Gavilo Random Facts
When Dr. Gavilo-Lane requires students to build 2×2 tables to interpret associations, she is emphasizing mastery of this fundamental epidemiologic measure used in case-control studies.
Odds Ratios
Relative Risk
PPV
NPV
Sensitivity
Specificity, etc.
Greg Martin: https://www.youtube.com/watch?v=-cQwiEJnPyA
Explain the difference between external vs. internal validity. How do you minimize these threats?
Internal validity is the extent to which the experiment is free from errors/bias, and any difference in measurement is due to an independent variable. Proper selection of study groups and randomization prevents internal validity problems.
External validity is the extent to which research results can be inferred to the world at large. Proper definition of a population parameter, sampling, and adjustments for confounding variables prevents external validity.
Measure used in case-control studies
What is the odds ratio?
OR > 1 indicates
Note: The closer the OR is to one, the less meaningful the OR.
Strongest evidence of causality from smoking data...
Nonspuriousness
Temporal relationship
Statistically significant
Dose-response relationship
Pick a number 1 - 23.
Give me either the chorus or the first two lines of the song...
https://music.apple.com/us/playlist/more-cowbell/pl.u-11zBXD3i80oEM2
This study starts with disease status and looks back at exposure.
What is a case-control study?
Can you define, identify strengths/weakness/limitations, statistical tests, and when you might deploy this study?
A randomized control trial comparing the efficacy of two drugs showed a difference between the two. Assume that in reality, however, it was a false negative. This is an example of the following:
a.Type I error (α error)
b.Type II (β error)
c.1 – α
d.1 – β
e.None of the above
b.Type II (β error)
A Type II error, or "false negative", occurs in statistics when you fail to reject a null hypothesis that is actually false.
Your entire group must demonstrate the sprinkler dance move and teach me...
Sprinkler dance move...
This measure compares the risk of an outcome between exposed and unexposed groups after controlling for other variables.
What is an adjusted risk ratio (ARR)?
Blinding test names (e.g., grading without student names) helps prevent:
What is observer bias (also called grader bias or detection bias)?
When an instructor knows a student’s identity, it can (even unintentionally) influence:
Blinding removes that information → promotes objective grading.
Know study biases for exam!
Main advantage of cohort over case-control
Answer: What is ability to measure incidence / temporal relationship?
Can you define, identify strengths/weakness/limitations, statistical tests, and when you might deploy this study?
Your group must ALL practice 3 new Ted Lasso dance moves...
Ted Lasso Dance Moves:
A study reports an adjusted risk ratio (ARR) = 1.25 (95% CI: 0.85–1.55). This suggests a 25% higher risk in the exposed group, but the confidence interval includes 1.
Answer:
What is not a statistically significant association between the exposure and outcome?
What are power calculations based on?
1. Effect Size: small, medium, large
2. Clinical outcome
3. Alpha (α):p < 0.01 , 0.05 , 0.10
4. Statistical power (1 - β) = 0.8 – 0.9
5. One-tailed or two-tailed test
6. Statistical test being used
According to student feedback on exams, these study resources were reported most frequently (50%) as helping prepare for the exam.
1. Case studies
2. PowerPoints
3. Videos
But don't forget:
5. Readings
6. Kahoots
7. Student presentations & papers
8. Studying w/ peers
Best study design for rare diseases
Case control.
GURL... you're going to calculate an OR or Odds Ratio.
Measure that reflects population impact.
What is risk difference?
Risk Difference (RD) — also called absolute risk difference — is the difference in the probability (risk) of an outcome between two groups.
RD=Risk(exposed)−Risk(unexposed)
Interpretation
Why It Matters
Risk difference tells you the actual impact on the population (how many extra cases occur), not just the strength of association.
Increasing OR across smoking categories suggests
What is dose-response relationship?
Explain type I and type II errors, as they relate to power analysis.
Alpha (α) or the p-value you select helps reduce Type I error or incorrectly rejection the null hypothesis, resulting in a false positive. Typically alpha selected at .05 or 95%.
Statistical power or (1 - β) is incorrectly failing to reject the null hypothesis a false negative. Typically selected at between .8 to .9
Researchers want to determine if the risk of potential health outcomes from heavy drinking among females is increasingly becoming a problem by monitoring a groups of women for 10 years.
Prospective cohort study.
P-values:
p <.05 - 95% significance
p <.10 - 90% significance
p <.01 - 99% significance
95% CI - upper and lower limit does not contain 1.
Key difference between statistical and practical significance
What is statistical ≠ meaningful impact?
Purpose of randomization
What is reduce selection bias / improve internal validity?