What kind of research is involved in this field of psychology? - Chapter 4

What is science?

The general purpose of science is to describe and explain phenomena. Scientific knowledge comes from a systematic formulation of a problem, observation and data collection and interpretation of research results. Theoretical assumptions and concepts are used to choose variables, procedures and research goals. Often hypotheses are tested that are derived from theories. Testing hypotheses is valuable because knowledge is then obtained in a systematic manner. When finished, a study does not prove that a hypothesis is true or false but it does offer evidence in favour or against the hypothesis. If a hypothesis is not supported, this can lead to an adjustment of the underlying theory.

What are factors of science?

Research participants

Researchers often try to make statements about their entire population of interest. Because it is not possible to examine everyone in a population, a representative sample is used. Representativeness can be achieved through random selection: every person from the population has an equal chance of being selected. Sometimes certain groups of people are systematically excluded from selection. But even though it’s handy, trying to get a true random sample is not always feasible because it is impossible to draw a random sample of, for example, all children with intellectual disabilities. However, there are ways to approach representativeness. The extent to which this is successful influences the interpretation of the research findings.

Research into psychological disorders often uses clinical populations: participants from, for example, hospitals or institutions. Such clinical populations are usually not representative of the entire population because they exclude children who, for example, cannot be treated due to financial circumstances. Clinical populations can also offer an overrepresentation of young people with more severe symptoms or with symptoms that affect the environment. In this case there is a selection bias.

Measurements

A researcher must make an operational definition of the behaviour or concept that is being studied, so that it becomes clear which observable behaviour or concept is involved in the research. For example, aggression can be operationalized as the frequency with which children threaten their peers.

Validity

A measurement must be valid. Validity refers to the accuracy of the measurement: the extent to which a measurement is an accurate indicator of the behaviour being studied. A distinction can be made between internal validity and external validity. Internal validity refers to the degree to which the statement/hypothesis is correct. Depending on the research design and research methods used, there are various factors that can threaten internal validity. The extent to which alternative explanations can be excluded determines the certainty that the research results are valid. External validity refers to generalizability: the extent to which the research results are applicable to other populations and situations.

A distinction can also be made between different forms of validity:

• Content validity: The extent to which the content of a measurement corresponds to the content of the characteristic being measured.
• Construct validity: The extent to which a measurement is related to the construct (concept) that underlies the characteristic being measured.
• Visual validity: The extent to which a measurement actually appears to measure what it should measure. (Aka measuring in pounds rather than kilograms)
• Competitive validity: The extent to which the scores on one measurement are related to scores on another accepted measurement of the characteristic being studied.
• Predictive validity: The extent to which the scores on one measurement are predictive of the scores on another measurement.

Reliability

In addition, observations must be reliable. Reliability refers to the consistency of results. This means that two different observations of the same behaviour under similar circumstances must lead to the same conclusions. Inter-assessor reliability refers to the degree of agreement between different assessors. This form of reliability is bigger if the observation requirements are specific and clear. If measurements are valid and reliable, this means that a study can correctly be replicated and it should lead to the same results.

Research methods

There are different ways to investigate a phenomenon. The choice of a method depends on the purpose of the research, but also on practical and ethical considerations. A distinction can be made between (1) descriptive, (2) non-experimental and (3) experimental methods.

• The purpose of descriptive methods is to describe a phenomenon. To achieve this goal, observations are made and analysed. The relationship between two or more variables is often described.
• In non-experimental investigations, correlational and multivariate statistical analyses are used to investigate complex relationships.
• Experimental studies can be randomized or quasi-experimental. A lot of value is placed on randomized experiments , because they come closest to establishing cause-effect relationships. A variable (A) is manipulated and the effect of this manipulation on another variable (B) is measured. A causal relationship exists when differences in A lead to differences in B. In order to exclude alternative explanations, it is important to randomly assign a research participant to the different conditions and to exercise control over procedures and external factors.

Quasi-experimental studies resemble randomized experiments. The difference is that research participants are not randomly assigned to the different conditions. This ensures that causal relationships can be discussed with less certainty.

What is a case study?

A case study is a descriptive, non-experimental method to study psychological disorders. The researcher focuses on only one person and describes the background, past and present living conditions, the functioning and characteristics of this person. A case study provides information about the nature, course, correlating factors, outcomes and possible causes of psychological problems. A case study is often used to investigate rare disorders. In addition, case studies can be used to formulate hypotheses that can be tested using other research methods.

