Psychology Research Methods: A Comprehensive Guide

# Chapter 1 Research Methods ## Learning Intentions In this chapter you will learn how to: - describe and use the concepts and terminology relating to research methods - describe, evaluate and apply the six main research methods - explain and use other concepts relevant to planning and conductin...

# Chapter 1 Research Methods ## Learning Intentions In this chapter you will learn how to: - describe and use the concepts and terminology relating to research methods - describe, evaluate and apply the six main research methods - explain and use other concepts relevant to planning and conducting studies and analysing their results - apply the concepts of research methods to novel situations - consider the debate surrounding ethics in psychology ## Introduction Psychological phenomena are explored using a 'research process' and the methods used to investigate questions in psychology are called 'research methods'. This chapter will help you to understand how those methods are used by psychologists to find out about human (and animal) cognition, emotions and behaviour. The chapter is divided into several sections, covering the basic research methods that you need to understand: experiments, self-reports, case studies, observations, correlations and longitudinal studies. Once you have studied these sections, you should be able to apply your knowledge of research methods to novel research situations. In addition, you will learn about the other features of the research process or 'research methodology'. These features are the steps a psychologist takes in developing and conducting studies to investigate a question or problem. This research process can be thought of as having several stages, the first of which is to develop an aim. Although developing an aim is one of many steps, it is useful to consider it here before moving on to the main part of the chapter. The aim is the intention of the study, the idea it is trying to test, problem it is intended to solve or the question it is attempting to answer. Consider the idea of different ways to help students to study, perhaps using mind maps or revision apps. Imagine that a psychologist, Dr Huang, asks a few of her psychology students which method they prefer, and finds that both are quite popular. Dr Huang wants to know which is most effective. This is Dr Huang's aim: to investigate whether mind maps or revision apps are more effective at helping students to learn. So, the aim tells you the purpose of the investigation: in this case, an experiment. It is generally expressed in terms of what the study intends to show. An aim can also express the intention to investigate a link or relationship between two variables, such as between the number of computer games a student plays and their final A Level grade: in this case, the study would be a correlation. This chapter therefore covers the whole of this research process: - development of an aim - selection of a research method and the designs, formats or techniques to be used within that research method - definition, manipulation, measurement, control of variables and variables in hypotheses - selection of participants - ethical considerations - analysis of data, including the drawing of conclusions - evaluation of research. To be trustworthy, research needs to be planned well and conducted effectively. Imagine an investigation comparing a new classroom technique to an old one. If the researcher didn't know how hard the students worked and used the new technique on a lazy class, and the old one on a highly motivated class, this would produce false results. Consider a study into consumer psychology that compared how many goods were sold with and without music playing in the store. If the researcher only played music at the weekends and played no music on weekdays, would you believe the findings of the study? ## 1.1 Experiments An experiment is an investigation that allows researchers to look for a cause-and-effect relationship. The researcher investigates the way one variable, called the independent variable, is responsible for the effect on another, the dependent variable. To test this, the researcher manipulates the independent variable (IV) to produce two or more 'levels' or conditions, such as creating 'bright' or 'dull' lighting or selecting 'early' and 'late' in the day. The effect of these conditions on the dependent variable (DV) is measured. For example, an IV of the brightness of lighting might affect attention, with people being better at paying attention when the light is bright. How well people pay attention would be the DV. If there is a big difference in the DV between the conditions, the researcher would conclude that the IV has caused the difference in the DV, i.e. that the brightness of light affects attention (Figure 1.2). ## 1.2 Self-reports In a self-report, the participant gives the researcher information about themselves directly. This is different from experimental tests or observations where the researcher finds the data out from the participant. There are two main ways to conduct a self-report: using a questionnaire or an interview. Both types of self-report allow the researcher to ask the participant questions. ### Questionnaires In a questionnaire, the questions are presented to the participant in written form. The two techniques for doing this are as a paper and pencil exercise or online. There are several different types of question. The two most important question formats are closed questions, which have a fixed set of possible responses, and open questions, which ask for descriptive answers in the participant's own words. **Closed questions** can take the form of simple choices, such as those asking for yes/no answers or items from a list. Other forms of closed questions include rating scales (where a number is chosen, e.g. between 0 and 5) and Likert scales, which ask the respondent to say how much they agree with a statement such as 'Obesity is not important' or 'Exercise is a necessity' using the choices 'strongly agree / agree / neither agree nor disagree / disagree / strongly disagree'. Some examples of closed questions are as follows: - What is your gender: boy or girl? - How do you travel to school? walk / bicycle/bus/ train / car - Indicate which animal(s) scare you: dog, spider, cat, rat, fish, rabbit, bird. [You may tick as many as you like] - How much do you like psychology on a scale of 0-4? (0 not at all, 4 = very much) **Open questions** prompt the respondent to give detailed answers, which may be quite long. They contain more depth than the answers to closed questions and are more likely to be able to explore the reasons behind behaviours, emotions or reasoning. They typically ask 'Why...?' or simply 'Describe...'