Factors influencing consumers’ purchase decisions regarding personal motor vehicle insurance in South Africa

Abstract

The South African short-term insurance industry is diverse and competitive. Consumers can choose from a wide variety of motor vehicle insurance policies. The aim of this paper was to determine the specific factors that influence individuals’ decision to purchase personal motor vehicle insurance in South Africa. A quantitative research approach was followed, which involved collecting data by means of a self-administered questionnaire. A three-phased analysis strategy, consisting of (1) exploratory factor analysis, (2) confirmatory factor analysis and (3) logistic regression model building, was followed to test and refine a consumer purchase decision model until an acceptable model could be established. The acceptable model identified specific internal and external factors, namely, marketing efforts (people, process and price of a policy); factors related to the sociocultural environment (items such as positive social class and negative reference family member); communication sources (paid-for social media efforts, buzz agents and customized messages); psychological attributes (items such as motivation and personality); and demographic variables (household income, highest level of education and age). It is recommended that insurance companies take note of the identified factors that influence the market since they are the policymakers who offer and market motor vehicle insurance products to consumers and potential consumers.

Keywords:

• consumer decision-making influencing factors
• internal influences
• external influences
• consumer buying behavior
• short-term insurance

1. Introduction

The South African motor vehicle industry reflects what is currently transpiring globally, specifically as far as the various factors that have an impact on consumers’ daily buying decisions are concerned. The number of motor vehicle accidents is increasing worldwide, and motor vehicle insurance has almost become a necessity. Although the African insurance sector remains poorly penetrated, South Africa has the highest insurance penetration (Olayungbo & Akinlo, 2016, p. 16). South Africa makes up 70% of the African continent’s total insurance premiums (Bagus et al., 2020). Motor vehicle insurance is the largest contributor to the non-life insurance market in Africa, but this is mainly driven by the compulsory minimum-level insurance (third-party insurance) requirement in countries such as Morocco, Kenya, Nigeria and Egypt (Bagus et al., 2020). South Africa is one of the few countries that do not obligate motorists to have motor vehicle insurance, not even third-party insurance. Only 30 to 35% of motor vehicles on South African roads are insured (Peyper, 2021).

Individuals’ purchase decisions are driven by their needs, and individuals assess and select insurance policies that will fulfill their needs. Consumer decision-making or the process of making purchase decisions, is a cognitive process that includes the mental activities that determine what actions individuals will undertake to eliminate a tension state caused by a need (De Mooij, 2019, p. 26). Although understanding the purchase behavior of consumers within a market is difficult, it is vital for any short-term insurance business.

Although every individual’s needs, demands and preferences are unique, individuals exhibit many similar behavioral patterns and follow more or less the same process when it comes to making buying decisions (Ragunathan, 2016, p. 182). Consumer decision-making changes with the type of purchase decision, which can be classified according to the degree of buyer involvement and the degree of differences among services (Lamb et al., 2018, p. 403). Individuals who consider purchasing motor vehicle insurance will engage in dissonance-reducing behavior since the primary cause of dissonance is the fact that they have to choose between numerous similar alternatives (Belch & Belch, 2015, p. 125). Individuals who engage in this type of behavior are highly involved in the purchase but are of the opinion that merely minor differences exist between the insurance products offered by the different companies (Techasurin et al., 2020, p. 2).

Motor vehicle insurance is usually seen as a grudge purchase (Albertse, 2022) and as something that is only needed in unfortunate situations (Spender et al., 2019, p. 16). Motor vehicle insurance consumers feel that they should have insurance cover for peace of mind should something go wrong. Every consumer has his or her own needs that should be fulfilled, and insurance companies have to keep in mind that external and internal factors influence a consumer’s need to purchase personal motor vehicle insurance.

Research has been conducted regarding the factors (internal and external) that have an impact on consumers’ purchase decisions in the general international insurance market (Ahmed & Hassan, 2020; Kaže, 2010; Souiden & Jabeur, 2015), the international life insurance market (Keat et al., 2020; Thoa & Le, 2021; Weedige et al., 2019) and the health insurance market (Mistry & Vyas, 2021). More specific research has been conducted regarding marketing mix elements or marketing efforts that influence consumer purchase decisions (Esau, 2015; Guan et al., 2020) and consumer beliefs that influence consumer attitudes and purchase intentions (Souiden & Jabeur, 2015) in the life insurance sector. In the general insurance industry, research was conducted to determine the demographic, economic, social and psychographic factors impacting consumer behavior in insurance uptake in Kenya (Langat et al., 2017).

In one study, factors affecting the purchase decisions of academic staff in Lagos state (Nigeria) in respect of third-party liability motor vehicle insurance were identified (Dansu et al., 2018). A more recent study focused on how advertising affects consumers’ decision-making in the auto/motor vehicle insurance market of the United States (US) (Tsai & Honka, 2019). In another study, the profile and the predisposition of consumers within the Brazilian motor vehicle insurance market were analyzed (Dallemole & Figueiredo, 2020). None of the studies highlighted above explored all external and internal factors that could influence consumers’ purchase decisions, specifically in the personal motor vehicle insurance industry. The aim of this paper is to determine the factors that influence individuals’ decision to purchase personal motor vehicle insurance in South Africa by means of a three-phased model fit analysis based on a newly developed conceptual model.

The paper reviews the literature on external and internal factors affecting the decision-making process in the personal vehicle insurance industry and presents a comprehensive methodology, as well as data analysis steps followed. Subsequently, the results of this paper are discussed and the conclusions and practical implications for short-term insurance companies are summarized. Finally, the limitations and recommendations for future research are presented.

