Pages: 75

No chapter abstract available.

Pages: 15

The scientific method is an approach to acquiring knowledge using systematic and objective procedures to understand how and why our world is the way it is. This text is devoted to the application of the principles of scientific methodology to the social and behavioral sciences. In this first chapter, we emphasize the basics – the scientific method, operationalization, and the place of theory in a developing science. The purpose is to acquaint readers with the fundamentals and basic vocabulary of scientific research, including how to generate research questions and hypotheses and how to convert conceptual ideas into research operations and empirical measurements. The chapter emphasizes the need for multi-methodological approaches to effectively advance knowledge across the social sciences as a whole. When American astronauts landed on the moon in the summer of 1971, their activities included an interesting and, for some, surprising demonstration. They showed that when the effects of atmospheric air friction are eliminated, a light object (a feather) and a heavy object (a hammer) will reach the ground at the same time if dropped simultaneously from the same height. This verification of a basic principle of high school physics delighted many viewers of the live televised broadcast, but probably few of them considered the fact that for hundreds of years before Galileo (who is thought to have originally proposed this test), Western scholars had accepted Aristotle’s hypothesis that heavy objects would always fall faster than lighter ones. For most of us, Aristotle’s assumption seems intuitively correct, even though we now know that it is incorrect and contrary to scientific theory. Not all scientifically demonstrated phenomena contradict “commonsense” intuitions in this way, but this case serves to illustrate the difference between science and intuition as bases of understanding the physical and social world.

Pages: 22

This chapter begins with a discussion of causation and causal hypothesis testing and then introduces the concepts of mediation and moderation of causal relations. Testing causal hypotheses in an experiment involves manipulating an independent variable (experimental treatment or intervention) and assessing effects on a dependent variable. Experiments can be conducted in laboratory or field settings for purposes of testing basic theory or achieving an applied goal. A good experiment will have high internal validity. That means if the independent variable results in changes in the dependent variable, we can conclude that there is a causal relation between them, if all irrelevant factors have been ruled out. Potential threats to internal validity are listed and discussed. Many of these threats can be ruled out if participants in the experiment are randomly assigned to experimental treatments. Random assignment is the essence of “true” experiments. The chapter closes with an introduction to external validity, the extent to which findings from a research study can be generalized to other populations and settings. A wide range of potential research designs will be covered in the chapters that follow. They can be categorized roughly into either experimental or nonexperimental strategies. Experimental methods involve research in which participant actions are limited or in some way constrained by the controlled manipulation of variables determined by the researcher. Ideally, experiments should be characterized by random assignment, a procedure used to place participants, by chance alone, into the different conditions, which enables strong inferences of causality. Quasi-experimental methods are a variant of experimental methods lacking in random assignment, but as in experiments, participants are exposed to some form of variable manipulation imposed by the researcher. For example, we might be interested in the effects of National Merit Scholarships on college performance. Obviously, a researcher cannot randomly assign high school seniors to receive or not receive a Merit scholarship, because of ethical concerns and the fact that receipt of the award is determined by scores on a test. Quasi-experimental methods (Chapter 9) are an attempt to make inferences about possible variables responsible for the outcome differences between recipients and non-recipients of the scholarship, and to disentangle the effects of the scholarship from other variables such as intelligence or work ethic.

Pages: 20

This chapter is concerned with the reliability or replicability of measures. It is based on classical measurement theory, which assumes that measures consist of random error and true score, the replicable part of the measure. Multiple operationism, using multiple items or measures to assess the same construct, is stressed. Different measures are susceptible to different forms of error, and because error is random, different errors may cancel each other. The distinction between random and systematic error (bias) is discussed. The chapter explores temporal stability, the test-retest relation between respondents’ scores of repeated administrations, and the internal consistency of the measure’s items (Cronbach’s coefficient alpha) as important methods of testing reliability. The chapter also introduces contemporary measurement theory, which extends its classical antecedents by relying on latent factors as analogues to the true score components of the classical model. The role of latent factors and exploratory factor analyses is discussed. The chapter ends with a consideration of item response theory, an important feature of contemporary measurement theory. Measurement is a form of operationalization in which abstract concepts are translated into observable and quantifiable information or data. In social research, various measurement techniques are used to assess differences among the people or units being studied. The objects of measurement can be individuals reporting on themselves, observations of behaviors, characteristics of situations, or records of past events. We adopt a classic measurement theory orientation, also known as psychometric theory, which is focused on any observation, ability, state, trait, or behavior. The classic theory approach assumes a consistent, stable underlying state that reflects the construct we seek to quantify.

