The desire Corporation desires to investigate the motivations of shoppers who purchase

Research objective

To study the prevalence of impulsive and compulsive shopping among consumers.

Scope

To study the impulsive and compulsive shopping behaviour of consumers in the apparel and accessories market (online as well offline).

Research questions

RQ1.

What is the role of social commerce in promoting impulsive and compulsive shopping?

RQ2.

Does the hedonic motivation of shopping promote impulsive and compulsive shopping?

RQ3.

The impact of COVID-19 on impulsive and compulsive shopping.

Hypotheses development

In accordance with the research questions, the following hypotheses are developed:

H1a.

Social media influence is significantly associated with impulsive buying.

H1b.

Social media influence is significantly associated with compulsive buying.

H2a.

Social media preferences are significantly associated with impulsive buying.

H2b.

Social media preferences are significantly associated with compulsive buying.

H3a.

Hedonic motivation is significantly associated with impulsive buying.

H3b.

Hedonic motivation is significantly associated with compulsive buying.

H4a.

Shopping in the times of COVID-19 is significantly associated with impulsive buying.

H4b.

Shopping in the times of COVID-19 is significantly associated with compulsive buying.

Research methodology

Quantitative research methods have been used for the purpose of this study, involving the use of statistical procedures for analysis (Onwuegbuzie and Leech, 2005). A close-ended questionnaire (Rossi and AB, 1983) was used for data collection with some previously proven constructs of social media influence, hedonism, impulsive buying and compulsive buying, as well as shopping in times of COVID-19 developed by the researcher. A questionnaire is a reliable instrument that is simple to administer, and an extensive amount of data can be generated in a cost- and resource-effective manner. The anonymity and confidentiality of the respondent are also respected (Welman and Kruger, 1999). The physical absence of the researcher also leads to non-biased responses. The apprehension of being judged by others would lead to Social Desirability Response (Mittal et al., 2018).

The method used for data collection was an online distribution of the questionnaire via google forms through social media channels, considering the restriction on movement because of COVID-19 efficiency and economic feasibility. This method also allowed accessibility to a larger sample and made it easier to collect and compile data (Metzner and Mann, 1952). The target population selected was online and offline shoppers above the age of 18 in Ahmedabad. A non-probability convenience sampling technique is used to collect data (Takona, 2002). Respondents were selected based on accessibility. However, because of certain categorical questions, judgement was used in selecting the final data. A total of 200 respondents were approached, out of which 146 questionnaires are completely filled and are valid.

Statistical package for the social sciences (SPSS) version 23 was used to analyse the collected data. The data was appropriately coded and questions that were negative in nature were appropriately reverse coded. Also, disguised questions were appropriately calculated with the related variable. All such variables are included in (Table 1).

The data was collected on a five-point Likert scale of agreement where 1 is strongly agree and 5 is strongly disagree. Based on their overall mean score, respondents were classified on whether their mean score was above or below 3.

Statistical analysis

• Descriptive statistics (Tables 2–4).

• A test reliability of scale to measure the consistency of the scale (Table 5).

• Followed by factor analysis with principal component analysis (PCA) method to find the latent variables (Tables 6–9).

• Structural equation modeling (SEM) in AMOS (statistical software) – SPSS is used for a diagrammatic representation of the relationship between variables (Figure 1, Tables 10–13).

• Regression analysis is further used to measure the relationship between predictor variables and dependent variables (Tables 14–21).

• Chi-square test is used to test the hypotheses (Tables 22 and 23).

Instrument

A questionnaire is formulated using various sub-scales of impulse buying, social commerce, compulsive buying and hedonic motivation of shopping as well as shopping in the times of COVID-19. Each construct is referenced with classic papers in the area of consumer behaviour as shown in Table 1. Some of the questions are developed by the researcher based on the unique situation created by COVID-19.

The questionnaire is created in Google forms with multiple response grids for the Likert scale, where 1 = strongly agree, 2 = agree, 3 = neutral, 4 = disagree and 5 = strongly disagree. Questions falling under the same construct are put together. However, some questions are interchanged and reversed to get an unbiased response.

