Abstract


This paper provides a critical review of the research, Color and store choice in electronic commerce: The explanatory role of trust, conducted by Lee, & Rao in 2010. The paper discusses the experimental design, hypothesis, variables, and limitations of the mentioned research. The paper then suggests an alternative approach that the authors could have used to answer their research questions.



1. Introduction


Researchers from the field of marketing, computer science, human-computer interaction, and advertising (Bellizzi et al., 1983; Kim, & Moon, 1998; Lee, & Rao, 2010; Middlestadt, 1990) have shown correlations between ambient color and purchase related variables. Lee & Rao (2010), with the goal to developed theoretical arguments to understand the effects of color in the electronic commerce, conducted a study to provide a relationship between color, trust, and electronic store choice. This paper discusses the authors’ approach and suggests an alternative approach that they could have used to answer their research questions.



2. Lee, & Rao (2010): Discussion


Based on the studies in marketing that show that cooler colors are more favorable for shopping-related behaviors (e.g., Bellizzi, et al., 1983; Bellizzi, & Hite, 1992), Lee & Rao formed their hypothesis from a theoretical perspective to answer the question ‘why will color affect store choice?’. They argued that color will influence store choice because the color will influence trust, and, differences in trust between stores will influence store choice. Figure 1 represents the theoretical model proposed in this study


2.1. Hypothesis


Based on the evidence provided in the literature, following hypothesis were formed for this study
Hypothesis 1: The web store associated with the color blue will be preferred to the web store associated with the color green.
Hypothesis 2: The store associated with the color blue will be trusted more than the store associated with the color green.
Hypothesis 3: The choice of the store will be correlated to the differences in trust between the two stores.
There’s a huge body of research that studies the influences of colors from the opposite end of the spectrum (e.g. Kim, & Moon, 1998; Bellizzi, & Hite, 1992). that people trust cool colors more than warm colors. On the contrary, it is interesting to note that, this study focuses on two cool colors, blue and green, which are very close to each other on the spectrum. Also, there are many studies that provide evidence of both, the correlation between color and purchase-related variables, and the correlation between trust and purchase-related variables. However, this study argues that web store colors produce a differential trust response, and the difference in trust between the two stores affects store choice.


2.2. Subjects


The experiment was conducted with two hundred and seventy-seven subjects who were enrolled in various sections of a junior level Introduction to Principles of Information Systems for Managers class. Below is the description of the sample as provided by the authors in the paper.
• Male to female ratio: 5:6
• Age range: 19 to 55
• Ethnicity: Hispanic (52%) and White Non-Hispanic (34%)
• Internet usage frequency: 84% used internet at least once a day
• Internet Shopping experience: 78% shopped less than once per month
As shown above, the paper doesn’t provide complete information of the sample. The ethnicity of 14% of the sample population is not provided. Likewise, internet usage frequency of 16% of the subjects; and internet shopping experience of 22% of the subjects are missing.


2.3. Experimental website


The websites used in the experiment were simulated web stores to buy textbooks. The two stores are identical in all respects other than the name and color differences. One website was blue in color and was named BlueStore.com whereas the other website was green in color and named GreenStore.com. In order to maintain the online textbook store consistency, 5 principal objects were included in the experimental website. The objects were the Store Name, the SparkNotes advertisement, the Free Shipping notice, the phrase New and Used, and the picture of the building with the fictitious address and phone number of the online store.

While the 5 principal objects included in the experimental website aptly represents online textbook store objects, they are not enough to completely replicate the online textbook store experience. Login/Account option in the header and search bar are two other principal objects common to online textbook stores that are missing in the experimental website. Lack of these objects can influence the trust variables, thus affecting the results.


2.4. Procedure


The subjects were assigned a unique identification number and provided with a hard copy of instructions. Each subject had to log in and fill a background questionnaire with demographic information, prior experience with online shopping, and prior experience with book buying. The subjects were also required to rank four colors (blue, green, red, and yellow) in order of preference (1=most preferred, 4=least preferred). Upon completion of the questionnaire, subjects with even identification numbers were presented with the blueStore.com followed by GreenStore.com. The websites were presented in reverse order to subjects with odd identification numbers. In each website, subjects had to perform certain tasks and then complete a questionnaire on perceived trust and risks in each store. In the final step, the subjects had to choose a store they would buy the book from.


2.5. Variables


Following variables were identified in this study:
Independent variables
1. Color: Approximately 25-30% of the website page was covered with either blue or green color.
2. Store choice: A value of 1 and 2 was associated with this variable. Store choice =1 for BlueStore.com or Store choice = 2 for GreenStore.com.
3. Items for perceived trust and perceived risk: A seven-point scale was used to measure perceived trust and risk variables (1=high and 7=low). The difference in perceived trust is the numerical difference calculated from the trust in blue store minus the trust in green store.
4. Order of presentation of the two stores: The value of this variable 0 if subjects viewed the BlueStore.com first and 1 if they viewed GreenStore.com first.
Dependent variables
1. Preferred color: The value of this variable was 0 if subject preferred blue to green, and, 1 if subject preferred green to blue.
The fact that the two web stores had different names adds an extra variable which wasn’t considered in this study. Many marketing researchers have provided evidence that store name information can influence buyers’ purchase intentions (e.g. Dodds, 1991; Grewal et al., 1998). This makes it imperative to identify and study store name as an independent variable.


