How do you work with research questions, validity and ethics? - Chapter 2

Research questions

The formulation and execution of a research within social psychology is (almost) always done with curiosity about the behavior of the human being. This curiosity usually comes from a certain social problem. Social psychologists search for generally applicable principles that can explain certain behavior. They therefore do not try to understand specific events or individuals, but look for general, useful and reliable theories about behavior. Such a scientific theory is a statement that must meet three requirements globally:

The theory is about constructs. Constructs are concepts that are abstract (such as "fear" or "aggression") and therefore can not be directly observed / measured.

It describes causal relationships between constructs, which means that as soon as a change takes place in one construct, the other construct is influenced by this. When we can make this connection, we can take practical steps called interventions, with which behavior can be changed or social problems can be solved.

The theory is general, so that it applies to different individuals at all times.

Do research from a theory

To validate a theory, there must be construct validity. Construct validity is the extent to which the independent and dependent variables in the research correspond to the theoretical constructs to be measured, or: do you really measure what you want to measure? The independent variables are the variables that are manipulated by the researcher(s); the dependent variables are the variables that depend on the manipulation and thus represent the effect of the manipulation. It is important that there is a considerable degree of construct validity because otherwise the results say nothing about the theoretical construct. Two things are important for this form of validity:

• The variables must be related to the theoretical constructs.

• The variables must not be associated with other constructs.

Nevertheless, it can be difficult to guarantee construct validity. There are many threats that jeopardize construct validity. A frequently occurring threat is the socially desirable answer bias. When this occurs, people behave in such a way that according to them is good to others. In this way, the observed variable is influenced by an unwanted factor, especially if researchers are looking for theoretical constructs that are generally seen as not socially desirable. Ensuring construct validity can be done in two ways: The use of the best type of measurement and the use of multiple measurements (as soon as different tests show the same results, a researcher can assume that the right construct has been measured).

The types of measurements that are frequently used in research are:

• Self-report measurements: a test subject indicates his or her behavior, thoughts and feelings. Such measurements are very sensitive to socially desirable answers, but represent intentions and attitudes very well.

• Observed measurements: measurements in which someone observes and maps the behavior of the subjects directly. These measurements often have good construct validity, unless the subjects know that they are being observed. (Then the socially desirable answer bias can play a role.)

• Performance measurements: test subjects are asked to perform a certain task as well as possible. This form of measurement often ensures that socially desirable answers are not applicable, because one is focused on the proper performance of the task.

• Physiological measurements: measurements based on a certain physiological process, such as heart rate or skin conduction (measuring the degree of sweating). Examples are the EEG (very accurately measures the electrical signals on the skull when specific neural activities take place) and the fMRI (indirectly measures the activation level of specific brain regions).

Internal validity

On the basis of internal validity it can be determined whether changes in the independent variable actually bring about changes in the dependent variable. There is a high internal validity if it can be reliably concluded that the change in the independent variable has brought about a change in the dependent variable. The question of whether such a conclusion can be drawn depends on the research design: a plan that specifies how participants will be selected and treated. Some types of designs offer a higher level of internal validity than other designs.

A design with a lack of internal validity is a non-experimental research design, also known as a correlational design. In such a case, both the independent and the dependent variables are only measured (not affected). This threatens the internal validity, because the research findings can be explained by other, unknown factors. With such designs, construct validity is high when the variables can be studied in their natural context. Internal validity can be guaranteed by setting up a good experimental research design. In such a design the independent variable is manipulated instead of simply measured. Here, the different groups of subjects can receive different treatments, as a result of which the effect can be attributed to the independent variable. The subjects are also randomly assigned to the conditions ('random assignment'), so that each subject has an equal chance to get into one of the experimental conditions. Experimental studies generally have a higher internal validity and often reflect causal relationships between constructs. Their construct validity is high when the manipulations can be carried out properly, but low when there are practical or ethical objections when performing the necessary manipulations. Correlational designs are therefore often used when investigating variables that can not be manipulated due to ethical concerns. It may also be that the manipulations are not strong enough so that the consequences are insufficiently perceptible.

