|
|
beta weights
|
Logged in as: Guest
|
|
Users viewing this topic:
none
|
|
Login  | |
|
beta weights - February 7, 2005 3:22:00 PM
|
|
|
Jon Newman
Posts: 1713
Joined: April 24, 2004
From: Amherst, WI
Status: offline
|
FYI,
In the most recent issue of PT journal, there is a report titled, "Relationship of physical examination findings and self reported symptom severity and physical function in patients with degenerative lumbar conditions"
Quite a title. I enjoyed the publishing of the beta weights in this study as it helps me understand the relative importance of the variables they studied. I'm still digesting the article.
jon
Still digesting....On second reading, I'm not sure they did report the beta weights, they reported the beta coefficients. Anyone want to help out here?
_____________________________
[URL=http://www.sonymusic.com/clips/selection/30/064887/064887_03_03_30.wav]Evidence[/URL]
|
|
|
|
|
Re: beta weights - February 14, 2005 1:09:00 AM
|
|
|
Jon Newman
Posts: 1713
Joined: April 24, 2004
From: Amherst, WI
Status: offline
|
Well I asked someone about beta weights versus coefficients and this, in short what I was told:
"Anyway, good stat's question. The quick answer is that beta coefficients are not beta weights. Beta coefficients are simply the amount of movement in Y (the DV, dependent variable, or response variable) given a unit movement in that particular independent variable (IV). Beta coefficients are also called regression coefficients. Beta weights are standardized regression coefficients. That is to say, as X (an independent variable) increases by one standardized unit, how much does Y move in standardized units. Because beta weights are standardized, you can compare the relative efficacy of each independent variable to each other. They are partial correlation coefficients which enables you to say that one IV explains more variance in the DV than another."
For those with an interest here's one link that speaks to useful statistics measures.
[URL=http://www-personal.umich.edu/~sdcamp/up504/module+regression.html]link 1[/URL]
here's another that compares two statistical procedures
[URL=http://www.kellogg.nwu.edu/faculty/weber/recanati/t-ratios.htm]link 2[/URL]
jon
_____________________________
[URL=http://www.sonymusic.com/clips/selection/30/064887/064887_03_03_30.wav]Evidence[/URL]
|
|
|
|
|
Re: beta weights - February 14, 2005 1:31:00 AM
|
|
|
PTupdate.com
Posts: 1490
Joined: October 8, 2001
From: Pittsburgh, PA USA
Status: offline
|
Jon,
This is where an interesting aspect of PT and EBM clash. I'd bet that 85% of all PT's practicing don't have a clue as to what you are talking about in this post, and another 14.9% might have a slight clue, but still don't get it.....and most likely never will! The typical "mind" that leans one into the sciences, including PT, is usually not the one that excels at math and statistics.
I think we have all heard it before: Persons who are good at math tend to stink at the sciences such as biology (bulk memorization stuff), and vice versa.
Because of this, we are stuck with a very large percentage of our practicing population being unable to determine how good the evidence really is. Just because it was published in a peer reviewed journal means nothing, UNLESS each article is followed by a critical analysis by someobdy proficient.
How many PT's stop doing some treatment because some article says it was not efficacious? Most probably just read the abstract on Pubmed, and those that read the whole article did not know to look for a power analysis and/or indication of a type II error.
I think somehow this all needs to be tied together, and doubt PT's will ever have the time to really get good at the stats stuff on their own...myself included.
John Duffy, PT OCS
[URL=http://www.PTupdate.com]www.PTupdate.com[/URL]
_____________________________
John M. Duffy, PT
Board Certified Orthopaedic Clinical Specialist
www.PTupdate.com
|
|
|
|
|
Re: beta weights - February 14, 2005 2:09:00 AM
|
|
|
Jon Newman
Posts: 1713
Joined: April 24, 2004
From: Amherst, WI
Status: offline
|
Hi Duffy,
I think you're correct. Whenever a new study comes out the only thing we can do is compare our current clinical experience to that study. This provides some reason to track our successes. However, this itself gets sticky as one person's success is not necessarily another's. If everyone is working on a standardized goal it certainly makes comparison much easier; kind of like beta weights.
I think abstract reading is useful in that it gives people of flavor of what is going on and a lot of abstracts can be read quickly. But you're correct, to drastically change your practice pattern without critical thought would be foolish.
While those who are good at math may not be good at biology, I would argue that those who excel at biology are probably good at math. Or at least the math that's relevant.
jon
_____________________________
[URL=http://www.sonymusic.com/clips/selection/30/064887/064887_03_03_30.wav]Evidence[/URL]
|
|
|
|
|
Re: beta weights - February 14, 2005 2:58:00 AM
|
|
|
JLS_PT_OCS
Posts: 1684
Joined: January 30, 2005
From: USA
Status: offline
|
I am definitely in the group Duffy cited as not having a clue what you're talking about.
Even the brightest minds in PT today, producing great research, work with statisticians.
We as a profession need to be better at stats, myself included.
I have found some other ways to have summaries kind of broken down for me... I think John's PTUpdate and John Child's evidence in motion are two good examples of that.
J
_____________________________
Jason Silvernail DPT, OCS, CSCS
"It isn't what you're able to do that requires your courage but rather what you have come to understand and are willing to express." - Barrett Dorko,PT
**I no longer post on RehabEdge**
|
|
|
|
 New Messages |
 No New Messages |
 Hot Topic w/ New Messages |
 Hot Topic w/o New Messages |
 Locked w/ New Messages |
 Locked w/o New Messages |
|
 Post New Thread
Reply to Message
Post New Poll
Submit Vote
Delete My Own Post
Delete My Own Thread
Rate Posts |
|
0.078
|
Printable Version
