Clinical trials- hypotheses, P values, Errors Flashcards
Clinical Trials & Chance results
Clinical trial attempts to tell you something about the general population. Therefore results will be an estimate for the population since if trial was conducted multiple times, results would not be exactly the same
**must include this uncertainty about the tested treatment in the results
Null hypothesis
The hypothesis that there is “no difference in population” between the treatments being tested
**Rejection of Ho =that one treatment is significantly different than the other
>can never actually prove or accept the null hypothesis is true, just reject it
Null hypothesis and chance
There is always a chance that a clinical difference may occur to reject Ho. However, as the outcome measures/difference become large enough, it is more likely that it is not just due to chance
Alternative Hypothesis
Alternative hypothesis is always that there is a difference between the treatments being tested
Accepted by default if the null hypothesis is rejected
Two tailed vs. one tailed test
Looks at whether something is better or worse.
Two tailed= both directions
One tailed= one direction
Which tail test should be used in a clinical trial?
Should always use a two tailed test to determine differences. Ensures that you are looking at the difference being both worse or better
Why would someone use a one tailed test?
-Allows the researcher to get a statistically significant result with fewer animals (Ex. more power)
**but very few situations where it would be considered valid
Hypothesis testing with statistical significance
Need a test of statistical significance AND a way to quantify the degree to which sampling variability may account for the results observed in a certain study
Ex. P value (standard distribution)
P-value
A probability value of getting the observed effect in the outcomes measured (what is the likelihood of seeing these results by chance alone)
**if null hypothesis is true
Very small P-value
Means that it is unlikely that we could have obtained the observed results if the null hypothesis was true
**Therefore reject Ho
Very large P-value
Means that there is a higher probability that we could have obtained the observed results if the null hypothesis were true
How small of a P-value is often needed for clinical trial?
-Needs to be decided prior to start of trial
-Typically use 5% of standard level (P=0.05) BUT this is arbitrary. Leads to question: why make it a yes/no? Maybe answer is it should be a continuum
Does significance mean the same as effect?
No. Just because there is no significance, does not mean there is no effect AND if significance is present, it does not indicate anything about magnitude of the effect
**many articles mistakenly misinterpret data and assume no effect at all
Comparing studies that are called statistically significant and statistically non-significant
Their significance does not necessarily mean they are contradicting studies/results.
>observed effect in two separate papers can be the exact same BUT one can be significant and the other can be non-significant
Types of errors in clinical trial data
Type I (alpha)
Type II (beta)
