Precedents of the compulsive use of a credit card: an analysis of university students' buying behavior

2.1 Hypothesis

Yamauchi and Templer (1982) initially developed the money attitude scale, which presents four dimensions related to attitudes toward money: power–prestige, retention time, distrust and anxiety. Three of these dimensions were adapted by Roberts and Jones (2001) and later used in the study by Felipe et al. (2017), who examined how attitudes of power–prestige, distrust and anxiety influence compulsive buying relationships among students. In addition to analyzing how power–prestige, distrust and anxiety influence buying behavior, our work observes how materialism can act as a predictor of compulsion.

The power–prestige construct represents the use of money as a way of impressing other people, with this financial resource being considered synonymous with power for some individuals (Roberts & Jones, 2001). Thus, there is signaling that current society considers the ostentation of material possessions as the most assertive parameter for power over the other (Sharif & Yeoh, 2018). Khare (2014) complements such thought and argues that compulsive buyers consider that acquiring material goods which are representative of wealth can influence the rise of social position.

Faber and O'Guinn (1988) report that people are predisposed to make judgments regarding possessions, and therefore they aim to acquire material goods with the intention of being judged positively. Individuals can not only buy because of the need for a material good, but also because of the feeling of superiority in their social circle, or because of the ability to influence others (Veludo-de-Oliveira et al., 2004). Both the material and the social aspects can induce people to use their credit card incorrectly (Veludo-de-Oliveira et al., 2014).

Smith (2002) corroborates this thought and adds that people act differently from what they would normally act when they gain something in return that they highly value, such as recognition, acceptance or something material. Palan et al. (2011) agree with this and argue that power–prestige, meaning money as a source of power or prestige, has induced young people to exaggeratedly use credit cards for purchases, especially university students (Wang & Xiao, 2009). These authors also suggest that one of the causes attributed to this misguided behavior is consumer culture, which involves both prestige power and materialism. Therefore, we tested the following hypothesis H1:

The anxiety construct refers to the feeling of nervousness when the individual faces situations that need to deal with money (Rimple et al., 2015). According to the study by Hayhoe et al. (2012), the relationship between savings and anxiety proved to be detrimental to attitudes regarding personal financial management, because people who had lower levels of anxiety were more assertive in their financial management behaviors. It should be complementarily emphasized that anxiety can be considered as an internal driving factor of individuals who buy uncontrolledly (Matos & Bonfanti, 2016).

The study by Roberts and Jones (2001) points out that the central motive of people who buy compulsively is an attempt to escape anxiety. However, when an individual seeks to eliminate anxiety through excessive shopping and spending, they acquire more and more objects, thus entering a vicious circle to feel good (Edwards, 1993). Accordingly, anxiety can be considered both an element that precedes the act of compulsively buying and a consequence of this scenario (Roberts & Jones, 2001). Therefore, this paper examines the following hypothesis H2:

Individuals with higher levels of distrust tend to have feelings of hesitation and doubt regarding contexts involving spending money (Yamauchi & Templer, 1982). Phau and Woo (2008) argue that more distrustful people can have difficulty in relying on their own ability in situations involving shopping and become insecure; therefore, they do too much research to achieve the ideal acquisition.

Cho (2006) suggests that distrust should play a critical role in consumer behavior, as distrust is a type of encoding that exists in the memory of these individuals and induces deviations from social norms. The theory is that people tend to make economic and financial decisions based on an assessment of potential gains or losses, with greater weight given to decisions in loss scenarios (Tversky & Kaneman, 1986). This reveals that negative information or messages can have a greater impact on personal behavior (Barros and Felipe, 2015).

Distrust is also closely related to the fear the individual has about the price paid for a material good and the benefits of obtaining it (Felipe et al., 2017). Roberts and Jones (2001) describe that price-sensitive people are less likely to become compulsive buyers since they are always questioning prices and are less likely to buy material goods in an exacerbated way. The price-sensitive dimension contains people for whom it is more common to compare prices before buying, and as a result they are able to calmly select items (Roberts & Jones, 2001). A good example of this is people who take advantage of promotions to accumulate things (Klontz, Britt, Mentzer, & Klontz, 2011). They do not necessarily keep their work environment or staff organized, but when they do, they realize how many unnecessary things they have (Rimple et al., 2015). In view of this, we tested the following hypothesis H3:

Another important factor to complement the study of precedents of the compulsive use of credit cards is materialism, defined as the emphasis that an individual gives to the possession of material goods in their life (Richins & Dawson, 1992). Materialism can be measured through the material values scale, developed by Richins and Dawson (1992) and later improved by Richins (2004). This measurement scale of materialism presents questions about centrality, happiness and success, as the three main domains of materialism (Richins & Dawson, 1992).