2. Literature review

2.1. Overview of the factors that influence the consumer decision-making process in the insurance industry

Various authors have discussed and described the consumer decision-making process (Acevedo, 2018; Belch & Belch, 2015, p. 111; Erasmus et al., 2019, p. 454; Kotler & Keller, 2012, p. 166; Lamb et al., 2017, p. 90). The general steps of the consumer decision-making process are need recognition, information search, evaluation of alternatives, purchase decision and post-purchase behavior. Most individuals in a market who have a need for a particular product or service will typically follow these steps.

The current models of consumer decision-making are not adequate for the financial services sector, necessitating the creation of a new conceptual framework (Mahmood & Baharun, 2019; Milner & Rosenstreich, 2013). Scholars have developed several models of consumer decision-making over the past 58 years by incorporating various constructs from the fields of economics and psychology. All the limitations of, and expansions to, the different models of consumer decision-making were considered to identify internal and external factors that influence individuals’ purchase decisions, with a view to including the factors in a new conceptual framework of consumer decision-making in respect of personal motor vehicle insurance.

A wide variety of external and internal factors influence individuals’ purchase decisions (Acevedo, 2018, p. 19). External stimuli, such as the marketing efforts of companies, factors related to the sociocultural environment and communication sources, represent external influences (Stankevich, 2017, p. 10). Internal influences include the personal and interpersonal aspects that are processed in individuals’ minds and that influence their purchase behavior (Acevedo, 2018, p. 19; Cunningham, 2018).

2.2. External influences

External influences consist of a number of factors under each variable. An insurance company’s marketing efforts involve the product/service (the policy offered), its price, its promotion and where it is sold (Lantos, 2011, p. 9; Stankevich, 2017, p. 10). Insurance is a financial service and, therefore, the other aspects that are included in a company’s marketing efforts are physical evidence (tangible components during service delivery), process (processes, transactions and communication during service delivery), people (employees who deliver the service) and partnerships or alliances or joint ventures with other businesses (Pride et al., 2018, p. 16). Factors related to the sociocultural environment include family, reference groups (friends and neighbors), other informal and non-commercial sources (comments of a friend, a newspaper article and the views of an expert on a blog or another internet platform), social class (a group of people who have the same prestige and status in society), culture and membership of subcultures (population groups, native language, religion and racial groups) (Adams & Lawrence, 2019; Schiffman & Wisenblit, 2019). The communication sources used to deliver the marketing mix and sociocultural influences are advertising, buzz agents, customized messages, paid-for social media and user-generated social media.

A company’s marketing efforts are successfully executed when individuals who need personal motor vehicle insurance are reached, informed and persuaded to purchase (Schiffman & Wisenblit, 2015, p. 19;, 2019, p. 375). However, these efforts can also be the reason why individuals decide not to purchase, or why consumers choose or do not choose a certain short-term insurance company.

Motor insurance policies that include telematics are becoming more popular. Consumers continuously show a positive attitude towards purchasing motor vehicle insurance that incorporates telematic devices that track their driving behavior (Alfiero et al., 2022, p. 10). Price bundling (combining different short-term insurance policy products in one bundle) is a popular mechanism that short-term insurance companies use to increase their revenue. However, new and loyal motor vehicle insurance consumers are price sensitive, and price is thus an important consideration in their purchase decisions (Dominique-Ferreira et al., 2016, p. 336).

Sales promotional activities can raise awareness among consumers and is an influential factor in a motor vehicle insurance consumer’s purchase decision (Azhar et al., 2021, p. 73). An insurer’s logo on emails is an example of visual communication that is used to create, develop and manage a brand in the eyes of consumers whilst providing them with information that will influence their purchase decisions (Trynchuk, 2017, p. 329). Obtaining quotations for motor vehicle insurance can be a lengthy and time-consuming process for individuals who are interested in purchasing insurance. Through the use of artificial intelligence (AI), insurance companies can automize the quotation pre-purchase search, allowing consumers to get an instant quotation by having a simple conversation (Riikkinen et al., 2018, p. 1153).

Insurance companies have thus facilitated instant quotations for motor vehicle insurance policies (Balasubramanian et al., 2021, p. 4). The automated processing of contacts can also speed up the process of searching for insurance, and a chatbot can provide consumers with instant answers to questions while they are searching for insurance (Eling & Lehmann, 2018, p. 367). The quality of the service provided by insurers’ employees changes consumers’ perceptions and influences their expectations (Abass & Oyetayo, 2016, p. 331), which will greatly influence consumers’ purchase decisions. Insurers’ partnerships with brokers, especially in respect of the claims management process, is another external factor that influences the purchasing process. If an online claims management service is offered, consumers will be more satisfied with insurers and brokers (Dominique-Ferreira, 2018, p. 1185).

The sociocultural environment can generate positive or negative opinions among consumers that influence their need to purchase products or services. A comment or an opinion (positive or negative) of a family member (a source from whom an individual learns values and who develops and shapes the individual’s personality) might influence an individual’s purchase decisions (Kotler & Armstrong, 2018, p. 164; Loacker, 2015, p. 111). Individuals’ reference groups can induce their values and attitudes or be the reason why they behave the way they do (Strydom, 2018, p. 307). Reference groups provide some points of comparison to consumers and potential consumers regarding their behavior, lifestyle or preferences (Kotler & Armstrong, 2018, p. 162).

Non-commercial sources can influence the purchase decision of an individual who is searching for short-term motor insurance (Ismagilova et al., 2017, p. 9). A previous study found that 80% of insurance buyers regard recommendations from family and friends as important when it comes to choosing between insurance brands and policies (Facebook IQ, 2019). This finding corroborates the influence that word-of-mouth from non-commercial sources can have on a consumer’s decision to purchase motor vehicle insurance. Individuals in the same social class give their opinions (whether positive or negative), which potential consumers or consumers consider in their decision to purchase motor vehicle insurance. People’s cultural and subcultural context shape their personal perceptions, disposition and behavior (East et al., 2017, p. 115). Individuals’ culture is the basic contributing factor in relation to what they want and how they behave.

