Psychological Research Methods

Research Designs

Developing Research Measures

All research designs require clear operational definitions of the variables being studied. Meaningful research results depend on the reliability and validity of the measures used.
Research designs in psychology help researchers achieve one of three goals: describing behavior, assessing the relationship between behaviors, or examining the influence of certain variables on behaviors. The optimal research method depends on the question being asked. Regardless of which method is used, researchers must accurately measure topics of interest using research tools, such as surveys or behavioral observations. For example, a psychologist might measure anger by asking people to complete a questionnaire about how they express anger or by recording their reaction to a frustrating situation.

An operational definition is the specific description of how a variable will be measured as determined by the researcher. It could be the score on a questionnaire or observations of a specific behavior. For example, if a researcher is interested in student study strategies, she must decide how to define studying. The researcher may define studying as any review of previously learned material. With that operational definition, reading a text for the first time would not count as studying, but rereading the textbook would. Operational definitions can differ across studies. Another researcher looking at study strategies might operationalize studying more broadly: reading a text for the first time, rereading, and reviewing class notes. Regardless of how a researcher chooses to define a variable, operational definitions must be clear and specific so other people know what the study measured.

Researchers must also consider the validity of their measurements. Validity is the extent to which an assessment tool accurately measures the variable of interest. For example, counting the number of hours students spend in the library is probably not a valid measure of studying. Students in the library might be studying, but they could also be checking social media or napping.

Lastly, researchers must also consider the reliability of their measures. Reliability is the extent to which assessment scores remain consistent across time, settings, or raters. Test-retest reliability is the reliability of a measure over time. For example, intelligence is thought to be a stable characteristic. A person's scores on an intelligence test today and again in three months should be very similar on a reliable measure of intelligence. Internal consistency is the extent to which items within a test measure the same thing. On a depression questionnaire, items asking about fatigue, lack of interest, and sadness would likely have high internal consistency. Adding questions about pet ownership would reduce internal consistency because they do not measure depression. Interrater reliability is the consistency between different raters' scores. If two raters graded a student essay independently, they would have good interrater reliability if they assigned the student the same grade.

Descriptive Research Designs

Descriptive research includes case studies, naturalistic observation, and survey research.
Because psychology is a science, psychological theories and hypotheses are based on evidence. Psychologists often obtain evidence by observing and describing naturally occurring behavior and mental states.

One way to acquire detailed information about behavior is through case studies. A case study examines a single person or group in detail over a period of time. Case studies typically involve a person who has a rare characteristic, such as damage to a certain part of the brain. Patient H.M., an American who underwent neurosurgery to treat severe epilepsy, is a famous example. H.M. had part of his brain related to learning and memory removed in 1953. Researchers studied H.M. throughout his life to understand how this brain region was essential to normal functioning. An advantage of case studies is the high level of detail regarding a specific person or condition. This information often provides researchers with insight to develop broader studies. A disadvantage is that case study may not be generalizable, or applicable, to other people.

With naturalistic observation, researchers directly observe behavior in a real-world environment without interacting with the participants. An advantage is the ability to document people's authentic behavior. A disadvantage is that observation alone does not help researchers understand why people behave the way they do. For example, if a researcher wanted to document how often mall shoppers make purchases, she could count the percentage of shoppers in a store who buy an item. However, the researcher wouldn't know why some people buy things and others don't.

In survey research, participants respond to questionnaires about their behavior, attitudes, and feelings. An advantage of survey research is that researchers can measure behavior and mental states they cannot observe directly. A disadvantage of this type of research is that the results can be easily influenced by the wording of questions or participants' willingness to answer honestly. For example, researchers might use a survey to ask people about sensitive topics, such as sexual behaviors or political beliefs. However, some participants may be uncomfortable providing this information, even anonymously. Results may underestimate or not accurately reflect the frequency of thoughts or behaviors people find embarrassing.

Correlational Research

Correlational research provides information about the strength and direction of a relationship between variables. Correlations can predict behavior but do not provide causal information.
Correlational research examines the strength and direction of a relationship between variables. The correlation coefficient is a numerical value between –1 and +1 that represents the strength and direction of the relationship between those variables. A positive value means the two variables are positively correlated. As the value of one variable increases, the value of the other variable increases as well. For example, age and height are positively correlated in children. Older children tend to be taller. A negative value means that the two variables are negatively correlated. As the value of one variable increases, the value of the other variable decreases. For example, as a person ages, their agility tends to decline. The closer the correlation coefficient is to either –1 or +1, the stronger the relationship between the variables. A correlation coefficient of zero indicates no relationship exists between the variables. Correlations can be depicted on a graph called a scatterplot.
Positively correlated variables move in the same direction-as one increases, so does the other. Negatively correlated variables move in opposite directions-as one increases the other decreases. Uncorrelated variables show no relationship to each other.
Correlations help researchers predict behavior. If two variables are strongly correlated, then knowing a participant's score on one variable allows researchers to predict the participant's score on the second variable. Two variables may be correlated because one variable causes the other. For example, smoking and lung disease are correlated because smoking is a cause of lung disease. In that example, the direction of causation could only run in one direction. Developing lung disease late in life could not have been the cause of the decades of smoking that came before the disease. However, sometimes when two variables are correlated, the cause-and-effect relationship is the opposite of what one might expect. This situation is caused reverse causality. For example, stress and symptoms of depression are positively correlated. People with high stress levels have more symptoms of depression. One might assume that this relationship exists because stress leads to depression. However, the causal relationship may run in the opposite direction. Depression may sap people’s ability to handle stress, making minor hassles feel overwhelming. Or depression may make it tough to stay on top of work and other obligations, leading to more stress.

