Chapter 12 Flashcards
(23 cards)
Null Hypothesis Significance Testing (NHST)
use of statistics and probabilities to evaluate the null hypothesis
Correlation family (or r-type)
category of effect size indices, including phi , point-biserial r, the alerting r, the contrast r and the effect size r
Difference Family
a category of effect size indices, including Cohen’s d and the risk difference
Ratio Family
a category of effect size (including the odds ratio and relative risk)
Null Hypothesis
generally implies no difference in the success rate between the experimental group and the control group
Alternative Hypothesis
generally implies difference in the success rate between the experimental group and the control group
Type I Error
mistakenly rejecting the null hypothesis when it is intact true and should not have been rejected
Type II Error
fail to reject the null hypothesis when it is false and should have been rejected
alpha
probability of making a Type I error
Significance Level
probability of a type I error
p value
probability value or level obtained in a test of significance
Beta
risk or probability of making a type II error
Two-tailed p value
applicable when the alternative hypothesis did not specifically predict in which direction or (tail) of the probability distribution the significance would be detected
One-tailed p value
applicable when the alternative hypothesis requires the significance to be in one tail rather than the other
Counter null Statistic
useful for minimizing two common error in thinking about effect sizes
- 1st common error: researcher mistakenly infers that failure to reject the null hypothesis also implies and effect size of 0
- 2nd common error: research mistakenly equates the rejection of the null hypothesis with having demonstrated a scientifically important effect
Statistical Power
the sensitivity of a significance test to provide an adequate opportunity to reject the null hypothesis when it warrants rejection
Power Analysis
used to see whether there was a reasonable chance to of rejecting the null hypothesis and whether the statistical power should be increased in any future studies to increase the sensitivity of the test
Power
1 - Beta
- the probability of not making a Type II Error
Odds Ratio (OR)
Relative indication of the odds for a binary variable in two groups of participants
example: adverse event present vs. not present and treated subjects vs. controlled subjects
Relative Risk (RR)
ratio of the incidence of a specified adverse event among participants administered a treatment to the incidence rate of such an event in the participants that were not given the treatment
Relative Risk Reduction (RRR)
reduction in relative risk as a percentage
Risk Difference (RD)
the difference between the risk of a specified adverse event in the treatment group and the control group
Number needed to Treat (NNT)
the number of patients needed to be treated by a drug for one patient to benefit
