to further investigate the influence of gender on the risk factors affecting online buying behaviour.

The influence of consumers’ age on the risk factors affecting online buying behaviour is important for retailers to develop sustainable strategies. According to Singh and Rana (2018:32), age has no influence on consumers’ perceptions towards adopting online shopping. In Marriott and Williams (2018:14) study investigating the moderating effect of age on perceived risk towards online shopping, age was found to have a significant influence on consumers’ adoption of online shopping. However, performance risk and time risk were found to have no significant influence on young and old consumers’ intention to shop online (Marriott & Williams 2018:14). Thus, the following hypotheses were formulated:

In order to achieve the objectives of this study, a survey research method was deemed appropriate to test the hypotheses as shown in Figure 1. Similarly, prior studies on online shopping adopted a survey method to determine risk factors influencing consumer attitude towards online shopping. A survey requires a respondent to choose a response from those provided and allows a researcher to compare responses. It was appropriate for this research to determine the risks influencing consumer attitude towards online shopping and to determine the moderating influence of demographic factors on risk influencing consumer attitude towards online shopping.

The target population for the present research comprised of retail consumers in SA who shops at the two malls: Cresta mall in Johannesburg and Sunny Park mall in Sunnyside, Pretoria. These shoppers must have access to the internet and have either shopped online or intend to shop online. The researchers were given permission by the shopping mall managers to intercept any consumers between the age of 10 and 65, visiting the malls at the time of data collection. The questionnaires were completed by 207 shoppers who visited the mall at the time of study. Participants were made aware that participating in the study was voluntary. Convenience sampling was used and fieldworkers were used to hand out questionnaires in the two malls. Convenience sampling was considered appropriate for the study because shoppers who visited the malls at the time of data collection were intercepted when entering the malls. Using convenience sampling made it possible to have easy access, obtaining respondents quicker and cheaper (Wiid & Diggines 2013). The two malls also have high congestion, which supports the adoption of a convenience sampling method. Of the 210 targeted users, 207 responded, culminating in a response rate of 98%, which is a high response rate.

The questionnaire was developed using the scales of prior research studies: Product risk (Aghekyan-Simonian et al. 2012; Ariffin et al. 2018; Hsieh & Tsao 2014; Javadi et al. 2012; Tandon et al. 2018; Tariq et al. 2016; Thakur & Srivastava 2015; Vijayasarathy 2004), privacy and security risk (Ariffin et al. 2018; Hsieh & Tsao 2014; Thakur & Srivastava 2015) and delivery risk (Hong & Yi 2012; Javadi et al. 2012; Tandon et al. 2018; Tariq et al. 2016). The study focused on two demographic factors: gender and age; however, additional 11 demographic questions were included in the questionnaire for the purpose of understanding the respondents better. In addition, 21 statements of the questionnaire covered risk factors influencing consumer attitude towards online shopping. The consumers were asked to rate each of the risk factors on a Likert scale from one to five, with one measuring ‘highly disagree’ and five measuring ‘highly agree’. The five-point Likert scale is the most commonly used and was deemed appropriate for this study to determine the risk factors influencing consumer attitude towards online shopping.

Data were collected from the two malls from March 2019 to April 2019. Shoppers who visited the mall during this period were intercepted by field workers and were asked to participate in the study. Respondents were made aware that completion of the questionnaire was voluntary and that no compensation would be offered to consumers who completed the questionnaire that was ethically approved by the academic institution. Two hundred and seven questionnaires were completed for the study.

The data were analysed with the aid of international business machine (IBM) statistical package for the social sciences (SPSS) for Windows, version 25. Various statistical analyses were conducted to achieve the objectives of this study, including descriptive analyses, such as mean and standard deviation, factor analysis, analysis of variance (ANOVA) and regression analysis.

The study received ethical clearance from the Department of Marketing and Retail Management Ethics Review Committee on 15 February 2019 with reference number: 2019_MRM_001.

As shown in Table 1, respondents consisted of more females (60.4%, n = 125) than males (39.6%, n = 82). Almost one-third of the respondents were aged 18–21 years (31.6%, n = 65), followed by 22–25 years (28.5%, n = 59). The 22–25 years were represented by 19% of the respondents (n = 40), whilst the 31–35 years was represented by 13% of the respondents (n = 17). Those over 36 were less represented with 15 respondents representing 7% of the population.

All respondents (100%, n = 207) had access to the internet. More than 90% (93.7%, n = 194) of the respondents use their cell phones to access the internet. The next more popular method for connecting to the internet is from a computer at home (45.9%, n = 95). More than 70% (72.8%, n = 150) of the respondents use the internet for communication, social websites and others. The next more popular reason for accessing the internet is finding information (57.3%, n = 118), followed by research, homework and study (48.1%, n = 99).

