Elea McDonnell Feit

Patients prioritize local recurrence risk over other attributes for surgical treatment of facial melanomas—Results of a stated preference survey and choice-based conjoint analysis

Background Surgical treatment options for facial melanomas include conventional excision with postoperative margin assessment, Mohs micrographic surgery (MMS) with immunostains (MMS‐I), and slow MMS. Patient preferences for these surgical options have not been studied. Objectives To evaluate patient preferences for surgical treatment of facial melanoma and to determine how patients value the relative importance of different surgical attributes. Methods Participants completed a 2‐part study consisting of a stated preference survey and a choice‐based conjoint analysis experiment. Results Patients overwhelmingly (94.3%) rated local recurrence risk as very important and ranked it as the most important attribute of surgical treatment for facial melanoma. Via choice‐based conjoint analysis, patients ranked the following surgical attributes from highest to lowest in importance: local recurrence rate, out‐of‐pocket cost, chance of second surgical visit, timing of reconstruction, travel time, and time in office for the procedure. Consistent with their prioritization of low local recurrence rates, more than 73% of respondents selected MMS‐I or slow MMS as their preferred treatment option for a facial melanoma. Limitations Data were obtained from a single health system. Conclusion Patients prefer surgical treatment options that minimize risk for local recurrence. Logistics for travel and treatment have less influence on patient preferences. Most survey participants chose MMS‐I to maximize local cure and convenience of care.

Bayesian Imputation for Anonymous Visits in CRM Data

Targeting individual consumers has become a hallmark of direct and digital marketing, particularly as it has become easier to identify customers as they interact repeatedly with a company. However, across a wide variety of contexts and tracking technologies, companies find that customers can not be consistently identified which leads to a substantial fraction of anonymous visits in any CRM database. We develop a Bayesian imputation approach that allows us to probabilistically assign anonymous sessions to users, while ac- counting for a customer’s demographic information, frequency of interaction with the firm, and activities the customer engages in. Our approach simultaneously estimates a hierarchical model of customer behavior while probabilistically imputing which customers made the anonymous visits. We present both synthetic and real data studies that demonstrate our approach makes more accurate inference about individual customers’ preferences and responsiveness to marketing, relative to common approaches to anonymous visits: nearest- neighbor matching or ignoring the anonymous visits. We show how companies who use the proposed method will be better able to target individual customers, as well as infer how many of the anonymous visits are made by new customers.

Reducing Data Complexity

Marketing data sets often have many variables—many dimensions—and it is advantageous to reduce these to smaller sets of variables to consider. For instance, we might have many items on a consumer survey that reflect a smaller number of underlying concepts such as customer satisfaction with a service, category leadership for a brand, or luxury for a product. If we can reduce the data to its underlying dimensions, we can more clearly identify the relationships among concepts.

Comparing Groups: Tables and Visualizations

Marketing analysts often investigate differences between groups of people. Do men or women subscribe to our service at a higher rate? Which demographic segment can best afford our product? Does the product appeal more to homeowners or renters? The answers help us to understand the market, to target customers effectively, and to evaluate the outcome of marketing activities such as promotions.

Association Rules for Market Basket Analysis

Many firms compile records of customer transactions. These data sets take diverse forms including products that are purchased together, services that are tracked over time in a customer relationship management (CRM) system, sequences of visits and actions on a Web site, and records of customer support calls. These records are very valuable to marketers and inform us about customers’ purchasing patterns, ways in which we might optimize pricing or inventory given the purchase patterns, and relationships between the purchases and other customer information.

Segmentation: Clustering and Classification

In this chapter, we tackle a canonical marketing research problem: finding, assessing, and predicting customer segments. In previous chapters we’ve seen how to assess relationships in the data (Chap. 4), compare groups (Chap. 5), and assess complex multivariate models (Chap. 10). In a real segmentation project, one would use those methods to ensure that data has appropriate multivariate structure, and then begin segmentation analysis.

Confirmatory Factor Analysis and Structural Equation Modeling

In this chapter, we discuss structural equation models in R. We show how R can be used for both covariance-based and partial least squares modeling, and present basic guidelines for model assessment. We also demonstrate the power of R to simulate data and to use simulation to inform our expectations.

An Overview of the R Language

In this chapter, we cover just enough of the R language to get you going. If you’re new to programming, this chapter will get you started well enough to be productive and we’ll call out ways to learn more at the end. R is a great place to learn to program because its environment is clean and much simpler than traditional programming languages such as Java or C++. If you’re an experienced programmer in another language, you should skim this chapter to learn the essentials.

Identifying Drivers of Outcomes: Linear Models

In this chapter we investigate linear models, which are often used in marketing to explore the relationship between an outcome of interest and other variables. A common application in survey analysis is to model satisfaction with a product in relation to specific elements of the product and its delivery; this is called “satisfaction drivers analysis.” Linear models are also used to understand how price and advertising are related to sales, and this is called “marketing mix modeling.” There are many other situations in which it is helpful to model an outcome, known formally as a response or dependent variable, as a function of predictor variables (also known as explanatory or independent variables). Once a relationship is estimated, one can use the model to make predictions or forecasts of the likely outcome for other values of the predictors.

Comparing Groups: Statistical Tests

In Chap. 5 we saw how to break out data by groups and inspect it with tables and charts. In this chapter we continue our discussion and address the question, “It looks different, but is it really different?” This involves our first inferential statistical procedures: chi-square, t-tests, and analysis of variance (ANOVA). In the final section, we introduce a Bayesian approach to compare groups.