Pages: 16

This chapter is concerned with measurement validity, which refers to the correspondence between variations in the scores on a measure and variations among respondents on the underlying construct being assessed. Measurement validity presupposes measure reliability. It is focused on the adequacy of the measure representing the construct it is designed to assess. Types of measurement validity discussed in the chapter include, among others, face validity, content validity, criterion validity (concurrent validity and predictive validity), and convergent and discriminant validity. The latter two are evaluated in the multitrait-multimethod matrix (MTMM), which also is discussed and illustrated in the chapter. The central lesson of the MTMM, that the variance of measures consists of trait and method components is stressed, and this view is related back to classical measurement theory presented in Chapter 3. Developing measures that meet the criteria of reliability in terms of internal consistency, temporal stability, or inter-observer agreement satisfies a basic requirement for the operationalization phase of a scientific investigation. In the first chapter of this text, however, we made a distinction between the adequacy of an operation in terms of replicability and its adequacy as a manifestation of a theoretical construct. This distinction marks the difference between the reliability of a measuring device and its validity. Whereas reliability has to do with the consistency of measurement, the validation of operations relative to the hypothetical concepts under investigation (constructs or latent factors) is crucial from the standpoint of theory development.

Pages: 142

No chapter abstract available.

Pages: 18

This chapter first describes variations on the basic experimental design with a single independent variable. Variations include the pretest-posttest control group design, the posttest-only control group design, and the Solomon four-group design. It then moves to discussion of factorial designs that combine more than one independent variable with multiple experimental treatments. Factorial designs allow for testing interactions between two or more independent variables, evidenced if the effect of one independent variable on the dependent variable is altered or moderated by variations in the level(s) of other independent variable(s). Different forms of interaction effects are described and interpreted. The chapter then discusses the use and implementation of blocked designs, where experimentally manipulated independent variable(s) are combined with nonexperimental factor(s) (usually individual difference variables, such as sex or political affiliation). The chapter concludes with discussion of how to design and conduct experiments with repeated measures, where participants are exposed to more than one treatment condition of the independent variable. Chapter 2 introduced the concepts of internal and external validity (two important forms of validity to be considered in experiments) and provided some information on the basic structure of an experiment. Here, we expand on this basic structure to consider variations in the ways in which experiments can be designed, set up, and executed.

Pages: 23

This chapter introduces readers to the practical details of planning and conducting laboratory experiments in social science research, ranging from identifying and recruiting participants, determining the number of participants needed to prepare experimental materials, obtaining approval from institutional review boards (ethical review), and setting up the laboratory environment. Different types of experimental manipulations are described and illustrated, including environmental manipulations, stimulus manipulations, social manipulations, instructional manipulations, and priming as an experimental treatment. A critical feature of true experiments is random assignment of participants to the different treatment conditions of a study, and this chapter describes the basic mechanics of accomplishing random assignment. Emphasis is given to the need to create experimental realism so that participants are fully engaged with the laboratory situation and attending to instructions and the task at hand. The chapter concludes with discussion of simulations and analogue experiments designed to mimic real-world situations in the laboratory setting. In the preceding chapter, we discussed the design of experimental studies at a relatively abstract level to introduce basic principles of planning and constructing experiments. In the present chapter, we will become somewhat more concrete, and consider how to implement an experimental design in terms of the basic construction of a laboratory experiment, and the different forms of experimental manipulations that could be pursued. Along the way, we also reflect on other aspects of the experiment that, while not formal features of the design, can nevertheless have a powerful impact on the interpretation of a study. Although the chapter focuses on constructing laboratory experiments, some of the concepts also are applicable to conducting experimental research outside the laboratory.

