Handbook of Market Research

The question of how a certain activity (e.g., the intensity of communication activities during the launch of a new product) influences important outcomes (e.g., sales, preferences) is one of the key questions in applied (as well as academic) research in marketing. While such questions may be answered based on observed values of activities and the respective outcomes using survey and/or archival data, it is often not possible to claim that the particular activity has actually caused the observed changes in the outcomes. To demonstrate cause-effect relationships, experiments take a different route. Instead of observing activities, experimentation involves the systematic variation of an independent variable (factor) and the observation of the outcome only. The goal of this chapter is to discuss the parameters relevant to the proper execution of experimental studies. Among others, this involves decisions regarding the number of factors to be manipulated, the measurement of the outcome variable, the environment in which to conduct the experiment, and the recruitment of participants.

Digitalization of value chains and company processes offers new opportunities to measure and control a firm’s activities and to make a business more efficient by better understanding markets, competitors, and consumers’ behaviors. Among other methodologies, field experiments conducted in online and offline environments are rapidly changing the way companies make business decisions. Simple A/B tests as well as more complex multivariate experiments are increasingly employed by managers to inform their marketing decisions.This chapter explains why field experiments are a reliable way to reveal and to prove that a business action results in a desired outcome and provides guidelines on how to perform such experiments step by step covering issues such as randomization, sample selection, and data analysis. Various practical issues in the design of field experiments are covered with the main focus on causal inference and internal and external validity. We conclude the chapter with a practical case study as well as a brief literature review on recent published articles employing field experiments as a data collection method, providing the reader with a list of examples to consider and to refer to when conducting and designing a field experiment.

Surveys represent flexible and powerful ways for practitioners to gain insights into customers and markets and for researchers to develop, test, and generalize theories. However, conducting effective survey research is challenging. Survey researchers must induce participation by “over-surveyed” respondents, choose appropriately from among numerous design alternatives, and need to account for the respondents’ complex psychological processes when answering the survey. The aim of this chapter is to guide investigators in effective design of their surveys. We discuss state-of-the-art research findings on measurement biases (i.e., common method bias, key informant bias, social desirability bias, and response patterns) and representation biases (i.e., non-sampling bias and non-response bias) and outline when those biases are likely to occur and how investigators can best avoid them. In addition, we offer a systematic approach for crafting surveys. We discuss key steps and decisions in the survey design process, with a particular focus on standardized questionnaires, and we emphasize how those choices can help alleviate potential biases. Finally, we discuss how investigators can address potential endogeneity concerns in surveys.

This chapter explains the need to conduct international market research, identifies the main challenges researchers face when conducting marketing research in more than one country and provides approaches for addressing these challenges. The chapter examines the research process from the conceptual design of the research model to the choice of countries for data collection, the data collection process itself, and the data analysis and interpretation. Challenges identified include differentiating between etic and emic concepts, assembling an adequate research unit, ensuring data collection equivalence, and reducing ethnocentrism of the research team. We draw on the extant literature to determine methods that address these challenges, such as an adapted etic or linked emic approach, to define the concept of the culti-unit, and to identify prominent approaches to cultural dimensions and collaborative and iterative translation and statistical methods for testing equivalence. This chapter provides researchers with the methods and tools necessary to derive meaningful and sound conclusions from research designed to guide international marketing activities.

This chapter introduces readers to applications of data fusion in marketing from a Bayesian perspective. We will discuss several applications of data fusion including the classic example of combining data on media viewership for one group of customers with data on category purchases for a different group, a very common problem in marketing. While many missing data approaches focus on creating “fused” data sets that can be analyzed by others, we focus on the overall inferential goal, which, for this classic data fusion problem, is to determine which media outlets attract consumers who purchase in a particular category and are therefore good targets for advertising. The approach we describe is based on a common Bayesian approach to missing data, using data augmentation within MCMC estimation routines. As we will discuss, this approach can also be extended to a variety of other data structures including mismatched groups of customers, data at different levels of aggregation, and more general missing data problems that commonly arise in marketing. This chapter provides readers with a step-by-step guide to developing Bayesian data fusion applications, including an example fully worked out in the Stan modeling language. Readers who are unfamiliar with Bayesian analysis and MCMC estimation may benefit by reading the chapter in this handbook on Bayesian Models first.

