Keywords

Urge to Buy Impulsively, Antecedents, E-impulse Buying, Moderating Factors.

Introduction

The emergence of cashless economy and the expansion of online businesses drastically change the shoppers’ purchase pattern as well as retail setting attributes. Researchers able to track the shopping patterns in online settings and found a significant number of purchases are on impulse. Adelaar, Chang, Lancendorfer, Lee & Morimoto, (2003) also advocated that consumers often encounter with urges to buy something impulsively and hence impulse purchases contribute magnificently in sales revenue. E-tailers in this internet prone era try to increase the sales revenue by creating more probability of encountering UBI on online shopping websites. And as only positive and satisfactory shopping experience lead to more impulse buying in future thus they are also trying to develop a paradigm that can provide higher shopping satisfaction. Therefore we can say that E-impulse buying is the phenomenon having utmost importance in this world of e-business so it is essential to get the deeper insights about this concept. This paper attempts to develop a comprehensive model for E-impulse buying by emphasizing over the all-important dimensions of UBI that ultimately leads to E-impulse buying.

Literature Review

E-impulse Buying

E-impulse buying can be defined as the spur-of-the-moment buying or a buying without a priori shopping intention in online context. Traditionally, Stern in 1962 derived major four types of impulse purchases: Pure impulse purchase; reminder purchase; suggestive impulse purchase; and planned impulse purchase. All these types of impulse purchases are linked with an unplanned buying behavior of shoppers. In online context several antecedents propel shoppers to buy something on impulse.

Urge to Buy Impulsively (UBI)

Urge to buy impulsively can be defined as the sudden desire to buy a product while encountering various stimuli (Rook, 1987). It is a state of mind to experience spontaneous urge to buy while encountering an object in an environment (Beatty and Ferrell, 1998). And then consumers try to fulfil their spontaneous desire by buying that product. Hence it can be concluded as all the actual impulse purchases are encountered only after experiencing UBI (Rook, 1987). It is not mandatory that all the urges to buy impulsively can be converted in to impulse purchases but the likelihood of actual impulse purchases can be increased by developing more chances of experiencing UBI (Beatty and Ferrell, 1998).

Contribution of Other Models in E-impulse Buying

Many researchers tried to capture the antecedents of UBI that further lead to E-impulse buying. Most of the studies in online context emphasize over the website related factors that create wide platform of experiencing UBI. Zhang, Prybutok & Strutton, in 2007 demonstrated the role of gender and subjective norms in influencing consumer impulsivity and actual impulse purchase. The impact of electronic service quality on the buying impulse shows significant results (Bressolles, Durrieu & Giraud, 2007). Parboteeah, Valacich & Wells, in 2009 also explored the significant impact of website quality in triggering impulse buying. The effect of online atmospheric cues also have significant impact on the online impulse buying (Floh and Madlberger, 2013; Liu, Li & Hu, 2013). Demographic factors like gender, marital status and income also influence impulse buying on internet (Jeffrey and Hodge, 2007; Kollat and Willet, 1967). Apart from website attributes and demographic factors, psychological attribute like impulse buying tendency, affective state and normative evaluation of an individual have a positive impacts on impulse buying (Dawson and Kim, 2009). Verhagen and Dolen in 2011 highlighted the positive impact of merchandise attractiveness, enjoyment and online store communication on impulse purchase. Other studies also reveal some important factors that may lead to E-impulse buying like-personality traits, emotions, social media, social capital and peer effect (Ozen and Engizek, 2014; Cheng and Tseng, 2014; Turkyilmaza, Erdem & Uslu, 2015; Huang, 2016).

