Social Psychology Research Methods Research Paper

Custom Writing Services

View sample Social Psychology Research Methods Research Paper. Browse other social sciences research paper examples and check the list of research paper topics for more inspiration. If you need a religion research paper written according to all the academic standards, you can always turn to our experienced writers for help. This is how your paper can get an A! Feel free to contact our research paper writing service for professional assistance. We offer high-quality assignments for reasonable rates.

Social psychological research methods involve methods in use throughout the behavioral and social sciences. The goal of social psychology is to understand human behavior as it is affected by and affects social contexts. Accordingly, social psychologists attempt to measure and explain characteristics of individuals as social actors and characteristics of the social environments in which they act. Importantly, they also attempt to understand the interactive effects of both individual and environmental variables. Social psychological methods are fundamentally empirical methods, using observation and data to guide theoretical development. Ultimately, however, those methods are simply tools in the development of theoretical insights into the nature of human social behavior.



1. Research Validities

Methods of research in social psychology are useful only in so far as they permit one to critically evaluate and develop theoretical claims. Accordingly, a set of research validities that represent criteria by which any given piece of research, and any program of research, can be evaluated is defined. Ultimately research methods that permit the accomplishment of these validities permit the gathering of data that will be maximally informative for theory construction. Importantly, however, as will be discussed, there are no perfect methods that simultaneously maximize all of the research validities. Ultimately, convergence on theoretical insights is only gained through the use of multiple methods of research, triangulating across individual studies and individual methods, to maximize the validity of an entire program of research (for general discussions of research validities, see Brewer 2000, Cook and Campbell 1979).

1.1 Construct Validity

Theoretical claims in social psychology typically involve statements about abstract and poorly defined constructs. So, for instance, small group research in social psychology has been interested in the role of alternative leadership styles on the successful functioning of the group. Researchers in intergroup relations have explored how levels of group cohesion affect the derogation of outgroups. Research in person perception has explored how information that is inconsistent with prior beliefs about someone is cognitively integrated into an overall impression of what that person is like. In each of these examples, the phenomena that social psychologists are interested in understanding are quite vague constructs, amenable to multiple alternative definitions.

To gather data that are informative for theory development, these constructs must be measured. The success of that measurement determines the construct validity of any piece of research. In other words, construct validity concerns the extent to which the constructs of theoretical interest have been successfully captured by the variables actually measured in research. The fundamental tenant of construct validity is that multiple operationalizations of any construct are mandatory. Any one measure is only an imperfect indicator of the construct it is supposed to measure. Confidence in measurement is produced only if multiple operationalizations tend to converge or triangulate upon the same answer.

1.2 Internal Validity

Theoretical claims in social psychology typically involve claims about the causal impact of constructs upon each other (e.g., social norms cause conformity, persuasive communications cause attitude change). Once the constructs have been operationalized, then research that is informative for theory construction and evaluation is research in which one can argue not only that one variable is associated with another, but also that one variable (the independent variable) is causally responsible for variation in another (the dependent variable). Research that is more internally valid is research where causal claims are more defensible.

Internal validity is affected by the nature of the research design that is employed. Ultimately, confidence in causal claims is only possible in the case of what are called randomized experimental designs, in which the units of observation (typically persons) have been randomly assigned to levels of the independent variable. For instance, to examine whether group sizes (the independent variable) influences the quality of a group decision (the dependent variable), people would be randomly assigned to groups of different sizes. Random assignment means that the researcher typically must control and manipulate the independent variable. Random assignment is importantly not the same thing as haphazard or uncontrolled assignment. It involves careful planning so that the probability of any unit occurring in a particular level of the independent variable is constant across all experimental units.

For many research questions in social psychology, randomized experiments are not feasible. Accordingly a variety of quasi-experimental designs can be employed. As is later described, these involve the collection of longitudinal data, without random assignment of units to levels of the independent variable. Finally, from the point of view of internal validity, cross-sectional studies (also known as correlational) studies are the least valid, although data from such studies can be quite informative for other reasons. Additionally, the absence of covariation between the independent and dependent variables in a correlational study would certainly lead one to doubt the existence of a causal relationship between the two.

1.3 External Validity

Data are always collected from specific units in specific settings. Obviously, theoretical conclusions require generalization from those units and settings to broader populations of theoretical interest. The external validity of any given piece of research indexes the extent to which this sort of generalization is feasible.

In general, generalization from specific samples to populations of interest is only possible if probability sampling methods have been used to select the units and settings from which to collect data. Although widely used in social psychology, nonprobabilitysampling methods (e.g., gathering data from participants who are given experimental credit in introductory psychology classes) pose serious threats to external validity.

