Applying an eMASS Customization Program as a Research Tool to Evaluate Consumer Benefits

Our protocol requires participants to experience mass customization before taking a survey, and uses the survey data to test the relationships between consumer responses. The protocol allows consumer data to be applied to structural equation modeling and latent means analysis to obtain sophisticated statistical data and to increase the validity of the research. Our study will benefit research investigating the effects of retail or marketing technology on consumer responses.

Although testing several invariance tests such as configural, matrix and scalar invariance may be difficult, we advise patience and following this procedure very carefully to increase the quality of the data. Demonstrating the procedure will be Hiyun Kim, a grade A student from my laboratory. Use an online survey to recruit female consumers who have an online apparel shopping experience.

Send an invitation email to qualifying participants that includes information regarding the purpose of the study and an assurance of the confidentiality of the responses. Send guidelines to those survey participants who agree to take part in a survey showing how to create trench coats using the customization program. Participants who agree to take part in the survey start the survey procedure.

They receive a link that is connected to the eMass Customization program in an existing shopping website, and are asked to imagine you are well off enough to purchase likable clothing and have to purchase a trench coat to attend an important meeting. You will want to create a unique trench coat. While browsing the internet, you come across the perfect apparel website that has a mass customization program.

24 hours after receiving the scenario, activate the survey link so that participants who have finished their trench coat and are ready to take the survey can click on the survey link. In the survey, have the participants upload the screenshot and price of the trench coat they created to the first page of the survey. Then, have the participants complete the online questionnaire regarding the perceived benefits and emotional attachment to the customized product, and attitude toward the customization program, loyalty intentions, and demographic questions.

Provide a monetary reward to those participants who complete the survey. When all of the surveys have been completed, save the survey data in an SPSS file containing of all the responses of the survey participants and use the cleaned data to conduct a structural equation modeling analysis. Using a median split, sum and average the scores of six items of fashion innovativeness and calculate the median score of fashion innovativeness.

Under the transform menu, click recode into different variables and code one for a low fashion innovative group if the mean score is lower than the median. Double click the fashion innovative group variable to move it to the split cases by field. Then, assign the output file directory location to save the files as Data_low fashion innovativeness.

sav and Data_high fashion innovativeness. sav in the assigned directory. To confirm the convergent validity, click select data files, Data_TOTAL.sav.

Develop the measurement model based on the research questions to include five latent variables and 17 observed variables. Set each of the variances of the latent variables as one and click calculate estimates. Then, check the fit indices of the measurement model from the results of the single group confirmatory factor analysis, goodness of fit index, adjusted goodness of fit index, normed fit index, Tucker Lewis index, comparative fit index and root mean square error of approximation.

In this representative analysis, a single group confirmatory factor analysis was conducted with five latent variables and 17 observed variables. All critical ratios of the factor coefficients were significant, implying that the convergent validity was achieved. The fit indices of a single group structural equation modeling revealed an acceptable fit.

Moving from the single group confirmatory factor analysis to multi-group confirmatory factor analysis to cross-validate the five factor measurement model for both groups reveals that configural invariance was achieved and that all of the critical ratios of the factor coefficients were significant. To test the metric invariance, the factor coefficients were constrained to be the same across two groups, and another multi-group confirmatory factor analysis was performed, indicating that a chi-square difference of 14.728 was not significant and that the metric invariance was satisfied. Since the metric invariance model was accepted, scalar invariance was tested.

Since the full metric/full scalar invariance model was nested in the full metric invariance model, a chi-square difference test was conducted, demonstrating that a chi-square difference of 11.18 was not significant, and that scalar invariance was satisfied. Given that the configural invariance, metric invariance and scalar invariance were achieved, a latent mean analysis was performed and the means of five latent variables for the high innovative groups were determined to be positive values that were significantly higher than those for the low fashion innovative groups. The mass customization program may seem difficult depending on an individual's perception of the task complexity.

Remember to provide enough time for the participants to acclimate to the customization program. To address group differences in the relationships among latent variables, multiple group structural equation modeling can be performed to compare past coefficients across the groups. Most researchers have used the multiple group structural equation modeling for group comparisons.

Our study provides another way for multiple group comparisons to be conducted in a social science area.