. Some examples of open questions are as follows: - Why do you believe it is important to help people who suffer from phobias? - Describe your views on the use of social media sites with regard to encouraging helping behaviour. - Explain how you would respond if you were told to hurt another person. ### Interviews An interview is a research method in which the researcher can use two alternative techniques. Typically they are face-to-face with the participant but the interview can also happen via the telephone. Interviews can, however, be conducted in other ways that allow real-time interaction, such as through a chat facility. The same formats of questions can be asked in interviews as in questionnaires, although often interviews use more open questions than closed questions. The schedule of questions, that is the range of questions that are asked and the order of them, differs between different interview formats. In a **structured interview**, the questions asked are the same for every participant and the order is fixed. There may even be instructions for the interviewer about their tone of voice, how to sit or how to dress so that the procedure is standardised each time data is collected. In an **unstructured interview**, in contrast, the questions asked depend on what the participant says, so the questions may be different for each participant. This is a very flexible format but it may be hard to compare data collected from different participants or by different researchers. A compromise is a **semi-structured interview**. Here, there are some fixed questions, which make sure that there is some similar information from every participant. This means that comparisons can be made between them, and averages can be calculated if this is appropriate. In addition, it is possible to ask some questions that are specific to individual participants. This allows the researcher to develop ideas and explore issues that are particular to that person. ## 1.3 Case Studies A case study is a detailed investigation of a single instance, usually just one person, although it could be a single family or institution. The data collected is detailed and in-depth, and may be obtained using a variety of different techniques. For example, in one case study, a participant may be interviewed, observed, given tests and asked to fill in questionnaires. Case studies are particularly useful for looking at rare cases where a detailed description is useful, and for following developmental changes where the progress of a child, or a person with a disorder, can be tracked through their improvement or decline. Case studies are therefore sometimes linked to therapy but it is important to remember that when the case study as a research method is being discussed, the therapeutic purpose is not the main aim. In addition to using different methods or techniques within a case study, different sources of evidence may be used. These could include interviewing the participant themselves, their relatives or other people such as colleagues to obtain a wide range of information about the individual's history, as well as accessing pre-existing information such as medical or school records. ## 1.4 Observations Observations involve watching human or animal participants. There are many different choices in the way that observations are conducted and we will consider each in turn. An awareness of these choices is important both for understanding and planning observational studies. The setting in which the observation takes place is one such choice. A **naturalistic observation** is conducted in the participants' normal environment, without interference from the researchers in either the social or physical environment. A **controlled observation** is conducted in a situation that has been manipulated by the researchers. This may be in terms of the social environment, such as varying group sizes or adding a model, or in terms of the physical environment, such as providing objects for play, different foods or new locations. Controlled observations can be done in either the participants' normal environment or in an artificial situation such as a laboratory. ## 1.5 Correlations A correlational analysis is a technique used to investigate a link between two co-variables (measured variables). Correlations are useful when it is possible only to measure variables, rather than manipulate them, i.e. when an experiment cannot be conducted. This may be because changing the variables would not be practical or would be unethical. For example, it would not be practical or ethical to conduct an experiment which controlled children's long-term exposure to television and it would not be ethical to increase real-life exposure to violent television programmes. Both of these could, however, be investigated using correlations. Real-life exposure of a group of children to violent television could be measured and correlated with another measured variable, such as how aggressive their behaviour was in school. To look for a correlation between