








USMLE Step 1 Forum USMLE Step 1 Discussion Forum: Let's talk about anything related to USMLE Step 1 exam 

Thread Tools  Search this Thread  Display Modes 
#1




Biostat
Researcher thinks that the new medication she was testing has a real effect upon blood pressure. sHe publishes her research, stating that the pvalue of her analysis was < 0.05 and thus the medication did affect blood pressure. In truth, the medication doesn't affect blood pressure, and the data simply represented a normal (although unusual) variation. What type of error did she commit?
Please some one explain this question. Thank You. 
#2




Quote:
So here's my quick and dirty way to do it, and it all revolves around the formula for power: Power = 1  beta Power: our ability(or probability) to detect something IF it actually exists. So what's beta? Well, if beta was 0, then according to the formula, our power would be 100%. So therefore, logically: Beta: probability we WON'T detect a difference even if it exists. We could use simple algebra to rearrange our formula as this: Power + Beta = 1 or Probability of detecting a difference + probability of failing to detect it = 1 To put it into a simple analogy: Probability of rain tomorrow + probability of no rain tomorrow = 1 So now we can say that beta means failing to find a difference that's actually there. So what's alpha? The opposite of course: alpha = probability the difference you found isn't really there. Does this sound familiar? It should, because that's exactly what a pvalue is. It's the probability that the result was simply a result of random chance in our sampling. In science and especially in medicine, we've all just decided that 5% or 0.05 is the level we call "significant." In the example above, the researcher found a difference that had a <5% chance of being due to random chance, but it turns out really was due to random chance. Even though it wasn't likely, she committed an "alpha" error after all. The only other terminology to remember is Type I vs Type II error, and these correspond very nicely with alpha and beta: Alpha = Type 1 Beta = Type 2 So our researcher in this question could also be said to have committed a type 1 error. Sorry for the long answer, but this is a common question I get from people I work with. 
Thread Tools  Search this Thread 
Display Modes  

Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
biostat  mysha  USMLE Step 1 Forum  1  03142013 07:30 PM 
biostat Q.  Drmedic  USMLE Step 1 Forum  1  12172012 10:56 PM 
Biostat  medhelp  USMLE Step 1 Forum  2  11032012 03:18 PM 
Biostat  p value  Rookie  USMLE Step 1 Forum  6  06202012 05:03 AM 
Biostat  dr.dhruvdesai  USMLE Step 2 CK Forum  1  03032012 05:31 PM 