1.4 Tradeoffs Among Research Validities

It is exceedingly likely that there are tradeoffs among these validities in any particular study. For instance, to effectively manipulate an independent variable in order to maximize internal validity by the use of a randomized experimental design, the construct validity of the manipulated independent variable may be seriously compromised. Historically, much of the research reported in the leading social psychological journals has relied upon randomized experiments conducted in laboratory settings with participants recruited from introductory psychology pools. Implicitly, such research has assigned internal validity paramount importance while compromising the other two validities. Although tradeoffs among the validities may be inevitable in any particular study or data collection effort, it is important to employ a variety of different research strategies and designs across a line of research, moving from the laboratory to the field, from experimental designs to correlational ones, from manipulated independent variables to carefully measured ones. In this way, across a series of studies, it becomes possible to maximize all three sorts of research validities.

2. Research Designs

As just defined, randomized experimental designs, in wide use in social psychology, involve the random assignment of units to levels of the independent variables. Most frequently, multiple independent variables are included in crossed experimental designs, so that their separate, as well as interactive effects, can be assessed simultaneously. The interactive effects of multiple independent variables have been of great theoretical importance in social psychology, as they suggest, for instance that the impact of situational or environmental factors may depend on characteristics of the units being assessed (i.e., person by situation interactions).

If the effects of independent variables are considered to be short-lived, then a within-subjects research design may be employed to maximize power. In other words, each experimental unit occurs in each level of the independent variable (or variables). Random assignment in this case means that the order of the levels of the independent variables to which units are exposed is counterbalanced, with units randomly assigned to the different possible orders.

Many experimental designs in social psychology involve some mixture of between-subject and within subject independent variables, and these are typically crossed so that their interactions can be assessed. The typical analytic approach to such data involves the use of analysis of variance.

Quasi-experimental designs involve the longitudinal assessment of units on the dependent measure(s) both before and after exposure to levels of the independent variable. The classic quasi-experimental design is the non-equivalent control group design in which units are assessed on the dependent variable twice, once as a pretest and once as a post-test. Some units are exposed to the experimental treatment and others to the control condition (i.e., the two levels of the independent variable), but the assignment process is typically an uncontrolled one in which the assignment rule, rather than being random, is unknown. Analyses typically involve the analysis of differences between conditions on the post-test, attempting to remove or adjust for any pre-existing differences on the pretest (through an analysis of covariance or some other adjustment procedure).

A more elaborate version of the nonequivalent control group design uses the same basic approach except that multiple pretests and multiple post-tests are assessed. Such designs permit greater confidence about treatment effects because differences in units in spontaneous growth can be modeled, and these differences can then be ruled out as competing explanations for any condition differences.

A particularly strong version of this quasi-experimental design would be one in which the treatment variable is turned on or off at different points along the time sequence during which the dependent variable is being assessed. There may be multiple measures of the dependent variable over time and at some time during that sequence the treatment gets turned on, but the time point at which that occurs varies between units (with some units serving as control units where the treatment never gets turned on).

The regression—discontinuity quasi-experimental design is one in which a known, rather than a random assignment rule is used to assign units to either the treatment or control conditions. Although seldom used in basic social psychology, this design has received attention in more applied research endeavors (e.g., deciding who will receive a new and scarce medical treatment based on the severity of symptoms manifested by patients).

And finally, as defined above, correlational designs involve the cross-sectional measurement of both the independent and dependent variables. (See Cook and Campbell 1979, Judd and Kenny 1981, Smith 2000, and West et al. 2000 for discussions of alternative research designs and their implications.)

3. Measurement

As discussed under Sect. 1.1, measured variables in social psychology are assumed to be only imperfect indicators of the constructs that are of theoretical interest. Classic measurement theory has focused on random errors of measurement, assuming that variation in measured variables derives from two sources: true score variation and random error variation. According to this classic approach, the reliability of a variable is defined as the proportion of variation in the variable that is true score variation. A variety of procedures for assessing reliability have been developed from this classic model, the most widely used being coefficient alpha which assesses the degree to which a set of items share true score variation.

More recent approaches to measurement take a more fine-grain approach, assuming those errors of measurement can involve systematic as well as random errors of measurement. Accordingly variation in any variable can be decomposed into three sources: variation due to the construct of interest; variation due to other constructs that represent sources of systematic error; and random error variation. Convergent validity assesses the contribution of the construct of interest; discriminant validity assesses the contribution of systematic sources of error variation; and reliability assesses the contribution of random errors of measurement. Procedures for disentangling these sources of error variation have been developed by collecting data according to the multitrait-multimethod matrix, where multiple measures of multiple constructs are assessed, with various measures sharing systematic sources of error, particularly due to measurement methods (Campbell and Fiske 1959).

In addition to these approaches to measurement, based on the presumption that measurement adequacy is best assessed by patterns of covariation across multiple alternative measures, some have argued for a more axiomatic approach to measurement (Dawes and Smith 1985). This approach is based on the presumption that regularities of individual responses should follow certain axiomatic principles deduced from the constructs that are assessed (e.g., transitivity). In general, this approach has not proved fruitful for empirical work since random error in responses often prohibits strong tests of these axiomatic principles. Nevertheless, a few examples, such as Guttman and Thurstone scaling, are of historical importance in the domain of attitude measurement (see Judd and McClelland 1998 for a general treatment of measurement in social psychology).