Research Model and Hypothesis

Literature review suggest the need of a comprehensive model that may include all the main dimensions of the antecedents of urge to buy impulsively that may ultimately lead to E-impulse buying. Different studies on impulse buying uncover several dimensions of UBI that stimulate E-impulse buying. The main dimension that is the foundation of UBI in online context is website attributes. In this study UBI is captured by considering all the factors that stimulate UBI in a comprehensive way. Thus UBI has been captured by five validated dimensions-website related factors; situational factors; promotional offers; impulse buying tendency and website communication. The UBI construct for this study will be captured by above mentioned five dimensions including sixteen items. E-impulse buying or actual impulse purchase on internet will be captured by two variables-EIB1: When I saw this product, I suddenly wanted to buy it; and EIB2: When I saw this product, I could not resist the desire to buy it. As it is obvious that E-impulse buying is preceded by urges to buy impulsively first hence it is clear that UBI has some relationship with E-impulse buying. Verhagen and Dolen in 2011, state the positive impact of UBI on impulse buying. Huang in 2016 also advocated UBI as the critical determinant of impulse buying in online context. Thus the first hypothesis generated over here is:

H1. Urge to buy impulsively is positively related to E-impulse buying

Researchers also argued over the impact of some demographic factors like – gender, marital status and income and on the relationship between UBI and E-impulse buying. Kollat and Willet (1967) advocated that consumer’s characteristics and demographics also influence the impulse purchases. Dittmar, Beattie & Friese, (1995) also introduced gender as a factor that influences impulse buying of a product. Zhang, Prybutok & Strutton. (2007) through his model of relationships between genders, subjective norms, consumer impulsivity, purchase intention and actual purchase behavior in online marketing environments, derived that male consumers are more impulsive online shoppers than female consumers. From a socioeconomic point of view, individuals with relatively low levels of household income indulge into impulse buying. Hence we can consider income as a significant factor that may affect the incidences of E-impulse buying. Kollat and Willet (1967) advocated that impulse buying increases irregularly with increase in number of years of marriage. Marital status influences the shopping pattern of individuals for both planned and unplanned purchases. It is also observed that the time spend on internet by individuals will lead to more probability of encountering urges to buy impulsively due to various online stimuli. Jeffrey and Hodge (2007) advocated that the likelihood of impulse buying on internet is positively correlated to the amount spent on website prior to see an impulse item. Internet addiction and its impact on E-impulse buying are also observed to be significant (Sun and Wu, 2011). Therefore we can conclude that higher usage of internet leads to higher possibilities of encountering urges to buy impulsively on the internet and it will ultimately leads to more chances of E-impulse buying. So finally the hypothesis formulated from the above mentioned rationales are as follow:

H2. Gender significantly moderates the relationship between urge to buy impulsively and E-impulse buying

H3. Income significantly moderates the relationship between urge to buy impulsively and E-impulse buying

H4. Marital status significantly moderates the relationship between urge to buy impulsively and E-impulse buying

H5. Internet usage significantly moderates the relationship between urge to buy impulsively and E-impulse buying

Therefore the proposed model for the E-impulse buying will be reflected as the E-impulse buying impacted by UBI consisting five dimensions (website related factors, situational factors, promotional tools, website communication and impulse buying tendency) and this relationship of UBI and E-impulse buying will be moderated by gender, income, marital status and internet usage (Figure 1).

Methodology

For this study sample data has been collected through online survey. The online questionnaire includes 18 items and few questions to collect demographic statistics of the respondents. Few questions were also asked to know about their internet usage and other related to shopping. 223 respondents effectively respond to this survey out of which 101 were female and 122 were male respondents (Table 1). To get the comprehensive scale for the antecedents of UBI first exploratory factor analysis was applied and to check the relationship of UBI with E-impulse buying confirmatory factor analysis and structural equation modelling was applied by using Amos 24. Moderating effects of other variables on the relationship of UBI and E-impulse buying was also checked during the modelling by using Amos 24.

Table 1: Demographic Characteristics of the Respondents
		GENDER
		Female	Male
	18 to 24 years	27	68
Age group	25 to 35 years	65	42
	36 to 50 years	8	10
	More than 50 years	1	2
Marital Status	Married	40	31
	Unmarried	61	91
Family Income (per annum)	Less than 5 lakhs	31	48
	5 to 10 lakhs	44	58
	10 to 20 lakhs	19	12
	More than 20 lakhs	7	4
Education	Bachelors or less	19	54
	Masters or less	60	55
	Doctoral or less	22	13

Data Analysis and Results

Sampling

Data was collected through online survey by using convenience sampling throughout the India from the online shoppers who often buy impulsively. The survey starts with an introductory note about the aim of the survey and it was directed that only those shoppers who often go for unplanned buying on the internet may carry forward with this survey. Finally 223 effective responses were received throughout the India. These respondents are from different regions of India and are from different demographic statistics.

