1.5 Situational stimuli that influence online impulse buying behavior

1.5.1 Variety of selection and product availability

Liu, Li and Hu [2013] define product availability as “the presence of a diversity of various products in the online store to satisfy a broad shopping interest of potential consumers”. They suggest that poor variety of selection may lead to customers experiencing feeling of boredom in case they do not find anything that might interest them. The scholars conclude that consumers perceive web stores with rich product selection as more visually appealing. According to Moe [2003], online shoppers are more inclined into experiencing feeling of joy while browsing web sites with a great variety of selection. This results in positive evaluation of the unplanned purchases by a customer and evokes further urge to buy on impulse. Chen-Yu and Seock [2002] proved that this stimulus is the one of the most important for impulsive shoppers.

However, according to Kahn [1998], huge variety of selection may confuse and even cause irritation in customers, who are pushed to reducing the number of alternatives and use simplified decision rules in order to make purchases. That is why Kahn distinguishes two types of variety: perceived variety and actual variety, although noticing that depending on the case they might be equal to each other. The scholar suggests that the perceived product availability is a more significant variable than the actual variety, because the more products the store offers the more evaluation they require, and that is why the actual variety has to be narrowed to perceived variety. For example, in an offline store an assistant may help a customer with that, and in online stores it can be done with the use of search mechanisms. Kahn claims that the larger the assortment the harder it may be to find a desired product.

1.5.2 Return policy

Minjeong and Johnson [2009] suggest that a liberal return policy can reduce the uncertainty about purchasing a product as well as it is able to decrease purchase risks, since the buyer is promised that the item he is buying can be returned later. Thus, lenient return policy appears to be encouraging for people to commit impulse purchases. The scholars proved this statement by finding correlation between impulse buying behavior and frequent returns of purchased goods. However, they also point out that participants in their research did not always consider the availability of liberal return policy and still conducted impulse purchases.

Janakiraman, Syrdal and Freling [2016] distinguish five return policy factors: 1) time, 2) money, 3) effort, 4) exchange, 5) scope. Time refers to the deadline when the product can be returned (the more time available to return the good the more liberal the return policy is). Money refers to the presence of any monetary restrictions - the customers either get full refund of the purchased product or only the portion of the money spent (such restriction may be imposed due to shipping costs, for example). Effort stands for difficulties consumers face when they try to return the product such as the need for the original receipt, undamaged package and so on. Exchange means that a store offers product exchange (or anything similar) instead of cash refund (return policy that proposes cash refund is considered to be more liberal). Scope refers to the availability of an item for return, since some products may not be “return-worthy”. The scholars found that money and effort factors of a return policy are the most significant ones from the consumer's perspective.

Overall, customers require returning the purchased goods due to different reasons (for example, if it is of bad quality, wrong color, size or model, etc.), that is why the return policy stimulates unplanned buying: customer is not afraid of losing money. Even in that case, some people feel inconvenient while returning goods, so they keep them even though they do not satisfy their actual needs [Mesiranta, 2009]. Respondents in Mesiranta's study stated that liberal return policy of an online store has influenced their decision to conduct an impulsive purchase.



2. Methodology

This study is mixed methods research, since it includes quantitative and qualitative methods of collection and analysis of data. The methods of collection are semi-structured interview and a survey. Both the semi-structured interview and the survey methods have trial stages. At these stages the methods of collecting the data are tested on small groups of people - that complies with the selection criteria, however it is not included in the sample. Subsequent data analysis is conducted with the use of Stata software and includes the use of such tools as multiple linear regression and binary logistic regression.

The stages of our empirical analysis are following:

1. Creating a trial version of Protocol for the semi-structured interviews in order to point out the most influential factors that affect online consumer behavior.

2. Testing the trial version of Protocol on a sample of young online customers.

3. Identifying inaccuracies and mistakes made in this version of Protocol and improving them.

4. Repeating steps 2 and 3 if needed.

5. Conducting the semi-structured interviews, according to the updated version of the Protocol.

6. Analysis of the semi-structured interviews results with the use of logistic regressions performed by Stata software.

7. Preparing questions for the quantitative stage of the empirical part basing on the results of the analysis (logistic regression). The aim is to gather information regarding the previously pointed out factors on a larger sample with the help of questionnaire.

8. Testing the questionnaire a few times on a trial sample of young online customers.

9. Identifying inaccuracies and mistakes made in this version of questionnaire and improving them.

10. Repeating steps 8 and 9 if needed.

11. Conducting the survey with the updated version of questionnaire and gathering the data.

12. Analyzing the results with the method of binary logistic regression with t-tests by using Stata software.

13. Presenting the resulting patterns.

Since apparel is one of the most popular products to be bought online in Russia, according to Yandex's recent research [“Розничная онлайн-торговля в России,” 2015], it is chosen as an example to study the online impulse buying phenomena.

