click below
click below
Normal Size Small Size show me how
PHC 6000: Testing
Introduction to epidemiology: Hypothesis testing
Question | Answer |
---|---|
What is Type I error? | Rejecting the null when it is true = alpha |
What is alpha? | Probability of rejecting the null when it is true = Type I error |
What is Type II error? | Accepting the null when it is not true = beta |
What is beta? | Probability of accepting the null when it is not true = Type II error |
What type of error is convicting an innocent person? | Type I error |
What type of error is letting a criminal go free? | Type II error |
What is the approx standard margin of acceptable Type I error? | 0.05 |
What is the approx standard margin of acceptable Type II error? | 0.20 |
What is power? | Probability of correctly rejecting the null |
What is the approx standard amount of minimum power? | 0.80 |
What is the relationship between alpha and beta? | Inverse relationship. Lower alpha = higher beta |
What is the effect of a lower alpha on the strength of statistical evidence necessary? | with a lower alpha it will take much stronger statistical evidence to ever reject the null hypothesis, even if it is false. |
When should one minimize the amount of Type I error? | When the cost of rejecting the null is high |
If there was a criminal trial or some other very serious life-changing event, would it be better to have a low or high alpha? | A lower alpha is better because the cost of rejecting the null is high |
Is it always important to minimize Type 1 error? | When there is an interest in changing the status quo and the cost of rejecting the null is low, a higher margin of Type I is acceptable. |
If there was a non-invasive treatment being tested, would it be better to have a low or high alpha? | A higher alpha would be acceptable |
What is the null value for risk ratio? | 1 |
What is the null value for risk difference? | 0 |
When should nonparametric tests be used? | Should be used on samples that are non-random (e.g., convenience samples) |
What is the definition of a 95% confidence interval? | If the research were done 100 times in the targeted population, the true effect will lie between the specified range 95% of the time |
What information does the p-value provide? | Whether or not there is a significant difference |
What information does the point estimate and confidence interval provide? | Size of difference, precision of the estimate, and likelihood of significant effect if sample size is increased |
How can one estimate the likelihood of significant effect if sample size is increased? | Look at the confidence interval; a disproportionately large interval on either side of the null tells us that if we increase the sample size, we will likely get a significant result. |
What is the difference you want to be able to detect called? | Delta/effect size |
Which type of study doesn't have an effect size? | Descriptive studies |
What is observing a difference when in truth there is none called? | Type 1 error |
What is failing to observe a difference when there is one called? | Type 2 error |
If two confidence intervals overlap, are they significantly different? | No |
What is the alpha-level of study? | The amount of Type 1 error that's tolerable for a given study |