Pages: 16

This chapter highlights generalizability, or the external validity of laboratory research, with special emphasis on the role of those who serve as study participants, the possibly distorting effects of experimenters, and ways to avoid them. The chapter argues that although generalizability is never established unequivocally, it can be improved. Over-dependence on college undergraduates as research participants is examined and found to be less a generalizability problem than sometimes thought. Crowdsourcing methods to recruit participants over the Internet have enlarged the participant pool, and research has shown it to be a reasonable addition to the scientific toolchest. Issues of participant roles and “types” (voluntary, involuntary, and nonvoluntary) are discussed, as is their impact on research validity. Experimenter expectancy, bias, and demand are considered, and their solutions include monitoring, blind procedures, and mechanization. The chapter concludes with description of three aspects of external validity: robustness (replicability), ecological validity (representativeness), and relevance. It often is thought that the use of laboratory experiments in research involves achieving internal validity at the cost of external validity, or generalization of results to the world outside the laboratory. This view betrays a misunderstanding of the logic and meaning of internal and external validity, but it is common, and because of this, critics within and outside the social sciences have argued that the experimental approach that structures so many of our research endeavors is inadequate or inappropriate for the study of social beings in natural settings.

Pages: 19

This chapter discusses features of experimental research conducted outside the laboratory. Such research may enhance construct, internal and external validities, and bolsters the recommended cyclical research process in which field research (including Internet research) adds value to laboratory findings, and vice versa. The value of field-lab replication research is noted, and exact and conceptual replication differences are discussed. The chapter examines operational features of conducting field research, including participant selection, experimental control, random assignment in the field, and outcome measurement. Uses of naturalistic observation and unobtrusive measures also are described. Internet-based research and conducting research online as well as reliability and validity considerations specific to this research context are examined. The chapter ends with a discussion on conducting online studies and their advantages and disadvantages. In the preceding chapters, we discussed ways experiments conducted in laboratory environments could be made as realistic, involving, and impactful as possible. Even when such methods are used, however, it still is desirable to move outside the laboratory into field contexts to extend the construct and internal and external validities of the results of any research program. As stated throughout this book, research should be viewed as a process in which initial observation of a phenomenon gives rise to theory development, which is followed by empirical research. Discrepancies between hypotheses and research results lead us back to further observations, to theory revision and refinement, and then to additional research, as shown in Figure 8.1. Neither laboratory nor field research is accorded primacy in this diagram. Each has its place, and each should be engaged only after appropriate observation and theory development. In this chapter, we support this view by considering ways experimental methods can be applied outside the laboratory, and how laboratory and field experiments can be used in conjunction with each other. In later chapters, we will take up nonexperimental research methods in field contexts.

Pages: 24

This chapter’s focus is on quasi-experiments, which are distinguished from true experiments by some form of experimental control but lack randomization of participants to conditions. This feature often precludes quasi-experiments from producing results, allowing for confident causal interpretation, but not always. Quasi-experiments are useful in evaluation and applied research contexts, where randomization may be neither feasible nor ethical. A common threat to interpretability in QE designs is the regression artifact, which is covered in some detail in this chapter with respect to its relation to measurement error, which circles back to the content of Chapter 3. Use of matching designs and propensity score analysis to offset lack of randomization is discussed, and the advantages of the latter are presented. The remainder of the chapter is devoted to consideration of different quasi-experimental designs, including nonequivalent groups designs, and ways of improving their interpretability, case-control designs, time-series designs (simple, multiple, and comparison), and regression discontinuity designs. In the preceding chapters we distinguished experimental from correlational research, in which the investigator’s role is largely that of an observer. In a real sense, all variables in correlational studies are dependent (or response) variables, and the researcher’s job is to assess their natural variation and discern the patterns and relations among them. In experiments, the researcher actively intervenes in the normal relational pattern by systematically controlling variation in the independent variable (or variables) to assess the causal impact of this variation on outcome variables. Controlled variation of the (presumed) causal variable and random assignment of subjects to research conditions are the necessary hallmarks of true experiments, which are the backbone of internally valid cause-effect analyses.