However, the limitations of case studies are reliability and validity. The question is how reliable and valid descriptions of past events are. In addition, the validity of interpretations can be questioned. In addition, research findings cannot be generalized and can’t usually be exactly replicated either.

What are various types of research?

Correlational research

Correlational studies are non-experimental and describe the relationships between two or more factors. The participants are not exposed to manipulation. These tests can be performed in the natural environment, but also in the laboratory. The question is whether there is a relationship between variable X and variable Y. Statistical procedures are used to investigate the strength and nature of the relationship. An example is that the relationship between self-confidence (X) and fear of failure (Y) is investigated. A score on X and a score on Y is collected for each participant. A correlation coefficient (Pearson's r ) is then calculated. Pearson r varies from -1 to 1. The direction of the relationship is indicated by the sign of the coefficient. A positive sign (+) indicates a positive relationship between X and Y: high scores on X go together with high scores on Y. This means that a lot of confidence goes together with a lot of fear of failure. A negative sign (-) indicates a negative relationship between the two variables: a lot of self-confidence goes hand in hand with little fear of failure. A negative relationship is also called a reverse or indirect correlation.

The strength of the relationship can be deduced from the absolute value of the correlation coefficient. The strongest relationship is expressed by an r of +1 or -1. The smaller the absolute value of the coefficient, the weaker the relationship. A correlation of 0 says that there is no connection between X and Y.

No cause-effect relationship can be deduced from a correlational study. We cannot therefore conclude that much self-confidence leads to little fear of failure, because it can also be the other way around. In addition, it is possible that a third variable is involved in the relationship between X and Y. For example, a third variable can be responsible for both little fear of failure and much self-confidence.

Randomized experiments

Randomized experiments are the best research method for drawing conclusions about causal relationships between variables. A controlled manipulation ( independent variable ) is offered to research participants, who are randomly assigned to the different research conditions. The result of the manipulation ( dependent variable ) is measured and then the differences between the conditions are studied. Apart from the independent variable, all conditions are treated as equally as possible, so that the group differences can be assigned to the independent variable. Random assignment of the participants to the groups increases the chance that differences are not caused by existing differences, but by the manipulation itself.

Single-case experimental designs

With single-case experimental designs , a variable is manipulated for just one or a few participants. These studies are also called time-series designs, because the dependent variable is measured repeatedly. Such a research design is often used to assess the effects of clinical interventions. The external validity of the method is not great but can be strengthened by repeating the research with other research participants or in other settings. The internal validity can be improved through specific characteristics of the research design, which check for the possibility of alternative explanations.

The ABA design is an example of a single-case experimental design, in which a problem behaviour is measured in different time periods and in which the participant is exposed to different conditions. An example is studying the playing behaviour of a child. During phase A (the baseline phase) it appears that the child plays little. In this first phase, the child is observed while there is no treatment. Subsequently, the child receives treatment in phase B, and it appears that the child is going to play more. The treatment is then removed, and the child returns to phase A. In some cases, the behaviour in the second baseline phase remains the same as in phase B. For example, if a treatment lesson leads to increasing academic skills, it may be that there is no decrease in skills if treatment is no longer given. From a treatment perspective, this is a positive outcome. However, from a scientific perspective it cannot be demonstrated that the positive behaviour is a consequence of the intervention.

In some research designs it is not appropriate to take a treatment away. In that case, use is made of multiple baseline designs. There are different baselines (measuring moments) in which different things can be measured: different behaviours of one participant, the same behaviour of one participant in different settings, the same behaviour of different participants, and so on. The intervention is then offered to observe the effects on one of the baselines. If there are effects, they are probably the result of the intervention rather than external factors. In this way, multiple baseline designs provide a basis for internal validity.

An example is that two different behaviours of a child are noted down. The child is then treated for only one behaviour. The other behaviour is also treated in the next phase. An example is that a researcher thinks that the child's anger attacks and throwing objects are maintained by the attention of parents to the behaviour. If this hypothesis is correct, not giving that specific attention should lead to a decrease in these behaviours. For example, first the parents no longer pay attention to the child when he or she throws things and then when he or she has a tantrum.

Cross-sectional research

In cross-sectional examinations , participants are examined once. Cross-sectional research is inexpensive and efficient. Groups are often compared with each other that differ in age or level of development. In this way, (the degree of) aggression can be compared between eight, ten- and twelve-year olds. However, if it turns out that eight-year-old children exhibit more aggression than twelve-year-old children, it cannot be concluded that children become less aggressive when they get older. This is because there is an age difference and not a development difference between these two groups.