The Instrument

For capturing data from the respondents regarding E-impulse buying a questionnaire of 18 items at seven point Likert type scale ranging from 1) “strongly disagree” to 7) “strongly agree was constructed. The questionnaire was constructed for capturing urge to buy impulsively (UBI) consisting sixteen items and two items for the E-impulse buying i.e., actual impulse purchase in online context. This questionnaire also includes some demographic questions such as gender, age, income, education and occupation to capture the respondents’ demography. Few other shopping related questions were also asked like internet usage, shopping probability and shopping experiences.

Sample Characteristics

Table 1 exhibits the demographic details of the respondents. Out of 223 respondents 101 were female and 112 were male respondents. These respondents belong to diversified age groups, out of which 25 to 35 years age group entailed maximum respondents. The survey revealed that 49.3 percent of respondent are pursuing bachelors or hold a degree of bachelors (Table 1). Most of the respondents are from the income group 5 to 10 lakhs per annum and maximum number of respondents is employed.

Exploratory Factor Analysis (EFA)

For evolving the dimensions of UBI principal component analysis was applied on sixteen variables of UBI revealing the five distinct dimensions out of the sixteen variables. The KMO value was found to be 0.798 that comes under the acceptable range i.e., more than 0.6 and also found to be significant (Table 2).

Table 2: Kmo And Bartlett's Test Result
Kaiser-Meyer-Olkin Measure of Sampling Adequacy		0.798
Bartlett's Test of Sphericity	Approx. Chi-Square	2540.863
	Df	120
	Sig.	0.000

Communalities for all the sixteen variables were also found in acceptable range i.e., more than 0.5. The value of Cronbach’s alpha for all the five dimensions is more than 0.7 that is quite good. Total variance explained (TVE) by these five dimensions is 80.478% (Table 3). In principal component matrix we avoid factor loadings less than 0.4. Table 3 reflects the clear cut five dimensions of UBI that may further impacts on E-impulse buying.

Table 3: Principal Component Analysis Results
	Variables		Component					Communality
Statements			1	2	3	4	5
For me planned buying is irritating. I just love to browse and buy immediately whatever I like	IBT1			0.895				0.820
If I like something then I have to buy it right now	IBT2			0.892				0.833
I always feel to shop online without any prior planning	IBT3			0.887				0.820
When I see flash sales on internet I feel to buy products	PT1				0.864			0.814
Flat rates on online products make me buy those products that are not in my shopping list	PT2				0.866			0.806
Special vouchers provided on online sites create urges to buy items without any prior purchase intention	PT3				0.795			0.751
Social media increases my urge to buy immediately from shopping websites	WC1						0.815	0.730
Recommendations on internet influences me for unplanned buying	WC2						0.774	0.751
I feel to buy product online due to a sudden urge after reading the reviews	WC3						0.795	0.708
Website is safe and secure	WF1		0.763					0.721
Website stick to their words whatever it is communicating to shoppers	WF2		0.943					0.926
I have a satisfactory previous shopping experience on this website	WF3		0.948					0.933
Return and refund policies of the website are transparent	WF4		0.931					0.903
I am in a bad mood	SF1					0.855		0.830
I have plenty of money to spend	SF2					0.856		0.790
I have sufficient time to browse and shop	SF3					0.810		0.740
	Variance %		21.497	15.957	14.673	14.521	13.831
Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization		a. Rotation converged in 6 iterations.

Therefore, we found five distinct dimensions of UBI website related factors (WF1, WF2, WF3 and WF4); situational factors (SF1, SF2 AND SF3); promotional tools (PT1, PT2 and PT3); website communication (WC1, WC2 and WC3); and impulse buying tendency (IBT1, IBT2 and IBT3) (Figure 2). Each dimension consists of three variables except website related factors that consists four variables. This scale for UBI with five distinct dimensions shows a comprehensive picture for capturing the impact of UBI on E-impulse buying.

Structural Equation Modelling (SEM)

Structural equation modelling was applied in this study to establish a comprehensive model for E-impulse buying. SEM is a multivariate technique that combines multiple regressions with confirmatory factor analysis to estimate simultaneously a series of interrelationship between the constructs of the hypothesized model. It basically consists of two models measurement model and structural model discussed later in details for this study.

