this is why the world is beautiful, maybe its just me but i find this cool as fuck
"Your kid says hi." -The sun
"Technology is ruining human connections between humans" said nobody who’s ever used it to affirm ties they might otherwise only experience on rare meatspace occasions and to bring everyone they love together.
Go sit down and write three lines of whatever it is you’re supposed to be writing.
Three lines done. I introduced an important character conflict!
Three lines in this code mages thing, because that was the first thing that came up.
(I really, really, really shoulod be working on White Lightning. Oops.)
Yeah, I did have that “wait, which of ALL the things I’m supposed to be writing is the thing I’m supposed to be writing” moment there. XD
This article talks about how advisors can help a grad student who’s become stuck/stymied in the writing process. I found it useful from a student standpoint as well. I would also recommend reading the comments at the bottom; there’s some good stuff there.
I can! First we have to understand two concepts:
- Null hypothesis - suggests there is no relationship or effect between the independent and dependent variable. So if we are testing an antidepressant, the null hypothesis suggests the medication has no effect on depression levels.
- Alternative hypothesis - suggests there is a relationship or effect between the independent and dependent variable. So for that antidepressant, this might suggest the medication has some effect on depression levels.
Statistical significance in its proper label is called null hypothesis significance testing. When testing for statistical significance between our independent and dependent variables, we follow a specific process:
- Assume the null hypothesis is true and that there is no effect (fun fact: this is never the case, but we assume it anyway).
- We apply a statistical model to our data in a way that represents the alternative hypothesis (i.e., that there is some effect) and see how strong of a fit it is while still assuming the null is true.
- We then calculate the probability of this new model “fitting” when we assume that the null hypothesis is true (i.e., “if there truly was no effect, then what is the probability of getting the results we see here?”)
- If that probability is sufficiently small (often when p < .05), then we assume that the alternative hypothesis’ model fits the data well and we can reject the null.
Ultimately, statistical significance suggests that the observed results are very likely to be inconsistent with the null hypothesis.
A lot of people think that our p value represents the probability that our results are due to chance, or that statistical significance means we can be “95% confident our results are accurate and not due to chance.” These are not exactly correct. Instead, it’s just the probability that our attained results would be very extreme or unlikely should the null hypothesis be true, and because of the low likelihood, we can probably conclude that there is some effect.
What’s important to remember is statistical significance in no way hints at the importance or size of the effect being observed. Some recent article quoted a researcher as calling it “statistically discernible,” and I think that’s a fantastic term. Something can be immensely significant but be completely meaningless, while something can be very insignificant but be incredibly meaningful. Statistical significance plays virtually no role in making that call.
This is a wonderfully useful post! Statistics is SO IMPORTANT for scientists to understand. I am on Team Statistically Discernible fo’ sho!
I’m heading into the start of my Statistics PhD program and already I can’t help but be overwhelmed by HOW INCREDIBLY USEFUL EVERYTHING IS FOR SCIENTISTS! I can’t wait for my head to be full of all the usefuls!