Experiments Flashcards
Field experiments
Field experiments are conducted in real-world settings rather than in controlled environments. Researchers manipulate one or more independent variables in a natural context to observe the effects on dependent variables. Researcher has less control than in laboratory experiments.
High external validity and moderate internal validity (low control). Medium N, case-representative samples.
Reliability: Possible to measure internal consistency if they incorporate multi-item scales. Difficult to do test-retest and parallel-test reliability due to the real-world setting and difficulties in maintaining consistent conditions
Validity: Strong in external, face, and content validity because of their real-world settings but may struggle with internal and construct validity due to less control.
Natural experiments
Natural experiments occur when researchers take advantage of a naturally occurring event or situation that affects some individuals but not others, allowing for the observation of causal effects without direct manipulation of the independent variable.
Weakness: No control of group assignment (potential selection bias) or intervention
Control group: Typically involves a comparison between groups that are affected by the event and those that are not.
Example: A researcher studying the impact of a new law (e.g., a smoking ban) on public health might compare health outcomes in a city that implemented the law to outcomes in a similar city that did not.
Reliability: Out of all of the types of field experiment, natural experiments are the most difficult to assess for reliability, as the conditions are not controlled and may not be repeatable
Validity: Stronger external validity, but struggle with internal validity due to confounding variables
Quasi experiments
Quasi-experimental designs involve the comparison of groups that are not randomly assigned. Researchers still manipulate an independent variable, but the lack of random assignment means that the groups may differ in significant ways.
Validity: Stronger external validity (if representative sample), but moderate internal validity -> requires statistical control
Randomized control trial (RCTs)
RCT’s experiments use random assignment to ensure comparability between groups. This design is commonly used in clinical research to assess the effectiveness of interventions, such as medications, therapies, or educational programs. Focuses on evaluating the effectiveness of interventions in real-world contexts.
Validity: External validity depends on sampling representation, strong internal validity due to random assignment
Survey experiments
Survey experiments combine elements of traditional surveys with experimental manipulation. Researchers present respondents with hypothetical scenarios or questions that include variations in the independent variable to measure their responses.
Moderate internal validity and moderate to high external validity (depends on sampling). Often large N.
Reliability: Parallel-test reliability through split-ballot designs and can effectively measure internal consistency
Validity: Aim for a balance of internal and external validity but may face challenges in content, face, and predictive validity due to their reliance on hypothetical scenarios and self-reported data.
Laboratory experiments
Laboratory experiments are conducted in controlled environments where researchers manipulate independent variables and measure their effects on dependent variables. The laboratory setting allows for precise control over extraneous variables. Focuses on understanding causal relationships.
Lower external validity (artificial), high internal validity (treatment controlled). Often small N.
Reliability: Best potential for test-retest and parallel-test reliability due to their controlled nature
Validity: Excel in internal and construct validity due to their controlled environment but may lack external and face validity.
Experimental designs
An experimental design determines how participants are allocated to the
different groups in an experiment.
Between-subject design
Between-subjects designs involve different participants in each experimental condition. For example, one group receives a treatment, while a separate group serves as the control. The comparison is made “between” these distinct groups to see if the treatment had an effect
Within-subject design
Within-subjects designs, expose each participant to multiple treatment conditions over time. In this design, the same individuals experience all levels of the independent variable, and comparisons are made “within” each subject across different conditions