This chapter provides a nontechnical summary of how to deal with endogeneity in regression models for marketing research applications. When researchers want to make causal inference of a marketing variable (e.g., price) on an outcome variable (e.g., sales), using observational data and a regression approach, they need the marketing variable to be exogenous. If the marketing variable is driven by factors unobserved by the researcher, such as the weather or other factors, then the assumption that the marketing variable is exogenous is not tenable, and the estimated effect of the marketing variable on the outcome variable may be biased. This is the essence of the endogeneity problem in regression models. The classical approach to address endogeneity is based on instrumental variables (IVs). IVs are variables that isolate the exogenous variation in the marketing variable. However, finding IVs of good quality is challenging. We discuss good practice in finding IVs, and we examine common IV estimation approaches, such as the two-stage least squares approach and the control function approach. Furthermore, we consider other implementation challenges, such as dealing with endogeneity when there is an interaction term in the regression model. Importantly, we also discuss when endogeneity matters and when it does not matter, as the “cure” to the problem can be worse than the “disease.”

Cluster analysis is an exploratory tool for compressing data into a smaller number of groups or representing points. The latter aims at sufficiently summarizing the underlying data structure and as such can serve the analyst for further consideration instead of dealing with the complete data set. Because of this data compression property, cluster analysis remains to be an essential part of the marketing analyst’s toolbox in today’s data rich business environment. This chapter gives an overview of the various approaches and methods for cluster analysis and links them with the most relevant marketing research contexts. We also provide pointers to the specific packages and functions for performing cluster analysis using the R ecosystem for statistical computing. A substantial part of this chapter is devoted to the illustration of applying different clustering procedures to a reference data set of shopping basket data. We briefly outline the general approach of the considered techniques, provide a walk-through for the corresponding R code required to perform the analyses, and offer some interpretation of the results.

Finite Mixture models are a state-of-the-art technique of segmentation. Next to segmenting consumers or objects based on multiple different variables, Finite Mixture models can be used in conjunction with multivariate methods of analysis. Unlike approaches combining multivariate methods of analysis and cluster analysis, which require a two-step approach, the parameters are then directly estimated at the segment level. This also allows for inferential statistical analysis. This book chapter explains the basic idea of Finite Mixture models and describes some popular applications of Finite Mixture models in market research.

Experiments are becoming increasingly important in marketing research. Suppose a company has to decide which of three potential new brand logos should be used in the future. An experiment in which three groups of participants rate their liking of one of the logos would provide the necessary information to make this decision. The statistical challenge is to determine which (if any) of the three logos is liked significantly more than the others. The adequate statistical technique to assess the statistical significance of such mean differences between groups of participants is called analysis of variance (ANOVA). The present chapter provides an introduction to the key statistical principles of ANOVA and compares this method to the closely related t-test, which can alternatively be used if exactly two means need to be compared. Moreover, it provides introductions to the key variants of ANOVA that have been developed for use when participants are exposed to more than one experimental condition (repeated-measures ANOVA), when more than one dependent variable is measured (multivariate ANOVA), or when a continuous control variable is considered (analysis of covariance). This chapter is intended to provide an applied introduction to ANOVA and its variants. Therefore, it is accompanied by an exemplary dataset and self-explanatory command scripts for the statistical software packages R and SPSS, which can be found in the Web-Appendix.

Linear regression analysis is one of the most important statistical methods. It examines the linear relationship between a metric-scaled dependent variable (also called endogenous, explained, response, or predicted variable) and one or more metric-scaled independent variables (also called exogenous, explanatory, control, or predictor variable). We illustrate how regression analysis work and how it supports marketing decisions, e.g., the derivation of an optimal marketing mix. We also outline how to use linear regression analysis to estimate nonlinear functions such as a multiplicative sales response function. Furthermore, we show how to use the results of a regression to calculate elasticities and to identify outliers and discuss in details the problems that occur in case of autocorrelation, multicollinearity and heteroscedasticity. We use a numerical example to illustrate in detail all calculations and use this numerical example to outline the problems that occur in case of endogeneity.

