regression - nonsig to sign

[Zdaniuk, Bozena-2]

Hello, when I enter predictor A by itself into regression model, it is not significant. But if I add predictors B and C, predictor A becomes significant. How can I interpret it?
thanks a lot.
Bozena

________________________________________

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Hector Maletta]

Bozena,
It may mean and B and C mask the effect of A. Once B and C are controlled
for, the net effect of A independently of B and C is significantly different
from zero.
Remember again that significance is not relevance or substantive
meaningfulness. Significance means that the size of your sample allows you
to reject the null hypothesis that the coefficient is zero in the
population, given the size of the coefficient in the sample. If you had a
much larger sample, the simple regression using only A may turn to be
significant (i.e. the small BETA(A) coefficient would be confidently said to
be different from zero in the population, with less than 5% probability of
error). This would not make it more relevant, more causal, or more
meaningful.

Hector




-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Zdaniuk, Bozena
Sent: 19 May 2008 10:35
To: [hidden email]
Subject: regression - nonsig to sign

Hello, when I enter predictor A by itself into regression model, it is not
significant. But if I add predictors B and C, predictor A becomes
significant. How can I interpret it?
thanks a lot.
Bozena

________________________________________

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Swank, Paul R]

There are two ways in which this can happen. First and most commonly, it
is as Hector has said, the variables B & C relate to variability in the
dependent variable (call it Y) that is not associated with A. Ten the
effect of A stands out more once this other variance is removed. This is
the reason we often have covariates in our models, the control for other
sources of variation in Y. So we increase the power of our test on A.
The other was is when B &/or C are related to A and reduce the
extraneous variance in A that is not related to Y. This is often called
a suppressor effect. Thus, it would help to look at the whole
correlation matrix of A, B, C and Y to see how these variables
interrelate.

Paul R. Swank, Ph.D.
Professor and Director of Research
Children's Learning Institute
University of Texas Health Science Center - Houston


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Zdaniuk, Bozena
Sent: Monday, May 19, 2008 8:35 AM
To: [hidden email]
Subject: regression - nonsig to sign

Hello, when I enter predictor A by itself into regression model, it is
not significant. But if I add predictors B and C, predictor A becomes
significant. How can I interpret it?
thanks a lot.
Bozena

________________________________________

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Zdaniuk, Bozena-2]

Thanks a lot, this is very helpful.
I think I have a bit of both. B & C are related to both Y and A, although a bit stronger to Y. In any case, I think I can conclude that adding B & C to the model allows for 'flushing out' the relation btw A and Y that would otherwise be masked, correct?
Bozena

Bozena Zdaniuk, Ph.D.
University of Pittsburgh
UCSUR, 6th Fl.
121 University Place
Pittsburgh, PA 15260
Ph.: 412-624-5736
Fax: [phone]
Email: [hidden email]

-----Original Message-----
From: Swank, Paul R [mailto:[hidden email]]
Sent: Monday, May 19, 2008 2:52 PM
To: Zdaniuk, Bozena; [hidden email]
Subject: RE: regression - nonsig to sign

There are two ways in which this can happen. First and most commonly, it
is as Hector has said, the variables B & C relate to variability in the
dependent variable (call it Y) that is not associated with A. Ten the
effect of A stands out more once this other variance is removed. This is
the reason we often have covariates in our models, the control for other
sources of variation in Y. So we increase the power of our test on A.
The other was is when B &/or C are related to A and reduce the
extraneous variance in A that is not related to Y. This is often called
a suppressor effect. Thus, it would help to look at the whole
correlation matrix of A, B, C and Y to see how these variables
interrelate.

Paul R. Swank, Ph.D.
Professor and Director of Research
Children's Learning Institute
University of Texas Health Science Center - Houston


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Zdaniuk, Bozena
Sent: Monday, May 19, 2008 8:35 AM
To: [hidden email]
Subject: regression - nonsig to sign

Hello, when I enter predictor A by itself into regression model, it is
not significant. But if I add predictors B and C, predictor A becomes
significant. How can I interpret it?
thanks a lot.
Bozena

________________________________________

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Swank, Paul R]

You can also check the regression coefficients and their standard
errors. If the coefficient when A is in the model is roughly the same as
when A, B, & C are in the model but the standard error is smaller, then
it looks like you are controlling for variance in Y. If, however, the
regression coefficient for A when B & C are in the model is quite
different than when A is alone, then it may be suppression.

Paul R. Swank, Ph.D.
Professor and Director of Research
Children's Learning Institute
University of Texas Health Science Center - Houston


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Zdaniuk, Bozena
Sent: Monday, May 19, 2008 2:12 PM
To: [hidden email]
Subject: Re: regression - nonsig to sign

Thanks a lot, this is very helpful.
I think I have a bit of both. B & C are related to both Y and A,
although a bit stronger to Y. In any case, I think I can conclude that
adding B & C to the model allows for 'flushing out' the relation btw A
and Y that would otherwise be masked, correct?
Bozena

Bozena Zdaniuk, Ph.D.
University of Pittsburgh
UCSUR, 6th Fl.
121 University Place
Pittsburgh, PA 15260
Ph.: 412-624-5736
Fax: [phone]
Email: [hidden email]

-----Original Message-----
From: Swank, Paul R [mailto:[hidden email]]
Sent: Monday, May 19, 2008 2:52 PM
To: Zdaniuk, Bozena; [hidden email]
Subject: RE: regression - nonsig to sign

There are two ways in which this can happen. First and most commonly, it
is as Hector has said, the variables B & C relate to variability in the
dependent variable (call it Y) that is not associated with A. Ten the
effect of A stands out more once this other variance is removed. This is
the reason we often have covariates in our models, the control for other
sources of variation in Y. So we increase the power of our test on A.
The other was is when B &/or C are related to A and reduce the
extraneous variance in A that is not related to Y. This is often called
a suppressor effect. Thus, it would help to look at the whole
correlation matrix of A, B, C and Y to see how these variables
interrelate.

Paul R. Swank, Ph.D.
Professor and Director of Research
Children's Learning Institute
University of Texas Health Science Center - Houston


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Zdaniuk, Bozena
Sent: Monday, May 19, 2008 8:35 AM
To: [hidden email]
Subject: regression - nonsig to sign

Hello, when I enter predictor A by itself into regression model, it is
not significant. But if I add predictors B and C, predictor A becomes
significant. How can I interpret it?
thanks a lot.
Bozena

________________________________________

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Ian Straus]

Bozena,
You interpret it in a straightforward manner.
Predictors B and C explain most of the variance from the mean.
Indeed, if they are not included in the model then their stronger
effects cover up A.
Once they have been used in the model, predictor A explains a
significant part of the residuals.

Ian Straus
Market Research Specialist
Public Affairs Division.
VIA Metropolitan Transit
San Antonio, Texas
[phone]



------------------------------

Date:    Mon, 19 May 2008 09:35:07 -0400
From:    "Zdaniuk, Bozena" <[hidden email]>
Subject: regression - nonsig to sign

Hello, when I enter predictor A by itself into regression model, it is
not significant. But if I add predictors B and C, predictor A becomes
significant. How can I interpret it?
thanks a lot.
Bozena

________________________________________



This transmission is for the intended addressee only. Unless you are the addressee (or authorized to receive for the addressee), you may not use, copy or disclose to anyone the information contained in the message. If you have received this message in error, please advise the sender by reply e-mail and delete the message.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD