Yozef Asks: Calculation of overall effect size in meta analyses using random and fixed effect modelĪs a pharmacist and non-statistician, I'm struggling to understand the interpretation of a 2019 meta-analysis on the effect of prazosin. However, I am worried that the data is interpreted as if, e.g., out of the 150 observations, all 150 chose 85/150 as their response instead of that 85 chose 1 and (150-85) chose 0 (which is correct) - which would impair test statistic. I tried to implement tests in STATA using frequency weights, which gets the number of individuals right. Running tests/regressions on this data feels weird. An illustrative excerpt of the data looks like that: (Treatment "A" Individuals 150 Actions (k) 85, Treatment "B" Individuals 76 Actions (k) 34, Treatment "A" Individuals 87 Actions (k) 50, Treatment "B" Individuals 98 Actions (k) 85). I have multiple observations which, however, differ in the number of treated individuals. My goal is to analyze whether the treatment leads to significant difference in actions. For instance, one observations shows that 150 individuals were treated and that out of these 150 individuals 85 chose a certain action (the row vector hence looks like (150 85). I am dealing with a dataset that contains at the observation level the arithmetic sum of some variables. What is the proper way to analyze data that is only available in a collapsed format for statistical significance? Beck Asks: Analyzing collapsed data for statistical significance
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. Archives
February 2023
Categories |