Questions like whether a customer is going to buy a product (purchase vs. non-purchase) or whether a borrower is creditworthy (pay off debt vs. credit default) are typical in business practice and research. From a statistical perspective, these questions are characterized by a dichotomous dependent variable. Traditional regression analyses are not suitable for analyzing these types of problems, because the results that such models produce are generally not dichotomous. Logistic regression and discriminant analysis are approaches using a number of factors to investigate the function of a nominally (e.g., dichotomous) scaled variable. This chapter covers the basic objectives, theoretical model considerations, and assumptions of discriminant analysis and logistic regression. Further, both approaches are applied in an example examining the drivers of sales contests in companies. The chapter ends with a brief comparison of discriminant analysis and logistic regression.

Many phenomena in marketing involve multiple levels of theory and analysis. Adopting a multilevel lens to marketing phenomena can often yield richer and more rigorous results. However, the consideration of multiple levels of theory and analysis often leads to the challenge to cope with nested data structures in which a lower level unit of analysis is nested within a higher level unit of analysis. Explicitly acknowledging such nested data structures is important as its analysis with single level analysis techniques may result in biased results and thus incorrect conclusions because nested data structures often violate assumptions of conventional single level analysis techniques. A methodological approach which explicitly accounts for multiple levels of analysis and thus the nested structure of data is referred to as multilevel modeling. This chapter attempts to help researchers and practitioners interested in investigating multilevel phenomena by providing an introduction to multilevel modeling. It therefore describes the theoretic fundamentals of multilevel modeling by outlining the conceptual and statistical relevance of multilevel modeling. Furthermore, it provides guidance how to build a multilevel regression model using a step-by-step approach. The chapter also discusses how to assess the fit of multilevel models, how to center variables at different levels of analysis, and how to determine the sample sizes to adequately estimate multilevel models. Moreover, it offers insights how the logic of multilevel regression analysis could be expanded to multilevel structural equation modeling, discusses different statistical software packages that can be employed to estimate multilevel models, and provides a detailed example of building and estimating a multilevel model.

The analysis of panel data is now part of the standard repertoire of marketers and marketing researchers. Compared to the analysis of cross-sectional data, panel data allow marketers to alleviate endogeneity concerns when linking an independent variable (e.g., price) to an outcome variable (e.g., sales volume). The more accurate estimates that result from panel data analysis help improve marketers’ decision-making in focal areas such as price setting and marketing budget allocation. Besides, panel data allow marketers to track customer behavior changes and distinguish real loyalty effects (i.e., same customer repeatedly buys a brand) from spurious effects (i.e., the same number of, but each time different set of, customers buys a brand). This chapter provides a nontechnical introduction to panel data analysis. Marketers will learn how to manage and analyze panel datasets in Stata. They will learn about the focal panel data estimators (pooled OLS, fixed effects, and random effects estimator), their underlying assumptions, advantages, and pitfalls. Besides, we introduce the between effects estimator, the combined approach, the Hausman-Taylor approach, and the first differences estimator as further techniques to analyze panel data. Finally, readers will receive an introduction to advanced topics such as dynamic panel models, panel data multilevel modeling, and using panel data to address measurement errors.

Time-series models constitute a core component of marketing research and are applied to solve a wide spectrum of marketing problems. This chapter covers traditional and modern time-series models with applications in extant marketing research. We first introduce basic concepts and diagnostics including stationarity test (the augmented Dicky-Fuller test of unit roots), and autocorrelation plots via autocorrelation function (ACF) and partial autocorrelation function (PACF). We then discuss single-equation time-series models such as autoregressive (AR), moving average (MA), and autoregressive moving average (ARMA) models with and without exogenous variables. Multiple-equation dynamic systems including vector autoregressive (VAR) models together with generalized impulse response functions (GIRFs) and generalized forecast error variance decomposition (GFEVD) are then discussed in detail. Other relevant models such as generalized autoregressive conditional heteroskedasticity (GARCH) models are covered. Finally, a case study accompanied by data and R codes is provided to demonstrate detailed estimation steps of key models covered in this chapter.