Data collection

The data is collected over a three-day period in the month of September 2020. Social media networks were used for getting the respondents to participate in the questionnaire (Wadhera and Sharma, 2019). Respondents were approached in accordance with the research methodology. A total of 200 respondents were approached, out of which 146 participated. All of them were valid with no missing fields.

Respondent characteristics

Table 2 shows the demographic profile of the respondents. They are mainly from the age group 30–39 (48%). These are the older millennials. About 75% were females, 85% married and 53.4% are postgraduates. About 53.4% of the respondents have a monthly family income of more than 200,000 and 38.4% are self-employed. These categories have relatively more engagement in shopping in general. The younger millennials (25–29 years) and Gen Z (less than 25 years) are 16.6%. The older millennials (30–39 years) have a higher income and as a result a higher spending capacity.

Descriptive statistics

The mean values of items used in the scale are illustrated in Tables 3 and 4. Table 3 contains the means values of four variables, namely, social media influence, social media preferences, hedonism and shopping in the time of COVID. Table 4 shows the mean values of impulsive and compulsive buying.

Reliability test

Straub (1989) states that constructs reliability shows the internal consistency of the scale items measuring the same construct for the data. Cronbach’s alpha is used to measure the reliability of the scale. Cronbach’s alpha was calculated for each construct. Here, the Cronbach’s alpha is 0.951, which is above the recommended value of 0.7 reflecting reliability of the scale as shown in Table 5. Thus, the measurement shows good reliability.

Factor analysis

Kaiser-Meyer-Olkin (KMO) measures the sampling adequacy, which should be close to 0.5 for a satisfactory factor analysis to proceed (Kaiser, 1974). It determines whether the responses given with the sample are adequate or not. A value of 0.5 is considered acceptable, 0.7–0.8 is considerate acceptable and above 0.9 is considered as outstanding. To test the sampling adequacy, KMO test was carried out and the resultant value is 0.871 as shown in Table 6. This is way above the recommended value of 0.5 and closer to outstanding value of 0.9. Thus, it can be considered as acceptable.

To remove the redundant variables and uncover the latent variables, all the 28 variables of factors influencing impulsive and compulsive buying are treated with PCA to identify closely related variables. Out of the 28 variables, six latent variables emerged on rotation of the variables using varimax method as shown in Table 8. This is done to make the interpretation of the analysis easier. Factor analysis shows that 66.87% of the total variance can be explained by classifying 28 variables into six components or factors as shown in Table 7. Only the variables with eigenvalue of more than 1 are accepted in the study. The six components are further named as shown in Table 8.

The model is an over-identified model with df being 1. The goodness of fit indices are acceptable (chi square = 8.007, p value = 0.005, root mean square error of approximation = 0.22, goodness of fit index = 0.987, normed fit index = 0.987, comparative fit index = 0.988).

Social media influence, social commerce, social commerce-electronic word of mouth (EWOM), hedonic happiness, hedonic fun and shop in COVID-19 are exogenous variables that predict the endogenous variables impulsive buying and compulsive buying. Error variables e1 and e2 are unique variables that could affect the endogenous variables. The predictor variables can predict the dependent variables up to 45% for impulsive buying, whereas 59% for compulsive buying as shown in Figure 1. Both values are above 30%, hence are considered acceptable.

Maximum likelihood estimates

For the endogenous variable impulsive buying, the most important predictors are social media influence (0.52) and shopping in COVID-19 (0.515) with significance levels of less that 0.005 as given in Tables 10 and 11.

For the endogenous variable compulsive buying, the most important predictors are hedonic happiness (0.38) and shopping in COVID-19 (0.348) at significance levels of less than 0.005 as given in Tables 10 and 11.

All the values (p ≥ 0.05) are acceptable, so there exists co-variance between all exogenous variables as shown in Table 12. Hence, social media influence, social commerce, EWOM of social commerce, hedonic happiness, hedonic fun and shopping in COVID-19 reflect co-variance among each other (Table 13).