Communication sources represent the communication from organizations to consumers and the mechanisms that are used to deliver the marketing mix and sociocultural influences to the market (Schiffman & Wisenblit, 2015, p. 340;, 2019, p. 376). Short-term insurance companies’ advertising, regardless of whether it is traditional (television and radio advertisements) or online (video advertisements and banner ads on websites), shapes the image of the insurance companies in consumers’ consciousness (Borda & Jędrzychowska, 2012, p. 28; Juska, 2018, p. 4). Advertisements significantly influence the purchase behavior of individuals (Haider & Shakib, 2017). Evidence suggests that organizations invest more in social media and mobile advertisements, which form part of digital marketing (Haider & Shakib, 2017). Previous research revealed that 33% of British insurance consumers and 37% of Australian insurance consumers show an interest in a motor vehicle insurance brand or policy once they viewed advertisements on Facebook and other social media platforms (Facebook IQ, 2019). Advertising efforts by insurance companies can create a buzz among consumers, which can drive traffic towards a certain motor insurance company and its policy offerings (Ekanayake, 2021, p. 238).

Buzz agents are individuals who are sourced by companies or individuals who are influential in their communities or social circles to serve as brand ambassadors, that is, to spread the word about the companies’ products (Kotler & Armstrong, 2018, p. 163). Consumers can also act as buzz agents since buzz is driven by word-of-mouth (Mohr, 2017, p. 11). Consumers who offer a product referral and tell others about a product or a service are regarded as buzz agents (Claro & Bortoluzzo, 2015, p. 210). Hence, anyone who tells others about motor vehicle insurance can be regarded as a buzz agent in the motor vehicle insurance market. Customized messages conveyed to consumers form part of the direct marketing component. Given that direct marketing is more targeted, a customized message is usually personalized and directed at a specific consumer or group of potential consumers (Kotler & Armstrong, 2018, p. 440). A distinct shift towards personalized marketing is occurring within the motor vehicle insurance market since personalized marketing improves relationship-building with consumers (Adamova et al., 2018) and serves as an effective way of reaching insurance consumers and potential consumers (Moorcraft, 2021).

Paid-for social media content relates to social media marketing activities, which include actions/posts on social media platforms that are aimed at encouraging individuals to choose products and brands (Bilgin, 2018, p. 129; Parkin et al., 2019). Advertising campaigns on Facebook are a form of paid-for social media content, but the new, dynamic way of paying for social media content is to utilize social media influencers (Appel et al., 2020, p. 82). Influencers who have large followings on social media are paid to mention brands or products/services in social media posts (Carter, 2018). An influencer’s opinion of a brand or a product/service, when shared on social media, is seen by a large audience, which is made up of individuals who trust the influencer (Barysevich, 2020). Individuals trust influencers because they can relate to the influencers on a personal level (Barker, 2020). Almost 40% of Twitter users in a previous study conducted, specified that they had made a purchase as a direct consequence of an influencer’s tweet (Santora, 2021). Previous research confirms that social media marketing activities greatly affect consumer brand awareness, brand image and brand loyalty (Bilgin, 2018, p. 142). User-generated social media content relates to instances where consumers share/post personal opinions about a company or a service, advice and recommendations on a social media platform. These posts are created by family members or someone from a reference group of an individual. Word-of-mouth via user-generated social media content is indispensable in the motor vehicle insurance market since it generates conversation about motor vehicle insurance products among consumers (Ekanayake, 2021). Online consumer reviews are an influential source of recommendations. Consumers’ purchase decisions are greatly affected by this electronic form of word-of-mouth (Teixeira et al., 2018, p. 361).

2.3. Internal influences

Individuals’ psychological attributes and demographics are the two internal factors that directly impact the recognition of a need to purchase personal motor vehicle insurance. Motivation, perception, learning, personality and attitudes are elements of psychological attributes (Schiffman et al., 2012, p. 36).

The first factor in the psychological attributes variable is motivation. Motivation is the driving force that drives individuals to fulfill their needs and wants (McIntee, 2015, p. 72; Rajagopal, 2019, p. 17). An individual who has a specific need will be motivated (Strydom, 2018, p. 306) and is ready to act (Kotler & Armstrong, 2018, p. 172). Consumers’ motivation to purchase insurance is shaped by two main components, namely, risk expectation and risk sensitivity. Risk expectation is the financial value of insurance for the item/object (motor vehicle) being insured. Risk sensitivity refers to the concerns that consumers have, which affect the moral benefit of having insurance or not (Meral, 2019). Individuals’ peace of mind in knowing that they will be “protected” in case of an adverse event (e.g., an accident or theft) is associated with risk sensitivity (Suter et al., 2017, p. 84).

Perception is a process in which individuals select, organize and interpret stimuli by means of one or more of the senses, that is, vision, hearing, taste, smell and touch (Hoyer et al., 2018, p. 80). Individuals act and react on the basis of their perceptions. Negative word-of-mouth through social media platforms can result in negative perceptions among consumers who are still deciding whether to purchase insurance or not (Arruda et al., 2021, p. 25).

Learning, as a psychological attribute, relates to how individuals purchase and consume information, as well as the experience that they can apply to future decisions (Erasmus et al., 2019, p. 455). Individuals learn as they execute the steps of the decision-making process (Kotler & Armstrong, 2018, p. 173). Past experiences with insurance products and insurance companies influence the purchase decision regarding a type of policy or a certain insurance company (Schmidt, 2019, p. 495).

