This preview shows pages 1–3. Sign up to view the full content.
JT: Ch 5 Q 12 or 15
Megan Goldman: Reliable also needs to take into account the correlation.
.. even if you're inside the range
of the original data, if the association is fairly weak your guess may be not that great.
Megan Goldman: If it's extrapolation, I'd go with "rather dubious."
On interpolation with strong
assocation (fairly high r), I'd go with "quite reliable".
Interpolation with a weaker assocation, I'd say
"possible justified".
Megan Goldman: r = .59 isn't anything to write home about.
JT: ok thanks
Tiffany: so megan is there a time when it does matter if regression of x on y or vice versa
Tiffany: like does it ever change any properties
Megan Goldman: It matters very much.
The slope will change!
Tiffany: bc of which one we divide by right
Megan Goldman: If you're predicting Y using X, the slope is (r * sdy)/sdx. However, if you're predicting
X using Y, then it's (r * sdx)/sdy.
.. the roles of x and y reverse.
Megan Goldman: Exactly.
Tiffany: oh okay well im going to read about ch. 6, i know you got a lot of questions about that
Megan Goldman: So, if you've done a regression and found Y = slope * X + intercept, and then you're
given a value of Y, do NOT use this equation to predict X!
Tiffany: is there any way to look at the earlier conversations
Tiffany: maybe someone ask the same question
Megan Goldman: You'd need to rerun and get X = slope * Y
+ intercept.
Megan Goldman: Not that I've found.
I'd imagine over the net day or so people will start asking Ch 6
questions on the discussion board, and so hints will start popping up there as well.
Megan Goldman: Hi melinda and Anna.
melinda: hi Megan. can i ask u a few questions about ch.4 hwk?
melinda: hi
Megan Goldman: Sure.
Anna Park: hello
Anna Park: i also have questions about the chapter 4 homework
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentMegan Goldman: Okay.
melinda: my first question is about q8.
..i have submitted four times and put that the graph was no b/c of
nonlinearity and association being weak which was wrong.
..i think it might also be hetero.
..not really
sure.
..the link for my graph is http://s294.photobucket.com/albums/mm118/melindakhoshabe/?
Megan Goldman: Let me try to pull that up.
melinda: okay thanks for the help
Anna Park: yeah my graph sort of looks like that and i'm having trouble with that question too
Tiffany: dont put weak association
Megan Goldman: heh, this is exactly the graph Serene was looking at earlier.
Megan Goldman: Tiffany is our resident expert on this problem. ;)
Tiffany: yeah i hate those graphs
melinda: could you help me in how to approach the answers to this problem?
Tiffany: it took 2 gsi and the professor
Tiffany: haha im very bitter about these graphs
Anna Park: i am nearing that sentiment
melinda: tiffany are u sure not to put weak assocaition, can u look at my graph
melinda: so what should i put for this problem?
Tiffany: if it is weak
This is the end of the preview. Sign up
to
access the rest of the document.
 Summer '08
 anderes
 Statistics, Correlation, Probability

Click to edit the document details