two variables, each variable must exist over a range, sometimes called 'continuous data', and it must be possible to measure them numerically. Suitable variables would include durations of time, totals of tallies, numbers from a rating scale or test scores. Several techniques can be used to collect data for correlations, such as self-reports, observations and different kinds of tests and tasks. It is important to recognise that any link found between two variables in a correlation cannot be assumed to be a causal relationship, that is, we cannot know whether the change in one variable is responsible for the change in the other variable (Figure 1.10). This is possible in the case of an experiment, as we can know that there is a causal relationship between the IV and the DV because changes in the IV are shown to be the cause of changes in the DV. However, we cannot say from one correlation that an increase in one variable has caused an increase (or decrease) in the other because it is possible that the changes in both variables could be the result of another factor. Imagine that two variables are being measured: attention in class and score on a test. If these co-variables correlate it would be tempting to say that paying attention in class is responsible for good test results but we cannot be sure of this. It could be that both of these factors depend on a third variable, such as how hard-working the individual student is. The sort of student who pays more attention in class might also study much harder for the test. All we can conclude is that the two factors we have measured vary together, not that there is a cause-and-effect or causal relationship between them. As a consequence, it is important that you refer to 'co-variables' or 'measured variables' in a correlation and not independent and dependent variables. To make judgements about causality, an experiment must be used, so that we can be more certain that it is the manipulation of one variable that is responsible for the change in the other. If, on the other hand, we conduct a correlational study and find that there is no link between two variables, then we can conclude that there is no causal relationship. The nature of the relationship between the two variables in a correlation can be described in terms of its direction. In a **positive correlation**, the two variables increase together. The change is in the same direction, so higher scores on one variable correspond with higher scores on the other. For example, in a positive correlation between exposure to aggressive models and violent behaviour, greater exposure to models would be linked to higher levels of violence. When two variables are **negatively correlated**, higher scores on one variable correspond with lower scores on the other. For example, a negative correlation might exist between number of years in education and level of obedience: people with fewer years of education are more obedient (see also Section 1.9 on how to draw a scatter graph and a discussion of the strength of a correlation). ## 1.6 Longitudinal Studies A longitudinal study is one that follows a single group of participants over time, studying one or more variables at intervals. By measuring variables over weeks, months, years or even decades, a longitudinal study can detect changes in individuals. People change over time, not just as they age but as they accumulate life experiences. In addition, longitudinal studies can explore the effect of specific 'experiences' such as interventions or events on development. The use of longitudinal studies is an alternative to cross-sectional studies as a way to look at effects over time. In a **cross-sectional study**, differences between people of differing ages, or at different stages (such as different periods after an intervention) are explored by comparing different groups at one point in time. For example, to explore changes in mindfulness with age, a cross-sectional study could look at 10-, 20-, 30-, 40- and 50-year-olds. Although a cross-sectional study allows researchers to investigate long-term influences, it is difficult to separate changes over time from differences due to the individuals growing up at different times. For example, if a cross-sectional study looking at helpfulness in people of different ages found a pattern in the results, this could be due to differences in upbringing at the time when they were children or the changing expectations of society rather than their age. Such problems are avoided in a longitudinal study as a single **cohort**, a group of participants all selected at the same age or stage, are chosen and tested at different time points. A cohort might be individuals born in a certain year, people about to embark on a course of treatment, such as a new drug regime or people experiencing a life change, such as at the beginning of a pregnancy. Longitudinal studies may have different intentions or designs. Some studies track patterns of change in variables to investigate them, such as changing emotions or beliefs as we get older. Other studies record longitudinal changes in two or more variables to investigate possible correlations between them, such as exposure to a multicultural society and prejudice. Yet others use an experimental design where one or more variables, measured over an extended period of time, are expected to change as a consequence of a deliberate manipulation. In the last case, the variables being measured are dependent variables and the testing at different time points in relation to the deliberate manipulation is an independent variable. To provide a baseline for tracking long-term change, the variable is usually measured at the outset of the study, and again after one or more periods of time. This procedure can be thought of as a very long-term repeated measures design, so is sometimes referred to as a 'quasi-experimental' design. This procedure can