Individuals’ personalities are based on their inner characteristics, namely, their specific qualities, attributes, traits, interests, drives and emotional patterns (Schiffman & Wisenblit, 2019, p. 90). Inner characteristics, in short, are the aspects that distinguish one individual from another. Attitude refers to a psychological tendency where an individual weighs a favorable or unfavorable judgment regarding an idea or an object (Ajzen in Alfiero et al., 2022, p. 3), in this case, a policy or an insurance company. Previous research showed that individuals’ attitudes regarding insurance are predominantly formed through past purchase experiences (Suter et al., 2017, p. 155).

Individuals change the services they purchase over their lifetime (Langat et al., 2017, p. 704), and the demographics of individuals influence the purchases they make during their lifetime. The demographic factors that influence purchase decisions, as identified in previous research, include age, gender, income, household size, education, ethnic background and family lifestyle (Boone & Kurtz, 2014, p. 170; Langat et al., 2017, p. 711; Willan, 2021).

All the internal and external factors discussed in this section were variables in the conceptual consumer decision-making model for personal motor vehicle insurance developed by Van Huyssteen (2022, p. 121).

3. Research methodology

The study employed a quantitative research approach to investigate the external and internal factors influencing consumers’ purchase decisions within the South African personal motor vehicle insurance industry. This approach was selected due to its capacity to provide statistically significant insights into variable relationships. The research targeted individuals who are vehicle owners and/or drivers (motorists) in South Africa, specifically those who had the option to either purchase or abstain from purchasing personal motor vehicle insurance.

3.1. Sampling and data collection

Convenience sampling and snowball sampling were used to select the appropriate respondents. The steps stipulated by Saunders et al. (2019, p. 316) were taken to select the appropriate non-probability sampling techniques. Convenience sampling was deemed feasible because no sampling frame was available (Silvia, 2020, p. 60). A database of motorists from the Automobile Association (AA) and a Facebook advertising/marketing tool were utilized to distribute an online survey. An advertisement with a link to the survey was posted to individuals who showed an interest in motor vehicles within the nine different provinces in South Africa. Additionally, snowball sampling was integrated by urging respondents to involve their peers in completing the online survey. Printed surveys were also distributed to vehicle drivers using a blend of snowball and convenience sampling methods.

Although the response rate was less than 1% (678 responses), online surveys commonly exhibit lower response rates compared to paper-based questionnaires (Ebert et al., 2018; Nulty, 2008, p. 303). Data were collected by means of a survey strategy, more specifically, online and paper-based surveys using a self-administered questionnaire. The questionnaire was newly developed according to the steps outlined by Barry et al. (2011, p. 98). Prior to commencing the pilot test and data collection, ethical clearance was secured from the Business Management Research Ethics Review Committee at the University of South Africa. A pilot study was carried out to determine the accuracy of instructions and to establish if there were any weaknesses in the design of the instrument (Bryman & Bell, 2015, p. 272). The weaknesses that were identified were improved. Data collected from January 2020 to November 2020 were processed using LimeSurvey.

4. Data analysis

The captured data were retrieved in an Excel spreadsheet (machine-readable form) before being uploaded to IBM SPSS (Version 27), the computer software used for data analysis. The data of this study were numeric and comprised statistical measures (Gliner et al., 2017, p. 9). Descriptive statistics were part of this study to provide a broad description of the dataset and descriptions of variables in the dataset.

The model fit analysis involved an exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and logistic regression. Data were analyzed using SPSS for EFA to discover the underlying structure of a relatively large set of variables (Hair et al., 2019, p. 124). The external and internal factors in the conceptual consumer decision-making model for personal motor vehicle insurance (gained from literature) compiled by Van Huyssteen (2022, p. 121) have not been tested before. Therefore, EFA is ideal for exploring and estimating the factor structure of the external factors of the theoretical consumer decision-making model (Matsunaga, 2010, p. 98). In previous research in the field of consumer decision-making model building it was also found useful to explore the factor structure (Tendhar et al., 2017; Wang & Yu, 2017). A holdout sample of 200 was performed to explore the factor structure of the external factors identified in the literature review.

CFA (using AMOS as the statistical software), which is a method of structural equation modelling (SEM), was utilized to test how well a pre-specified measurement theory, composed of measured variables and factors, fits the reality as captured data (Hair et al., 2019, p. 660). In previous research it was found satisfactory to use CFA as a follow-up strategy to EFA to confirm the existence of a specific factor structure (Nayeem & Casidy, 2015, p. 70). A validation sample of 478 was performed to determine a model fit of the external factors.

The findings of the CFA were analyzed using SPSS to perform logistic regression as a model-building strategy. Logistic regression, a widely used model-building strategy, was proven to be useful for investigating the independent effect of variables on dichotomous outcomes in studies on consumer decision-making in the global insurance market (Huber et al., 2015; Leiria et al., 2021; Tan & Lim, 2017). Given that the overall sample in the present study exceeded the recommended 400 (Hair et al., 2019, p. 557), this model-building strategy was deemed fitting.

Internal consistency was assessed to ensure reliability. Internal consistency was evaluated using Cronbach’s alpha coefficient (Cooper & Schindler, 2014, p. 260; Cooper et al., 2019; Zikmund et al., 2017, p. 274), while reliability in CFA was ensured through the reporting of composite reliability (CR) values (Hair et al., 2019, p. 763). Content validity was assured through expert validation of the questionnaire. Criterion validity was achieved by assessing the questionnaire’s relevance, bias freedom, reliability, and availability. Construct validity was verified in the EFA, where discriminate validity was used (Fornell & Larcker, 1981; Hair et al., 2019, p. 761), and in the CFA, three validity types included are construct, convergent, and discriminant.

5. Results

5.1. Demographic profile

The demographic profile of the respondents is indicated in Table 1.