Measurement Model

Confirmatory factor analysis was used to assess the validity and reliability of the constructs in the model. To get the satisfactory viability of the model CFA was applied by using Amos 24. Literature suggested the positive relationship of the UBI with E-impulse buying hence the measurement model includes five dimensions of UBI evolved through EFA and E-impulse buying as the main constructs of the model (Figure 3).

The reliability, convergent and discriminant validity of the model was also evaluated (Table 4). Internal consistency was also evaluated by calculating the value of Cronbach’s alpha for each construct. All the values of average variance explained (AVE) are more than 0.5 and less than the value of Cronbach’s alpha. Maximum shared variance (MSV) and average shared variance (ASV) values are less than AVE. Thus we can conclude that discriminant validity of the model has been proved. Composite reliability (CR) is also evaluated to measure the reliability of a construct in the measurement model.

Table 4: Reliability Test
	Composite Reliability (CR)	Cronbach’s alpha	Average Variance Explained (AVE)	Maximum Shared Variance (MSV)	Average Shared Variance (ASV)
WF	0.947	0.892	0.817	0.226	0.115
IBT	0.894	0.943	0.737	0.184	0.089
PT	0.860	0.808	0.671	0.232	0.131
WC	0.810	0.852	0.587	0.303	0.201
SF	0.856	0.855	0.666	0.222	0.137
EIB	0.867	0.865	0.766	0.303	0.187

It offers a more retrospective approach of overall reliability and estimates consistency of the construct itself including the stability and equivalence of the construct (Hair, Anderson, Babin & Black, 2010). A value of 0.70 or greater is deemed to be indicative of good scale reliability (Fornell and Larcker, 1981; Nunnally and Bernstein, 1994).

The entire model fit indices of the measurement model reflected satisfactory results. Comparative fit index (CFI) was 0.936; normed fit index was .899; goodness-of-fit index (GFI) was .882; and Tucker-Lewis Index (TLI) was .919. The root mean square error of approximation (RMSEA) was .082 that is less than 10% i.e., within the acceptable range (Table 5). CMIN/DF value was 2.501 i.e., less than 3 percent that shows an acceptable range i.e., less than 3. These values depicted that measurement model must be accepted as per the fit indices.

Table 5 : Model Fit Indices Of The Measurement Model
Model	CMIN/DF	CMIN	RMSEA	CFI	NFI	GFI	TLI
Five dimensions model	2.501	300.062	0.082	0.936	0.899	0.882	0.919

Invariance Test

To assess the effect of groups, invariance test was applied on the measurement model for four factors gender (male and female), income (high and low), marital status (married and unmarried) and internet usage (high and low). Chi-square difference values shown significant results and model fit indices for both constrained and unconstrained models were satisfactory hence it may conclude from the analysis that groups show satisfactory invariance results. Therefore it was indicated that all the groups of respective four factors were different. As these groups of factors were found to be variant therefore further the effects of these factors may be checked for the structural model.

Structural Model

To check the hypothesized research model SEM was applied by using Amos 24. All the factor loadings for the constructs of the model are found to be significant. The factor loading of the relationship between UBI and E-impulse buying (EIB) was 0.72 that was found to be quite good and significant too (Figure 4). The structural model represented in Figure 2, shows satisfactory factor loading and results.

The model fit indices of the structural model were quite satisfactory. CMIN/DF value was 2.578 i.e., less than 3 comes under the acceptable range. RMSEA was found to be less than 10 percent i.e., 0.084. Other fitness indices CFI, GFI, NFI and TLI having values 0.928, 0.871, 0.888 and 0.941 respectively all values found to be satisfactory (Table 6). Therefore we can conclude from the analysis and results that as per the fitness indices values structural model is acceptable.

Table 6: Model Fit Indices Of The Structural Model
Model	CMIN/DF	CMIN	RMSEA	CFI	NFI	GFI	TLI
Five dimensions model	2.578	332.596	0.084	0.928	0.888	0.871	0.914

Moderation Effect

The moderation effect was checked for the different groups and it was found that all the groups for four factors show satisfactory results. Gender, income, marital status and internet usage moderates the relationship of UBI and E-impulse buying in the above model. Male shoppers possess more impact than female shoppers over this model as estimates for both male and female respondents are found to be significant along with the significant z-score (Table 7). Similarly for other factors high income level, married shoppers and high usage of internet moderate the relationship of UBI and E-impulse buying more effectively rather than low income level, unmarried shoppers and low internet usage respectively (Tables 8-10).