For the study, we chose people at the age of 16 to 30 living in Saint-Petersburg, which is supported by the study of a Russian scholar [Белый, 2012], who distinguishes at least three sub-groups of young people - from 14 to 18 years, from 18 to 22-25 years and from 25 to 30 years. Since we operate with such independent variables as Employment and Income and the minimum age when a person can legally get a job is 16 in Russia, we took that age as the lower threshold. Besides, this group age tends to averagely conduct online purchases more than any other age. As far as they spend a lot of time in the Internet and mostly have financial freedom, we believe that their online behavior will be significantly affected by the factors we point out. Moreover, their web-activity will also help us, because they are more willingly agree to participate in the interview and survey. The respondents for the both semi-structured interviews were found in social groups dedicated to online shopping and online shops on vk.com. Such approach simplified the process of finding respondents, because the search engine on this web-site allows to filter the group members by age, gender and etc, so it allowed us to diversify our sample. It was also convenient to contact people via private messages and ask them to take the survey/semis-structured interview. In case of semi-structured interviews, we asked our respondents to use the VK's phone calls or any other convenient tool of their choice.

Our sample size is counted according to the main formula. First of all, according to Petrostat data [Петростат, 2017], our general population equals to 1.030.000, confidential interval - 95%, the accuracy of the estimation is in the frameworks of +-5%, thus the minimal sample size is 384. As far as we will use online-questionnaire with a responsive rate of approximately 25%, our requirement sample size is 1536 and the potential sample error equals to 5, according to the formula.

On the first stage we conduct 40 interviews. The chosen number of interviews, which constitutes 40, is justified by obtaining some curious trends and patterns that showed up at this point. In case of wrong sampling, we added filtering questions regarding age of the respondent and his experience in conducting impulse purchases online. The list of questions may be found in the Appendix 1. We have built up our questions aiming to analyze the influence of 11 factors on impulse buying behavior. The coded semi-structured interviews can be found in Appendix 2.

On the second stage, we use online survey that can be found in Appendix 3. For the quantitative part of the research, we used online survey and sent 596 online forms. We gathered replies from 401 Russian young people. The replies were divided into 134 responses from rational online shoppers, who have bought apparel online rationally - not on impulse, and 267 responses from respondents who have conducted impulsive purchases of apparel online. The first question in our survey allowed us to do that. Multiple linear regressions were built for the gathered data. After discussion of the analysis results, the differences between the two sets of data are examined.



3. Results

Summing up the results of the qualitative part of the research (the semi-structured interviews), during which the responses from 40 respondents were gathered, Figure 1 represents the amount of respondents that mentioned the influence of a particular stimulus on their online impulse buying behavior during the interview. Numbers from 1 to 11 stand for the eleven stimuli that were discussed above: 1 is Price, discount and sales, 2 is Bonus and promotion, 3 is Scarcity, 4 is Atmospheric cues and web site quality (store content, ease of use and navigation, visual appeal), 5 is Use of recommendation agent, 6 is Social presence, 7 is Telepresence, 8 is Personalization and customized view, 9 is Use of search mechanism, 10 is Variety of Selection, 11 is Return policy. The numbers on the vertical axis represent the number of respondents that agreed that one or another stimulus matters for them.

Figure 1 Number of respondents per stimulus

Without preliminary analysis it can be stated that such factors as Price, discount and sales, Atmospheric cues (store content, ease of use and navigation), Social presence, Use of search mechanism and Return policy were the most repeatedly named by the respondents during the semi-structures interview and, thus, they are included in the quantitative part of this study.

Out of 40 interviewees 37 told that Price, discount and sales matter to them. One of the reasons why this stimulus is so influential was described by one of the 37 interviewees, whose name is Julia: “I often visit online shops just to check for bargains - price is definitely one of the most influential factors for me. However, sometimes there are so many products with discounts that I at points get bored, but still continue browsing the web stores in case I find anything that fascinates me”. Julia admitted that she would not necessarily buy the clothes she has purchased if not the price. However, Polina, who is one of the three interviewees that replied negatively to the question regarding the importance of this stimulus stated: “The price does not really matter for me. If I like the product, I will buy it anyway”, - the exact same position is shared by other two respondent. Still, as it is seen, this is not a popular opinion.