Table 1. Demographic profile (n = 678)

Gender	Female	47.5%
	Male	52.5%
Age group	18–30	22%
	31–40	25.5%
	41–50	18.5%
	51–60 and over 60	34%
Marital status	Married and living with partner (but not married)	67%
	Single, divorced and widowed	33%
Highest level of education	Less than Grade 12 and Grade 12	30.6%
	Post-matric certificate, higher certificate, diploma and higher diploma	29.6%
	Post higher diploma, bachelor’s degree, honors degree, master’s degree and doctorate	39.8%
Employment status	Permanently employed on a full-time basis	54%
	Student, unemployed and retired	29%
	Temporarily employed and permanently employed on a part-time basis	17%
Household income	R0–R16 000	29%
	R16 001–R46 000	29.5%
	R46 001 or more	27%
	Prefer not to say	14.5%
Personal income	R0–R2 500	18.5%
	R2 501–R16 000	20.5%
	R16 001–R46 000	31.5%
	R46 001 or more	19%
	Prefer not to say	10.5%

More males (52.5%) participated in the study than females (47.5%). The 31–40 age group represents the biggest age group among the respondents. The majority (67%) of the respondents indicated that they are married or living with a partner (but not married). A total of 39.8% of the respondents indicated that they have a post higher diploma or a higher qualification. Most respondents (54%) indicated that they are permanently employed on a full-time basis. Almost 60% of the respondents indicated that they have a household income of less than R46 000 per month. A total of 31.5% of the respondents indicated that their personal income is R16 001–R46 000, which represents the largest personal income group.

5.2. Purchase decision descriptive statistics

More than half of the respondents (55.9%) indicated that they currently have personal motor vehicle insurance and 44.1% of the respondents specified that they currently do not have personal motor vehicle insurance. This question in the questionnaire served as the dependent variable.

5.3. EFA

The EFA (with an n = 200 exploration sample) was performed using principal component analysis for extraction and varimax for rotation. The results of 0.847 for the Kaiser—Meyer–Olkin (KMO) test and 0.00 for the Bartlett sphericity p-value verified that the data were appropriate for factor analysis. The minimum factor loading criteria was set to 0.40, taking into consideration that a sample size of 200 is needed for significance (Hair et al., 2019, p. 152). The results show that all the communalities were over 0.40. Seven factors that were identified in the initial study explained 59.985% of the variance among the items. Six of these factors can be regarded as external and internal factors that influence consumers’ purchase decisions (illustrated in Table 2).

Table 2. Factor findings and loadings from the EFA

Factors identified from EFA	Constructs	Items/variables	Factor loadings
1	Psychological attributes	Motivation Perception about motor vehicle insurance Perception about a certain insurance company Learning Personality Attitude towards motor vehicle insurance Attitude towards a certain short-term insurance company	0.690 0.758 0.715 0.626 0.684 0.782 0.653
2	Marketing efforts	Types of policies offered Price of a policy Distribution channels Physical evidence Process People Partnerships	0.620 0.602 0.644 0.479 0.711 0.679 0.512
4	Negative sociocultural environment	Negative family member Negative reference person or group Negative non-commercial sources Negative social class	0.798 0.859 0.722 0.719
5	Communication sources B	Promotion Buzz agent Customized messages User-generated social media posts Paid-for social media efforts	0.510 0.761 0.690 0.558 0.642
6	Positive sociocultural environment	Positive family member Positive reference person or group Positive non-commercial sources Positive social class Cultural and subcultural factors	0.628 0.604 0.634 0.694 0.446
7	Communication sources A	Traditional advertisements Online advertisements	0.742 0.551

The six factors identified as part of the EFA are aligned with the theoretical proposition in the research. The six factors were used in the CFA phase.

5.4. CFA

The model specification in the CFA was firmly grounded in existing theory and previous empirical research (Hahs-Vaughn, 2017, p. 447). Thus, the external and internal factors that are part of the conceptual model of consumers’ purchase decisions regarding personal motor vehicle insurance (Van Huyssteen, 2022, p. 121) were developed from existing theory. The EFA ensured a firm substantive and empirical basis for guiding the model specification of the CFA. The indicators (previously referred to as items/variables) that explain the latent variables (previously referred to as constructs) gained from the EFA are illustrated in Table 2.

The indicators in the “marketing efforts” latent variable scored relatively high, with the exception of the “physical evidence” and “partnerships” indicators, suggesting that these two indicators have a lower influence on the purchase decision. All the indicators in the “negative sociocultural environment” latent variable exhibit a high influence on the purchase decision. The indicators, “promotion” and “user-generated social media posts”, exhibit a lower influence on the purchase decision than the other indicators in the “communication sources B” latent variable.

The model was identified on the basis of the condition that one parameter (factor loading) relating to the latent variable with an item was set to 1.00 (Hahs-Vaughn, 2017, p. 449). Maximum likelihood estimation (MLE), a commonly used estimation method for the CFA model, was used for model estimation. The model fit statistics revealed that the Tucker Lewis Index (TLI) and the Comparative Fit Index (CFI) fell outside the prescribed threshold; therefore, this model had to be modified.

The indicator with a factor loading below 0.5, namely, “Types of policies offered” was removed. After assessing and identifying the indicators with high values of standardized residual covariance, the following indicators were removed: Physical evidence; user-generated social media posts; promotion; distribution channels; and partnerships. After these modifications had been made, a new, revised model was generated. The external and internal factors of the purchase decision elements gathered from the CFA revised model fit and the index values are presented in Table 3.