2.6. Results and conclusion


Hypothesis 1: One hundred and seventy-eight (178) subjects chose the blue store over the green store, while ninety-nine (99) subjects chose the green store over the blue. Thus, supporting the hypothesis that significantly more subjects preferred the blue store to the green store
Hypothesis 2: The mean difference in trust was - 0.046 (on a scale of 1 to 7), The difference in perceived trust is the numerical difference calculated from the trust in blue store minus the trust in green store. Thus, a negative difference means that the blue store is trusted more than the green store providing support for hypothesis 2.
Hypothesis 3: The results provide evidence that the difference in trust between the two stores is highly significant in explaining store choice. The results also indicated that other covariates such as the difference in perceived risk between the two stores, the order of presentation, and preferred color of the subject do not explain store choice. The results thus support hypothesis 3.
It is interesting to note that even though the mean difference in trust was -0.046 (marginally significant), 178 subjects (65%) chose the blue store over the green store. The study, thus, shows that small differences in trust can lead to big differences in store choice



3. Alternate Approach


In order to improve the overall quality of the research, an alternate approach is discussed below that the authors can use to answer their research questions.


3.1. Test Design


In the experiment, the subjects were asked to rate the trust variables for both the websites as part of trust/risk questionnaire immediately after visiting each website. One cannot be 100% sure about the accuracy of self-reported data. Such data is exposed to certain limitations due to the way that subjects generally behave. Self-reported data can be exaggerated or influenced by various biases, like response bias. Self-report studies are inherently biased by the person's feelings at the time they filled out the questionnaire. If a person feels bad at the time they fill out the questionnaire, for example, their answers will be more negative. If the person feels good at the time, then the answers will be more positive. Also, People interpret and use scales differently, what I might rate as ‘8’ on a 10-point scale, someone with the same opinion might only rate as a ‘6’ because they interpret the meanings of the scale points differently.

Thus, to capture accurate data (in this case to check if they actually trust the website or not), the test needs to be designed in a way such that it eliminates self-reporting. Keeping this in mind, an alternate test-design is proposed.

Provide them with pre-paid coupon codes to buy textbooks and instructions to first, investigate specific books on both the websites (task 1) and second, to make the purchase through one of the 2 websites they feel comfortable with (task 2). The first task will help the subjects get accustomed to both the websites and the second task will provide the evidence of their preference. It is imperative that the sequence of presentation of both the websites in the first task, be reversed for alternate subjects (similar to the test conducted by Lee and Rao).

In order to capture accurate preference of the subjects, it is imperative to consider two things.
1. The timing of the test is very critical in this approach. It is important that the subjects are tested just before the commencement of their semester when they actually intend to buy textbooks.
2. Textbooks that they buy during the test should be part of their incentive.
This test can also be seen as an alternate test to assess the consistency of the results, thus measuring parallel reliability.


3.2. Control Extraneous variable


The results were interpreted “…small differences in trust can lead to big differences in store choice”. However, one can argue that the big difference in store choice can be caused by multiple small differences. In other words, there can be few independent variables involved in the experiment that weren’t accounted for. Once such variable is store name. As discussed in the ‘Variable’ section above, store name information can influence buyers’ purchase intentions. Hence, in order to make sure that only independent variables are causing the changes in the dependent variables, store name must be controlled. Doing so would also increase the internal validity of the research.

The variable store name can be controlled in two possible ways. First, include store name as an independent variable in the experiment and study its effects on store choice. Second, Make the store name constant by keeping it same for both the websites. Since the goal of the research is to prove that color will influence store choice because the color will influence trust, and, differences in trust between stores will influence store choice, it would make sense to keep the store name constant. This would also eliminate the complexity of manipulating and studying an additional independent variable.


3.3. Test-retest


To obtain a higher measure of reliability, I’d suggest administering the same test twice over a period of time to a similar group of individuals under same conditions. The results from both the tests can then be correlated in order to evaluate the test for stability over time.

It is important to note that a perfect correlation between the test and the retest is impossible as one cannot remove all the confounding factors completely. However, the effects of these factors can be minimized by replicating experiment conditions and involving similar subjects. Also, since the experiment was conducted with large subject groups (277 subjects), the chances of few subjects skewing the results reduces as the extremes are drowned out providing more accurate results.


3.4. Involve diverse subjects


Since the subjects were students from North America, a Westernized country, the generalizability of the results should be limited to countries with an occidental (Westernized) culture. The fact that all the subjects involved in the study were students of Introduction to Principles of Information Systems for Managers class also limits the generalizability the results. Another critical factor to consider would be the ethnicity of the subjects. Research has shown that color has cultural significance. The experiment only involved Hispanic and White Non-Hispanic subjects which also limits the application of the results.

In order to improve the generalizability of the results, a more diverse population of subjects needs to be tested. The diversity of subjects must span across ethnicity, profession, and geography.



References


Bellizzi, J. A., Crowley, A. E., & Hasty, R. W. (1983). The effects of color in store design. Journal of retailing.

Bellizzi, J.A. and R.E. Hite, "Environmental Color, Consumer Feelings, and Purchase Likelihood," Psychology & Marketing, Vol. 9: 347-363,1992

Dodds, W. B. (1991). In search of value: how price and store name information influence buyers′ product perceptions. Journal of Consumer Marketing, 8(2), 15-24.

Grewal, D., Krishnan, R., Baker, J., & Borin, N. (1998). The effect of store name, brand name and price discounts on consumers' evaluations and purchase intentions. Journal of retailing, 74(3), 331-352.

Middlestadt, S. E. (1990). The effect of background and ambient color on product attitudes and beliefs. ACR North American Advances.

Kim, J., & Moon, J. Y. (1998). Designing towards emotional usability in customer interfaces—trustworthiness of cyber-banking system interfaces. Interacting with computers, 10(1), 1-29.

Lee, S., & Rao, V. S. C. (2010). Color and store choice in electronic commerce: The explanatory role of trust. Journal of Electronic Commerce Research, 11(2), 110.