The majority of the respondents (83.1%, n = 172) possess a bank card and buy clothing and accessories online (61.1%, n = 118). The next most popular type of purchase is books (32.6%, n = 63), followed by electronic goods (30.6%, n = 59). Almost 60% (58.5%, n = 121) of the respondents do not buy electronic products online. Of those respondents who do not buy electronic goods online, almost one third (32.5%, n = 40) indicated that they may buy electronic goods online in the next 12 months. Of those respondents who currently buy electronic goods online, more than 40% (43.2, n = 38) do so once a year. Of those respondents who buy electronic goods online, 36.8% (n = 75) visit one to three online stores before making the purchase.

The most popular online shop to purchase electronic goods from is Takealot (70.4%, n = 119), followed by Makro (38.5%, n = 65).

Principal component analysis (PCA) with IBM SPSS Statistics 26 was used to examine patterns of correlations amongst the questions used to assess the respondents’ perceptions regarding the consumer risk of online buying in SA.

The factorability of the correlation matrix was investigated using Pearson’s product-moment correlation coefficient. Preliminary distribution analyses indicated that the assumptions of normality, linearity and homoscedasticity were not violated. The correlation matrix (Table 2) demonstrated some of coefficients of 0.3 and above. The Kaiser–Meyer–Olkin value was 0.825, well above the recommended minimum value of 0.6 (Kaiser 1970, 1974) and the Bartlett’s test of sphericity (Bartlett 1954) reached statistical significance, p < 0.001. Thus, the correlation matrix was deemed factorable.

Twenty-one items were initially subjected to PCA and this resulted in a four-factor solution that explained 61.515% of the variance in the data. Six of the variables had to be excluded because they were not contributing to the solution. The remaining 15 items resulted in a three-factor solution explaining 59.864% (Table 2) of the variation in the data.

As is visible from Table 3, the Cronbach’s alpha for the three factors ranged from 0.736 to 0.877, which is satisfactory because they are over 0.70. The overall Cronbach’s alpha was 0.713, also higher than 0.70 recommended by Malhotra (2010). To determine the validity of the instrument, the threshold of 0.490–0.740, was maintained on the communalities, as well as a cut-off point of 0.30 on the Pearson’s correlations, as suggested by Kim and Mueller (1978). Product risk loaded six items whilst privacy and security loaded five item and delivery risks loaded four items. The mean scores for the factors were 2.135, 2.735 and 3.051, respectively, which means that the respondent agreed more on product factors impacting on attitude towards online factors followed by privacy and security factors and delivery factors, respectively.

To establish the effect of the predictors on attitude towards online shopping, multiple ordinary least squares (OLS) regression was used, with the Enter method for adding predictors. As shown in Table 4 and Table 5, the model is significant (F [7] = 12.424, p < 0.001), meaning that at least one of the predictors has a significant relationship with attitude towards online shopping. From the adjusted R-square value of 0.282, the model explains at least 28% of the variation in attitude towards online shopping.

Collinearity statistics (Table 6) indicate that multicollinearity among predictors is not a problem (VIF < 10). The only predictors that are significant in the model, however marginal, are product factors (0.031), privacy and security factors (0.000) and age (0.033) with levels of significance lower than 0.05. According to Tabachinik and Fidell (2001:32), a hypothesis with a significance level below 0.05 should be accepted.

The only predictors that are significant in the model are product factors, privacy and security factors, and age. The constant in the linear equation is also significant. Gender has no moderating effect on the relationship between risk factors and attitude towards online shopping. However, age has the moderating effect on the relationship between product factors, privacy and security factors and attitude towards online shopping. The privacy and security factors have a stronger effect on attitude towards online shopping with a beta coefficient of 0.508 compared to −0.143 for products effects, which is both lower and negative. Existing findings reported contradicting results with Dai and Chen (2012) supporting the findings of this study, while Nawi et al. (2019) findings do not support the findings of this study. Aghekyan-Simonian et al. (2012) support this study’s findings that product influences consumers’ attitudes towards online shopping. Thakur and Srivastava (2015) found that security and privacy risk weigh relatively lower as compared to product performance risk, which also contract this study’s findings.

The above findings indicate that there is no influence of gender towards risk factors influencing attitude on online shopping. The results are supported by Makhitha et al. (2019) who found that gender has no influence towards online shopping. Other studies on online shopping also did not find a significant impact of gender towards online shopping (Brashear et al. 2010; Chiu, Lin & Tang 2005). However, some studies such as those of Slyke, Comunale and Belanger (2002)and Venkatesh and Morris (2000) found online shopping behaviour to vary across gender of consumers.

Existing studies also reported conflicting findings with some reporting age to have influence towards online shipping (Makhitha 2014; Makhitha et al. 2019; Rudansky-Kloppers 2016), while others reported that age has no influence towards online shopping (Kim 2006).