Valuing intervention and observation | JoVE Visualize

Area of Science:

Cognitive psychology
Causal inference
Decision-making

Background:

Causal inference is crucial for action but often complicated by confounding variables.
Understanding how people value different data sources is key to improving research methods.
Previous research has not fully explored preferences for intervention versus observational data when confounds are present.

Purpose of the Study:

To investigate how individuals value data from hypothetical interventions versus observations.
To determine if preferences for data collection methods change based on the presence or absence of a potential cause.
To examine confidence in causal judgments and individual differences in data source preferences.

Main Methods:

Three experiments were conducted using hypothetical scenarios involving a possible cause (bottled water), a confound (food), and a context (restaurant).
Experiment 1 used a between-subjects design to test preferences for intervention vs. observation data.
Experiments 2 and 3 examined confidence ratings and individual differences in data source preferences, respectively, using between-subjects and within-subjects designs.

Main Results:

Experiment 1 confirmed that participants preferred intervention data when the cause was present and observational data when the cause was absent, aligning with the goal of specific causal inference.
Experiment 2 found no significant difference in confidence between intervention and observation data, but confidence was influenced by the probability of an effect given the cause.
Experiment 3 revealed individual differences, with participants differing in whether they perceived intervention or observation data as more confounded.

Conclusions:

Participants' evaluation of data sources is nuanced and depends on the specific causal question and the presence of confounding factors.
While confidence in causal judgments may not differ between methods, the perceived value and preference for intervention versus observational data can vary significantly.
These findings highlight the importance of considering how individuals assess evidence when designing studies and interpreting results, particularly in research involving human behavior.