Looking at the graph, I don’t see a qualitative difference between 1000mg and 500mg. The gap in mean `responder rate` is about twice as high for 1000mg, indicating—insofar as the study indicates anything—a linear relationship between how much was taken and how effective pain relief was.
“Statistically significant” is a criteria for reporting results, meant to reduce junk science output, rather than a threshold for whether an effect is real or not. p=0.1 doesn’t mean “this effect doesn’t exist”, it means “there’s a 90 percent chance our results indicate this effect is real, but we’ll need to raise our N and try again if we want to publish a paper that says so”.
Looking at the graph, I don’t see a qualitative difference between 1000mg and 500mg. The gap in mean `responder rate` is about twice as high for 1000mg, indicating—insofar as the study indicates anything—a linear relationship between how much was taken and how effective pain relief was.
“Statistically significant” is a criteria for reporting results, meant to reduce junk science output, rather than a threshold for whether an effect is real or not. p=0.1 doesn’t mean “this effect doesn’t exist”, it means “there’s a 90 percent chance our results indicate this effect is real, but we’ll need to raise our N and try again if we want to publish a paper that says so”.