Lecture2_Notes.txt

Biost544 Lecture 2

Simulation of data using distribution probability values for Treatment and Contro responses.

How do you account for uncertainty in estimation of response proability in control group?

Check if piC is less than or equal to piT ; simulate with this pair of pi, and see if the simulated data is consistent with the data —> or MOST consisdtent with the data

find most consisten values where piT < piC ; the measured proportions do not satisfy this statement

The simplest instance where “response prob on treatment (piT) is no greater that response probability on control (piC)” is the case where piT=piC. —> Given a situtation where this is true (piT=piC), and we have an observed dataset where piT>piC ; how likely is it that what we observed is consistent with the null distribution, i.e. is it an expected fluke/randomness is the data, or does it represent something special and meaningfull —> is the obsrved data significantly different from the expected null.