Regarding Bonus and promotion, 23 people pay their attention to this stimulus, and 17 respondents do not. These are the words Yana, who was positive about this factors: “Some of the purchases I have made were dictated by the fact that I will get free delivery if the total cost of the apparel bought is equal to some specific sum”. Indeed, it turned out that more than the half of the respondents pay attention to such bonuses as free shipping, free delivery, discount coupons and similar that comes along with some product. However, as some of the 17 respondents have noted, this factor is not able to influence their purchase decision, since it is quiet inferior to the other stimuli such as Price, discount and sales.

Opinions regarding Scarcity are split in half - 20 interviewees agree that scarcity stimulates them to conduct a purchase, and 20 other respondents do find this factor to be stimulating. Roman, one of the respondents for whom this stimulus turned out to be significant told us: “My last online purchase of apparel was influenced by a special offer that was limited in time. That happened right before the New Year”. This is a great example of how continuing time scarcity stimulus works - the time pressure stimulates people to buy products in hurry, before the offer ends. However, for the other half of the respondents this factor is not that crucial. For example, Anastasia, who reported negatively to the importance of this factor for her, states: “Personally, when I see that there are only few items of clothes are left in stock, I think to myself that if I buy this item the chances that there are many people wearing the exact same clothes is pretty high. That is why I will not buy the product even if I like it”. In this case, we see that even though the product is popular (and this might serve as a strong social proof), Anastasia believes that this item of apparel will not make her feel or look special, so she does not fall for the quantity scarcity stimulus. We suggest that Scarcity deserves more attention and should be studied form the three different perspectives that were discussed in the literature review: the continuing time scarcity, quantity scarcity and frequency scarcity.

38 respondents stated that Atmospheric cues is a significant stimulus for them. Marina, who considers this factor to be the most important for her, said during the semi-structured interview: “I do not shop on web sites that look cheap, because I am not sure if I can trust it - it can be a scam. If visually it looks nice and browsing its sections does not require much effort, the chances that I will not leave it soon are very high”. 2 interviewees out of 40 took a completely different position regarding trust and told that this stimuli is not crucial for them, because they always do research about the web sites they visit before conducting purchases there.

Use of recommendation agent is an important stimulus for 19 respondents. 21 respondents out of 40 has never bought any apparel that was proposed by the web store's recommendation agent. They either have paid attention to the recommendations and browsed them without buying anything or just have not looked at them at all. However, Anastasia has made a remark during the semi-structured interview: “I have a favorite online shop that introduces new lookbooks each season, and I buy some items presented there every time. It is very convenient, because such recommendations allow you to take a look on how the clothes look in combination with other pieces of apparel”. There were also some other respondents who noted the convenience and usefulness of lookbooks for shopping online.

Social presence was found to be important by 30 interviewees. Most of these people mentioned that they have conducted impulse purchases of apparel online, because there were positive reviews of the products they liked. Another opinion belongs to Olga, who expressed her opinion about artistic description of a product: “If the item of apparel is expensive and represents some well-known brand, then the artistic description is able to influence my purchase decision, because it would boost my self-esteem, but in general I do not pay attention to descriptions or reviews, so I cannot say that this factor is important for me”. Polina also stated: “Web sites that sell clothes and have online assistants are just bad”. Such people as Polina believe that the presence of online consultants shows that the web-site is not able to increase shoppers' trust and loyalty in any other way, so it uses assistants, “who would type with mistakes and ask if you need any help every time you enter the online store”. Still, only 10 people expressed their negative view of this stimulus, so it is kept for the research in the quantitative part of the study.

24 respondents mentioned that Telepresence contributes to their decision to buy apparel online impulsively. For example, Alexey said during the semi-structured interview: “Since I can't touch the fabric or try the item on, I examine it carefully with the tools provided by the web shop”. 16 interviewees have stated that the use of 360-degree view or 3D model of a good can be helpful, but it is not necessary.

Use of search mechanism was found to be as influential as Price, discount and sales - 37 respondents agreed that it matters to them and 3 disagreed with that view. According to the semi-structures interviews, these 3 people tend to browse web stores in search of bargains or they just do that occasionally as a hobby and conduct precise searches only if they are planning to buy something particular - such behavior is not impulsive in its nature. For 37 interviewees the possibility of structuring and limiting the number of displayed goods is important. Most of them stressed the importance of sorting by the products by size, color and brand, because it helps to save time by not looking at products that will not fit or suit them.

