The researchers Holzwarth, M., Janiszewski, C. & Neuman, M. present an adequate rationale for conducting the study “The influence of online avatars on online consumer shopping behavior”. They talk about the main problem in internet selling; that is, low conversion of shoppers to purchasers. They even quote statistics on the percentage of online visitors who fail to buy anything from a website. After identifying the key problem with this sales channel, they then look at justifications for this behavior. Holzwarth et. al. (2006) suggest the use of avatars as a potential solution to low conversion after studying buyers’ justifications. Therefore, the authors have done a good job of identifying a problem in marketing and providing a solution for it through their research.
It is likely that this study will lead to greater use of avatars among online companies. In turn, this will enrich consumer experiences and thus influence consumers shopping behavior more profoundly. Companies may achieve a higher conversion of online visitors to buyers through such a strategy. In the field of marketing, the study will enable sellers to reap maximum benefit from a relatively new sales channel. Not only does the research suggest that the use of avatars can increase online shopping purchases, but it teaches internet companies how to effectively use that strategy.
The literature review is also quite thorough and comprehensive. Holzwarth et. al. (2006) started with basic descriptions of common concepts like social response theory and computer-mediated technology. In this regard, they wanted to illustrate how avatars affect the psychology of buyers. They then looked at qualities inherent in avatars like reciprocity and how this increases compliance among users. Holzwarth et. al (2006) drew similarities between avatars and sales agents, then used those qualities to make their hypotheses. Issues such as the effect on purchase intention, entertainment value, satisfaction with the seller and attitude towards the retailer could all be affected by the presence of avatars based on the literature review.
This was a very convincing literature review; however, they would have improved upon it if the authors also mentioned the negative elements of personal sales agents and hence avatars. The authors only selected articles that demonstrate positive relationships between persuasion and customer purchases. It would have been insightful if they also chose others that did not support this point of view. Consumers sometimes have a tendency to block out unconcealed persuasive tactics.
Holzwarth et. al. (2006) also did a good job of defining critical concepts such as avatar, and computer-mediated communication. On the other hand, the authors assumed that all readers would know what purchase intention means. Purchase intention could denote actual purchases or willingness to buy. None of these meanings were clarified or related to high conversion rate.
The researchers also mentioned previous methods that support the purpose of their study. For instance, they mention the usefulness of personal communication over mass communication and relate this to their study of avatars. They also show how people visit websites in order to derive entertainment from them, and relate it to their analysis.
Evaluation of the purpose statement and the hypothesis
The purpose statement is not clearly defined, but can be deduced from the hypotheses as well as the research topic. Holzwarth et. al. (2006) wanted to find out whether avatars influence online shopping behavior but did not explicitly say this in the literature review or methods section. However, from the abstract at the beginning of the paper as well as the general discussions in the literature review, it is possible to know what they are trying to achieve through their research. Therefore, the purpose statement of the research is as follows: “To determine whether and how an avatar leads to a)greater satisfaction with a retailer b)a positive attitude towards the product and c)greater purchase intention.” The component of ‘how’ was analyzed by examining the characteristics of avatars.
The purpose statement is clearly derived from the arguments in the literature review as the authors actually include the statement in the literature review. Every time they mention a hypothesis, they also back it up with assertions from the literature concerning similar occurrences.
Variables of interest have been clearly identified in the research. Independent variables include presence of an online avatar sales agent, attractiveness of avatar, and expertise of avatar, while dependent variables include retailer satisfaction, purchase intention, website’s entertainment value, website’s information value and attitude to products.
The authors use alternative directional hypotheses for their work. For instance, the first hypothesis is “Avatar mediated communication has a positive effect on perceived entertainment value of a website” (Holzwarth et. al., 2006, 20). There is a clear rationale for the direction of the hypothesis based on the literature review. Each hypothesis is derived from similar types of work carried out by individuals within their field. For instance, the second hypothesis on entertainment value of a website is derived from previous literature on the search for pleasure from shopping experiences.
In the research, the hypotheses are both testable and falsifiable. Attractiveness and expertise of avatars could be easily affirmed through their credentials, appearance and the way they give information. A general consensus exists in western societies concerning such parameters so they are testable. Additionally, quantitative information is valuable on purchase intention. Participants can also state whether they were entertained, informed and satisfied with a website or product. Therefore, these hypotheses are testable. The hypotheses are also falsifiable because one can eliminate the presence of avatars in general and analyze what happens. Few mediating factors can lead to similar reactions between avatars and online consumers. Therefore, one can assert that the hypotheses are falsifiable. It was tested with clarity and brevity through administering a pseudo online sale in the presence of an avatar and without one. Participants also gave feedback on how they rated avatars. Since the questions were direct, it is plausible to deduce that the hypotheses were clearly tested.
Evaluation of the methods section
The authors did not specify the nature of sampling method used. They simply stated that participants were recruited through different offline and online methods. It is thus difficult to judge whether these were appropriate methods or not.
Holzwarth et. al. (2006) did a god job of defining the relevant characteristics of the population. They talked about the mean age of the shoppers. This is a relevant factor because one’s sample must mirror the mean age of most online shoppers. Additionally, they talked about the education level, sex, and exposure to online shopping among the participants. These are crucial characteristics that may affect outcome.
The sampling techniques were clearly not appropriate (even though they were not outlined) because they resulted in samples that did not mirror the average German, online population. The authors themselves stated that 66% of the participants were college graduates, yet online shoppers do not have such a high education level; this was not representative. In addition, the age of the participants may also not be an accurate representation of the sample population, so their sampling method should have been revised.
