Ooh.. I have never investigated what the rendering looks like on an android tablet. In the browser, the typesetting is almost always fixed by refreshing the page. Can you try that on your tablet and let me know it fixes the problem or not?

(If it doesn't, I'm going to figure out how to file a support ticket....)

Perhaps this just illustrates my inexperience, but how do you know what is the value of p? 95% seems the default rather than verified. Maybe to put a different way, what happens when the distribution is nonnormal? Do confidence intervals for, say, skewed distribution differ from distributions with long tails, vs normal distribution? Asking as a radiologist.

Great question! Tomorrow, I'll talk about this 95% convention. It's just a tradition that stuck and we can't escape from. I think life would be better if we reported 5-9 confidence intervals (99.999%), but this wouldn't fix all of our statistical issues.

With regards to other distributions, I'll try to touch on those as well later in this series. Confidence intervals for non-normal distributions are much weirder than you might think.

Great topic, Ben, looking forward to reading more!

But math typesetting in Substack makes me sad as in MS Word sad.

I know, I know. I'm trying to figure out how to fix it. Are you reading on a web browser or a phone?

Android tablet here Ben

Ooh.. I have never investigated what the rendering looks like on an android tablet. In the browser, the typesetting is almost always fixed by refreshing the page. Can you try that on your tablet and let me know it fixes the problem or not?

(If it doesn't, I'm going to figure out how to file a support ticket....)

And now after a reload, like this

https://drive.google.com/file/d/1lgdWBEhC8KR80UclBMiEawPT7noThvfU/view?usp=drivesdk

this looks ok, right?

I've been having the problem sometimes where I see the formula next to the markup. but this goes away when I refresh.

The set $$S$$ looked like plain text when I assumed you wanted it in math typeface.

Now I see the math typeset well, but also the markup for the formula next to it, separated by a "/" character. Does that make sense?

Perhaps this just illustrates my inexperience, but how do you know what is the value of p? 95% seems the default rather than verified. Maybe to put a different way, what happens when the distribution is nonnormal? Do confidence intervals for, say, skewed distribution differ from distributions with long tails, vs normal distribution? Asking as a radiologist.

Great question! Tomorrow, I'll talk about this 95% convention. It's just a tradition that stuck and we can't escape from. I think life would be better if we reported 5-9 confidence intervals (99.999%), but this wouldn't fix all of our statistical issues.

With regards to other distributions, I'll try to touch on those as well later in this series. Confidence intervals for non-normal distributions are much weirder than you might think.