Table 7: Moderation Effect Of Gender On The Model
			Female		Male
			Estimate	P	Estimate	P	z-score
EIB	<---	UBI	1.723	0.000	2.419	0.000	2.025**

Table 8: Moderation Effect Of Income On The Model
			Low income		High income
			Estimate	P	Estimate	P	z-score
EIB	<---	UBI	1.700	0.000	2.621	0.000	2.16**

Table 9: Moderation Effect Of Marital Status On The Model
			Married		Unmarried
			Estimate	P	Estimate	P	z-score
EIB	<---	UBI	2.040	0.000	1.667	0.000	2.05**

Table 10: Moderation Effect Of Internet Usage On The Model
			Low internet usage		High internet usage
			Estimate	P	Estimate	P	z-score
EIB	<---	UBI	1.829	0.000	2.703	0.000	2.61**

Discussion and Recommendation

This study provides interesting insights through the comprehensive model for the E-impulse buying. First this model inhibits a five dimension scale for the urges to buy impulsively (UBI) that finally leads to E-impulse buying. E-tailers may focus over the comprehensive set of antecedents of UBI to increase or optimize the chances of E-impulse buying. They utilize this model by emphasizing over the antecedents of UBI to generate high sales revenue by creating more urges for E-impulse buying (Adelaar, Chang, Lancendorfer, Lee & Morimoto, 2003). Secondly e-tailers also need to focus over the moderating impact of demographic factors as well as level of internet usage (Zhang, Prybutok & Strutton, 2007; Kollat and Willet, 1967; Sun and Wu, 2011; Jeffrey and Hodge, 2007). From this study it is found that male shopper exhibits higher level of impulsivity for shopping as compare to female shoppers (Zhang, Prybutok & Strutton, 2007) thus e-tailers require to modify their website practices gender wise. High income group of shoppers also have more affordability to convert their urges to buying impulsively into actual E-impulse buying thus e-tailers may customize their practices and features as per shoppers’ income level (Jeffrey and Hodge, 2007). Shoppers who are married indulge more in impulse buying as compare to unmarried shoppers (Kollat and Willet, 1967). Thus e-tailers may utilize this model to create more urges for married shoppers. As same antecedents of UBI encountered by the married and unmarried shoppers but conversion into actual impulse purchase is higher in case of married shoppers. Sun and Wu in 2011 suggested that high internet users have the high probability of encountering urges to buy impulsively and hence the probability of actual E-impulse buying is more in the case of high internet users rather than low internet users. Thus e-tailers can modify their features and practices as per the inferences of this study to generate higher sales revenue along with higher customer satisfaction. Also this study fills the academic gap and thus academician and practitioners may utilize this study to better fill the knowledge gap of this concept.

Conclusion, Limitation and Further Direction Of Research

This study is able to understand the phenomenon of E-impulse buying with the virtue of comprehensive scale for the urge to buy impulsively. This study contributes in the prior studies by developing a model with comprehensive approach to capture E-impulse buying. Similar to the literature support this model also reveals significant positive relationship between UBI and E-impulse buying. This study may also provide insights regarding the moderating factors gender, income, marital status and internet usage on the relationship of UBI and E-impulse buying. Like male shopper possess high impulsivity than female shopper. Married shoppers are more intend to buy impulsively as compare to unmarried shoppers. High income group also indulge in more impulse buying rather than low income group. Internet users who indulge in activities on internet are more prone to feel urges to buy impulsively due to more chance of encountering urges to buy impulsively.

The limitations of this study exhibits the geographical and demographic sample constraints. As this study only reveals the inferences with respect to Indian online shoppers thus the generalization of the study may be checked for other countries or the sample size may be wide in terms of demography of the online shoppers across different nations. Culture as the affecting factor for the E-impulse buying is not taken in to consideration for this study. The effect of sub-cultures within India were not analysed therefore a cross-cultural study may be done in future within sub-cultures or within different cultures for digging out more inferences related to E-impulse buying. Some other extraneous factors may also be included in this model as moderating or mediating factors. And may be in future due to changing paradigms of online retailing more dimensions or variables may be added to the scale of UBI.