Retrospective longitudinal study

A retrospective longitudinal study goes back in time. Information about their previous characteristics and life experiences is collected from a certain group of adolescents, for example with a specific disorder. A control case study is often retrospective in nature. A comparison is made between a group of people with and without a certain disorder. The purpose of this method is to formulate hypotheses about the predictors or causes of the disorder. However, the reliability of retrospective investigations is questioned, because memories are often used in these investigations. It often happens that memories are incorrect or incomplete.

Prospective longitudinal research

In a prospective longitudinal study , individuals are repeatedly observed over time. In this way information is obtained about the nature and progress of the development. For example, children with language disorders can be tested repeatedly over time to investigate how these disorders develop. However, this research method has a number of limitations:

• It is an expensive method that requires a lot of effort from the scientists.
• The sample can be biased due to the failure of research participants .
• It is possible that by repeatedly measuring research participants 'learn' what the correct answers to test questions are, or how to go about the test. This can be checked by using different measuring instruments, but this makes it difficult to compare research findings from different measuring moments with each other.
• Changes can be the result of social and historical events and not of developmental factors. If you have been following a group of people for thirty years, then found changes may be the result of social changes. These types of changes are called generational (or cohort) effects.

Accelerated longitudinal research

From a speed (accelerated), longitudinal study is the case where cross-sectional and longitudinal study research are combined. For example, it is possible to measure the aggression of three groups of children (aged eight, ten and twelve) at measuring moment I (now), at moment II (three years later) and moment III (six years later). Cross-sectional comparisons can be made at any measuring moment. In addition, the children in the three groups were longitudinally studied over a period of six years. In this way a distinction can be made between age differences and developmental changes. In addition, generational effects are taken into account.

Qualitative research

Many of the methods discussed are quantitative in nature: they collect and analyse numerical data. Qualitative research , on the other hand, collects and analyses non-numerical information and prefers methods such as interviews and intensive case studies. Another commonly used method within qualitative research is observation. In naturalistic observations, individuals are observed in their own living environment. In participant observations, the observer becomes part of the setting. For example, a researcher can join an extreme right-wing group to see how group processes develop there. Observations have a narrative form and strict coding schemes are not used.

Qualitative research collects large amounts of written data that must be conceptualized, analysed and interpreted. Thus, comments or written observations are coded or categorized. The categories are often based on the data instead of being determined in advance.

Qualitative research has a number of limitations. For example, the sample is often small, it is difficult and expensive to analyse large amounts of data and the reliability and validity are questioned. An advantage is that qualitative research increases the basic knowledge, facilitates the formation of new hypotheses and illustrates and enriches findings from quantitative research.

Qualitative and quantitative research can be combined. If qualitative and quantitative studies result in the same results, then the researcher gets more confidence in the conclusion he or she draws.

What are the ethical issues?

The individual rights of participants in scientific research must be protected. Governments and professional organizations have therefore developed ethical guidelines. There is a lot of overlap between the guidelines developed by different organizations. The guidelines of the 'Society for Research in Child Development' focus on research among young people.

Research proposals must be approved by Institutional Review Boards (IRBs). These committees consist of professionals who pay attention to, among other things, the risks and benefits of research proposals. They pay extra attention to vulnerable people, such as young people, people with intellectual disabilities and people with an economic disadvantage.

Fundamental to most ethical guidelines is voluntary informed consent: the voluntary participation of individuals in a study, who are fully aware of the purpose of the study, the procedures, the risks and benefits and the possibilities to cease participation at any time. They must be able to understand the information and assess the risks and benefits. For minor children, the permission comes from their parents / guardians.

Sometimes participants have to provide personal information for an investigation. The principle of confidentiality states that participants have the right to control the extent to which personal information is made public. There are a number of issues regarding children and adolescents. Parents and schools may be interested in the research results. The researcher can limit the transfer of information to parents, for example when it entails a risk for the child. In other cases, the child may benefit from the transfer of information, for example if he / she uses illegal drugs. Participating youngsters, parents and relevant institutions must have insight into the limitations of confidentiality.

The principle of non-maleficence refers to the protection of research participants against negative consequences. The principle of beneficence states that the benefits of a study must be maximized for each individual.