Biostats 2- Inferential Statistics Flashcards
This is the concept when there is variation in the values of a sample when compated to other samples or to the population, usually just by unfortunate random chance.
Random sampling variation
What are the 2 methods for investigating the role of random sampling variation in a study’s results?
Hint: 1 is the range for a level of precision, the other is the testing of significance & p-values.
Confidence Intervals
Hypothesis testing
This is the transformative statment based on the clinical question of interest (research), or a claim or statement about a property of a population (stats).
Hypothesis
This is the type of hypothesis where there is no difference, and the calculated difference is due to chance.
Null Hypothesis (Ho)
If we end up FAILING to reject Ho, what does that mean?
That the difference is due to random chance alone (random sampling variation)
(like if the Ho assumes a conclusion of no therapy effect of a drug, and we reject Ho, we are saying that there is no difference in the effectiveness in the drugs therapy and the drug sucks balls)
This is the type of hypothesis where a real difference exists, and the difference is not due to chance.
Alternative hypothesis (Ha)
If we end up accepting Ha, what does that mean?
It’s basically the same thing as rejecting Ho, where there is a difference in the hypothesis.
What is the test for the null hypothesis?
Test of significance
This is when there is statistical evidence that a real, meaningful difference exists (not due to random sampling variation).
Statistically significant difference
This is the type of significance test used when the dependent variable is continuous AND data is normally distributed.
Parametric tests
This is the type of significance test used when the dependent variable is categorical OR continuous but fails the normality assumption.
Nonparametric tests
What is the test used to measure the temp of each great lake?
1-way ANOVA
This is the probability that the difference is due to chance (random sampling variation).
P-value
Do higher or lower P-values show the more likely the differenece id eo to chance (therefore supporting Ho).
High P-values –> support Ho
What is the usual value for the cutoff for P-values (alpha)?
0.05
What happens to alpha if you use a 2-tailed test?
You gotta split it in 1/2 for each tail so each tail now has an alpha = 0.025
What is necessary for clinical significance, for the decision to modify treatment?
Statistical significance
What does the absolute risk include that the relative risk doesnt include?
Context/prevalence
This si the type of error where u mistakenly reject the Ho when it is actually true (false +).
Type I (alpha) error
What should you try to do to alpha to limit type I errors?
Lower it (usually to 5%)
What happens in type II (beta) errors to give u a false negative?
You accept the Ho when its actually FALSE
What is the normally beta values?
0.20
What is the eqn for Power?
Heavy weights + muscle milk = Power
What is the eqn for statistical Power?
Power = 1- beta
What is the purpose of statistical power?
To avoid false negatives (errors) and finding true negatives when one actually exists.
What is the % goal for adquate power?
0.80
What are the 3 ways of increasing statistical power and thereby avoiding false negatives?
- ↓ alpha
- ↓ variation (↑ sample size and good data collection)
- ↑ effect size (magnitude of a relationship or difference between 2 mean values)