4. Analysis

As discussed above, historically the dominant empirical approach in social psychology has involved the use of experimental designs, with independent variables orthogonally manipulated either within or between participants. Accordingly, the dominant analytic approach to data has been the analysis of variance. As researchers have increasingly extended their bag of research tools to include field studies and other sorts of research designs, more general approaches to data analysis, based on multiple regression and the general linear model, have become more prevalent. Increasingly, researchers are employing multiple regression as a general data analytic approach, incorporating analysis of variance through the use of contrast-coded predictors (Cohen and Cohen 1983, Judd and McClelland 1989).

Longitudinal data structures as well as more sophisticated measurement models have resulted in increasing use of latent variable structural equation modeling procedures (Bollen 1989, Kline 1998). In general these procedures allow researchers to examine and test models of data that incorporate both systematic and random measurement errors, estimating the linear relations among latent constructs that are each assessed with multiple indicators. Although sample size requirements and other restrictions have placed limits on the use of these procedures in social psychology, the general data analytic approach that they represent is particularly promising.

Data analysis strategies that allow for correlated errors of measurement are particularly promising in the subfield of social psychology devoted to the study of dyadic and small group interaction. Traditional inferential procedures normally assume that observations are independent of each other and this assumption is nearly always violated in dyadic and small group research where individual persons interact and are jointly influenced by situational factors. Accordingly, considerable work has been devoted in recent years to procedures for the analysis of dependent data structures. The most general of these approaches involve multilevel or hierarchical linear models, where individual observations are nested under factors whose levels are themselves treated as random observations. Traditional analysis of variance and regression procedures do not allow for this sort of hierarchical data structure with random factors (see Bryk and Raudenbush 1992, Kenny et al. 1998).

Beyond the domain of small group and dyadic research, these hierarchical models are also particularly useful in longitudinal data structures, where individual observations are repeatedly taken from participants over time and one is interested in modeling growth or change in response to various situational factors that also vary across time. Data from various quasi-experimental designs are particularly likely to exhibit this sort of structure.

5. Conclusion

Although experimental laboratory research procedures have been the dominant approach to research in social psychology, increasingly researchers are using a wide range of research designs, measurement procedures, and data analytic techniques to address the multifaceted research questions of social psychology. Certainly there will remain an important role for laboratory experimental work in social psychology. However, the wide range of methods now used throughout social psychology offers tremendous potential for addressing the diversity of theoretical questions addressed by social psychology in very rich ways.


  1. Bollen K A 1989 Structural Equations with Latent Variables. Wiley, New York
  2. Bryk A S, Raudenbush S W 1992 Hierarchical Linear Models: Applications and Data Analysis Methods. Sage, Newbury Park, CA
  3. Brewer M B 2000 Research design and issues of validity. In: Reis H T, Judd C M (eds.) Handbook of Research Methods in Social and Personality Psychology. Cambridge University Press, New York
  4. Campbell D T, Fiske D W 1959 Convergent and discriminant validition by the multitrait-multimethod matrix. Psychological Bulletin 56: 81–105
  5. Cohen J, Cohen P 1983 Applied Multiple Regression Correlation for the Behavioral Sciences, 2nd edn. Erlbaum, Hillsdale, NJ
  6. Cook T D, Campbell D T 1979 Quasi-experimentation: Design and Analysis Issues for Field Settings. Rand McNally College Pub. Co., Chicago
  7. Dawes R M, Smith T L 1985 Attitude and opinion measurement. In: Lindzey G, Aronson E (eds.) The Handbook of Social Psychology, 3rd edn. Random House, New York
  8. Judd C M, Kenny D A 1981 Estimating the Effects of Social Interventions. Cambridge University Press, New York
  9. Judd C M, McClelland G H 1989 Data analysis: A ModelComparison Approach. Harcourt, Brace, Jovanovich, San Diego, CA
  10. Judd C M, McClelland G H 1998 Measurement. In: Gilbert D T, Fiske S T, Lindzey G (eds.) The Handbook of Social Psychology, 4th edn. McGraw-Hill, Boston
  11. Kenny D A, Kashy D A, Bolger N 1998 Data analysis in social psychology. In: Gilbert D T, Fiske S T, Lindzey G (eds.) The Handbook of Social Psychology, 4th edn. McGraw-Hill, Boston, MA
  12. Kline R B 1998 Principles and Practice of Structural Equation Modeling. Guilford, New York
  13. Smith E R 2000 Research design. In: Reis H T, Judd C M (eds.) Handbook of Research Methods in Social and Personality Psychology. Cambridge University Press, New York
  14. West S G, Biesanz J C, Pitts S C 2000 Causal inference and generalization in field settings: Experimental and quasi-experimental designs. In: Reis H T, Judd C M (eds.) Handbook of Research Methods in Social and Personality Psychology. Cambridge University Press, New York
Sociological Social Psychology Research Paper
Social Psychology Research Paper


Always on-time


100% Confidentiality
Special offer! Get discount 10% for the first order. Promo code: cd1a428655