Term
| why are statistics necessary |
|
Definition
| to understand the potential influence on experimental results and show that the treatment and only the treatment cause the effect or that the frequency of an event is attributable to a given factor. |
|
|
Term
| what does an experiment tell us about the effect of a treatment |
|
Definition
| it can only ESTIMATE the underlying effect of a treatment |
|
|
Term
|
Definition
| a clear statment of what the investigators expect the study to find, testable |
|
|
Term
| what used to be the paradigm for statistical inference |
|
Definition
|
|
Term
| what is the null hypothesis |
|
Definition
| there is no difference between outcomes due to the treatments being compared, in most research this is what you start out with, what you set to test |
|
|
Term
| what is the goal of hypothesis testing |
|
Definition
| calculate the probability of getting the observed results if the null hypothesis is true |
|
|
Term
|
Definition
| in reality the null hypothesis is true but in experiment it suggests that an alternative hypothesis is true |
|
|
Term
| what does the P value show |
|
Definition
| tells how unlikley it is that the result occured simply by chance if the null hypothesis is true |
|
|
Term
| what P values are considered statistically significant |
|
Definition
|
|
Term
|
Definition
| in reality there is a difference between the two groups but when you do the research it says there isnt |
|
|
Term
| what happens to error when the sample size is too small |
|
Definition
| increased risk of type 2 error |
|
|
Term
| what happens to the error when the sample size is large |
|
Definition
| less risk of type two error |
|
|
Term
|
Definition
| the probability of not comitting type 2 error |
|
|
Term
| what are the limitations to hypothesis testing |
|
Definition
| limits decision to yes or no, choice of p value significance seems arbitrary, cant inform as to wether or not the sample size is sufficient or if the results are definitive |
|
|
Term
| what is a confidence interval based on |
|
Definition
| observed result and the size of the sample |
|
|
Term
| what information does a confidence interval provide, how is it presented |
|
Definition
| gives range of possibilities in which the true probability would like within the 95% of the time |
|
|
Term
| what is the relation of confidence intervals to power |
|
Definition
| can determine if there is adequate power to make the results definitive, lets you know if there was enough people in the sample to make the study meaningful |
|
|
Term
| what does it mean when a conficence interval is not difinitive |
|
Definition
| the range is less than you stated. For example: you said it would be greater than a 5% difference but the CI shows a 2-10% difference |
|
|
Term
| what does it mean when a confidence interval is difinitive |
|
Definition
| the results fall within your projected value. for example: you say there 5% or greater difference and the CI shows 5-10% |
|
|
Term
|
Definition
| if the lower boundry of a CI is greater than the smallest difference clinically significant, it is difinitive. if it is less it needs more trials |
|
|
Term
|
Definition
| if less than the smallest difference that is clinically significant, it is definitive. if greater then more trials are needed |
|
|
Term
|
Definition
| risk of the event after treatment as a percentage of the origional task |
|
|
Term
| what is absolute risk reduction |
|
Definition
| difference in the risk of the outcome between the two groups, aka risk difference |
|
|
Term
| what is relative risk reduction |
|
Definition
| estimates the percent of baseline risk that is removed as a result of the therapy |
|
|
Term
| what are the measures of association |
|
Definition
| absolute risk reduction, relative risk reduction, relative risk, odds ratio |
|
|
Term
| what measures of association stay the same no matter what group you look at (high risk or low risk patients) |
|
Definition
| relative risk reduction and relative risk |
|
|
Term
|
Definition
| estimates the odds of an event occuring |
|
|
Term
| what types of studies are odds ratio used in |
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
| the ARR, absolute risk reduction |
|
|
Term
| what does a NNT of 6 mean |
|
Definition
| for every 6 good procedures you do you prevent a death from the old bad one |
|
|
Term
| what is a case control study |
|
Definition
| participants are selected based on an outcome or if they have experienced an event |
|
|
Term
| what does an odds ratio of 1 mean |
|
Definition
| there is no difference between the groups |
|
|
Term
| what does an odds ratio of > 1.2 mean |
|
Definition
| statistically signifianct difference between groups |
|
|
Term
|
Definition
| you are more likley to disclose or remember information based on your disease |
|
|
Term
| what measure of association is the best in randomized and cohort studies |
|
Definition
|
|
Term
| what do RR and RRR not count for |
|
Definition
|
|
Term
| which measure of association is the best for case control studies |
|
Definition
|
|
Term
| what does correlation and regression help us understand |
|
Definition
| relationship between variables, enables us to make perdictions about a given patient |
|
|
Term
|
Definition
| the magnitude of the relationship between different variables, neither variable is the target |
|
|
Term
|
Definition
| makes a perdiction or casual inference, one variable is the target |
|
|
Term
| what does a correation of -1 mean, what does it look like |
|
Definition
| strong negative correlation, line from upper left to bottom right |
|
|
Term
| what does a correlation of 1 mean, what does it look like |
|
Definition
| strong positive correlation, bottom left to top right |
|
|
Term
| what does a correlation of 0 mean, what does it look like |
|
Definition
| no correlation, scattered |
|
|
Term
| what can a correlation not distinguish |
|
Definition
|
|
Term
| what is a multivariate regression |
|
Definition
| considers all of the independet variables in a mathematical model |
|
|
Term
| what conclusions should we take from the lecture (3) |
|
Definition
| make friends with statiticians, dont take statistical competence or honesty of authors for granted, shoot for proficiency in hypothesis testing, confiedence intervals, measures of association, and correlation and regression |
|
|