Time-series data include repeated measures of marketing activities and performance that are typically equally spaced in time. In the context of such data, Vector Autoregressive (VAR) models are uniquely suited to capture the time dependence of both a criterion variable (e.g., sales performance) and predictor variables (e.g., marketing actions, online consumer behavior metrics), as well as how they relate to each other over time. The objective of this chapter is to provide a foundation in VAR models and to enable the readers to apply them in their own research domain of interest. To this end, the chapter will discuss both the underlying perspectives and differences among alternative VAR models, and the practical issues with testing, model choice, estimation, and interpretation that are common in empirical research in marketing.From a marketing strategy perspective, both managers and academic researchers pay attention to whether a performance change is temporary (short-term) or lasting (long-term). Establishing the distinction between short-term and long-term marketing effectiveness is central to the understanding of marketing strategy and its implications, which this chapter aims to do. The interaction among appropriate marketing phenomena, modeling philosophy, and contemporary substantive topics sets this work apart from previous treatments on the broader topic of econometrics and time-series analysis in marketing (e.g., Dekimpe and Hanssens, Persistence modeling for assessing marketing strategy performance. In: Lehmann D, Moorman C (eds) Cool tools in marketing strategy research. Marketing Science Institute, Cambridge, MA, 2004; Hanssens et al., Market response models: Econometric and time series analysis. Springer Science and Business Media, 2001; Pauwels, Found Trends Market 11(4):215–301, 2018).

This chapter presents an overview of the process of structural equation modeling, involving the steps of model specification, model estimation, overall fit evaluation, model respecification, and local fit assessment (including interpreting the parameters of the model). Various extensions of the core structural equation model are described to enable more general representations of measurement and latent variable models as well as applications of the model to heterogeneous populations. An empirical example is provided to illustrate the process of structural equation modeling and to demonstrate some of the complexities that may arise in practical applications.

Partial least squares structural equation modeling (PLS-SEM) has become a popular method for estimating path models with latent variables and their relationships. A common goal of PLS-SEM analyses is to identify key success factors and sources of competitive advantage for important target constructs such as customer satisfaction, customer loyalty, behavioral intentions, and user behavior. Building on an introduction of the fundamentals of measurement and structural theory, this chapter explains how to specify and estimate path models using PLS-SEM. Complementing the introduction of the PLS-SEM method and the description of how to evaluate analysis results, the chapter also offers an overview of complementary analytical techniques. A PLS-SEM application of the widely recognized corporate reputation model illustrates the method.

The amount of text available for analysis by marketing researchers has grown exponentially in the last two decades. Consumer reviews, message board forums, and social media feeds are just a few sources of data about consumer thought, interaction, and culture. However, written language is filled with complex meaning, ambiguity, and nuance. How can marketing researchers possibly transform this rich linguistic representation into quantifiable data for statistical analysis and modeling? This chapter provides an introduction to text analysis, covering approaches that range from top-down deductive methods to bottom-up inductive methods for text mining. After covering some foundational aspects of text analysis, applications to marketing research such as sentiment analysis, topic modeling, and studying organizational communication are summarized and explored, including a case study of word-of-mouth response to a product launch.

Recent technical advances and the rise of digital platforms enhanced consumers’ abilities to take and share images and led to a tremendous increase in the importance of visual communication. The abundance of visual data, together with the development of image processing tools and advanced modeling techniques, provides unique opportunities for marketing researchers, in both academia and practice, to study the relationship between consumers and firms in depth and to generate insights which can be generalized across a variety of people and contexts.However, with the opportunity come challenges. Specifically, researchers interested in using image analytics for marketing are faced with a triple challenge: (1) To which type of research questions can image analytics add insights that cannot be obtained otherwise? (2) Which visual data should be used to answer the research questions, and (3) which method is the right one?In this chapter, the authors provide a guidance on how to formulate a worthy research question, select the appropriate data source, and apply the right method of analysis. They first identify five relevant areas in marketing that would benefit greatly from image analytics. They then discuss different types of visual data and explain their merits and drawbacks. Finally, they describe methodological approaches to analyzing visual data and discuss issues such as feature extraction, model training, evaluation, and validation as well as application to a marketing problem.

The increased awareness about the presence of social effects in consumer networks has inspired marketers to better understand and address the needs of their consumers through network analyses. In this chapter we consider network analyses as a set of techniques which allows researchers to analyze how the social structure of relationships around consumers affects their attitudes and behavior, and vice versa, how attitudes and behavior may affect the social structure. We focus on the types of network analyses that are currently most prominent within the field of marketing. We provide basic network theory and notation with references to key publications in the field. We also provide suggestions for software (packages) and useful functions including code snippets to support researchers and practitioners in setting up their first social network analyses. At the end of the chapter we discuss several more advanced network analysis methods and list several resources that might be useful to the interested reader.

