Getting Grounded in Social Psychology

The Essential Literature for Beginning Researchers

Todd D. Nelson, Todd D. Nelson

• Publisher: Psychology Press
• Year: 2017
• ISBN:
  9781351694667
• Edition: 1
• Topic: Psychology
• Subtopic: Social Psychology
• Index: Psychology

This volume provides a fast and efficient way for undergraduate and graduate students to gain a solid understanding of the social psychology literature. Each chapter reviews a major subsection of research in the field, written by a leading social psychology researcher in that area. Coverage includes all the major empirical, theoretical and methodological developments in its subfield of social psychology. Beginning social psychologists, as well as those who may have emerged from their formal training with a less-than-solid grounding in the research literature, will find this volume invaluable. It is the book all social psychologists wished they had access to when they were getting grounded in the research literature!

1
Creating Social Psychology

Some Ideas on the Research Process

Charles Stangor and Noah Wolf

Social psychology, like any other scientific endeavor, involves a semi-organized and extended process of learning about a topic through the collection and analysis of data. Our topic is the study of people and their social interactions, and our research methods are based, primarily, on quantitative analysis of data.

The business of social psychology is research. Social psychologists also define themselves through their teaching and mentoring, through their theory building, and through their social interactions with other scientists—but, bottom line, for most of us it is our research that matters most to us and to the academic community.

The importance of research is perhaps overrated to social psychologists, and some of us have argued that we should think more broadly about how we can contribute to understanding social behavior (Gergen, 1973, 1985). Despite such suggestions, social psychologists do not generally allow the presentation of ideas or theoretical models without accompanying data. Social psychology is its data and its interpretation of that data. Indeed, social psychology exists as a field thanks to the thousands of researchers and hundreds of thousands of research participants—mostly undergraduate students—who have participated in our research programs.

Successful research careers are valued in part because they are difficult to accomplish. Research requires extended periods of dedication to hypothesis development, testing, refinement, and replication. Because many of our journals require multiple experiments per publication, the timeline for developing even a single successful paper is long. Careers in social psychology are created through the development and execution of hundreds or thousands of experiments, many of which are unsuccessful. This creates a very high bar indeed, and it is no wonder that graduate students frequently find research to be the most stressful part of the graduate experience.

Because the “scientific method” is a set of guidelines and not an absolute template, it is subject to interpretation. Different scientists will make different decisions, even about the same topics, and this is to be expected. In short, data is messy—and because data is collected by, analyzed by, reported by, and cited by people, it gets even messier. Learning to understand, have confidence in, and nevertheless love our data requires large doses of patience and fortitude.

In this chapter we will discuss some of the many aspects of doing research in social psychology. The topic is much broader than we can cover here, and there are excellent textbooks available. We’ll provide some thoughts about what seem to us to be the most critical aspects at each stage of the research process and then address some relevant current topics in the field.

Getting Ideas

Social psychology is in good part based on common sense. And common sense tells us that we could develop ideas for research through the process of induction—drawing broad conclusions from the perception of individual behaviors. Many classic social psychological research programs were developed through induction: Festinger’s (1957) Cognitive Dissonance theory and Milgram’s (1963) study of obedience are a couple of classic examples.

Students may often develop their research ideas through induction in part because they have personal interests in certain research topics and in part because they have not yet committed themselves to a particular theoretical orientation.

Induction is not to be discounted as a source of ideas, but working the other way—using deduction—is probably better. Using deduction, we generate more specific predictions on the basis of broader theories or principles. For one, because deductive research is built on prior theorizing, the a priori probability of results working out is likely higher. Working blindly without the guidance of theory leads to many false positives (Stroebe, 2016). Secondly, working from existing research programs allows us to draw from the knowledge already gained by prior researchers. When we work from theory and prior research we can choose paradigms, manipulations, and methods that have already worked for others.

No matter their source, we should spend a good part of our research time getting ideas and researching them. The press of graduate training generally makes it difficult for graduate students to think broadly, because they are primarily focused on learning content in their own specific research areas. But we should all be encouraged, at every stage of our career, to relentlessly reach out to other fields. This is not always easy—the rhetoric and research paradigms vary widely across areas. But what you want to study may have already been done, and done well, in a different field.

A question is the degree to which our hypotheses should predict unexpected or surprising findings. Our social perceptions are pretty good, and often accurate. It is not surprising to say that people like other people who are similar to them and whom they perceive as attractive. We know this from copious amounts of social psychological research—but it’s also pretty obvious to a casual observer.

Other things are not as obvious (at least initially), and these topics receive the lion’s share of our work and our accolades. The tendency to overestimate personal causes for the behavior of others—the correspondence bias—is not clear to the lay observer, but social psychological research suggests that it is so. The findings that attitude change can be greater under small rather than large reward, that we can form accurate impressions of others in several seconds, and that priming us with smart people can make us smarter are even more surprising.

Working with unexpected hypotheses is risky, however. When the prior probability of the hypothesis being true is low, the likelihood that a significant result is a Type 1 error is high (Stroebe, 2016). A strategy for beginning researchers then is to choose a judicious combination of risky and less risky hypotheses.

Basic Versus Applied Research

Another difficulty of the social psychological endeavor is that there is no clear end state. Astronomers may work to prove Einstein’s theories or to discover the planets in the solar system. Cancer researchers strive to slow the growth of malignancies. These represent clear end states that make headlines when they are successful. The end states of social psychological research are more poorly defined. We want to, for instance, understand prejudice, to learn how close relationships are developed and maintained, and to determine how much of our everyday behavior is automatic.

