Social Psychology

Student Learning Program

Chapter 2: Testing theories: From theory to research (pp. 2947)

Ask Yourself?

In this topic

  1. Construct Validity and Approaches to Measurement (pp. 3033)
    1. Threats to construct validity
    2. Ensuring construct validity
  2. Internal Validity and Types of Research Design (pp. 3336)
    1. Threats to internal validity
    2. Ensuring internal validity
    3. Experimental versus nonexperimental research designs
  3. External Validity and Research Populations and Settings (pp. 3643)
    1. Generalizing to versus generalizing across people and places
    2. External validity and research participants
    3. Cultures and external validity
    4. External validity and laboratory research
    5. External validity and nonlaboratory research
    6. Ensuring external validity
  4. Evaluating Theories: The Bottom Line (pp. 4347)
    1. The importance of replication
    2. Competition with other theories
    3. Getting the bias out
Construct Validity and Approaches to Measurement

Construct validity: If a research study has strong construct validity, both the independent and the dependent variables of the study, or the events that occur in the study, should correspond to/reflect the intended theoretical constructs.

Independent variable: causal factors; factors that are manipulated by the researcher.

Dependent variable: variables representing effects (they depend on the causal/independent variable).

There are two ways of ensuring construct validity has been met:

Threats to construct validity

It is often difficult to ensure construct validity, as the variables being tested might be affected by other influences/constructs besides the one being measured.

One common threat to construct validity is the social desirability response bias, which means that people will answer questions, or act, in ways that make them look good.

Social psychologists must always be on guard against biases to construct validity, especially when they are researching attitudes or behaviors that are not socially approved of.

Ensuring construct validity

The two main ways of ensuring construct validity are (1) to use the best measures for the purpose, and (2) to use multiple measures.

(1) Using the best measure for the purpose: Three different types of measure categories can be used in research:

(2) Using multiple measures: The different measures mentioned above each have different strengths, so it is best to use multiple measures to ensure construct validity.

Internal Validity and Types of Research Design

Internal validity has to do with knowing whether the measured changes in one variable (the independent variable) actually caused the changes in another variable (the dependent variable). Research only has high internal validity if the researcher can confidently conclude that a change in the independent variable caused a change in the dependent variable.

Threats to internal validity

The major threat to internal validity is that factors or variables other than the independent variable are causing the observed changes in the dependent variable.

Nonexperimental research designs are designs in which researchers simply measure the independent variable and the dependent variable. These are very vulnerable to threats to internal validity, as it is harder to state the actual cause of the observed difference/change in the dependent variable.

Ensuring internal validity

Internal validity can be ensured by using an experimental research design. In an experimental research design: (1) participants are randomly assigned to experimental groups, and (2) the independent variable is manipulated (intentionally), rather than simply being measured.

Random assignment gives every participant in the experiment an equal chance of ending up in either experimental group.

The independent variable is intentionally manipulated, so that the different groups are exposed to different treatments. Because the groups are equal before the manipulation, as a result of random assignment, the observed changes/differences in the groups are assumed to be due to the independent variable.

Experimental versus nonexperimental research designs

Experimental research designs offer higher internal validity, and permit stronger tests of causal relations between constructs, because random assignment and manipulation of the independent variable allow the researcher to rule out alternative explanations for the observed differences between groups.

Nonexperimental research designs are used: (1) because some important variables cannot be intentionally manipulated; (2) for ethical reasons, because some variables shouldn't be manipulated; and (3) because sometimes experimental manipulations are not as strong as, or cannot replicate, the effects found in everyday life.

External Validity and Research Populations and Settings

External validity is the extent to which the results of a research study can be generalized to other people, settings, and times. External validity can be ensured by conducting repeated tests of a theory in different populations or settings, or by using different participants.

Generalizing to versus generalizing across people and places

Applied research is an example of research in which the researcher's primary interest lies in a specific target population and setting.

However, the research goal in social psychology is usually a broader sort of generalization across factors. The key question in these studies is: What aspects of the research conclusions will successfully generalize across other settings and populations?

It is not crucial whether the study's specific findings can be generalized.

The generalization that is expected lies at a higher theoretical level: that of causal relationships between abstract constructs. So researchers hope that the underlying principles of their research will generalize across various populations and settings.

