Experiemental Studies, Evaluating Associations, & Ajusted Rates

What is an experimental study?

Researchers assign people to groups to test an interven

What is random assignment?

Participants are placed into groups by CHANCE.

What is blinding?

Keeping participants or researchers unaware of who receives the intervention to reduce BIAS.

Single-blind vs. Double-blind?

• Single: ONE side doesn't know.

• Double: BOTH participants and researchers don't know.

What is the null hypothesis (H₀)?

There is NO difference between groups.

What is the alternative hypothesis (H₁)?

There IS a difference between groups.

Null vs. Alternative hypothesis example?

• Null: Snacking does NOT change procrastination.

• Alternative: Snacking REDUCES procrastination.

What is a p-value?

How surprising the results would be IF the null hypothesis were true.

Smaller p-value = Stronger evidence against the null.

What does p < 0.05 mean?

Statistically SIGNIFICANT → Reject the null hypothesis.

What does p ≥ 0.05 mean?

NOT statistically significant → Fail to reject the null.

Does NOT automatically mean there is no effect.

What is statistical power?

The ability to DETECT a real effect if one exists.

What increases statistical power?

• LARGER sample size

• BIGGER effect

• More common condition

• Appropriate study design

What is a 95% confidence interval?

A range of values used to estimate the TRUE population value.

Confidence interval

Range of likely values.

When is a 95% CI statistically significant for OR or RR?

• Does NOT include 1 → SIGNIFICANT

• Includes 1 → NOT significant

Statistical vs. Clinical significance?

• Statistical: Unlikely to be explained by chance under the null.

• Clinical: BIG enough to matter in real life.

Statistical = Numbers. Clinical = Real-life importance

What are adjusted rates?

Rates changed to make comparisons between populations FAIR.

Why do researchers adjust for age?

Different age groups have different disease and death rates.

What is internal validity?

Are the study's results ACCURATE for the people studied?

What is external validity?

Can the results apply to OTHER people?

What is random error?

Mistakes or differences caused by CHANCE.

What is systematic error (bias)?

A consistent problem that makes results WRONG.

What is sampling error?

Your sample differs from the population because of CHANCE.

What is selection bias?

The people selected or remaining in a study are systematically DIFFERENT from those not included.

What is information bias?

Exposure or disease information is measured INCORRECTLY.

What is recall bias?

People remember past events INACCURATELY.

What is prevarication?

Participants LIE or hide information.

What is interviewer bias?

Interviewers question or treat groups DIFFERENTLY.

Reliability vs. Validity?

• Reliability: CONSISTENT results

• Validity: ACCURATE results

What is confounding?

A THIRD factor affects the relationship between exposure and disease.

What TWO things must a confounder have?

• Associated with EXPOSURE

• Risk factor for DISEASE

How can researchers control confounding?

During study design:

• Randomization: Assign groups by chance.

• Matching: Match people with similar traits.

During analysis:

• Stratification: Separate people into groups.

• Multivariable analysis: Account for other factors statistically

What is causality?

Back: When one factor actually CAUSES an outcome.

Remember: Association does NOT always mean causation.

What is strength of association?

A STRONGER relationship provides more evidence for causation.

Example: Higher exposure linked to much higher disease risk.

What is dose-response?

MORE exposure → MORE disease risk

What is specificity?

An association is linked to a PARTICULAR disease or outcome.

What is temporality?

Exposure must happen BEFORE disease.

Cause FIRST → Effect SECOND.

What is biological plausibility?

The relationship makes sense SCIENTIFICALLY.

What is consistency?

MULTIPLE studies find similar results.

What is experimental validation?

Experiments provide evidence supporting a cause-and-effect relationship.

What are the 5 questions to evaluate an association?

• Chance? → Could results be random?

• Bias? → Were there study errors?

• Confounding? → Was a third factor involved?

• Generalizable? → Does it apply to others?

• Causation? → Does one actually cause the other?