What can I say? I am still working on my writing style. Still trying to find a style that will appeal to more people. Even a very helpful blogger, Andrea, who actually gave a me a good review (holy crap! I actually got one of those?) told me that I could greatly benefit from a writing class and I am seriously looking into the ones near my area. As I said before, I am still trying to find my voice and experimenting with new formats.
So in an attempt to see if I could myself make a scientific notion more popular among the masses with a new style, I have decided to start with my dissertation. Please bear with me and just picture me chomping strawberry bubble gum while twirling my hair in a bedroom (pink, of course!) somewhere in suburban Encino as you read this.
So, um, like I’m going to explain my dissertation to you now and like my like dissertation advisor can’t like kill me because all three of our papers were like accepted for publication anywaaay. Although like looking like at my book sales, like him killing me like wouldn’t be all that like bad. Like OhMyGod. Tee,hee.
But like anyway, so like have you like ever tried to explain multiple imputation to a non-statistician or like even to a statistician who never like imputed like data? It’s like OhMyGod. They’re like “Um, so you’re like make up data and stuff to like fill in like missing values.” And you’re like, “Um, actually, we are like drawing values from like a plausible distribution.” And they’re like “But you’re like making up those values.” And you’re like “Um, but the distribution is like plausible! Hello!” So finally you like give up and say like “Okay, fine, we like are like making up values. Whatever.” So, like, anyway like most imputation models depend on the assumption that the data is like normally distributed. But like in most like cases data are not normally like distributed. I know, right? Like Gauss like totally like lied to us. Like he must have been like the like original Trickster Guy. Like, OhMyGod! That’s like so mean. And mean people are MEAN. So anyway like we came up with like a way to transform the values to like normally distributed and then like impute them and then like back-transform them. But like the transformations themselves are like … um, I mean, they do not like depend on the distribution of the data, so it’s like totally like cool. But anyway, like I hope like you got all of that and like next time, maybe we’ll like get more into like parameters and like algorithms and like even talk about trachyons from like quantum physics and stuff. And like I want to get like Tim Blais to like help me like explain those. Cuz he’s like a total hottie! So toodles until then! Tee,hee.
- Something else that’s not working (irenehelenowski.wordpress.com)