Table 3. External and internal factors of the purchase decision elements gathered from model fit of the CFA revised model and the index values

Latent variables	Observed variables	Index values
Psychological attributes	Motivation Perception about motor vehicle insurance Perception about a certain insurance company Learning Personality Attitude towards motor vehicle insurance Attitude towards a certain short-term insurance company	0.72 0.81 0.82 0.81 0.68 0.78 0.73
Marketing efforts	Price of a policy Process People	0.67 0.71 0.78
Positive sociocultural environment	Positive family member Positive reference person or group Positive non-commercial sources Positive social class Cultural and subcultural factors	0.76 0.82 0.84 0.79 0.54
Negative sociocultural environment	Negative family member Negative reference person or group Negative non-commercial sources Negative social class	0.84 0.89 0.86 0.82
Communication sources group A	Traditional advertisements Online advertisements	0.78 0.89
Communication sources group B	Buzz agent Customized messages Paid-for social media efforts	0.76 0.73 0.80

The model fit indices of the revised model implied an acceptable fit with the data. The TLI value was 0.89, just under the 0.9 threshold, but a TLI value close to 0.9 can still be deemed acceptable (Schumacker & Lomax, 2016, p. 112). The square root (SQRT) of the average variance extracted (AVE) per latent variable was 0.765 (psychological attributes), 0.721 (marketing efforts), 0.758 (positive sociocultural environment), 0.850 (negative sociocultural environment), 0.835 (communication sources group A) and 0.763 (communication sources group B). Thus, the reliability and validity results of the revised model yielded no concerns, and this model was found fit and could be accepted. The external and internal factors of the purchase decision elements (in Table 3) were regarded as acceptable to use in the third phase.

5.5. Logistic regression model building

A holdout sample of n = 200 was used to ensure model exploration. Table 4 presents a summary of the statistics of the univariate analysis for purposes of identifying the important covariances.

Table 4. Univariate analysis summarized to identify the important covariances

Variable	Type	T-test p-values	Chi-square p-values	r-value	P-value	Test	Include in baseline model
Original variables
Marketing efforts	Interval	0.000		0.363	0.000		Yes
Positive sociocultural environment	Interval	0.179		0.098	0.168		Yes
Negative sociocultural environment	Interval	0.131		0.109	0.123		Yes
Communication sources A	Interval	0.000		−0.292	0.000		Yes
Communication sources B	Interval	0.000		−0.299	0.000		Yes
Psychological attributes	Interval	0.000		0.365	0.000		Yes
Demographic variables
Age	Interval	0.000			0.331	Pearson	Yes
Age regrouped	Ordinal		0.000	0.000	0.281	Kendall’s tau-b	Yes
Gender	Nominal - Dichotomous		0.588	0.588	−0.048	Phi	No
Marital status regrouped	Nominal - Dichotomous		0.000	0.000	0.348	Phi	Yes
Highest level of education	Ordinal				0.427	Kendall’s tau-b	Yes
Highest level of education regrouped	Ordinal		0.000	0.000	0.446	Kendall’s tau-b	Yes
Employment status regrouped	Ordinal		0.004	0.004			Yes
Household income	Ordinal				0.486	Kendall’s tau-b	Yes
Household income regrouped	Ordinal		0.000	0.000	0.658	Kendall’s tau-b	Yes

The T-test and the Pearson correlation were used to analyze the original variables (latent variables from CFA). As revealed in Table 4, the T-test of the “positive sociocultural environment” and “negative sociocultural environment” variables shows that nothing distinguishes these two groups from one another. There is no relationship between these two variables and the dependent variables since the p-values are higher than 0.05. Consequently, these two variables can be included in the baseline logistic regression model. The statistics presented in Table 4 show that the “gender” variable is not important for inclusion in the baseline logistic regression model.

A holdout sample of n = 478 was used to conduct a logistic regression estimation. Omnibus Tests of Model Coefficients indicated that the Sig. value (x2/p-value) should be below 0.05. The Cox and Snell R-square and Nagelkerke R-square values were between 0 and 1 and could therefore be accepted. The p-value in the Hosmer and Lemeshow test was higher than 0.05, indicating that the model is worthwhile. The classification table drawn from the statistics revealed that 89.6% could be predicted from this baseline logistic regression model. The variables in the equation for the baseline logistic regression model indicated that the p-value for “communication sources A” (0.709), “marital status” (0.293), “employment status” (0.239), “employment status 1” (0.092) and “employment status” (0.441) was above 0.05, indicating that it did not significantly contribute to the predictive ability of the model. It was therefore removed.

Collinearity statistics were obtained and the two values indicated were tolerance and variance inflation factor (VIF). None of the tolerance values were below 0.1 and none of the VIF values were above 10. Therefore, no multicollinearity was assured among the variables for the final logistic regression model. All the covariances had a linear relationship, which was ensured with the Box—Tidwell procedure and the acceptance of the assumption of linearity.

In assessing the overall model fit of the final logistic regression model, the Omnibus Tests of Model Coefficients showed that the final model performs well overall. The Cox and Snell R-square and Nagelkerke R-square values were between 0 and 1 and could therefore be accepted. The p-value in the Hosmer and Lemeshow test was above 0.05, indicating that the model is worthwhile. The classification table drawn from the statistics revealed that 88.9% could be predicted from the final logistic regression model. The variables in the equation for the final logistic regression model are presented in Table 5. None of the Wald values scored very low; therefore, all the covariances indicated that the values are important predictive variables.

Table 5. Variables in the equation for the final logistic regression model

	B	S.E.	Wald	df	Sig. p-value	Exp(B)	95% CI for EXP(B)
							Lower	Upper
Marketing efforts	0.707	0.218	10.516	1	0.001	2.027	1.323	3.108
Positive sociocultural environment	−0.978	0.254	14.772	1	0.000	0.376	0.229	0.619
Negative sociocultural environment	0.423	0.181	5.464	1	0.019	1.526	1.071	2.176
Communication sources B	−0.932	0.221	17.707	1	0.000	0.394	0.255	0.608
Psychological attributes	0.568	0.226	6.308	1	0.012	1.765	1.133	2.749
Age	0.061	0.013	23.270	1	0.000	1.063	1.037	1.090
Highest level of education	1.108	0.236	22.011	1	0.000	3.028	1.906	4.810
Household income	1.570	0.271	33.697	1	0.000	4.808	2.830	8.171

All the covariances are variables that significantly contribute to the predictive ability of the final model.

