Authors

Ross Metusalem

JMP

Muralidhara A

JMP

Objective

Apply measurement and structural models to survey responses from online shoppers to build and evaluate competing models.

Background

Online shopping has influenced the world of marketing significantly. Various online retailers across the globe sell a multitude of products and services, and consumers often prefer the convenience of shopping from home. While it provides an opportunity to shop 24/7, online shopping also exposes customers to increased risk. They can fall victim to fraud through fake websites set up to steal personal information or through the theft of their information from a seller’s database. Even without fraud, the relatively lower overhead of setting up an online store gives easy access to the marketplace to sellers who lack experience or skill, creating the risk of incorrect, incomplete, or lost orders. Therefore, to promote sales, online sellers have a strong incentive to encourage customers to view them as trustworthy. 

The Task

Anna, a marketing research analyst, is studying how different aspects of shoppers’ trust in an online seller influence their intent to purchase a product from that seller. Through prior knowledge and research literature, she has two competing models that she wants to test. In the first model, she theorizes that shoppers’ intent to purchase (Purchase) is driven in part by trust in the seller (Trust), which itself is driven by shoppers’ perceptions of their personal privacy during the shopping experience (Privacy), the reputation of the seller (Reputation), and the perceived security of the online shopping experience (Security). In addition, Security is proposed to directly influence Purchase, because there may be elements of perceived security unrelated to trust in the seller (e.g., the shopper is using public Wi-Fi) that influence a shopper’s intent to purchase. However, an alternative model proposes that Reputation also has a direct effect on Purchase because, theoretically, a positive perception of a seller’s reputation may increase intent to purchase without necessarily causing increased trust in that seller.

Here, Purchase, Trust, Security, Reputation, and Privacy are considered latent variables or constructs, meaning that they cannot be observed directly and must be inferred from objective measurements.

Anna’s competing theories are presented in more detail by the path diagrams in Exhibit 1 (see PDF). Both make testable claims regarding the influence of Privacy, Reputation, Security, and Trust on Purchase. In Model 1, Privacy, Reputation, and Security are all proposed to influence Trust, which in turn influences Purchase. In addition, Security is proposed to influence Purchase directly. Finally, Privacy, Reputation, and Security are all proposed to covary with one another. Model 2 includes all the effects of Model 1 while adding a direct effect of Reputation on Purchase.

Anna has collected survey data with the goal of testing this theory. Respondents rated a series of statements regarding a recent online shopping experience. For each of the five latent variables in her theory, Anna asked respondents to rate between three and five individual statements on a 0-5 scale. For example, one of the Privacy items stated, “I was not asked to provide any unnecessary personal information.” A response of 5 indicated strong agreement. 

Anna performs the following tasks using structural equation modeling (SEM):

  • -Conduct confirmatory factor analysis (CFA) to validate the measurement model. 
  • -Build competing structural equation models to test the theory. 
  • -Compare the models to identify which better fits the data.

Use the links below to read the full case study and download the data files