These basic research questions are important, and they underlie fundamental desires on the part of social psychologists to make a difference. But although our focus is on basic research, we ignore application at our own peril. Social psychology will have a bigger impact when our research addresses applied problems, even though the solutions to those problems may seem a long way off. Interesting recent examples of putting basic research to an applied test include Banaji’s work on predicting suicide using the IAT (Nock et al., 2010) and Keith Payne’s work on the “shooter bias” using police officers (Payne, 2006). Funding agencies generally support this belief—our chances at grants are greater when the project has an applied focus.

Hypothesis Development

The best hypotheses are the simplest ones. We are often trained to favor interaction designs because they help us understand the important why? and when? questions and allow us to assess the limiting conditions of an effect (see later sections of this chapter). But interaction tests are notoriously difficult to get, statistically, and hoping for them frequently leads to a lot of failed studies and slows us down. Main effect tests are not irrelevant and should not be ignored. A good strategy is to first conduct a simple main effect test to be sure you can get the basic result that you want. Only then should the moderating variables be added in constructive replications.

Again, working from existing (or potentially new) theory is most important. As an example, the general idea of communal relationships (Clark & Mills, 1979) has been important in helping us understand long-term relationships. But the concept has been expanded and developed over the years (Park, Troisi, & Maner, 2011). Sometimes these contributions seem incremental, but they can make a huge difference in the long run.

Sampling

Social psychologists most often use college students as their research participants, and there is no clear reason not to. Although college students have characteristics that make them unique (Henrich, Heine, & Norenzayan, 2010), they are nevertheless a good place to start testing our theories. The expectation is that our theories are general—they apply to everyone—and so we can test them on whomever happens to be convenient. If a result is confirmed using college students, and if there are clear a priori reasons to move to other samples, researchers can then replicate quickly using another sample from another population.

Research based on college students is frequently criticized using the argument that it will not generalize. This argument holds the research process to an unacceptably high bar. There is no guarantee that a therapy administered in one hospital will generalize when it is administered in another hospital, or that a training program that is effective in one organization will generalize to another organization. External validity is always a matter of doubt and should be tested based on theoretical reasons. If you have a theoretical reason to believe that college students may be a poor sample for a test of your theory, then replicate with another population. If there is no such theory, then no replication is necessary.

Measurement

New students of social psychology are often less concerned about measuring their dependent variables than they should be. They may carefully consider their research hypothesis and the operationalizations of the independent variable, appropriately creating the most powerful and theoretically relevant manipulation they can develop. But then they may add a dependent measure without a lot of consideration. What many new students do not realize is that many experiments fail, not because the hypothesis is wrong, but because the dependent measure is not accurately assessed.

Our measures should be reliable, in the sense that they give the same or similar measures each time they are taken. They should be sensitive such that they appropriately vary with changes in the conceptual variable. And they should be construct valid, meaning that they actually measure what we say they are measuring. Developing a new measure takes time and patience and should consume a good deal of our research effort.

One difficulty is that measures change with the development of new technology. Just as our models became computational with the development of the personal computer in the 1980s, they have become biological with the development of brain imaging techniques (EEG, fMRI) and biological assays (cortisol, testosterone, and oxytocin are examples) in the 21st century.

It is often claimed—usually without much evidence—that the new measures are better than the old ones are and as a result, feeling thermometer measures of prejudice are replaced with Implicit Association Tests (IATs), self-report measures of anxiety are replaced with cortisol assays, and behavioral measures of aggression are replaced with fMRI scans.

Students should be wary of developing their research only around the assessment of new and relatively untested dependent variables. There are many dependent variables that we already know are useful—in addition to the standard repertoire of self-report measures, social psychologists use a wide variety of behavioral and well-established psychophysiological measures (heart rate; SCR). Again, a judicial balance is in order—do not miss the “wave of the future” by ignoring new approaches, but remember that these new techniques are not the only approach to measurement. The IAT and the fMRI take substantial time for participants to complete, and yet quick behavioral measures may be just as effective. When possible, use a previously validated measure rather than developing a new one.

Although social psychology is the study of “thoughts, feelings, and behavior,” it is the latter that really matters and that has the most impact. The research that has defined our field assesses behavior. These include the Milgram (1963) studies of obedience and conformity, Asch’s (1955) line-matching studies, The Robbers Cave studies of Sherif (1958; Sherif et al., 1961), Zimbardo’s prison investigations, stereotype threat research (e.g., Steele & Aronson, 1995), and staying or leaving a marriage partner (Rusbult & Martz, 1995). You are well advised to collect behavioral measures in your research programs.

Reporting Research Honestly

Since the reporting of the fraudulent activities of Dietrich Stapel in 2011 (https://en.wikipedia.org/wiki/Diederik_Stapel), there has been a renewed interest in social psychology about the validity of our research—and particularly the possibility that many of our findings are actually Type 1 errors (false rejections of the null hypothesis). This concern has led to a sizeable number of publications on this topic (Finkel, Eastwick, & Reis, 2015; Ledgerwood, 2014; Sakaluk, Williams, & Biernat, 2014; Stangor & Lemay, 2016), as well as some sharp debates among scientists. Approaches to the crisis have ranged from continuing “business as usual” (e.g., Baumeister, 2016) to calls for a vast overhaul of the entire process of research and dissemination.

The concerns about Type 1 errors were given concrete form when an attempt to conduct exact replications of 100 psychological experiments resulted in the successful replication of only 36 studies (Open Science Collaboration, 2015). These low replication rates have a ...