Bayesian models have become a mainstay in the tool set for marketing research in academia and industry practice. In this chapter, I discuss the advantages the Bayesian approach offers to researchers in marketing, the essential building blocks of a Bayesian model, Bayesian model comparison, and useful algorithmic approaches to fully Bayesian estimation. I show how to achieve feasible Bayesian inference to support marketing decisions under uncertainty using the Gibbs sampler, the Metropolis Hastings algorithm, and point to more recent developments – specifically the no-U-turn implementation of Hamiltonian Monte Carlo sampling available in Stan. The emphasis is on the development of an appreciation of Bayesian inference techniques supported by references to implementations in the open source software R, and not on the discussion of individual models. The goal is to encourage researchers to formulate new, more complete, and useful prior structures that can be updated with data for better marketing decision support.

Conjoint analysis is one of the most popular methods to measure preferences of individuals or groups. It determines, for instance, the degree how much consumers like or value specific products, which then leads to a purchase decision. In particular, the method discovers the utilities that (product) attributes add to the overall utility of a product (or stimuli). Conjoint analysis has emerged from the traditional rating- or ranking-based method in marketing to a general experimental method to study individual’s discrete choice behavior with the choice-based conjoint variant. It is therefore not limited to classical applications in marketing, such as new product development, pricing, branding, or market simulations, but can be applied to study research questions from related disciplines, for instance, how marketing managers choose their ad campaign, how managers select internationalization options, why consumers engage in or react to social media, etc. This chapter describes comprehensively the “state-of-the-art” of conjoint analysis and choice-based conjoint experiments and related estimation procedures.

This chapter gives an introduction on how to exploit data from field experiments and aims to provide an intuitive understanding for managers and researchers alike. We outline the relevance and hurdles in identifying causal effects compared to observing purely correlational associations in studies which take place in the real world. We further provide a framework to classify different kinds of field experiments, such as quasi field experiments and natural field experiments. The core of this chapter focuses on giving an understanding of three standard econometric methods to exploit data from field experiments: difference-in-differences, regression discontinuity, and instrumental variables. For each method, we provide an intuitive understanding of the core features and its critical assumptions. We complement those explanations with an in-depth look at one practical application of each method in a field experiment setting and with a variety of practical examples from recently published research. Lastly, we provide a brief overview on how to implement each method in standard software packages such as STATA, R, and SPSS.

This chapter introduces the conceptual and statistical basics of mediation analysis in the context of experimental research. Adopting the respective terminology, mediation analysis can be referred to as an array of quantitative methods developed to investigate the causal mechanism(s) through which an independent variable influences a dependent variable. The chapter takes a regression-based approach to mediation analysis and focuses on mediation models likely to be tested in experiments (i.e., the single mediator model, parallel and serial multiple mediator models, and conditional process models). Yet, the scope of mediation analysis beyond an experimental setting will also be touched upon. Furthermore, the chapter addresses the question how to strengthen causal inference in mediation analysis through design, the collection of additional evidence, and statistical methods. It closes with a discussion of common topics of relevance when implementing mediation analysis such as sample size and power, mean centering in conditional process analysis, coding of categorical independent variables, advantages and disadvantages of a regression-based approach to mediation analysis, and software options to perform mediation analysis.

Measuring customer satisfaction and customer loyalty represents a key challenge for firms. In response, researchers and practitioners have developed a plethora of options on how to assess these phenomena. However, existing measurement approaches differ substantially with regard to their complexity, sophistication, and information quality. Furthermore, guidance is scarce on how firms can leverage and combine these approaches to implement a state-of-the-art satisfaction and loyalty measurement system. This chapter attempts to address this vacancy. The authors first define and conceptualize customer satisfaction and customer loyalty. Next, the authors provide an overview of the different operationalization and measurement approaches that companies face when designing a customer satisfaction and loyalty measurement system. The authors also discuss some of the common modeling challenges associated with measuring loyalty, namely, dealing with self-selection bias. Finally, the authors project what the future holds in this area.

Market segmentation describes the practice of grouping consumers that are alike concerning specific characteristics. The idea is that firms can better identify and target attractive segments and customize marketing actions for each segment. Equally important, segmentation allows firms to avoid consumers that are unprofitable or otherwise incompatible with its marketing strategy. Like other marketing concepts, market segmentation has changed over the years with increasing globalization and digitalization. But the concept of market segmentation has been and will continue to be one of the key concepts in marketing practice. In this chapter, we define market segmentation along with its key characteristics, describe the process by which it unfolds, outline the main traps to avoid, and provide an outlook into the future. A key concern of this chapter is also to reflect the key challenges in business environments, such as the abundance of data, globalization, as well as the acceleration of different trends.