Importantly, two variables can be correlated without any cause-and-effect relationship between them. Correlation does not imply causation because the relationship between two variables may be caused by some other factor. This is called a confounding variable or the third-variable problem. One of the most famous examples of this is the correlation between ice cream consumption and swimming. Eating ice cream does not cause people go swimming. Nor does swimming cause people to eat ice cream. Instead, hot weather (the third variable) makes both ice cream and swimming especially appealing.

Within psychology, correlational relationships are often complex. For example, studies designed to explore causal relationships between stress and depression show that stress can cause depression. They also show that depression can cause stress and that third-variables like poverty and racism predict both stress and depression.

Reasons for Correlations

A correlation between variables A and B may occur because A causes B or because B causes A (reverse causality). However, a correlation does not necessarily indicate a causal relationship. Variables A and B may only appear related because a third variable, variable C, causes both A and B.

Experimental Research

Experimental designs allow scientists to draw conclusions about causes of events. Experiments require random assignment of participants to an experimental or control condition.
Experimental research is the only research design that allows researchers to draw conclusions about cause and effect. The independent variable is the variable that researchers change or manipulate in an experiment. In a study investigating the potential effectiveness of a medication, the independent variable would be a treatment condition, such as medication versus waiting list. In lab research on spider phobia (arachnophobia), the independent variable might be placing people in a room with or without a caged tarantula.

The dependent variable is the variable being measured by the researcher. The name comes from the expectation that scores on this variable will depend on the independent variable. In a treatment outcome study, the dependent variable would be symptoms. In the spider phobia study, it might be anxiety level or heart rate.

In experiments, participants are placed in different groups. The experimental group is the group containing participants who receive treatment or participate in an experimental condition. The control group is the group containing participants who do not receive treatment or participate in the experimental condition. To produce accurate results, researchers use random assignment to decide which participants will be placed into experimental or control groups. Random assignment is a process in which each participant has an equal chance of being placed in either the experimental or the control group. The goal of random assignment is to minimize the impact of a confounding variable, an uncontrolled factor that a researcher did not account for in a study, but that could influence study outcomes. With random assignment, individual differences in background, personality, behaviors, and experiences have a better chance of being distributed across both groups. This minimizes the influence of individual characteristics on the outcome of the experiment.

The high level of control in experiments allows researchers to attribute different group scores on the dependent variable to the independent variable.

For example, a psychologist might conduct an experiment to examine the effect of caffeine on test taking: The researcher hypothesizes that drinking caffeine prior to an exam will improve test performance. The researcher recruits 100 students in a large section of introductory psychology to participate in the study. These participants are referred to as the sample. The researcher then randomly assigns 50 of them to the experimental group and 50 to the control group. In the experimental group, participants drink 12 ounces of caffeinated coffee 1 hour prior to their final exam for the course. In the control group, participants may drink nothing before the final exam. An even better version of the control group would have participants drink 12 ounces of decaffeinated coffee. After the exam is over, the researcher compares scores on the final exam in the experimental group to scores in the control group. Higher scores in the experimental group would support the researcher's hypothesis.

Factors other than caffeine consumption will influence exam scores. For example, grades will be influenced by how much each student studied for the exam. Random assignment increases the likelihood that people who studied a great deal will be roughly equally distributed across groups. It helps handle the risk that an unmeasured variable influences results by reducing the likelihood of the variable contributing to differences between the control and experimental groups.

Steps in Experimental Research

In experimental research, researchers manipulate an independent variable to measure the influence it has on a dependent variable. This is the only research design that allows researchers to make inferences about causation.

Overview of Research Approaches

Research Method What Is It? Advantages Disadvantages
Case studies Detailed study of one individual over a long period of time Descriptive; forms the basis for future research Cannot be generalized to a larger population
Naturalistic observation Record/description of natural behavior in real-world settings Participants behave normally Cannot infer the cause of people's behavior
Survey research Questionnaires for people to self-report on behavior, attitudes, and feelings Measures variables that are subjective or difficult to observe Self-report can be inaccurate for many reasons (e.g., participants respond to questions based on what they think the researcher wants to hear vs. how they truly feel)
Correlational research Examination of the relationship between two variables Describes and predicts behavior Cannot determine cause and effect
Experimental research Manipulation of one variable to measure the effect on another variable Determines cause and effect Controlled settings may not mimic behavior in the real world

No one research method is right for all topics. Each method has advantages and disadvantages.