6. Discussion

A summary of the findings regarding factors that influence an individual’s personal motor vehicle insurance purchase decision is presented in Table 6.

Table ‎6. Summary of findings regarding factors that influence an individual’s personal motor vehicle insurance purchase decision

Constructs	Item structure was confirmed with the CFA	According to the logistic regression
Marketing efforts	• Types of policies offered • Price of a policy • Distribution channels • Physical evidence • Process • People • Partnerships	• Price of a policy • Process • People
Sociocultural environment factors	Positive sociocultural environment • Positive family member • Positive reference person or group • Positive non-commercial sources • Positive social class • Cultural and subcultural factors Negative sociocultural environment • Negative family member • Negative reference person or group • Negative non-commercial sources • Negative social class	Positive sociocultural environment • Positive family member • Positive reference person or group • Positive non-commercial sources • Positive social class • Cultural and subcultural factors Negative sociocultural environment • Negative family member • Negative reference person or group • Negative non-commercial sources • Negative social class
Communication sources	Communication sources A • Traditional advertisements • Online advertisements • Promotion Communication sources B • Buzz agent • Customized messages • User-generated social media posts • Paid-for social media efforts	• Buzz agent • Customized messages • Paid-for social media efforts
Psychological attributes that influence need recognition	• Motivation • Perception about motor vehicle insurance • Perception about a certain short-term insurance company • Learning • Personality • Attitude towards motor vehicle insurance • Attitude towards a certain short-term insurance company	• Motivation • Perception about motor vehicle insurance • Perception about a certain short-term insurance company • Learning • Personality • Attitude towards motor vehicle insurance • Attitude towards a certain short-term insurance company
Demographic variables		• Age • Highest level of education • Household income

The marketing effort elements include the following: Price of a policy; process; and people. These items will increase the probability that an individual will purchase personal motor vehicle insurance. Respondents who had personal motor vehicle insurance were more likely to report a higher influence of marketing effort elements than those who did not have personal motor vehicle insurance. The way in which employees of insurance companies interact with consumers appeared to be the most important marketing effort element among the respondents in this study. This finding speaks to literature that suggests that consumers are price sensitive in respect of motor vehicle insurance policies (Dominique-Ferreira et al., 2016, p. 336). Furthermore, this finding supports insurance companies’ shift towards providing instant quotations for motor vehicle insurance policies (Balasubramanian et al., 2021, p. 4). This finding also corresponds with literature that suggests that the quality of the service provided by insurance company employees changes consumers’ perceptions and influences consumers’ expectations (Abass & Oyetayo, 2016, p. 331).

The results indicate that the “negative sociocultural environment” variables have a more important influence on the purchase decision than the “positive sociocultural environment” variables. The literature suggests that insurance buyers regard recommendations from family and friends as important when it comes to choosing between insurance brands and policies (Facebook IQ, 2019). Conventional thinking dictates that negative opinions of family members, reference persons or groups, non-commercial sources and members of the social class will discourage consumers from purchasing personal motor vehicle insurance. The findings of this paper suggest the opposite. The question regarding this construct in the questionnaire was an influence-scale question. Each statement measured the extent of influence of the positive or negative comments/opinions of a family member, a reference person/group, a friend or a newspaper article, or the views of an expert on a blog or another internet platform and individuals in the same social class, in general, and not specifically comments or opinions about motor vehicle insurance or short-term insurance companies. Thus, if a consumer’s family member makes a general comment about the increase of motor vehicle theft in the area, it could be perceived as negative, but it may alert the consumer to potential risk and convince him/her to purchase insurance. However, a positive comment regarding, for example, a decrease in motor vehicle theft might lead the individual to believe that his/her risk is low, resulting in a decision not to purchase personal motor vehicle insurance. Given that insurance is a grudge purchase (Albertse, 2022), this construct shows that consumers think in this way when they purchase personal motor vehicle insurance. The opinions gathered from participants indicate that the “negative sociocultural environment” variables have a more important influence on the purchase decision than the “positive sociocultural environment” variables. Respondents who had personal motor vehicle insurance were more likely to report that negative opinions/comments from the sociocultural environment greatly influenced their need to purchase such insurance than those who did not have personal motor vehicle insurance. In contrast, respondents who did not have personal motor vehicle insurance were less likely to report that positive opinions/comments from the sociocultural environment greatly influenced their need to purchase such insurance than those who had personal motor vehicle insurance.

According to the tested model, communication sources decrease the probability that consumers will purchase personal motor vehicle insurance, which is an interesting finding since one would think that the customized messages and the paid-for social media efforts of short-term insurance companies are aimed at increasing the likelihood that individuals will purchase insurance. This question in the questionnaire also tested the influence scale of each statement. The descriptive statistics revealed that two (buzz agents and paid-for social media efforts) out of the three communication sources included in the final model do not influence the respondents’ purchase decisions to a large extent. This result explains why the communications sources identified in the final model decrease the probability that consumers will purchase motor vehicle insurance. One would expect user-generated social media posts to have a much greater influence on the need to purchase motor vehicle insurance. Social media is a commonly used platform where consumers express their experiences of products and services. Word-of-mouth using user-generated social media is essential in the motor vehicle insurance market to enable individuals to talk about an insurance company’s motor vehicle insurance products (Ekanayake, 2021; Teixeira et al., 2018, p. 361). However, it was found in this study that word-of-mouth using user-generated social media does not influence the need to purchase personal motor vehicle insurance in South Africa to a large extent. Although it appears that buzz agents do not influence the respondents’ need to purchase motor vehicle insurance to a large extent, short-term insurance companies may not be using buzz agents enough or correctly. Globally, customized messages are seen as an effective way of reaching motor vehicle insurance consumers (Adamova et al., 2018; Moorcraft, 2021). The findings of this study correspond with international findings that customized messages, which represent an external factor, influence individuals’ need to purchase personal motor vehicle insurance, in this instance, in the context of South Africa. Respondents who did not have personal motor vehicle insurance were less likely to report a large influence of communication sources than those who had personal motor vehicle insurance.