Pages: 20

The chapter offers an introduction to correlational research designs and analyses. In many areas of research, experimental control or manipulation over variables is impossible, unethical, or impractical. Thus, researchers may pursue correlational designs, which illustrate the association between two or more variables. The family of bivariate correlations (Pearson correlation, point-biserial correlation, phi correlation) and their applications are described. Multiple regression is an extension of bivariate correlations to permit one or more predictors on an outcome variable. The elements that constitute unstandardized and standardized regression equations are defined and explained. The four-step approach for evaluating mediational processes using multiple regression is described. Finally, the differences between statistical control used in correlational research and experimental control used in experimental research are described. Furthermore, the issues that ought to be pondered when conducting correlational research are emphasized: linearity of relations, restriction of range, underspecified models, and hidden third variables are factors that can produce misleading correlational results. Previous chapters of this book were focused on considerations relevant to the development and application of experimental designs. In these designs, participants’ exposure to a treatment or manipulation is controlled by the researcher. The major purpose of such control is to reduce or eliminate the plausibility of confounds or alternative explanations for effects on the dependent variable. In experimental research, a clear distinction is made between the independent (or causal) variable, which is controlled or manipulated by the experimenter, and the dependent (outcome) variable, which is allowed to vary freely. In many areas of research, however, experimental control over variables is impossible, unethical, or impractical.

Pages: 20

This chapter extends upon basic correlational designs and focuses on advanced multivariate and multivariable correlational approaches. Multilevel models permit the evaluation of statistical relations involving predictors of an outcome when data are based on a nested hierarchy of grouping units. Features of the intraclass correlation and two-level and three-level multilevel models are described. Structural equation models permit the estimation of multiple predictors and multiple outcomes in a single analysis and comprise two subtypes, path analysis and latent structural equation models, which are considered in detail. Terminology for understanding structural equation models is defined, including exogenous variables, endogenous variables, predictive errors, measurement errors, latent factors, and model fit indices. Cross-lagged panel models are pursued to test the temporal directionality of variables assessed longitudinally using correlational data. Growth curve models are applied to evaluate and identify temporal trends based on longitudinal data. Examples of growth curves involving linear, quadratic, and cubic trajectories are provided. The previous chapter described the concepts for interpreting and understanding basic correlational designs involving the Pearson correlation, simple regression, and multiple regression. This current chapter continues with the discussion of correlational designs by focusing on approaches that permit the estimation of more complex relations. The focus will be on common multivariable and multivariate correlational designs: multilevel models, structural equation models, cross-lagged panel models, and growth curve models. Multilevel models are an extension of multiple regression that permits the evaluation of predictors on an outcome when data are collected based on a nested hierarchy. Structural equation models are an extension of multiple regression that permits the estimation of multiple predictors and multiple outcomes in a single analysis. Structural equation modeling has two subtypes: path analysis, which does not use latent factors, and latent structural equation models, which involve latent factors. Cross-lagged panel models are used to unravel the direction of the temporal relations between variables using longitudinal data. Growth curve models are applied to evaluate and identify temporal trends based on longitudinal data.

Pages: 176

No chapter abstract available.

