Biostats Unit 3 Exam

Experiments

These are designed to find causal relationships where we assign participants to groups and treatments

Observational

These studies identify associations that may signal a causal relationship, but we can only observe what participants chose.

Confounder

This is a variable that changes the response variable and is linked to the explanatory variable

Example:

You study whether ice cream sales (X) cause drowning (Y)

Hidden variable:

Temperature (Z)

Temperature affects BOTH X and Y

Conclusion:

• Temperature = __________

• It creates a fake association between ice cream and drowning

Mediator

This explains why a variable affects the outcome.

You study whether exercise (X) lowers blood pressure (Y)

Mechanism:

Weight loss (M)

• Exercise → weight loss

• Weight loss → lower blood pressure

Weight loss is part of the causal chain

Conclusion:

• Weight loss = __________

• It explains HOW exercise affects blood pressure

Cross Sectional Study

This is an observational study design where the data for explanatory variable and response variable are taken down right away/at the same time.

Cohert Study Design

This is an observational study where the explanatory variable is exposed to participants but the data for response variable is recorded over a period of time. It’s usually prospective.

Usually require large sample size

Case Control Studies

This is an observational study type that’s typically retrospective. Participants are chosen based off of an already available response outcome while the explanatory variable is what is being explored/recorded

Sample size is usually on the smaller side.

Odds Ratio

Case control studies use _______ instead of the RR. One issue is though is that this exaggerates data, making it be father from 1. But, the greater the sample size, the closer this is to RR.

Group Selection

This is a type of threat to the causality argument:

How similar or different are groups"?

Drop out Differences

This is a type of threat to the causality argument:

Are the rates and reasons different between groups for dropping the study?

Attrition

Refers to the people who drop out of the study, either don’t follow up or quit.

Test Familiarity

This is a type of threat to the causality argument:

Are individuals improving on their scores/data because they are learning how to do the test?

Big problem with pre-post experimental design

Setting effects

This is a type of threat to the causality argument:

Does the setting/experimental/environmental differences affect the response?

Independence

This is a type of threat to the causality argument:

Are participants interacting? Is each response unique and affected?

sample size variability

Power increases as the ____ ____ increases and random sources of ________ decreases.

Statistical power

Probability of detecting a true effect. Risk of type two error.