The logistic regression indicated that all seven psychological attributes, as identified in the initial conceptual model, increase the probability that consumers will purchase personal motor vehicle insurance. Respondents who had personal motor vehicle insurance were more likely to report higher levels of agreement with the psychological attributes’ influence on need recognition than those who did not have personal motor vehicle insurance. As indicated in Table 3, the “personality” variable scored the lowest of all the psychological attributes, which means that it has the least influence on individuals’ purchase decisions. Perception about a certain insurance company has the highest index value in this construct, which indicates that it has the greatest influence on individuals’ purchase decisions within the “psychological attributes” construct.

Age, highest level of education and household income significantly contribute to the predictive ability of the model. All three of these demographic variables increase the probability that consumers will purchase personal motor vehicle insurance. The respondents who had personal motor vehicle insurance were more likely to report higher ages, educational levels and household incomes than those who did not have personal motor vehicle insurance.

7. Conclusions and practical implications

It is a complex and challenging endeavor to gain an understanding of the process involved in individuals’ purchase decisions within a market since there are internal and external factors that could influence consumers’ purchase decisions. The aim of this paper was to determine the factors that influence the purchase decisions of consumers and potential consumers regarding personal motor vehicle insurance in South Africa. This paper not only contributes to the body of knowledge on individuals’ purchase behavior in the personal motor vehicle insurance industry of South Africa but also offers practical implications for short-term insurance companies. Therefore, the main conclusions drawn from this study include: (1) The marketing efforts employed by short-term insurance companies, encompassing policy pricing, process, and personnel, significantly influence consumers’ decisions to purchase personal motor vehicle insurance. (2) Notably, the study identified that variables associated with a “positive sociocultural environment,” such as positive opinions from family or reference groups, non-commercial sources, social class, and cultural factors, decrease the likelihood of individuals purchasing such insurance. Conversely, “negative sociocultural environment” variables, including negative opinions from family or reference groups and social class, increase the probability of purchase. (3) Communication sources like paid social media efforts, buzz agents, and customized messages utilized by short-term insurance companies reduce the probability of individuals opting for motor vehicle insurance. (4) Psychological attributes, notably motivation, perception of motor vehicle insurance and certain short-term insurance companies, learning, personality, and attitudes towards both insurance and certain companies, significantly elevate the likelihood of purchasing personal motor vehicle insurance and thus impact purchase decisions. The motivation for financial protection and security emerges as a driving force for such decisions, influenced by respondents’ risk expectations and sensitivity. The purchasing decision process is steered by individuals’ perceptions of insurance and specific companies. Previous experiences and attitudes towards motor vehicle insurance shape future purchasing decisions, with attitudes exerting a greater influence than attitudes towards specific companies until the purchase decision point. (5) Demographic factors, specifically age, highest education level, and household income, notably influence purchase decisions concerning personal motor vehicle insurance.

The practical implications for short-term insurance companies encompass: (1) Enhancing sales and reducing policy cancellations hinges on skilled employees adept at delivering satisfactory consumer interactions during sales and claims procedures. (2) Streamlining service delivery through an automated quotation process tailored to market preferences could bolster productivity. (3) Effective pricing strategies, coupled with appealing prices, enable adaptation to economic challenges and ensure competitiveness. (4) While cultural and subcultural factors remain beyond insurers’ control, acknowledging evolving cultural trends is vital for tailoring policies to market demands. (5) Exploring the benefits of digital marketing and buzz agent strategies are avenues worth exploring. (6) Effectively crafting positive attitudes and perceptions through transparent communication, top-quality service, and competitive policies is paramount. (7) Notably, demographic factors—age, education level, and household income—emerged as influential in purchase decisions. Segmenting the market based on these demographics offers insurers the opportunity to optimize policy offerings and pricing for specific consumer groups.

8. Limitations and future research

Notwithstanding its considerable significance, this study does have certain limitations. The low response rate can be regarded as a limitation. Generalizability could be limited owing to the sampling size and the characteristics of the research sample, as well as the sampling methods used. However, the sample of n = 678 was sufficient for the performance of a statistical analysis that could lead to significant conclusions. Another limitation is that the findings are based on data relating to one industry, namely, the personal motor vehicle insurance industry, which is only a part of the larger motor vehicle insurance industry of South Africa. These limitations do not prevent this paper from contributing to the body of knowledge on individuals’ purchase behavior in the personal motor vehicle insurance industry of South Africa.

Further research could be done on price as a factor that influences consumers’ purchase decisions in respect of motor vehicle insurance, with specific reference to the post-COVID-19 financial obstacles that consumers face. Additionally, exploring new variables and dimensions that can potentially impact the consumer insurance purchasing process, like ethnocentrism, is recommended in future research. A pertinent aspect to consider when making insurance-related decisions is whether consumers exhibit a preference for domestic products over those offered by foreign companies. This exploration would contribute to a more comprehensive understanding of the multifaceted factors shaping consumer choices in the insurance market, thus paving the way for a nuanced and insightful analysis. Research could also be undertaken regarding consumers’ purchasing decision behavior in relation to other types of insurance, such as medical or household insurance and, more specifically, different insurance products.

Disclosure statement

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