Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.
|
|
|
|
[ap-stat] Re: regression questions......
Posted:
Oct 12, 2005 3:26 PM
|
|
Michael Berkley wrote:
>hello all.....
>i am a first year ap stat teacher and have a few questions that my
>kids do also.......
>
>1. why do we take log of y-values and then transform to exponential
> equation? it seems to me that we could just do a exponential
> regression and skip all short work....
>
(1) The AP syllabus specifies that we teach power and log
transformations to achieve linearity.
(2) Converting a logy = a+bx equation into exponential form is not
really the same as exponential regression, even though the calculator
calls it that.
(3) Students are not required to make that conversion. They do need to
be able to work with the log y equation.
>
>2. Same question for power regression, why take log of x's and y's
> then convert? why not just do power regression and skip work?
>
Same answers. And again, don't convert. Kids need to be able to use the
logy = a+blogx equation to make numerical estimates.
>
>3. Can r-values from different regressions simply be 'compared'?
> namely, can we just do a bunch of regressions and then pick the
> one with the best r value?
>
With what objective? First, r applies only to linear regressions. And a
high r-value does not mean that the model is appropriate. Only
randomness in the residuals can address that issue.
>
>4. What is the round-off rule? most of the time i use one more
> decimal than what is given in data, this is what my stat book for
> college does, is this standard? is there a guideline for ap exam?
>
There's no fixed rule. One needs to keep enough decimal places to make
answers reasonably reliable, but it's silly to report answers to the
apparently-precise-but-meaningless number of places the calculator
coughs up. On the AP exam we care far less about the accuracy of the
decimal places than about the proper application of statistical methods
and clear interpretation of results.
- Dave Bock
>.
>
>
>
====
Course related websites:
AP Central: http://apcentral.collegeboard.com/stats
FAQ: http://www.mrderksen.com/faq.htm
Archives: http://mathforum.org/kb/forum.jspa?forumID=67
ap-stat is an Electronic Discussion Group (EDG) of The College Board, 45 Columbus Avenue, New York, NY 10023-6992
To UNSUBSCRIBE or CHANGE YOUR EMAIL ADDRESS, please use the list website:
http://lyris.collegeboard.com/cgi-bin/lyris.pl?enter=ap-stat
Send questions about the list to owner-ap-stat@lyris.collegeboard.com
|
|
|
|