Only 18 people out of 40 find Variety of Selection and Product Availability to be an important stimulus, and that makes this stimuli to be the least “popular” and influential, according to the respondents. Kristina in her interview has expressed an opinion that is shared by the most of the respondents and it perfectly explains why this stimulus has the least number of positive answers: “The more products there are, the harder it is to make a decision”. This is a fascinating remark, because it can be suggested that the more time consumers spend taking the decision, the less the possibility of the purchase. However, Anastasia has a different view on variety of selection: “The more products are available for purchase, the higher the chances that there are not many people wearing this particular item of clothes that I liked. For example, people, who shop in the same store that belongs to one popular brand, dress the same way, and I try to avoid that”.

Regarding return policy and product availability, it is a significant stimulus for 32 respondents out of 40. Sonya explained why: “If I see that the online store allows customers to return the purchased goods without any problems, then I will buy as many products as I want and later return those items that do not fit or suit me”. The same opinion was expressed by many of those who agreed about the importance of this stimulus. On contrary, Galina, who is one of the eight respondents who did not find this stimulus important noted: “If I liked an item of clothes and decided to purchase it immediately and impulsively, then I will not pay attention to the return policy. However, if I am making a planned purchase, then I will consider it”.

Table 2 represents Pearson correlation matrix. In this part of the empirical study, we decided not to use the independent variable Employment, since a number of the interviewees are self-employed, some do voluntary jobs and some receive social assistance. This independent variable is used in the quantitative part. Instead we used independent variable Earnings, which means that a person has any source of income, which is not limited by salary gained from the job.

It can be noted that there are a negative correlations between Gender and such factors as Scarcity and Telepresence and between Earnings and Social presence. The correlation coefficient of Gender and Scarcity equals to -0.408248, the correlation coefficient of Gender and Telepresence equals to -0.479167, the correlation coefficient of Earnings and Social presence equals to -0.302614. These correlations are worth paying attention to, because they show a good result for a study that aims to explain human behavior. It was decided to build simple logistic regressions for these variables to measure the variation, because the respondents' answers were “no” or “yes”, which are “0” and “1”.

Table 2

Correlation matrix

Stimuli	Gender	Earnings	Income
Pricе, discount and sales	0.038749	-0.208947	-0.048691
Bonus and promotion	-0.020646	-0.100727	0.025943
Scarcity	-0.408248	0.000000	-0.102598
Atmospheric cues	-0.046829	-0.168345	-0.117688
Recommendation agent	0.143066	0.068224	-0.077045
Social presence	0.235702	-0.302614	0.177705
Telepresence	-0.479167	-0.171184	0.209427
Personalization and customized view	-0.123876	-0.100727	-0.077829
Search mechanism	0.232495	-0.208947	0.146072
Variety of selection	-0.225668	-0.073750	-0.051557
Return policy	-0.229640	-0.104828	0.256495

Table 3 displays logistic regression of dependent variable Scarcity and all three independent variables. Pseudo R-squared equals to 0.1444, in case of Gender p-value equals to 0.010 and the coefficient is negative.



Table 3

Logistic regression of dependent variable Scarcity

Logistic regression				Number of obs	=	40
				LR chi2(3)	=	8.01
				Prob > chi2	=	0.0459
Log likelihood =	-23.7225			Pseudo R2	=	0.1444
Scarcity	Coef.	Std. Err.	z	P>z	[95% Conf.	Interval]
Gender	-1.93669	.755972	-2.56	0.010	-3.41837	-.455016
Earnings	.201673	.744735	0.27	0.787	-1.25798	1.66133
Income	-.762834	.758452	-1.01	0.315	-2.24937	.723705
_cons	2.34315	1.84795	1.27	0.205	-1.27876	5.96506

Table 4 shows logistic regression of dependent variable Telepresence and all three independent variables. Pseudo R-squared equals to 0.2217, in case of Gender p-value equals to 0.006 and the coefficient is negative.

Table 4

Logistic regression of dependent variable Telepresence

Logistic regression				Number of obs	=	40
				LR chi2(3)	=	11.94
				Prob > chi2	=	0.0076
Log likelihood =	-20.9524			Pseudo R2	=	0.2217
Telepresence	Coef.	Std. Err.	z	P>z	[95% Conf.	Interval]
Gender	-2.12597	.768299	-2.77	0.006	-3.63181	-.620133
Earnings	-.858696	.819165	-1.05	0.295	-2.46423	.746837
Income	.948234	.800387	1.18	0.236	-.620495	2.51696
_cons	-.194034	1.87384	-0.10	0.918	-3.86669	3.47863

Table 5 demonstrates logistic regression of dependent variable Social presence and all three independent variables. Pseudo R-squared equals to 0.2141, in case of Earnings p-value equals to 0.049 and the coefficient is negative. Besides, it also worth mentioning that p-value in case of Gender equals to 0.073 and the coefficient is positive.