External validity and research participants

The first major threat to external validity is the type of people studied in the research. Usually college students are used in psychological research; the specific findings of this research cannot then be generalized to other populations.

Generalizing about causal explanations is more justifiable than generalizing about specific thoughts and behaviors.

The most basic social-psychological processes are most likely to operate in similar ways across different groups of people.

The only way to ensure that research conclusions can be generalized to other settings and populations is by repeating the research with different types of participants at different times.

Cultures and external validity

Different cultures exist across the world; this limits the generalizability of social-psychological research, as social psychology remains largely a Western phenomenon.

Since culture strongly dictates thoughts and behavior, research on these specific things cannot be easily generalized to other populations. However, the process by which thoughts and actions are developed and used is more generalizable.

External validity and laboratory research

A third factor that threatens/limits the generalizability of research is the setting, or place, in which the research is conducted.

Most social-psychological research is conducted in laboratories, as these are the easiest places in which to maintain high internal validity.

The downside of laboratory research includes: (1) short time span, (2) the artificial quality of the lab, and (3) participants pay closer attention to what is happening around them during a short laboratory session than in their everyday lives, because they expect it will be important, and so different attention processes are operating.

When different processes are operating inside and outside the laboratory, theories and research cannot easily be generalized from one setting (the lab) to another setting (people's everyday lives).

An important disadvantage of laboratory settings for research includes demand characteristics: participants' behaviors are influenced by their perceptions of the research purpose, or what the researcher wants or expects from them. This is also a threat to construct validity.

Case study: Demand characteristics

External validity and nonlaboratory research

Field research takes place outside the laboratory, and can often complement laboratory research. It frequently has good construct validity, and has a nonexperimental research design.

Field research does not always lead to higher external validity; research conducted in one setting may often not generalize to another setting.

Ensuring external validity

If the goal of the research is to generalize to a specific target population and setting, the participants and the setting must be representative of the target.

If the goal of the research is to generalize across different populations, settings, and times, then external validity can be ensured by repeating the research in multiple settings, and with multiple populations (and cultures).

Evaluating Theories: The Bottom Line

Theories only become generally accepted as better than other competing theories if multiple studies have produced the same results.

Each type of validity is crucial for research to provide support for a theory.

The importance of replication

Theories cannot be supported by the results of a single study alone, so researchers need to carry out a replication of (or reproduce) their results in other studies, using different manipulations, in different settings, and with different populations.

Case study: The importance of replication

A meta-analysis is used to compare the results of many different studies testing one theory, and it allows researchers to examine the generalizability of results across these different replications.

Research activity: The importance of replication

Competition with other theories

A theory is never said to be "proven"; instead it is said to be "generally accepted," which means that a social consensus exists among the community of scientists.

Often, theories that seem to compete to explain results of different research studies turn out to complement each other. Competing theories might actually work best together to provide a better understanding of the research findings.

Getting the bias out

Research methods are designed to exclude as many biases and errors as possible from all three forms of validity.

The three forms of validity include:

Do not assume that scientists can always exclude all bias and error from their research methods. Scientists are human, and are often affected by their own cultural assumptions. There are no guarantees that scientists will always be objective, but the research techniques devised are the best ways to eliminate bias and error.

So what does this mean?

In order to test their theories, social psychologists must design research that is valid, according to three types of validity. Construct validity, which is the extent to which the independent variable and the dependent variable actually measure the intended theoretical construct, is often threatened by the social desirability response bias, in which people act in ways that they think are socially desirable. Internal validity ensures that changes in the independent variable actually caused the changes in the dependent variable.

Nonexperimental research designs often lack internal validity because the variables are merely measured (not manipulated); however, experimental research designs using random assignment of participants to groups, and manipulations of variables, have high internal validity. External validity is ensured by conducting replications of studies. Theories become more generally accepted if multiple replicated studies show they are more valid than competing theories; but, often, rival theories end up complementing each other to provide the best explanation for research findings.

Next topic

The role of ethics and values in research

In this chapter

  1. Chapter 2 introduction
  2. Research questions and the role of theory
  3. Testing theories: From theory to research
  4. The role of ethics and values in research
  5. Chapter overview (PDF)
  6. Fill-in-the-blanks
  7. Multiple-choice questions