Pages: 27

This chapter deals with survey methodology, whose principal focus is external validity, which prompts a discussion of distinctions between respondent selection and assignment to conditions. Census and survey are differentiated, their respective advantages noted, and the derivation and interpretation of precision are explained. Simple random (survey) samples and sampling frames are discussed and illustrated with numeric examples. Other random sampling models are introduced, including systematic sampling and stratified sampling, cluster sampling, multistage sampling, and two-phase sampling. Panel surveys also are introduced, as are the uses of oversampling and stratification (proportionate and disproportionate). Non-random sampling methods also are discussed, including convenience, snowball, and quota sampling techniques, and contexts in which they can prove useful. The final section is concerned with sampling issues, including considerations of necessary sample size, sample restriction, nonresponse, and applying sampling weights to datasets. The chapter ends with a focus on survey methods such as random digit dialing, face-to-face interviewing, Internet surveys, and how to deal with issues of missing data. In fundamental ways, survey research is different from experimentally oriented methods that occupied our attention in the text’s initial chapters. In this chapter, the term survey refers to the process of polling, or surveying, an identified group of respondents on topics of interest – people’s attitudes, perceptions, intentions, behaviors, and so on. Surveys may involve mailed questionnaires, telephone interviews, face-to-face interviews, or even online pop-ups. Most survey research is intended to use obtained answers to survey items to estimate those of a larger group or population. Therefore, a central concern in surveys is concerned with the adequacy of a subset of individuals in representing those of the population from which they were drawn. Generally, survey researchers are not overly concerned with issues of internal validity (i.e., Is an experimental manipulation responsible for the obtained findings?), because surveys are primarily used in descriptive or nonexperimental research. Although some researchers use surveys as a vehicle for experimental studies, the central goal of most survey research is to provide sample estimates of population values that are as accurate as possible.

Pages: 26

This chapter discusses systematic observational research that usually occurs outside the laboratory, is naturally instigated, places few restrictions on respondents’ observable behaviors, and makes use of a reliable system for coding complex events. Such studies are prized for their naturalism, which involves observation of natural behavior and responses to natural events that occur in natural settings. Different levels of participant observation along with the problems they might introduce are presented. Less intrusive observation techniques such as the experience sampling method also are discussed. Data coding issues are discussed in terms of choice of structured versus unstructured methods and use of category systems. Choice of scoring unit is discussed, as is the advisability of constructing one’s own category system. Features of coding systems involve their intensity, category number, category utility, and degree of inference required. The final part of the chapter is concerned with issues of interrater reliability, including consideration of Cohen’s kappa and Fleiss’ kappa, and the intraclass correlation coefficient (ICC). In the previous section of this book, we were concerned with general strategies used in the design and implementation of different types of research studies. We now turn attention to methods of assessment and measurement used to collect the data that constitute the output of our research operations. We begin this section with a general consideration of systematic observational methods. This is appropriate, because all the assessment or measurement techniques to be presented require some form of systematic observation. Although all science is bound fundamentally to observation, the term “systematic observational methods” in our view has come to refer to a diverse set of techniques that are employed to study behavior that: (Usually) occurs outside the formal boundaries of the laboratory (Usually) is naturally instigated, i.e., does not make use of a controlled experimental treatment (Usually) places few restrictions on the allowable responses of the persons under observation (Usually) focuses on observable behaviors rather than internal, cognitive processes (Usually) entails a replicable system of codifying observed events This last requirement for observational data separates systematic observational techniques from unsystematic research approaches that may play a role in hypothesis generation or theory development, but do not constitute formal data collection.

Pages: 18

Content analysis is a method for classifying, evaluating, and studying recorded communications systematically and objectively. Information culled from media sources, such as newspapers, song lyrics, television shows, billboard advertisements, memes, online message forums, social media profiles, and websites, can be examined using content analysis. Distinctions are made involving qualitative versus quantitative content analysis and inductive versus deductive content analysis. The steps for conducting a content analysis are provided: identifying the textual, auditory, visual, or multimedia source or medium; creating the codebook of coding variables and operational definitions; conducting a pilot study to test the codebook; developing the coding variables (e.g., word, theme, spatial, and temporal); obtaining a random or non-random sample of units; applying established coding systems, software; and statistical analyses of results. Representative examples drawn from the published content analysis literature are described and focus on topics concerning “who,” “what,” “to whom,” “how,” and “with what effect.” Content analysis refers to a research method for classifying, evaluating, and studying recorded communications in a systematic and objective manner. Information, messages, and content are embedded in recorded communications that offer valuable research insights about culture, social norms and values, and human activity. Our human ancestors communicated through recorded mediums such as cave drawings, carvings on pottery and ceramics, and other ancient artifacts. These antiquities have been studied by modern researchers using content analysis approaches. Today, people communicate and exchange informational content through modern recorded mediums such as newspapers, photographs, videos, and the Internet. All these sources have been subjected to content analysis research. Information from media sources – images, photographs, newspapers, audio clips, song lyrics, television shows, billboard advertisements, memes, tweets, online message forums, social media profiles, and websites – can be examined using content analysis (Lac, 2016; Stemler, 2001; Neuendorf, 2011). Additionally, raw unstructured information collected through research methods such as observational information, interviews, and focus groups can be used in content analysis to make sense of the recorded data.