Table 5

Logistic regression of dependent variable Social presence

Logistic regression				Number of obs	=	40
				LR chi2(3)	=	9.63
				Prob > chi2	=	0.0220
Log likelihood =	-17.6769			Pseudo R2	=	0.2141
Social presence	Coef.	Std. Err.	z	P>z	[95% Conf.	Interval]
Gender	1.72287	.961559	1.79	0.073	-.161754	3.60749
Earnings	-2.32584	1.18246	-1.97	0.049	-4.64341	-.008265
Income	1.30262	.924492	1.41	0.159	-.50935	3.11459
_cons	-.59825	2.21868	-0.27	0.787	-4.94679	3.75029

The analysis results demonstrate that Scarcity and Telepresence have an influence on women, even though both of them are not among the most repeatedly named stimuli that are important for online impulse shoppers. This means that people are not aware of the stimulating effect of these two factors that actually contribute to online impulse buying behavior. Moreover, the analysis shows that Social presence affects male respondents and people who do not have own source of income.

Basing on the responses, Price, discount and sales, Atmospheric cues and web site quality (store content, ease of use and navigation, visual appeal), Social presence, Use of search mechanism and Return policy were kept for the quantitative stage of the research. Since there were only 40 observations analyzed the results might not be representative and that is why it was decided to exclude Scarcity and Telepresence from the next stage of the research, even though these two stimuli demonstrated a connection to respondent's gender. Instead, the quantitative stage will focus on the top five stimuli. However, the findings of the qualitative part will be discussed in last chapters of this study.

Some certain outcomes were expected basing on the qualitative stage of the research, since Social presence demonstrated influence on male respondents and people who do not have own source of income. However, there were no other significant connections found between the independent variables and the other mostly mentioned dependent variables in the semi-structured interview analysis. Still, semi-structured interviews gave us understanding what stimuli are the most paid attention to and why. More demonstrative results were obtained in the quantitative part due to the larger number of respondents and the absence of the interviewer who asks questions.

The obtained results assist in understanding the significance of the listed factors, which affect the buying behavior of our targeted auditory. Furthermore, they elucidate the importance of the elements pertaining to the merchandising practices followed by the departmental stores and the manipulative tendency of those elements in determining the frequency of purchases made by a consumer. For the quantitative stage it can be concluded that:

1. Scarcity and Telepresence influence online impulse buying behavior of women;

2. Social presence influences online impulse buying behavior of men and people who do not have own source of income.

Next step is the quantitative part of this study, which is represented by the online survey. The used online survey can be found in Appendix 3. At this stage we gathered replies from 401 Russian young people, which were divided into 134 responses from rational online shoppers and 267 responses from shoppers with impulse buying experience. The respondents were split in these two groups according to how they replied to the first question of the survey. Multiple linear regressions were built for the gathered data. The discussion of the regression analysis of both groups, is followed by the comparison of the gained results. Since we create explanatory models, the fact that the R Squares are not high is not critical. Such small R-squared values can be explained by a large number of other factors that affect human behavior at the exact moment of browsing an online store. These factors include intention to buy, mood, credit card usage, personal traits, time availability and many other. Our study focuses on particular external stimuli that can be provided and managed by a web-store and any significant connections found will be discussed.

The first data set analyzed is of impulsive buyers. The analysis is represented by Table 6.

Table 6

Multiple regression for impulsive buyers data set

	(1)	(2)	(3)	(4)	(5)
VARIABLES	Price, discount and sales	Atmospheric cues	Social presence	Use of search mechanism	Return policy
Gender	0.250*	-0.481***	0.817***	0.419***	0.067
	(0.137)	(0.125)	(0.148)	(0.129)	(0.147)
Employment	-0.350**	0.025	-0.179	0.177	-0.218
	(0.164)	(0.149)	(0.177)	(0.154)	(0.176)
Education	-0.211	0.128	-0.040	0.084	-0.090
	(0.132)	(0.120)	(0.142)	(0.124)	(0.142)
Income	-0.509***	0.203**	-0.356***	-0.075	-0.461***
	(0.104)	(0.094)	(0.112)	(0.097)	(0.111)
Constant	5.505***	3.323***	3.848***	3.430***	5.286***
	(0.277)	(0.252)	(0.298)	(0.259)	(0.297)
Observations	267	267	267	267	267
R-squared	0.199	0.086	0.152	0.051	0.117
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

Several conclusion can be made basing on the results of analysis:

1. Price, discount and sales has a weak influence on male respondents, a moderate influence on unemployed respondents and a strong influence on people with lower income.

