Experimental Design and Ethics

Vocabulary flashcards based on lecture notes about experimental design and ethics.

Blind Observation

Experimental design where scientists and participants should not be able to identify which group an individual is in to eliminate bias.

Bias (in experimental design)

When something in our experimental design favors some data and not others.

How to Eliminate Bias

Using simultaneous control groups, random assignment, and blinding (hiding information).

Model Equation

Observation = Model + random chance + bias.

Pseudo-replication

Treating non-independent observations as if they were statistically independent.

The Experimental Unit

The independent unit to which treatments are assigned.

Blocked Experiment

Multiple treatments happen in each experimental unit. Common in expensive, large scale manipulation experiments and Natural experiments

Nested Experiment

Each experimental unit holds one treatment group and is common in natural experiments.

Paired Experiment

Each experimental unit contains one of each treatment; is a powerful way to determine the amount of variation the model predicts; ideal design for small manipulation studies.

Morally Responsible Science

Recognizing the draw that information has over us; considering the price we are willing to pay for discovery; involves statistical and humanitarian aspects.

Statistical Ethics Issues

Pseudo-replication, fabrication of data, altering data collection mid-measurement, modifying data in a dataset, early termination of a project, misuse of statistical tests.

Humanitarian Ethics in Science

Minimizing risk, obtaining informed consent, and guarding privacy.

IRB

Institutional Review Boards which ensure informed consent, aims, methods, funding, conflicts of interest, research institutions, benefits, risks, discomforts, and post-study provisions are declared and acceptable.