According to Burroughs and Rindfleisch (2002), a materialistic individual tends to associate increased well-being with material possession. Furthermore, Medeiros et al. (2015) add that materialistic people consider the acquisition of material goods as being primordial to being happy and having a successful life. In this sense, especially when it comes to happiness, what distinguishes the materialist is that this person aims for happiness through acquiring objects and not through lived experiences (Richins & Dawson, 1992). In addition, it is important to point out that higher levels of materialism can influence the difficulty of controlling the act of buying in the face of external stimuli (Santini et al., 2019). From this, we tested the following hypothesis H4:

Thus, it is predicted that the power–prestige, anxiety, distrust and materialism constructs may exert some influence on the compulsive use of credit cards. Therefore, the presentation of the constructs of this study and their respective hypotheses were gathered in an initial theoretical model (Figure 1).

4.1 Analysis of the sample

The survey sample was composed of 506 respondents, of which 63% were women and 37% men. The majority of respondents were between 18 and 29 years old at the moment of data collection, composing 79.6% of the total sample; 12.5% were between 30 and 36 years old; 5.3% were between 37 and 50 years old and 2.6% were over 50 years old. Of the respondents, 81.4% were single; 14.4% were married; 1.4% were divorced and 2.8% signaled as others. Regarding monthly income, the majority did not have any income, integrating 37% of the total sample; 20.7% had a monthly income of up to R$998.00; 22.9% had a monthly income of from R$998.01 to R$1,996.00; 8.5% had a monthly income from R$1,996.01 to R$2,994.00; 2.8% had a monthly income from R$2,994.01 to R$3,992.00; 3% had a monthly income from R$3,992.01 to R$4,990.00 and 5.1% earned above R$4,990.00.

Of the respondents, 63% are women, 79.6% of the sample were young and 37% of the sample has no income. These data should indicate that women are more prone to indebtedness (Vieira et al., 2014), which could be because they may be more likely to engage in compulsive buying behaviors, according to Faber and O'Guinn (1992). Furthermore, young people are more prone to indebtedness, an argument that is in line with a study by Ponchio (2006). These young people are also at the beginning of their financial lives, and in many situations, without income or with family financing (Bricker et al., 2012).

4.2 Precedents of compulsive use of the credit card

As suggested by Hair et al. (2009), a range of ±2.58 for asymmetry and ±3 for kurtosis was adopted as the parameter. Based on the values obtained, it was decided to preliminarily exclude the following variables: P3 (“I use money to influence other people to do things for me”), P4 (I seem to find that I show more respect to people with more money than I have”) and C3 (“I frequently use available credit on one credit card to make a payment on another credit card”), because they did not follow a normal distribution (Hair et al., 2009).

We calculated Cronbach's alpha to evaluate the reliability of the scales used, as it is a traditional and widely used measure to verify the internal consistency of the variables of a study. Thus, as recommended by Hair et al. (2009) and Hair et al. (2017), the Cronbach's alpha value should be higher than 0.6 or, being more conservative, 0.7. After measuring this value, it was possible to conclude that the adopted scales obtained satisfactory internal consistency (Table 1).

Once the normality of the data and the reliability of the scales were confirmed, the first model consisting of five latent variables was elaborated: power–prestige, anxiety, distrust, materialism and credit card use, totaling 33 observed variables. We calculated the latent variables representative of the constructs studied using CFA. Thus, the scores of the latent constructs were the factorial scores. The factor loads in this model were examined to ensure that the fit quality indexes were satisfactory, considering that the factor loads should be higher than 0.5. However, it was found that the fit quality indexes would be unsatisfactory, so we initially discarded the first model.

A second model was tested promoting some modifications in the variables observed as satisfactory for the fit quality indexes. These modifications refer to excluding the following items in relation to the initially tested model: P2 (“People I know tell me that I place too much emphasis on the amount of money a person has as a sign of success”), A4 (“I spend money to make myself feel better”), A5 (“I am bothered when I have to pass up a sale”), A6 (“It is hard for me to pass up a bargain”), D1 (“I automatically say, ‘I cannot afford it’ whether I can or not”), D2 (“When I make a major purchase, I have a suspicion that I have been taken advantage of”), D5 (“I hesitate to spend money, even on necessities”), C1 (“My credit cards are usually at their maximum credit limit”), C2 (“I have too many credit cards”), C4 (“I worry how I will pay off my credit card debt”), C5 (“I often make only the minimum payment on my credit card bills”), C9 (“I rarely go over my available credit limit”) and C10 (“I seldom take cash advances on my credit cards”).

Thus, five measures of adjustment quality were tested to validate this model: χ² normed, GFI, AGFI, RMSEA and SRMR. It was subsequently verified that the values obtained in the quality of fit index measures were in line with what the literature reports. However, some values obtained when calculating the AVE were not satisfactory, and so therefore it was decided to not follow this model because the minimum acceptable value is 0.5 (Hair et al., 2017).

A third model was sequentially tested in which some tests were performed with the observed variables, and then it was necessary to exclude the observed variable M2 (“I like a lot of luxury in my life”) and the latent variable prestige-power (H1) in order to increase the AVE of latent variables (maximum likelihood for both models). As a result, we obtained satisfactory values for seven measures of fit quality in this model GFI, AGFI, NFI, CFI, RMSEA, TLI and SRMR, as shown in Table 2.