2. Atmospheric cues have a strong influence on female respondents and a moderate influence on people with higher income.

3. Social presence has a strong influence on male respondents and respondents with lower income.

4. Use of search mechanism has a strong influence on male respondents.

5. Return policy has a strong influence on people with lower income.

The finding of previous stage that Social presence influences online impulse buying behavior of men is now proved.

The second data set analyzed is of rational buyers. The analysis is represented by Table 7.

Table 7

Multiple regression for rational buyers data set

	(1)	(2)	(3)	(4)	(5)
VARIABLES	Price, discount and sales	Atmospheric cues	Social presence	Use of search mechanism	Return policy
Gender	0.004	-0.327*	-0.146	0.074	-0.345*
	(0.205)	(0.171)	(0.230)	(0.207)	(0.192)
Employment	-0.439**	0.102	-0.379	0.178	-0.267
	(0.220)	(0.184)	(0.247)	(0.222)	(0.206)
Education	0.066	0.029	0.085	0.065	0.239*
	(0.141)	(0.118)	(0.158)	(0.142)	(0.132)
Income	-0.419***	0.041	-0.010	0.111	-0.188
	(0.137)	(0.114)	(0.153)	(0.138)	(0.128)
Constant	4.721***	3.881***	3.492***	3.283***	4.315***
	(0.346)	(0.288)	(0.388)	(0.348)	(0.324)
Observations	134	134	134	134	134
R-squared	0.099	0.029	0.025	0.020	0.063
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

Here we conclude:

1. Price, discount and sales has a moderate influence on unemployed respondents and a strong influence on people with low income.

2. Atmospheric cues have a weak influence on female respondents.

3. Return policy has a weak influence on female respondents and a weak influence on people with higher education.

Table 8 compares the most influential factors for impulsive and rational buyers. “!” stands for a weak influence, “!!” is for a moderate influence and “!!!” is for strong influence.

As it can be seen, Price discount and sales have a moderate influence both on unemployed impulsive and rational shoppers, and Atmospheric cues affect both impulsive and rational buyers that are female. These are the only similarities.

It also can be noted that impulsive buyers are more inclined towards being influenced by one of the five stimuli in comparison to rational shoppers.

Table 8

Comparison of significant factors for impulsive and rational online shoppers

	Price, discount and sales	Atmospheric cues	Social presence	Use of search mechanism	Return policy
Impulsive buyers	female		!!!
	male	!		!!!	!!!
	unemployed	!!
	higher education
	lower income	!!!		!!!		!!!
	higher income		!!
Rational buyers	female		!			!
	male
	unemployed	!!
	higher education					!
	lower income
	higher income



4. Discussions and conclusions

With this research we wanted to shed light on the phenomenon of impulse buying behavior by examining the relationships between a number of independent variables (gender, education, employment, income) and various stimuli that were to be important by a numerous studies. Based on the literature review, we hypothesized that the tendency of individuals to purchase impulsively is associated with their exposure to certain stimuli. The results were obtained through a semi-structured interview and online survey. Results emerged from the analysis have proved to be largely consistent with our hypotheses. Therefore, in the light of the relationships observed, our study seems to provide support for the theories, considered in literature review, according to which the tendency the impulse purchase has a basis consisting of the individual personality. It seems that Internet as a shopping environment contains elements that both encourage and discourage impulse buying.

In this chapter, we provide a summary and discuss the findings and the results of our research, what theoretical contribution do they bear, what is their managerial implication, what limitations does this study have and what might be the focus for the further researches in this field.

4.1 Limitations and future research

Every research has limitations that are able to influence generalization of the results and findings. The most crucial and central limitations of this study are discussed here.

To begin with, the findings are limited by the chosen age group of 16-30 years old people. It was decided to focus on young people, because it is expected that it is the younger audience that has online shopping experience as well as the experience of conducting impulse purchases, since young consumers are more open to taking risks in comparison to older ones. [Chandon et al., 2000]

We contend that the chosen respondents were appropriate, since they were able to understand and classify the stimuli, which have affected their online buying behavior. We suggest that future studies may broaden the samples in order to get results that are more valuable. It can be done by gathering data from several cities, regions or even countries and, thus, get wider picture of online impulse buying behavior of young people.

Secondly, since this research was conducted in Saint-Petersburg, Russia, there is a potential bias that may occur due to the possible differences in customer profiles in comparison to the e-commerce markets of other cities, regions and countries. In other parts of the world cultural traditions and values are different and they have to be considered. Thus, future studies may use the similar methodology that is presented in our study to find own unique patterns in the online shopping behavior of local people.

