Well, confidence interval and p value aren’t the same thing. They’re related, but different. You’ve identified that there’s a big confidence interval. But p value is what’s really important, because that tells you if the results are statistically significant. Now here’s a maths trick: if the confidence interval of the null hypothesis overlaps with the confidence interval of the result, then it won’t be significant. But if the confidence intervals don’t overlap, then your p value is smaller than 0.05; it’s significant.
Now here’s the data from the study:
The black circles represent years where the suicide attempt rate was not significantly different from baseline. The white circles are years where there was a significant difference to the baseline rate. So you can see that before these laws are passed, suicide rates are pretty much holding steady, and then on the second anniversary of the law’s enactment, it’s way up.
Now here’s the trick. That 7-72% is not a confidence interval. So it’s not actually related to significance. See, in the first year after the anti-trans laws were passed, for the teenage sample group, there was actually a significant effect. a 7% rise. Just very barely. You can see how the confidence interval line goes nearly all the way down to baseline. Second year, that’s way up. 72% up. So that’s the 7-72%. 7% is the first year, 72% is the second year.
So yeah, we’re pretty fucking sure this is because of the laws.
thanks for the graphics, I wasn’t accessing the study. i’m familiar with confidence intervals and p values but this is great for anyone who isn’t.
the first year’s plot looks within range, but the 72% point is a major change! something definitely happened with 13-24 year olds that year, but especially 13-17 year olds.
I see a definite occurrence that the suicide rates increased after those laws. and i’m wondering how we can be sure that it was because of the laws and not something else that happened around the same time?
i think that a reliable study reasonably isolates out other possible factors.
Well, confidence interval and p value aren’t the same thing. They’re related, but different. You’ve identified that there’s a big confidence interval. But p value is what’s really important, because that tells you if the results are statistically significant. Now here’s a maths trick: if the confidence interval of the null hypothesis overlaps with the confidence interval of the result, then it won’t be significant. But if the confidence intervals don’t overlap, then your p value is smaller than 0.05; it’s significant.
Now here’s the data from the study:
The black circles represent years where the suicide attempt rate was not significantly different from baseline. The white circles are years where there was a significant difference to the baseline rate. So you can see that before these laws are passed, suicide rates are pretty much holding steady, and then on the second anniversary of the law’s enactment, it’s way up.
Now here’s the trick. That 7-72% is not a confidence interval. So it’s not actually related to significance. See, in the first year after the anti-trans laws were passed, for the teenage sample group, there was actually a significant effect. a 7% rise. Just very barely. You can see how the confidence interval line goes nearly all the way down to baseline. Second year, that’s way up. 72% up. So that’s the 7-72%. 7% is the first year, 72% is the second year.
So yeah, we’re pretty fucking sure this is because of the laws.
thanks for the graphics, I wasn’t accessing the study. i’m familiar with confidence intervals and p values but this is great for anyone who isn’t.
the first year’s plot looks within range, but the 72% point is a major change! something definitely happened with 13-24 year olds that year, but especially 13-17 year olds.
I see a definite occurrence that the suicide rates increased after those laws. and i’m wondering how we can be sure that it was because of the laws and not something else that happened around the same time?
i think that a reliable study reasonably isolates out other possible factors.