Analyzing data requires more than simply running one model and reporting the results. Getting trustworthy results requires careful checking of the data, including checking for and addressing missing values, nonlinearities, and collinearity, and generating any necessary composite or recoded variables such as interaction terms, scales or indices, or even lagged variables. Finally, problems occur when messy real-world data meet assumption-laden models. Endogeneity, simultaneity, omitted variable bias, fixed effects, and dichotomous DVs all violate key assumptions of the OLS model and require appropriate statistical and/or theoretical adjustments to produce trustworthy results.

Now equipped with broader participant samples and more diverse stimuli, we can create Big Data experiments. This chapter reviews research methods involved in running Big Data surveys and experiments. The chapter discusses overt and covert measurements that we can collect via online experiments. The chapter then discusses practical logistics to keep in mind when running a Big Data experiment, including experimental design decisions, and a behind-the-scenes look at how data is saved online via server-side coding. Next, once you have the data from an experiment, how do you clean the data and how do you visualize it? The chapter ends with discussion on the ethical implications of collecting covert measures and the useful applications of web-coding skills to create public-facing websites.

Raw data require a great deal of cleaning, coding, and categorizing of observations. Vague standards for this data work can make it troublingly ad hoc, with much opportunity and temptation to influence the final results. Preprocessing rules and assumptions are not often seen as part of the model, but they can influence the result just as much as control variables or functional form assumptions. In this chapter, we discuss the main data processing decisions that analysts often face and how they can affect the results: coding and classifying of variables, processing anomalous and outlier observations, and the use of sample weights.

Once the information about nodes, links, and their substantive attributes has been collected, a bit more work is needed to prepare to use the data. This chapter covers this intermediate step, with tips for organizing and cleaning the data. Reading this chapter before collecting the data in the first place will help avoid some serious pitfalls. It covers ethical issues pertaining to collecting names (a necessary step in most methods of network elicitation), a method for automating the cleaning of name data, and robustness checks that can be done to assess the cleaning.

High-quality data are necessary for drawing valid research conclusions, yet errors can occur during data collection and processing. These errors can compromise the validity and generalizability of findings. To achieve high data quality, one must approach data collection and management anticipating the errors that can occur and establishing procedures to address errors. This chapter presents best practices for data cleaning to minimize errors during data collection and to identify and address errors in the resulting data sets. Data cleaning begins during the early stages of study design, when data quality procedures are set in place. During data collection, the focus is on preventing errors. When entering, managing, and analyzing data, it is important to be vigilant in identifying and reconciling errors. During manuscript development, reporting, and presentation of results, all data cleaning steps taken should be documented and reported. With these steps, we can ensure the validity, reliability, and representative nature of the results of our research.

The questionnaire, one of the most frequently used methods in the study of language attitudes, can be used to elicit both qualitative and quantitative data. This chapter focuses on the questionnaire as a means of eliciting quantitative data by means of closed questions. It begins by examining the strengths of doing this (e.g. the fact that the resulting data can easily be compared and analysed across participants) as well as the limitations (e.g. the fact that issues unforeseen by the researcher usually do not come to the fore). The chapter then discusses key issues in research planning and design: for example, question types, question wording, question order, reliability and validity, and more general issues regarding questionnaire design. The chapter also considers questionnaire distribution. The exploration of data analysis and interpretation focuses on data cleaning and coding, statistical analyses, and some points of caution regarding the interpretation of findings from questionnaire-based studies. A case study of language attitudes in Quebec serves to illustrate the main points made in the chapter. The chapter concludes with further important considerations regarding the context-specificity of findings and the benefits of combining questionnaires with other methods of attitude elicitation.

This chapter presents a detailed example that applies the compensation analytics concepts developed in Chapter 6. The reader is assumed to be a compensation consultant charged with evaluating whether gender-based discrimination in pay is present in a public university system in the sciences. Section 7.1 walks through the analysis step-by-step, from formulating the business question, to acquiring and cleaning data, to analyzing the data and interpreting the results from voluminous statistical output in light of the business question. Section 7.2 covers exploratory data mining, causality, and experiments. Exploratory data mining covers situations in which the manager does not know in advance which relationships in the data will be of interest, in contrast to the example in section 7.1 in which a statistical model and specific measures could be constructed that were directly tailored to address the business question at hand. Section 7.2 covers the challenges associated with establishing causality in compensation research and how experiments can sometimes be designed to address those challenges. Randomization and some pitfalls associated with compensation experiments are also covered

This chapter responds to the growing importance of business analytics on "big data" in managerial decision-making, by providing a comprehensive primer on analyzing compensation data. All aspects of compensation analytics are covered, starting with data acquisition, types of data, and formulation of a business question that can be informed by data analysis. A detailed, hands-on treatment of data cleaning is provided, equipping readers to prepare data for analysis by detecting and fixing data problems. Descriptive statistics are reviewed, and their utility in data cleaning explicated. Graphical methods are used in examples to detect and trim outliers. The basics of linear regression analysis are covered, with an emphasis on application and interpreting results in the context of the business question(s) posed. One section covers the question of whether or not the pay measure (as a dependent variable) should be transformed via a logarithm, and the implications of that choice for interpreting the results are explained. Precision of regression estimates is covered via an intuitive, non-technical treatment of standard errors. An appendix covers nonlinear relationships among variables.