External validity

External validity has to do with the generalizability of the results from a research. When research is done, generalizability is sought about people, places and times in terms of constructs that have been measured. This ensures that results actually say something about a large group (as indicated in the first chapter). This external validity can be guaranteed if the studies are replicated over different populations and / or situations.

Threats to external validity:

The subjects themselves. For generalization it is important that representative subjects participate in the research. Often, psychological examinations are carried out at universities and a specific group of subjects participate in the study. This ensures that the findings of such a research are difficult to generalize to other groups of people. It is often assumed that the basis of most social psychological processes work in the same way for different groups of people.

The difference in cultures. It is very difficult to generalize research results obtained on the basis of a group of people with a particular culture to people with a different culture. Western cultures have an individualistic culture in which they define themselves in terms of uniqueness. Japan and many other Eastern countries have a collectivist culture, in which people see themselves in relation to others. Since thoughts, feelings and behaviors are largely influenced by culture, it is difficult to generalize. The underlying process by which these thoughts, behaviors and feelings are developed can be generalized.

The place where an investigation was conducted. It has been mentioned before that most social psychological research is conducted at universities. This means that the research is often done in a laboratory. This gives the advantage that there is a higher degree of internal validity, but this can stand in the way of external validity. The laboratory research is often conducted in a short time and the subjects are aware of the fact that they are participating in an investigation. This can cause them to think more about things than they would normally do or take into account what they think is being investigated and adjust their behavior accordingly. The latter is also called demand characteristics. As soon as it is not known whether processes work differently outside a laboratory, this gives problems with the generalizability. Field research, which takes place outside the laboratory, can contribute to increased external validity. However, there is also a greater chance of undesirable influence from the environment, which can lead to distorted results.

The key to obtaining external validity depends on the underlying research goal:

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

• If the goal is to generalize across people, places and times, the research must be repeated in multiple settings and with multiple populations (with different backgrounds).

Evaluate theories

Theories are only generally accepted if the results of multiple valid studies show that this theory is better than rival theories. Since no single research is sufficient to support a theory, replicating research is of crucial importance. Replication is the execution of new research in an attempt to provide (supporting) evidence for the same theoretically predicted relationships found in previous research. Researchers are obliged to report the procedures and methods in their reports in such a way that the research can be replicated. Comparing the results of several studies that jointly test a theory is done through a meta-analysis: a systematic technique for searching studies with the same subject and summarizing their results. It is very important to realize that theories are not proven, but that there is a consensus about a certain theory among scientists.

Sometimes theories seem competitive, but they are actually complementary and can therefore be merged. Social psychological research does not escape the influence of the personal convictions and cultural values ​​of the researchers. By making thorough use of research methods (and carefully looking at construct validity, internal validity and external validity), we have the most chance of biases and errors that pollute everyday (common-sense) thinking, kept out of the research.

Values ​​and ethics

Meanwhile, the scientific community agrees that subjects should be treated fairly in an investigation. Prior to the study, the subjects are often told what awaits them and the test subject volunteer to participate in the study (this is the so-called informed consent). Sometimes, however, it is important that the researchers do not give away everything about the research. In order to prevent biases such as the socially desirable answers bias and demand characteristics in research with sensitive subjects, researchers are sometimes forced to deceive the subjects. In these studies it may be that test subjects are not informed or are deliberately misled about certain aspects of the research. After completion of the research, they will be fully informed of the (misleading) procedures and objectives of the research, something that is done in the so-called debriefing. In a debriefing, which takes place as soon as possible after the investigation, people can also ask questions about the research and the researcher can observe and address any negative effects of the research.

The research ethics is supervised by various committees. As soon as it is not ethical to carry out an investigation, these committees can stop an investigation or ask for adjustment. Nowadays it is mandatory that all research proposals first go through an ethics committee before they can be executed.