Thirdly, in this study we focused on eleven stimuli, ten of which were discussed in the literature analysis of Chan, Cheunga and Lee [2017] plus Return policy. We recognize that this number of factors may not represent all possible stimuli that are able to influence online shopping behavior. We suggest that future studies may include larger number of factors, especially into the group of the situational stimuli, which in our study only consisted of Return policy and Variety of selection. Hence, further researches may provide more insights if a larger variety of stimuli will be taken into consideration.

Fourthly, in this study we focused on following independent variables: gender, education, employment and income. Future studies may also consider more variables, for example, martial status, age, online shopping experience, profession and other to gain more specific and complete understanding of the online impulse buying phenomena and how it is connected to the socio demographic background of people and other determinants.

Fifthly, in qualitative part of the work we used a yes-no model to gather data regarding the importance of each stimulus. This was done to simplify and shorten the questionnaire, and to increase the involvement of the participants. However, there is a possibility that the respondents could perceive the significance of each stimulus in a different way, that is why the quantitative part included questions that asked to evaluate each stimuli on a rate from one to five. Besides, the answers of the semi-structured interview respondents could be distorted by the interviewer's presence.

Lastly, in our questionnaires (both in qualitative and quantitative parts of the research) we asked the respondents about their online impulse purchases of apparel products. That is why it has to be mentioned that the findings might not be applicable to other goods such as food, household appliances, pharmaceutical products, cosmetics and perfumery, books and so on. We justify the choice of apparel products as the main focus of our research by the fact that it is one of the most popular category of goods to be purchased in online environment in Russia, according to Yandex's recent research. [“Розничная онлайн-торговля в России”, 2015] We suggest that further studies should pay attention to other categories of goods as well to check if the patterns match for each of them or if there are significant differences that depend on the type of the product.

4.2 Theoretical contribution

Theoretically, this study contributes in a number of ways. This research provides the scientific community of consumer researchers a deeper understanding of the impulse buying behavior of consumers in the online context. Specifically, the contribution is represented in the following ways.

First of all, we have not focused only on one category of factors, but paid attention to all three types of stimuli: marketing stimuli, web-site stimuli, situational stimuli, with a number of factors in each category. Previous studies focused only on a single or several stimuli belonging to either one or two separate categories, so it was decided to broaden the scope of stimuli that influence online impulse buying behavior and see how they affect young Russian consumers.

Secondly, in this study, we do not only present data that represents the patterns in the online buying behavior of impulsive customers, but also we used our questionnaire from the quantitative part of the research to gather data from rational shoppers. That allowed us to compare how different or similar are the findings in these both cases.

Thirdly, in our qualitative part we examined actual, lived experiences of our respondents who had conducted impulse purchases of apparel in online environment. This allowed us to select only those factors that were not just repeatedly named by our interviewees, but those that were proved to be important for them. The clarification of such importance was achieved with the help of the openness provided by the semi-structured interview method, which was chosen for the quantitative part of this study. Such format allowed us to ask additional clarifying questions during the interviews to gain better understanding of how respondents value one particular stimuli in comparison to other. As a result, we managed to point out five stimuli that are paid the most attention from young Russian online customers, who conduct impulsive purchases. Besides, this study also contributes by focusing only on those online shoppers, who in their view have conducted impulsive purchases on the Internet.

4.3 Managerial implications

In addition to the theoretical contribution, this research provides important practical implications.

Consequently, companies started not just to seek for ways to expand physically, but also started to recognize the importance of the B2C Internet commerce and began to expand into the virtual space. Nowadays web browsing is considered to be a significant part of the online shopping experience, which has become a mundane activity for many people, especially the younger ones. This fact made many e-retailers to focus on conversion rates - the percentage of the web store visitors who make actual purchases. However, we suggest that online stores should focus on understating their online customers instead of the sales figures and conversion rates. Behind the figures, there is consumer behavior that generates these indicators and profits as well, that is why it is important to understand what drives customers' decision to make a purchase.

The study focuses on eleven stimuli that were proved by other researches to impact online shoppers' impulse buying decisions. These stimuli are: Price, discount and sales, Bonus and promotion, Scarcity, Personalization and customized view, Social presence, Telepresence, Use of recommendation agent, Use of search mechanism, Atmospheric cues (store content, ease of use and navigation, visual appeal), Variety of selection and product availability, Return policy. The findings of this research provide a number of managerial implications for web retailers regarding the ways to encourage consumers' impulse buying behavior in the online environment.

In the qualitative part of this study we found that such factors as Price, discount and sales, Atmospheric cues (store content, ease of use and navigation), Social presence, Use of search mechanism and Return policy are the most pointed out to be significant by online shoppers who have conducted impulse purchases of apparel products online. Basing on this data, we suggest several implications for each factor.