According to the literature, the GFI should be close to 0.90–0.95, with the GFI obtained being 0.93. AGFI should also be close to 0.90–0.95, and the AGFI obtained was 0.90. As recommended by Byrne (2010), the NFI and the CFI should be close to 0.95, with an NFI value of 0.92 and the CFI of 0.94 being obtained, respectively. The acceptable RMSEA should be between 0.05 and 0.08, and the RMSEA obtained is 0.07. Also, the TLI should be close to 0.90–1.00, and the TLI obtained was 0.92. To finish, the SRMR must be less than 0.08, and the SRMR obtained was 0.05. Thus, all the values found indicated compliance with the established parameters, indicating that this model has a satisfactory quality of fit (Figure 2).

Next, AVE and the CR were calculated to analyze the convergent validity of the adopted model (Table 3). The AVE represents the amount of variation of the observed variables of the respective constructs (Hair et al., 2009). The CR consists of an internal consistency measure of the variables observed concerning each construction (Hair et al., 2009). As suggested by Hair et al. (2017), the AVE and CR values should be respectively equal to or higher than 0.5 and above 0.7.

The Fornell–Larcker criterion was used to verify the discriminant validity which makes a comparison between the square root of the AVE and the correlations of each construct (Yee, Seong, & Chin, 2019). The discriminant validity (Table 4) is obtained when the square root of the AVE of each construct is greater than the correlation between these constructs analyzed in the study.

The values achieved by AVE, CR and discriminant validity were satisfactory. Moreover, all factor loads were higher than 0.6, and as proposed by Hair et al. (2009) and Hair et al. (2017), this value must be higher than 0.5 for the groupings to have practical significance. It should also be noted that the last model presented obtained the best fit quality indexes.

At this point, the p-values of the relationships between the observed variables and their respective latent variables were verified to validate the performed CFA. We also confirmed that all the indicated groupings were significant (level of 10%). Therefore, the performed CFA (Table 5) can be validated based on the concomitant analysis of AVE, CR, discriminant validity and p-values obtained.

After confirming the validity of the CFA, the tested relationships between the latent variables were examined through path analysis, which can be verified through the p-values obtained and the beta coefficient. First, the relationship between anxiety and compulsive use of the credit card was examined, which obtained a p-value of 0.001, proving its statistical significance. As the beta coefficient was 0.247, we noted that this was a directly proportional relationship (i.e. the greater the anxiety of the consumer, the greater the compulsive use of their credit card). From this, we understand that H2 was supported.

The finding that anxiety positively influences the compulsive use of credit cards can be reinforced through the study of Veludo-de-Oliveira et al. (2014). In this study, interviewees demonstrated that the anxiety level generated in daily life could be mitigated through purchases. Thus, it is considered that more anxious people tend to buy and spend too much. This evidence converges with the arguments in the study by Hayhoe et al. (2012), in which individuals with higher levels of anxiety were less willing to save some money. Veludo-de-Oliveira et al. (2014) also confirm that anxiety and credit cards act together in increasing uncontrolled purchases.

The relationship between distrust and compulsive use of credit cards was subsequently verified, in which we found a p-value of 0.238 and a beta coefficient of 0.062. Therefore, it was not possible to attest to its statistical significance, and we can therefore reject H3. This result differs from similar studies such as Roberts and Jones (2001), in which it was found that distrust of situations involving money reduces the possibility of individuals compulsively buying. However, the results regarding distrust are similar to those obtained by Sharif and Yeoh (2018).

The relationship between materialism and the compulsive use of credit card was analyzed, which reported a p-value of 0.000. As the beta coefficient was 0.299, we understand that this is a directly proportional relationship, meaning that the greater the materialism present in the individual, the greater the compulsive use of a credit card tends to be. As a result, the H4 was sustained. Confirmation of this finding can be explained by the fact that materialistic people tend to buy excessively (Medeiros et al., 2015), and this reinforces their behavior of uncontrolled acquisition of goods and attachment to material things (Moulding, Duong, Nedeljkovic, & Kyrios, 2017).

Our finding converges with the study of Limbu, Huhmann, and Xu (2012), in which it they found that young university students with higher levels of materialism used credit cards more often than less materialistic young people. In a complementary way, the studies of Pirog and Roberts (2007) and Sidoti and Devasagayam (2010) suggest that the greater the materialism, the greater the misuse of credit cards. As compulsive buying behavior has grown (Santos et al., 2021), we understand that our research has interesting implications for addressing compulsive credit card use. We point out the need for education of credit card users in relation to the responsible, conscious and balanced use of this means of payment. We believe that in order to formalize such action, it would be essential that colleges and universities include courses and lectures on financial management, and not only address this issue in elective subjects but also in mandatory subjects. We also emphasize the need for parents to help their children with the use of credit cards. We know that learning from healthy financial habits is not easy, takes time and requires efforts, so we believe that companies should also play a role in promoting the conscious and proper use of credit cards.