Pages: 22

The research interview, the focus of this chapter, involves some form of verbal communication between researcher and respondent to obtain data. Modes of administration include face-to-face, online, and telephone. When developing an interview, important decision points include question content, interview format, and interviewer behavior. The chapter includes considerable advice about item development, including using brief questions, avoiding subtle shading and complex language, and pilot testing interview schedules. Types of information usually available are various demographic data, reconstructions of past events, attitudes, intentions, and behavioral reports. Question specificity and constraint are discussed, as are techniques to deal with sensitive issues. Issues of interview structure and their constraints on closed- or open-ended questions are also treated. Conducting the interview requires obtaining entry and to establish and maintain rapport with interviewees. Informal interviewer tactics discussed are initial question sequence, selective use of leading questions, degree of directiveness, and informal behaviors, including verbal and nonverbal interviewer communication. The final part of the chapter is devoted to group interviews and focus groups. A research interview is a data collection method in which participants verbally communicate information about their behavior, thoughts, or feelings in response to questions verbally posed by an interviewer. Unlike most of the methods discussed in the two preceding chapters, interviews always involve some form of verbal interaction in which the investigator requests information from a respondent. This feature distinguishes the technique from self-administered questionnaire methods, where respondents may never interact personally with a researcher. The interactive process of the interview, and its dependence on verbal or linguistic responses, constitutes both its major strength and its major drawback as a method of scientific social research.

Pages: 19

Chapter 16 deals with the development of rating scales (multi-item measures amenable to reliability analysis) and questionnaires, which usually use one or two closed-ended or open-ended items to assess constructs. Thurstone’s method of equal-appearing intervals is outlined to provide a historical perspective on the scale development process. Likert’s method of summated ratings shortened the procedures needed to develop a rating scale and is widely used today. The chapter points out that internal consistency reliability, a critical indicator of measure quality, cannot be assessed on Thurstone scales, but it can be used with Likert scaling, and with Osgood’s semantic differential approach, which is also reviewed in the chapter, as well as Guttman’s scalogram measurement model. Methods of enhancing scale reliability and the use of exploratory and confirmatory factor analysis are discussed. Details of the scale development process, including number and extremity of response options, item reversals, and use of unipolar or bipolar scales, occupy the final section of the chapter. Studying people’s beliefs, attitudes, values, and personalities is a central research preoccupation of the social sciences. Typically, we use questionnaires or scales to measure these internal states or dispositions. Such measures rely on self-report, by asking respondents to provide answers to a set of questions or scale items that inquire about their personal thoughts, feelings, or behaviors. Using these measures requires that we consider variations in responses among respondents as meaningful, and not attributable to mere measurement error. In theory, all participants are expected to interpret the questions or items used in a scale or questionnaire identically. That is, a given questionnaire or scale item is assumed to mean the same thing to all participants, and differences in their responses to these hypothetically identical items are assumed to reflect real differences in their underlying dispositions.

