Saturday, August 29, 2009

I can't believe I'm not even halfway through

arrrrrrr. still in the grammar mines. working my way slowly toward the midpoint.
I'm going to let off some steam here, just to stay sane.
this is a textbook chapter where we have something to say, supported by the data in the medical literature.
But. Question of style. My colleague writes Smith et al did a study, gives a sentence about the study design, gives a sentence covering the major results. Goes to the next paper, same thing. Some sections are more of an annotated bibliography than a synthesis with something global to say. It takes some rearranging, but I can smooth that out. Conjugating verbs and choosing prepositions is the easy part.
And then, what keeps me flipping over to facebook or Free Cell or Blogspot instead of staying the course, there's the bias. My friend is an enthousiast. He wants hypnosis to work. He thinks it does work; it just hasn't been studied well enough to prove it. Usually, he's a good enough author to keep most of his personal perspective out of it, and just tell us what's known, what worked, what didn't, where the grey zones are. But not always. And since I don't share his belief but am rather waiting to be convinced by the evidence, this drives me nuts.
to wit:
'Somehow, a recent study of Classen et al (2008) attempting to determine ..... ' found no effect.
Alright, using 'somehow', which conjures an image of shaking one's head and wondering how on earth they came up with this, it must surely be wrong, might just be a language thing.
A little further on, there's a study that found a small and not statistically significant decrease in measures of depression in patients undergoing therapy using hypnosis. It was a 5 to 8% difference, and only one study, but the next sentence waves that aside and hails the therapy as a 'strong alternative to conventional psychotherapy'. Let's slide a may be in there, and cut the strong.
I will get through it, and we will meet in the middle and get a good, balanced, chapter out of it. But it's slow going, frustrating, and I'm very, very glad this is not my real job.


steven said...

so is that the book half done or the book half unfinished?! hmmmm. i say be picky picky picky. go with your intuition and thank yourself later of no one else does!!! have a peaceful evening. steven

EastwoodDC said...

I occasionally encounter someone with the wrong idea about the interpretation of their data. Not that they are being intentionally deceptive, but it can require some patience to educate them about why it means something else.

Hold it ... my statistics-sense* just kicked in ... the difference in that study of 5% to 8% is an amazingly narrow confidence interval for that type of study, because it implies a huge sample size (>1000). I would be highly skeptical of including that result in a book.

I could take a look at that paper and give you my opinion if you like. If you send me the citation, I can probably access it from work on Tuesday.

* Spider-Man has a Spider-Sense, statisticians have this. ;-)

NanU said...

Hiya, mistertomatoman!
5 - 8 % wasn't a range; there were three measures that improved by 5, 6 and 8 %. I just didn't go into it for the blog. I've eliminated most of the numerical details from the chapter - if a reader wants the guts, the reference is there - and we're sticking to more reasonable language. The few studies that have been done suggest...
It's only occassionally that my coauthor slips into avocacy. He's our staff statistician!

EastwoodDC said...

Back-at-ya SG!

That seems reasonable, and my Staty-Sense is now satisfied.

I see your blog has picked up a lot of followers, and I can't say I am surprised. It's about time they found you.