Price, discount and sale: Web stores should provide a section with bargain products that would be easy to detect and access, the assortment also should be updated frequently.

Atmospheric cues (store content, ease of use and navigation): Design that would please the eye of a customer is crucial here (for example, it may include proper use of fonts, neutral colors on the background or photos of people wearing the clothes sold). Minimalistic layout is able to contribute not only to the visual appeal of the web store, but also to the ease of use and the simplicity of navigation. Besides, it also matters what information the e-retailer includes - it is very important to provide consumers with the product illustrations, terms of purchase, number of payment and shipping/delivery options. Technologies that may improve the usability of the web site and make it easier to use are tree testing, heuristic evaluation, card sorting and other.

Social presence: In this case, extensive information about products in the assortment is crucial. The more detailed their description is, the more it feels like it was written by a person with feelings and emotions, the higher is the perceived social presence, which generates trust towards the web store. Besides, it is also advisable to implicate feedback section on product's page, so users could leave their commentaries and opinions regarding the good. If a customer sees that he can trust this product, it may trigger his impulsive buying behavior, if not, then he will probably continue his searches on this web site.

Use of search mechanism: We suggest that web stores, depending on the assortment of the products, should provide consumers with relevant search filters. They may include search by type, color, size, brand, price and so on.

Return policy: It is common that every e-retailer has own policy regarding the returning the purchased goods. The main point here is to give consumers the right to return unwanted goods and to make returning as smooth and convenient as possible. To be precise, for customers it is especially important when they save money and effort when they want to return what they have bought. A web store that liberates its return policy that way has an advantage over those that do not, since in that case online shoppers will be more prone to conduct impulsive buying decisions.

Besides, in the qualitative part of the research it was also found that Scarcity and Telepresence affect women's impulsive buying behavior in the web stores. Regarding Scarcity, it can be suggested for web stores that are oriented on female audience to make scarce promotions and offers more visible. For example, if there are only one or two items of some product left in the stock, it should be displayed on the product's page or in the list of the search results that demonstrates the products. Besides, such web stores also can use strategy of frequency scarcity, which means that the web store starts some event for a limited period of time at the same date, for example, every year. Such approach was discussed earlier in the literature review. Regarding Telepresence, it is important for female-oriented web stores to provide customers with possibilities to look at the 3D model of products, use 360-degree view and develop online fitting rooms that would allow customers to upload their photos in order to try on the desired products.

Regarding the quantitative part of this research, there are several conclusions that can be made regarding the behavior of both impulsive and rational shoppers in the online environment.

Focusing of impulsive buyers, it was found that:

1. Price, discount and sales has a weak influence on male respondents, a moderate influence on unemployed respondents and a strong influence on people with lower income.

Young Russian men are found to be more price-sensitive than women - for web stores that sell goods for men it means that discounts should be displayed in a such way, so they would be easily noticed. Besides, people without job and low income pay much attention to that factor. It can be suggested that online stores that are oriented on low price segment should make their promotional offers more obvious and fitting to the needs of unemployed people and those who have low income.

2. Atmospheric cues have a strong influence on female respondents and a moderate influence on people with higher income.

We suggest that web sites that sell apparel for women and people with high income should make sure that they provide relevant content, make the web site easy to use and navigate and take care of the design of the store. This will make them more inclined towards making impulsive purchases.

3. Social presence has a strong influence on male respondents and respondents with lower income.

It was found that young Russian men and people with low income are affected by the ability to read other shoppers' feedback on the bought products as well as the ability to communicate with online store assistant, who can provide you more detailed information regarding payment methods and delivery options. There are ways to implement Social presence into a web store that should be taken into account by online shops that are focused on low price segment and male audience.

4. Use of search mechanism has a strong influence on male respondents.

It means that web stores that are oriented on selling apparel to young men should make provide users with extensive options for choosing desired search elements such as category, color, brand and so on and then put the results of the search in a convenient and structured order.

5. Return policy has a strong influence on people with lower income.

Web stores that want to sell more goods to people with low income should consider adopting liberal return policy that would reduce the uncertainty about purchasing a product as well as to decrease purchase risks. It will make young Russian people with low income to make more impulsive purchases.

Focusing on rational buyers, it was found that:

1. Price, discount and sales has a moderate influence on unemployed respondents and a strong influence on people with low income.

The recommendation is, basically, the same as for the impulsive shoppers - to make the discount and sale offers more visible and useful for unemployed people and those who have low income.

2. Atmospheric cues have a weak influence on female respondents.

To make rational young Russian woman to buy something online the web store should focus on store content, ease of use and navigation, v...