QUESTION 1: NXP

NXP is an internationally renowned digital consultancy. For a one of their largest clients they have developed a study looking into the potential effects of personalization on sales (in €). The client is an online retailer operating in a pan European context. The NXP research team designed an experimental study using a between-subjects design with two factors: (1) product type [Product] (1=’functional products’; 2=’hedonic products’) and (2) personalization [Personalization] (1=’none’; 2=’static’; 3=’dynamic’). Basically, no personalization compares versus static personalization based on the customer’s profile information in the online store account and dynamic personalization on the basis of the customer’s order history. Essentially, NXP’s client is mainly concerned with the effectiveness of personalization strategies across functional products and hedonic products. In total 720 respondents were randomly assigned to the cells of the design. The experiment was run for 3 weeks and sales (in €; [Sales]), buying intention [Intention] and privacy concerns [Privacy] were recorded. Additionally, gender (Gender), age (Age) and the length of the relationship with the online store [Relationship] were made available from the client’s records. The variables for the NXP study are summarized in Exhibit 1.

Exhibit 1: Variables NXP Study

Experimental factor product type [Product]:

(1) Functional products.

(2) Hedonic products.

Experimental factor personalization [Personalization]:

(1) None.

(2) Static (based on account profile).

(3) Dynamic (based on order history).

Buying Intention [Intention]:

I would definitely buy this product [INT01]. a

This product would have my preference [INT02]. a

Buying this product is a good idea [INT03]. a

Sales [Sales]: in € provided from client records.

What is you gender? [Gender; provided from client records]

(1) Male

(2) Female

What is your age? In years [Age; provided from client records]

Length of relationship with client: In years [Relationship; provided from client records]

Privacy concerns (based on Smith et al., 1996 [Privacy]):

I am very sensitive about online stores handling personal information [PRIV01]. b

I find personal privacy a major concern [PRIV02]. b

I find it very important to maintain my personal privacy from online stores [PRIV03] b

Reference

Smith, H.J., Milberg, S.J. and Burke, S.J. (1996). Information Privacy: Measuring Individuals’ Concerns about Organizational Practice. MIS Quarterly, 20(2), 167-196.

a Measured on a 7-point scale (1= Very unlikely; 7= Very likely).

b Measured on a 7-point Likert scale (1=Strongly disagree; 7=Strongly agree).

(a) Determine the measurement level for the variables in the NXP study provided in Exhibit 1. Provide a rationale for your answers

(b) In order to determine the sample size necessary to obtain the desired power of 0.8 for the study using a significance level () of 0.05 G*Power 3.1 was used. For the main effects of product type and personalization a (partial) 2 of 0.20, respectively 0.25, was assumed. For the interaction of product type and personalization a (partial) 2 of 0.03 was assumed. In order to calculate the sample size for the study the NXP research team argues that they would basically only need the sample size for the interaction effect, as it is assumed to be the smallest effect size. Please, indicate whether this rationale is correct. Moreover, to calculate the sample size for the interaction effect of product type and personalization the NXP research team would need your assistance for obtaining the numerator degrees of freedom and the number of groups. Please, provide the numerator degrees of freedom and the number of groups to the NXP research team and provide a rationale for your answer. The relevant menu for the sample size determination using G*Power is depicted in Exhibit 2, Panel A (See highlight). Finally, Benjamin et al. (2017) have argued that a significance level of 0.05 would not be sufficiently stringent and they suggest a significance level of 0.005. In Exhibit 2, Panel B, a G*Power plot was prepared to assess the impact of a more stringent significance level on the required sample size. Indicate what would be the potential impact of a more stringent significance level (from 0.05 to 0.005) on sample size and provide a rationale for your answer.

(c) The NXP research team would like to assess whether the manipulation of personalization [Personalization] consisting of three levels, (1) none, (2) static and (3) dynamic was effective for the different levels of personalization. Toward that end a single question (MC.Personalization: “I receive personalized offers from store XYZ” on a 7-point scale [1=’Strongly disagree’; 7=’Strongly agree’]) was included in the design to allow for testing the extent to which the personalization manipulation achieved its objectives. This is typically referred to as a manipulation check. The NXP research team conducted a oneway ANOVA to assess the effectiveness of the manipulation for personalization. The SPSS output is provided in Exhibit 3. Assess the assumptions for the analysis, formulate the null and alternative hypotheses for the test, interpret the SPSS output, discuss the findings and implications using the SPSS output in Exhibit 3. Refer to specific elements in the output you have used to reach your conclusions. You may assume a significance level of 5% (α=0.05)

(d) The NXP research team conducted a two-way ANOVA with Product and Personalization as factors (independent variables) and Sales (in €) as dependent variable. The SPSS output for the GLM Univariate procedure is included in Exhibit 4. Assess the assumptions for the analysis, formulate the null and alternative hypotheses for the relevant tests, interpret the SPSS output and discuss the findings and implications using the SPSS output in Exhibit 4. Refer to specific elements in the output you have used to reach your conclusions. You may assume a significance level of 5% (α=0.05)

(e) The NXP team considered that privacy concerns [Privacy] might affect online sales and would like to control for them. As a result, they argued for the inclusion of a measurement instrument for privacy concerns (See Exhibit 1). The NXP research team suggested to calculate a mean score of the three items for privacy concerns [PRIV01, PRIV02, and PRIV03] and use it as a covariate in the analysis of the experimental design. Discuss the rational and the assumptions underlying the inclusion of a covariate in an ANOVA design. You may assume a significance level of 5% (α=0.05)