The researcher does not control, alter, or manipulate either the variables or the subjects. Instead, the researcher typically uses surveys, interviews, or case studies to collect data.

Correlational

For non-experimental research projects, participants typically fill out surveys and the researcher uses a correlational designs to determine whether a relationship exists between two or more variables. In order to use a correlational design, both the Independent and Dependent variable(s) must be continuous.

This means that the variables used must represent measurements on a continuous scale. For example, suppose your study sample consists of 50 firefighters who answered demographic questions about their weight, height, age, gender, and marital status. The firefighter’s weight, height, and age of respondents in a survey would represent continuous variables. However, the firefighter’s gender and marital status are categorical (or discrete) variables: There is no continuous scale, either a person is male or female, never married, married, or divorced, etc. You would also use a correlational design if you wanted to test personality variables using survey methods (ex. Narcissism, compassion, extraversion, agreeableness, and neuroticism). It is important to note that you cannot assume that one variable causes another variable in a correlational design. Correlation does not imply causation!

Example

A woman wants to test the hypothesis that narcissism is associated with posting selfies on Facebook. She decides to use her boyfriend and his friends as participants. She runs a correlational design by giving the boys surveys which test how narcissistic they are. She then asks the boys to report how many selfies a day they post on Facebook. Girlfriend runs the analyses using the two continuous variables (narcissism scores and number of selfies posted) and finds that the more selfies the boys posted the more narcissistic they were.

A true experimental design where the researcher manipulates the Independent Variable (or variables) to determine whether it has an effect on the Dependent Variable (usually a behaviour or cognitive process). In a true experiment, the researcher randomly assigns participants to a control group and a treatment group (ex. Draw names from a hat and put 10 participants in the control group and 10 participants in the treatment group) in order to control external factors from influencing the results. True experimental design must have manipulated independent variable(s) as well as random assignment, otherwise you cannot establish cause and effect relationships.

Treatment group

Group of items or participants that have received experimental treatment or manipulation.

Control group

Is the comparison group. Group of items or participants that are matched with the test group as closely as possible (for example, same age, occupation, ethnicity etc.), but have not been exposed to any experimental treatment.

Example

Researcher wants to test the effects of caffeine on blood pressure. Researcher gathers a group of 20 students to participate in the study. Researcher randomly assigns 10 students to the treatment group and 10 students to the control group by pulling names from a hat. Researcher gives the treatment group a cup of coffee and gives the control group a cup of water. Researcher then takes the blood pressure of all 20 participants and compares the scores of the treatment group to the scores of the control group. Researcher finds that students who drank the cup of coffee had higher blood pressure in comparison to the students who drank the cup of water.

Very similar to an experimental design (randomized-control trials), however, lacks one thing: random assignment. Quasi-experimental designs are used when items or subjects cannot be manipulated by the researcher. Instead, the researcher uses natural occurring treatment groups. Quasi-experimental designs are used when the researcher cannot possibly randomly assign participants to groups. For example, if a researcher wants to study the effects of smoking on cognition, the researcher cannot ask participants to smoke cigarettes as this is against human ethics (could danger their health). Instead, the researcher may choose to use a quasi-experimental design and ask both current smokers and non-smokers in the community to complete cognitive tasks. Because this design lacks random assignment, the researcher cannot conclude any casual relationships.

Example

A researcher is interested in comparing drinking levels of college students at two different colleges in Ontario. The researcher picks a college without a drinking awareness program (natural control group) and a college with a drinking awareness program (natural experimental group). Then the researcher compares the drinking levels of students from each college. The researcher finds that the college without the drinking awareness program had higher drinking levels in comparison to the college program with the drinking awareness program.