All Pearson’s correlations values are above the recommended value of 0.3, hence, independent variables (social media influence, social commerce, EWOM of social commerce, hedonic happiness, hedonic fun and shop in COVID-19) and dependent variable (impulsive buying) are correlated to each other as shown in Table 14.

Regression analysis

Regression analysis is used to test the significance and the relationship between dependent and independent variables. The model summary shows R = 0.667 and R2 = 0.445 as given in Table 15 . This shows dependent variable impulsive buying can be explained by the two factors by 66%. It also means social media influence and shop in COVID-19 contribute significantly and predict 44.5% of the variation in impulsive buying.

The F test states that the regression model predicts the outcome significantly as shown in Table 16. The level of significance is 0.000, which means the model can predict impulsive buying.

The t-values should be df − 2, which is 4 in this case. Table 17 shows social media influence (t = 5.749) and shop in COVID-19 (t = 3.9), which is almost 4. Social media influence and shop in COVID-19 have emerged as the strongest predictors for impulsive buying. This can also be re-affirmed with SEM shown in Figure 1.

All Pearson’s correlations values are above the recommended value of 0.3, hence, independent variables (social media influence, social commerce, EWOM of social commerce, hedonic happiness, hedonic fun and shop in COVID-19) and dependent variable (impulsive buying) are correlated to each other as shown in Table 18.

For all correlations above 0.3, endogenous and exogenous are correlated and independent and dependent are correlated.

Regression analysis is used to test the significance and the relationship between dependent and independent variables. The model summary shows R = 0.769 and R2 = 0.592 as shown in Table 19. This shows dependent variable compulsive buying can be explained by the two factors by 76%. It also means hedonic happiness and shop in COVID-19 contribute significantly and predict 59.2% of the variation in impulsive buying.

The F test states that the regression model predicts the outcome significantly as shown in Table 20. The level of significance is 0.000, which means the model can predict compulsive buying.

The t-values should be df − 2, which is 4 in this case. Table 21 shows hedonic happiness (t = 4.5) and shop in COVID-19 (t = 6.7). Both values are above the recommended value of 4. Hence, hedonic happiness and shop in COVID have emerged as the strongest predictors for compulsive buying. This can also be re-affirmed with SEM as shown in Figure 1.

Hypotheses testing

To test the hypotheses, chi-square test is conducted as well as phi and Crammer’s V are calculated. All results are displayed in Tables 22 and 23. The Pearson’s coefficients are highly significant with all (p = 0.05). Thus, all null hypotheses are rejected except H1b. Therefore, H1a, H2a, H3a, H4a, H5a, H6a and H2b, H3b, H4b, H5b and H6b are accepted. Thus, social media preferences, hedonic motivation and shopping in times of COVID-19 are significantly associated with impulsive and compulsive buying. However, social media influence significantly associates with impulsive buying but not with compulsive buying (H1b is rejected as p ≥ 0.05).

Findings and recommendations

All the variables of the six factors show a positive correlation with impulsive buying and compulsive buying. However, the regression analysis illustrated social media influence and presence of COVID-19 pandemic as the strongest predictors for impulse buying.

Impulsive buying is defined as unplanned purchases made on the spur of the moment (Rook and Fisher, 1995). Customers who feel inspired by social networks, purchases of their contacts, recommendations of friends and influencers on social media are most likely to go in for impulse purchases. Social media feeds many images on different platforms. The mentions and tags done by people on the contact list allows other users to see what brand they are purchasing. Tagging and retagging the mentions and pictures make the images available for more and more users (Çelik et al., 2019). After buying a particular brand that others on one’s contact list have purchased, users tend to mention and tag the same brand that others are tagging, thus keeping up with the trend. Such customers do not even need to read more about the brand. When someone from the social media friends buys it, the prospective customer reads this as an approval. The ease of clicking on a friend’s post takes the customer to a virtual shop – which allows buying in seconds facilitates impulsive buying. Online shopping has seen an upward trend since a few years now. With the current situation of COVID-19, online shopping has been a more preferred approach (Assomul, 2020; Ingaldi and Brozova, 2020). Most retailers have social media pages and transact through the same pages. Links for buying on the same page facilitates the impulse felt in that moment. Attractive marketing and promotion acts as an encouraging factor in such a situation to facilitate an unplanned purchase.

