Reply
Sun 4 May, 2014 11:16 am
I have a technician who handed me his worksheet consisting of test results he claims to have obtained for 34 sequentially collected samples, each independently and on a different day. I have reason to believe that the technician is dishonest and has been fabricating the data sheets by cutting and pasting old data from other spreadsheets. As I look at the 34 results I notice something strange. A chemical marker that is normally found in only 3% of all samples and distributed randomly in the population is found on days 11, 23, 29, 30, 31 and 32 in his sheet. I have noticed similar odd results previously. So I run an exact binomial test and find that the probability of obtaining 6 of 34 markers (at p = 0.03) is P = 0.0004. Because I have never seen more than 2 markers on a list this size and never found any sequentially listed, this seems not to express the overall probability. How do I calculate the overall probability of the binomial distribution as well as the probability of obtaining this sequence; finding four of these markers sequentially from 29-32?
This is your homework, right?
@contrex,
In a way! But in truth, this is real issue and a real technician I am dealing with and I'm a real scientist. I can see how to do both binomial and sequential (Runs test) but not get a combined probability.