Pages: 15

Chapter 17 discusses the construction of psychophysical scales, which include measures derived from pair comparison and rank order scaling. The chapter points out that while valuable in and of themselves these methods serve as a foundation for today’s more common techniques of unidimensional and multidimensional scaling. Stimulus scaling is shown to bring the measurement logic of physics to the social laboratory, where unlike physics, there exist no obvious physical metrics for their measures (e.g., of beauty, taste, and preferences). How methods of scaling stimuli represent the combined judgments of large groups of respondents with high accuracy is discussed. The chapter emphasizes the distinction between stimulus scaling and the more usual scales of individual differences, as it is concerned with judgments of different stimuli, not with the persons who made the judgments. Methods of pair comparison and rank order scaling and the strengths of each are presented. The process of combining respondents’ judgments to provide an interval scale are explained. Occupying the final chapter section is discussion of the role of stimulus scaling in development of multidimensional scaling. Developing and using scales of high utility, generality, and psychometric quality (i.e., of high reliability and validity) is a common part of the social researcher’s job description. As discussed in Chapter 16, questionnaires and rating scales are constructed to measure differences among individuals. For example, a scale of attitudes toward nuclear energy, assessed with multiple items, has as its purpose the classification of individuals along a continuum, so that those with the highest scores are defined as most in favor of adopting nuclear energy, and those with the lowest scores the least favorable. Those scoring in the middle are considered intermediate on their attitudes toward this energy source. In this form of scaling, the items are used to compute a summary score for the purpose of arranging people at some point on the scale: All items are designed to tap the same construct; that is, the items are assumed to differ only in terms of measurement error. So, in our attitudes toward nuclear energy scale, we assume that the questions (items) we pose to respondents, and that are used to create our summary score, all tap the same underlying construct (in this case, attitudes toward nuclear energy). Measures of this variety are called tests of differences among individuals, or more commonly, tests of individual differences. Individual difference assessments are the most common form of measures used in contemporary social research.

Pages: 26

The chapter describes modern methods that have been developed for using indirect techniques to assess implicit beliefs and affect (feelings and emotions) without relying on conscious self-report from research participants. Many of these techniques derive from experimental research on social cognition and rely on assessing elements of basic information processing such as attention and memory. Implicit measures involving attention include visual gaze tracking, interference (Stroop effect) as an indicator of automatic attention, and processing time as an indicator of interest and cognitive effort. Memory measures include recognition and recall as assessments of what individuals have encoded about past experiences or stimulus presentations. The chapter also discusses various priming methods, including subliminal priming, that are used to activate or engage specific mental concepts in participants’ minds without conscious awareness of the influence. It also discusses how reaction time and response interference can be used to assess implicit associations that are automatically activated by specific cues. These methods include the implicit association test for assessing category stereotypes and automatic evaluations. Finally, the chapter describes various advances in the use of physiological and neurological measures to assess cognitive and affective processes. As in any field of science, the use of the scientific method in social research requires that our assertions be based on observable phenomena (see Chapter 1). Whatever inferences we wish to make about the causes and processes underlying social behavior must first be grounded in observations that can be recorded and replicated. Research that employs the experimental method involves manipulating some aspect of the physical or social environment, and then observing and recording some type of response of participants as the dependent variable. The observed response could be an overt behavior or action of some kind (e.g., stopping to give help, pressing a button to deliver an electric shock to another person, choosing a gift). More often, however, the observed response is a written or oral report from a participant of his or her reactions to the situation, a judgment, or a decision. Similarly, in survey research involving interviews or questionnaires, the observations consist of respondents’ self-reports of their behaviors, feelings, or beliefs. Because inner experiences – personal feelings and mental life – are not directly observable, social researchers often must rely on people’s introspective reports of their private experiences to acquire data that are amenable to measurement and quantification.

Pages: 21

The focus of this chapter is on collecting and analyzing data describing social pairs and groups of individuals. Dyadic variables are characteristics of the relationship between two persons and their joint outcomes, whereas group variables refer to characteristics of a set of three or more interacting persons and their collective products. The chapter describes the basic elements of a dyadic interaction, measures of reciprocity between members of a dyad, and methods for analyzing the contributions of each person and their relationship to their joint outcomes. The chapter then goes on to describe designs for studying group structures, including round robin designs, sociometric methods, social network analysis, and nested group designs. The final section of the chapter deals with methods for assessing group processes, including interaction process analysis, group cognition, and computer-mediated group interaction. The preceding chapters on measurement and social cognition dealt with methods for assessing characteristics and behaviors of individual persons. Monadic variables are characteristics of single persons (e.g., age). A person’s attitude toward abortion, for instance, is considered monadic because it refers only to that specific individual’s own attitude. In many areas of social science, however, we are concerned with persons who interact in dyads (pairs) or groups. In this case, we are not assessing properties of the individuals separately, but the nature, relationship, structure, process, or outcomes of the dyad or group as a unit. Dyadic variables are characteristics of the relationship between two persons or their combined outcomes (e.g., level of warmth of their interaction). Group variables refer to characteristics of a set of three or more persons (e.g., total number of widgets created by a workteam). In this chapter, we consider the assessment of group or dyadic variables. Measures of behaviors or attitudes of interacting persons are a special case because the assessments taken from each of the group members are interdependent, insofar as each group member’s score is expected to be related to or influenced by that of the other member(s).