It is ideal for marketers to capitalize on this spontaneous shopping. Ease of payment provided by credit cards and other modes as well as fast and convenient home delivery further attracts the customers. Impulsive buying is significantly associated to influencers too. Micro-influencers have been sought after world over because of the niche groups they influence (Dhanesh and Duthler, 2019). Because of the small size of the groups, their reach and effectiveness could be more than celebrities. Marketers should identify and hire such influencers who are congruent to their brand to increase customer engagement leading sales. Retailers should further engage existing and prospective customers by reminding them to tag and mention their purchases with them. They could disguise it with a contest or give away alert. This would also help encourage others on their friends’ list to go in for an impulsive purchase. These spontaneous purchases should be shown as a new way of shopping as a complete contrast to planned researched shopping, highlighting the fact that when one trusts the retailer, one can buy anytime and not actually plan and research for it.

Compulsive buying is defined as an uncontrolled urge to buy regularly (Faber and O’Guinn, 1989). It is different from impulsive buying. It is also not a higher spectrum of the same scale (Flight et al., 2012). Hedonic motivation is seen as the main predictor for compulsive buying in this study. Impulse shopping and acquiring new products is a central activity in their lives. They shop to relive stress. Their happiness is highest when they shop. They feel it absolutely necessary to keep up with trends (Kukar-Kinney et al., 2016).

For marketers, it is easiest to tap into this segment. They love shopping and are doing so on a regular bases; if such customers are buying a certain brand, they must be retained. Hedonic happiness is of maximum importance to them, hence, they would love a great in-store experience or an advertisement driven towards hedonism in online shopping scenario. They shop not for buying the product but for enjoying the experience (Langrehr, 1991). They feel delighted with the compliments they get online or in person when they use their newly purchased products. Artificial intelligence directed towards fit and virtual body avatars (Tandon, 2021) could further engage these shoppers.

Both the impulsive buyers and the compulsive buyers have shown interest in shopping during these uncertain times. The presence of COVID-19 makes people find solace in shopping. They find it prudent to spend instead of saving. They shop in spite of a lot of things unused from last purchases. A small purchase also gives them happiness – the lipstick effect (Jamal, 2020). The continued presence of the pandemic has mandated customers to stay home to stay safe. Social media platforms have allowed them to wear and flaunt new shopping via pictures and posts even from the comfort of their homes.

Though many governments are trying to normalise the offline shopping experience, most companies had to pivot to online shopping to save the day. Ease of payment and contact-less delivery further encourage online shopping. COVID-19 has put social media and online shopping in the driver’s seat in the marketing game. To further leverage the situation and maximise sales, customers should be encouraged to make impulsive and compulsive purchases. Shopping in the time of COVID-19 should be promoted as comforting and way to cope with the uncertainty of the pandemic.

Limitations and further scope of research

Research is conducted in the city of Ahmedabad among SEC A and B in the age group of 25–70, though a large number of respondents fall under the age group of 30–50. Similar research can be conducted with the younger millennials and Gen Z. Most of the respondents are from the cities of Ahmedabad and Mumbai, however, research can be replicated for other cities of India. Apparels and accessions being global products, research can be conducted in any city around the world. A larger sample can be studied as well. The influences and preferences of social media have been studied in this paper. A focussed study on the moderating role of influencers and the impact of different platforms of social media can be conducted.

Because of the presence of the pandemic COVID-19, meeting the respondents was not possible for safety. However, in-depth interviews with the respondents could lead to more in-depth understanding of the same.