Confounding refers to a mixing or muddling of effects that can occur when the relationship we are interested in is confused by the effect of something else. It arises when the groups we are comparing are not completely exchangeable and so differ with respect to factors other than their exposure status. If one (or more) of these other factors is a cause of both the exposure and the outcome, then some or all of an observed association between the exposure and outcome may be due to that factor.

In this chapter, we look at the analytic studies that are our main tools for identifying the causes of disease and evaluating health interventions. Unlike descriptive epidemiology, analytic studies involve planned comparisons between people with and without disease, or between people with and without exposures thought to cause (or prevent) disease. They try to answer the questions, ‘Why do some people develop disease?’ and ‘How strong is the association between exposure and outcome?’. This group of studies includes the intervention, cohort and case–control studies that you met briefly in Chapter 1. Together, descriptive and analytic epidemiology provide information for all stages of health planning, from the identification of problems and their causes to the design, funding and implementation of public health solutions and the evaluation of whether these solutions really work and are cost-effective in practice.

In the previous chapter we alluded to what is sometimes called ‘secondary’ prevention, where instead of trying to prevent disease from occurring, we try to detect it earlier, in the hope that this will enable more effective treatment and thus improved health outcomes. This is an aspect of public health that has great intuitive appeal, especially for serious conditions such as cancer, where the options for primary prevention can be very limited. However, screening programs are usually very costly exercises and they do not always deliver the expected benefits in terms of improved health outcomes. In this chapter we introduce you to the requirements for implementing a successful screening program and to some of the problems that we encounter when trying to determine whether such a program is actually beneficial in practice.

The overarching goal of public health is to maximise the health of the population, and to achieve this we need evidence about what works and what does not work. Good studies are difficult to design and implement, and interpretation of their results and conclusions is not always as straightforward as we might hope. How, then, can we make the best use of this information? In the next three chapters we look at ways to identify, appraise, integrate and interpret the literature to generate the evidence we need to inform policy and practice. In this chapter we focus on interpreting the results from a single study, because if they are not valid they will be of limited value. The central question we have to answer when we read a study report is, ‘Are the results of the study valid?’

Network studies follow an explicit form, from framing questions and gathering data, to processing those data and drawing conclusions. And data processing leads to new questions, leading to new data and so forth. Network studies follow a repeating lifecycle. Yet along the way, many different choices will confront the researcher, who must be mindful of the choices they are making with their data and the choices of tools and techniques they are using to study their data. In this chapter, we describe how studies of networks begin and proceed, the life-cycle of a network study

At the core of epidemiology is the use of quantitative methods to study health, and how it may be improved, in populations. It is important to note that epidemiology concerns not only the study of diseases but also of all health-related events. Rational health-promoting public policies require a sound understanding of causation. The epidemiological analysis of a disease or activity from a population perspective is vital in order to be able to organize and monitor effective preventive, curative and rehabilitative services. All health professionals and health-service managers need an awareness of the principles of epidemiology. They need to go beyond questions relating to individuals to challenging fundamentals such as ‘Why did this person get this disease at this time?’, ‘Is the occurrence of the disease increasing and, if so, why?’ and ‘What are the causes or risk factors for this disease?’

How easy is it to repeat a previous corpus-based study? Repetition is a basic demand of scientific investigations. Hence our focus in this chapter is on an attempt to repeat some studies by Geoffrey Leech of modal verbs and word frequencies. In failing to repeat a number of observations we note the difficulty of repeating studies in corpus linguistics, in spite of the field having proposed that the ability to do this is one of its distinct strengths.

This chapter throws into relief the importance of the link between corpus hits and their sources (i.e. texts or speakers) by comparing study designs uninformed and informed by such metadata. The authors’ argument draws attention to the consequences of uneven distributions of observations across the basic text units of a corpus. In the case study (a distributional analysis of the use of actually in the BNCs of 1994 and 2014), it is demonstrated that disproportionate word counts contributed by individual sources (in this case, speakers) will distort estimates for relevant subsections of corpus data (in this case, demographic groups defined by age and gender) if the analysis assigns the same weight to every observation. The proposed solution is to factor in the text (or speaker) level, but this hinges on the availability of the relevant metadata. Moreover, insights into the hierarchical structure of corpus data are shown to benefit the design stage of a study. Thus, if manual post-processing steps preclude an exhaustive analysis of corpus hits, insights into the organization of data points can direct down-sampling strategies to generate a statistically efficient subset of tokens.

Statistical issues are prominent in Alzheimer’s disease (AD) clinical trials due to the enormous challenges in this disease. The complexity of the disease and intervention pathways challenge drug discovery efforts, but measurement and analysis complexities and subjective outcomes also interfere with successful drug development. Variability across disease stage, disease sub-types, comorbidities and concomitant treatments (between-subject), and non-equivalent forms, good and bad days, and rater inconsistencies (within-subject) increase the chance of failure. AD-specific statistical expertise is critical for success, in contrast to most disease areas that require less disease-specific statistical knowledge. Use of global statistical tests and composites, correcting for covariates, and model selection increase the chance of a clearly positive outcome for active treatments and a clearly negative outcome for inactive or harmful treatments.

This short chapter focusses on studies of semen quality, particularly the standardization of methods, the need for training of technicians, and study design.