Measuring accurate willingness to pay (WTP) is essential for designing pricing policies, particularly for pricing new products. Neglecting consumers’ WTP may lead to unexploited surplus when prices are set too low or to low demand when prices are set too high. Additionally, information on consumers’ WTP serves as valuable input to estimate sales and for use in optimization models, thus, to maximize profit. To date, various approaches to measure WTP exist that differ regarding their elicitation approach (direct vs. experimental) and whether they rely on stated or revealed preferences (hypothetical vs. actual WTP). This chapter provides an overview of the most common methods for measuring WTP and further discusses determinants of WTP.We further provide a practical illustration of WTP measurement. Therefore, we collected data on consumers’ WTP for a hypothetical new product offer using two stated preference approaches (open-ended questions and dichotomous choice method following a sequential monadic approach) as well as one revealed preference approach (BDM mechanism). We compare the results of these different methods and discuss how to apply WTP measures in practice.

Customers represent the most important assets of a firm. Customer lifetime value (CLV) allows assessing their current and future value in a customer base. The customer relationship management strategy and marketing resource allocation are based on this metric. Managers therefore need to predict the retention but also the purchase behavior of their customers.This chapter is a systematic review of the most common CLV, retention, and churn modeling approaches for customer-base analysis and gives practical recommendations for their applications. These comprise both the classes of deterministic and stochastic approaches and deal with both, contractual and noncontractual settings. Across those situations, the most common and most important approaches are then systematically structured, described, and evaluated. To this end, a review of the CLV, retention, as well as churn models and a taxonomy are done with their assumptions and weaknesses. Next, an empirical application of the stochastic “standard” Pareto/NBD, and the BG/NBD models, as well as an explanatory Pareto/NBD model with covariates to grocery retailing store loyalty program scanner data, is done. The models show their ability to reproduce the interindividual variations as well as forecasting validity.

In this chapter, we describe an approach to estimating the total long-term impact of brand perceptions on financial performance. The approach relies on modeling the stock market reactions to changes in brand perceptions and allows estimating their total impact even with limited time-series data. We present an application of the method to the Y&R Brand Asset Valuator (BAV) data. The analyses show that, on average, the bulk of brand impact on financial performance is realized in the future and the contemporaneous effects reflect only a small portion of the total impact. The analyses, however, also show considerable heterogeneity across industries: while in some industries the whole impact of brand asset occurs in current period only (restaurants), in other industries it occurs in future periods only (high-tech). Further, some components of consumer perceptions have differential effects in different industries. Returns to brand building, and to marketing efforts in general, should not be evaluated based on contemporaneous outcomes, but should rather be evaluated over a long-time horizon.

Sales promotions are an important marketing tool for both manufacturers and retailers. They include, for example, temporary price reductions, coupons, features, displays, sampling, and premiums. The bad news about promotions is that many of them are not profitable. The good news is that promotion effectiveness can be measured so that managers can identify the promotions which generate a profit and eliminate the ones that do not. This chapter presents data and models that can be used for this purpose. It focuses on panel data which is available at the aggregate (i.e., store) level and at the disaggregate (i.e., consumer) level. While aggregate data is more readily available and easier to analyze, disaggregate data allows for more detailed analyses. Several examples illustrate how models build on these data to measure promotion effectiveness. Since panel data has its limitations, it is often useful to complement it with surveys and/or experiments.

The proliferation of marketing media, especially since the advent of digital media, has created an urgent need for marketers to understand their relative importance in generating revenue for their brands. Ultimately, this understanding should result in managers’ ability to project returns from their media investments. This chapter will focus on quantitative methods that enable such media return calculations. We begin with a definition of “return on media” and show how it connects to the need of estimating top-line lift, i.e., consumer response to media, from various data sources. We introduce the standard media-mix response model and discuss the estimation of media response elasticities. We extend these models to include brand-building and customer-equity effects and intermediate-performance variables. Finally, we address return to media in the digital era, with specific reference to path-to-purchase models, and we describe how media returns are derived from sales response models.