be used, for example, to measure the consequences of interventions such as educational, health, mental health or occupational ones that aim to improve outcomes. One level of the independent variable is the baseline test or 'pre-intervention' time point and one or more repeat testing sessions occur, providing a 'post-intervention' level of the independent variable and possibly more time points during or long after the intervention as a follow-up. ## 1.7 The Definition, Manipulation, Measurement and Control of Variables Variables are factors that change or can be changed. In experiments, these are the independent and dependent variables as well as variables that are or are not controlled. In correlations, there are two measured variables called co-variables. Experiments look for changes or differences in the dependent variable (DV) between two or more levels of the independent variable (IV), which are set up by the experimenter. It is important that the IV is operationalised, so that the manipulation of the conditions represents the intended differences. At the beginning of this chapter, we discussed the idea of the aim of a study, that is, what the researcher is trying to find out. The aim refers to the variables but not necessarily very clearly. Operational definitions of variables help to add clarity to the intention expressed in the aim. Consider a study testing the effect of age on susceptibility to false memories. The IV would be age, with, for example, 'young', 'middle-aged' and 'old' groups. It is important to know how old the people in the groups are; this is operationalisation. You might operationalise 'young' as under 20 years old, 'middle-aged' as 40-50 years old and 'old' as over 70. The DV must also be operationalised, so it can be measured effectively. We could operationalise the DV by counting the number of details 'remembered' about the false memory or how convinced the participants were that it was true. ## 1.8 Sampling of Participants A **population** is a group of people (or animals) with one or more characteristics in common. For example, the population of a country is all the people who live there, the population of internet users is everyone who can access the internet. A population could also be people who share a particular interest, such as 'all football supporters' or who have a particular feature, for example all left-handed people'. The **sample** is the group of people who participate in a study. They are taken from a population and should ideally be representative of that group so that the findings will be generalisable. Details about the sample, such as age, ethnicity and gender are often important in investigations, so are commonly reported in the core studies. This is because these features may have an influence on psychological differences. Other characteristics of the sample, such as socio-economic status, education, employment, geographical location or occupation, may also be relevant. The size of the sample also matters. Small samples are less reliable and are likely to be less representative. This is because a smaller group of people is unlikely to contain all the differences or variations that exist with the populations and some of those will be important to any study. The different sampling techniques described next produce samples that differ in terms of how well they represent the population. The extent to which they are representative of the population determines how effectively generalisations can be made. ## 1.9 Data and Data Analysis Psychologists, like all scientists, often produce numerical results from their investigations. These results are called the 'raw data'. As it is difficult to gain an understanding from large amounts of figures, the results are often simplified mathematically so they can be represented visually on graphs. This makes the meaning of the results easier to interpret. We will discuss a range of analytical techniques in this section. It is important that you are confident that you can count up scores, find the mode or the range of a data set, make simple comparisons and interpret data from tables or graphs. ## 1.10 Ethical Considerations As you will see from the discussion of the core studies, psychologists need to consider ethical issues when they conduct research. It is important that they take steps to make sure that their research follows ethical guidelines. In this section, we will look at some of these ethical issues and consider the ethical guidelines that help psychologists to deal with these issues effectively. ## 1.11 Evaluating Research: Methodological Issues As well as evaluating research in terms of ethics, it can also be considered in terms of whether it is 'good science', i.e. by looking at methodological issues. There are several key methodological issues that you have encountered elsewhere in the chapter, such as reliability, validity and generalisability. ## 1.12 Activity 1.17 - Research Methods in Practice - Dr Singh is concerned about the generalisability of his findings. He has two ideas for changes to the procedure: conducting the same test in a village as well as his town and repeating the procedure using a female traffic warden in both conditions. Explain how each idea would improve generalisability. - The proposed study has high ecological validity. Explain why. - Dr Singh thinks that the observations in one of the behavioural categories, visibly reducing speed, could be subjective. Explain why this is likely. - Dr Singh wants to measure the inter-observer reliability of his four observers. Explain why this is important. - In the final part of the study, some participants find out that they have been in a study. Suggest one ethical problem that could arise from this. - By giving the drivers a number to call, rather than taking their number and calling them, Dr Singh is giving the participants their right to withdraw. Why is this important?