Sample biases clearly exist because this research did not represent typical exposure to online shopping. The authors stated that most of the shoppers had a lot more experience with internet purchasing than the average online shopper. Therefore, the results in the study may not necessarily mirror what goes on in typical internet websites.
The sample size is large enough for the study because 400 consumers were consulted. Internet shopping is quite a relevant phenomenon in several developed nations. In order to obtain useful information, one must have a sample size that is will allow one to estimate the variables of interest within a reasonable budget. On the other and, it would have been perceptive to use a scientific method to calculate a sample size, such as the probability statement equation.
Holzwarth et. al. (2006) did not state their research design, so one can only deduce it from other assertions. It is likely that they used the quasi experimental design. Subjects were not randomly selected because their behavior was beyond the investigators’ control. No mention has been made of how the study was conducted but it is likely that it depended on a first come first serve basis. Additionally, the researches also considered the ability and exposure of the individuals to online shopping before including them in the research.
The quasi experimental design is consistent with the hypothesis and purpose of the study. Holzwarth et. al. (2006) wanted to determine whether avatars’ affect online shopping behavior. The variables in the analysis cannot be randomized because subjects possess certain characteristics that make them favorable for the research over others.
The independent variables in the analysis are clearly defined and they include presence of an online avatar sales agent, attractiveness of avatar, and expertise of avatar while dependent variables include retailer satisfaction, purchase intention, website’s entertainment value, website’s information value and attitude to products. At some point, the authors revealed how one dependent variable in a hypothesis can become an independent variable in a hypothesis. For instance perceived attractiveness of an avatar (independent variable) increases entertainment value (dependent variable). Entertainment value in another hypothesis can lead high satisfaction with the retailer or a greater purchase intention. This makes it confusing to replicate the same in another experiment.
The potential threats to internal validity in this analysis include prior exposure to avatars as this may affect one’s judgment of the same. Some respondents may have had negative experiences with avatars or vice versa and this may affect their perception of the same. All the participants were aware that they were taking part in an experiment, so their responses may have been altered. Selection biases may also threaten internal validity because a disproportionate number of internet shoppers were selected and this may affect their behavior.
Threats to external validity include problems in sample selection as participants knew they were in an experiment. They may have altered their behavior to meet investigators’ expectations. Cultural biases may also come in the way as the German sample may only be displaying behavior that is typical of the German population. It may be difficult to generalize these findings to persons outside the country.
The authors outlined their data collection methods as self reports through structured questionnaires. Participants were required to rate their attitude towards the product, purchase intention, attitude towards the retailer. They were also required to rate the effect of the avatars with regard to their appearance as well as their level of information. The structured questionnaires used Likert scales.
Major challenges involved in the use of such a research instrument (that is, self reporting) is the lack of objectivity that an external observation accords. Participants can change their answers to impress the investigator or avoid extra questions. They may give inaccurate responses because they simply do not understand their own behavior. Regardless of these challenges, it is still appropriate because measuring behavioral elements like attitude towards retailers are rather difficult.
The reliability and validity of the instrument were partially reported. Internal validity was mentioned by the concerned individuals who stated that the quality of consultations in the first study may have reduced in the second and thus altered the answer. External validity was not reported, yet it was a factor since special samples were collected. Reliability was reported concerning inter-observer reliability. This was the degree to which measures were consistent among different observers. The authors repeated elements of the experiment using different raters in order to deal with this form of reliability.
Results section
Main descriptive statistics in the research is the mean value of the responses. These were appropriate because the answers were all ratings of research variables. It was imperative to get a measure of central tendency that represented average outcomes. Since the population was large, the mean was an effective measure. The samples involved in these calculations were accurately summarized.
Holzwarth et. al. (2006) relied on linear regression analysis as well as correlation analysis as inferential statistics. T tests and p values were the main parameters of comparisons. It was appropriate to use statistical methods that focus on correlational relationships. The analysis method as clearly defined although the authors would have included more of their detailed work in the appendix.
Discussion section
The authors clearly stated what they found with regard to the first, second, third, forth and fifth hypothesis. However, because their purpose statement was not clearly defined, then one has to deduce the meaning of the same from what they wrote.
The authors have clearly talked about the implications of their findings and even related it to previous findings. For instance, they stated that avatars are effective persuasion instruments and that they increase customer satisfactions./they related to previous work on persuasion.
The authors have also identified the potential limitations of their study. For instance, they have stated that while attractiveness of avatars is more important in low level involvement, it does not negate the usefulness of expertise. Therefore, the mediation of the two factors altered results. Additionally, they realized that sales data would have made the research richer. Alternatively, they could have used real websites with real items for sale. They also realized that an avatar’s effect would reduce as familiarity with transaction increased. Additionally, no reciprocal communication existed.
Some of the possible directions for future research in the paper include analyzing what makes avatars evade typical persuasive messages and understanding the factors that lead to reception of avatars as sales agents.
References
Elston, J. (1994). Populations, samples and study design. Philadelphia: Davis.
Frankfort-Nachmias, C. & Leon-Guerrero, A. (2006). Social statistics for a diverse society. Thousand oaks: Pine Forge Press.
Holzwarth, M., Janiszewski, C. & Neuman, M. (2006). The influence of online avatars on online consumer shopping behavior. Journal of Marketing, 70, 19-36.