Pages: 39

No chapter abstract available.

Pages: 20

The chapter focuses on replication and research syntheses, with a special emphasis on how to conduct a meta-analytic review. Meta-analysis is a quantitative approach for summarizing and interpreting primary studies that have addressed the same research question. The advantages of performing a meta-analysis compared to a narrative review include clearly stated inclusion and exclusion criteria and the use of statistical metrics that objectively combine findings across primary studies. The stages used in conducting a meta-analytic review are described: understanding the literature, defining study inclusion and exclusion criteria, searching for primary studies, computing the effect size for each primary study, deciding on the method for calculating the summary effect (unweighted, fixed-effect, or random-effects model), and interpreting the meta-analysis results. The proper interpretation of a research synthesis requires evaluating the heterogeneity for the distribution of effect sizes, examining the role of meta-moderators, scrutinizing the extent of publication bias using tests such as the funnel plot, and understanding the magnitude of the summary effect. All sciences progress through the gradual accretion of knowledge. Although surely inspired, Einstein’s classic formula relating energy to a function of mass and velocity was not developed out of thin air. His creative insight was based on a synthesis and extension of the available knowledge of the time (Clark, 1971; Kuhn, 1970). Over the years, researchers have depended on a similar process of integration and consolidation of the prior research literature to develop new insights, which could then inform the development of additional knowledge. A primary study is an original study reporting on the results of data analysis collected from a sample of participants (e.g., journal article, research report, thesis, dissertation). It serves as the principal unit of empirical knowledge. All the chapters covered to this point have described strategies for designing and conducting primary studies. However, we also have emphasized the importance of programmatic research and multiple studies directed at testing the same conceptual hypotheses. The current chapter is devoted to methods for integrating the findings from such cumulative research efforts.

Pages: 17

This chapter provides a comprehensive consideration of ethical issues in research with human participants and considers the social responsibility that social researchers have to their participants, to science, and to society. Research ethics apply to both the conduct of research and the reporting and dissemination of research findings. With respect to research practices, laboratory experiments may involve the use of deception and the chapter discusses how a cost-benefit analysis can be applied to evaluate whether some degree of deception of participants about the purposes of the research may be justified, and discusses the necessity of post-session debriefing to assure that participants are fully informed at the end of the study. Both laboratory and field research entail protection of participants’ privacy and confidentiality, and these issues are discussed next in the chapter. This is followed by discussion of the role of ethics codes and regulatory agencies in establishing and maintaining ethical practices in research. With respect to data reporting, the chapter covers recent issues in the replicability of social research findings and the move toward open science and transparency in reporting methods and results of empirical studies. Finally, the chapter discusses whether and to what extent researchers are responsible for how their findings are interpreted and used to impact the society at large. The social sciences share with all scientific endeavors the need to balance scientific zeal with other values that derive from the social context in which all scientific work takes place. To some extent, the scientific ideal of objectivity inevitably conflicts with humanistic values, and all scientists must at some time come to grips with this conflict. The issue, however, is particularly acute for social scientists because the focus of their research is the behavior of other human beings, and thus not only the goals of research but the very process of data collection is subject to value conflicts. In this chapter we consider first those ethical issues that are related specifically to research with human respondents; we then focus on the more general ethical issues of concern to all scientists in their choice of research problem and strategy of data reporting.