References

1. Adelaar, T., Chang, S., Lancendorfer, K. M., Lee, B. & Morimoto, M. (2003). Effects of media formats on emotions and impulse buying intent. Journal of Information Technology, 18(4), 247-266.

2. Beatty, S.E. & Ferrell, M.E. (1998). Impulse buying: Modelling its precursors. Journal of retailing, 74(2), 169-191.

3. Bressolles, G., Durrieu, F. & Giraud, M. (2007). The impact of electronic service quality's dimensions on customer satisfaction and buying impulse. Journal of Customer Behaviour, 6(1), 37-56.

4. Chang, C.C. & Tseng, A.H. (2014). The post-purchase communication strategies for supporting online impulse buying. Computers in Human Behavior, 39, 393-403.

5. Dawson, S. & Kim, M., (2009). External and internal trigger cues of impulse buying online. Direct Marketing: An International Journal, 3(1), 20-34.

6. Dittmar, H., Beattie, J. & Friese, S. (1995). Gender identity and material symbols: Objects and decision considerations in impulse purchases. Journal of Economic Psychology, 16(3), 491-511.

7. Floh, A. & Madlberger, M. (2013). The role of atmospheric cues in online impulse-buying behavior. Electronic Commerce Research and Applications, 12(6), 425-439.

8. Fornell, C. & Larcker, D.F. (1981). Structural equation models with unobservable variables and measurement error: Algebra and statistics. Journal of marketing research, 382-388.

9. Hair, J.F., Anderson, R.E., Babin, B.J. & Black, W.C. (2010). Multivariate data analysis: A global perspective, 7. Upper Saddle River, NJ: Pearson.

10. Huang, L.T. (2016). Flow and social capital theory in online impulse buying. Journal of Business Research, 69(6), 2277-2283.

11. Jeffrey, S.A. & Hodge, R. (2007). Factors influencing impulse buying during an online purchase. Electronic Commerce Research, 7(3), 367-379.

12. Kollat, D.T. & Willett, R.P. (1967). Customer impulse purchasing behavior. Journal of Marketing Research, 21-31.

13. Liu, Y., Li, H. & Hu, F. (2013). Website attributes in urging online impulse purchase: An empirical investigation on consumer perceptions. Decision Support Systems, 55(3), 829-837.

14. Nunnally, J.C. & Bernstein, I.H. (1994). The assessment of reliability. Psychometric theory, 3(1), 248-292.

15. Ozen, H. & Engizek, N. (2014). Shopping online without thinking: Being emotional or rational? Asia Pacific Journal of Marketing and Logistics, 26(1), 78-93.

16. Parboteeah, D.V., Valacich, J.S. & Wells, J.D. (2009). The influence of website characteristics on a consumer's urge to buy impulsively. Information Systems Research, 20(1), 60-78.

17. Rook, D.W., (1987). The buying impulse. Journal of consumer research, 14(2), 189-199.

18. Stern, H. (1962). The significance of impulse buying today. The Journal of Marketing, 59-62.

19. Sun, T. & Wu, G. (2011). Trait predictors of online impulsive buying tendency: A hierarchical approach. Journal of Marketing Theory and Practice, 19(3), 337-346.

20. Tifferet, S. & Herstein, R. (2012). Gender differences in brand commitment, impulse buying and hedonic consumption. Journal of Product & Brand Management, 21(3), 176-182.

21. Turkyilmaza, C.A., Erdem, S. & Uslu, A., (2015). The effects of personality traits and website quality on online impulse buying. Procedia-Social and Behavioral Sciences, 175, 98-105.

22. Verhagen, T. & van Dolen, W. (2011). The influence of online store beliefs on consumer online impulse buying: A model and empirical application. Information & Management, 48(8), 320-327.

23. Zhang, X., Prybutok, V.R. & Strutton, D. (2007). Modelling influences on impulse purchasing behaviors during online marketing transactions. Journal of Marketing Theory and Practice, 15(1), 79-89.