Pesky Statistical Interactions

[Gerónimo Maldonado]

Hi All,

I've constructed a binary regressional model (interactions included) and regretfully found 2 significant (p<0.05) interactions among my predictors. Question is... How to deal with these interactions??, I mean, should I leave them in my model???, should I leave them in my model and included them in my results??, is it bad to have them in my model???. I know how to interpret them, my question is really some technical stuff.

Thanks in advance.
--

[Steve Simon, P.Mean Consulting]

On 9/16/2010 8:03 AM, Gerónimo Maldonado wrote:

> I've constructed a binary regressional model (interactions included) and
> regretfully found 2 significant (p<0.05) interactions among my
> predictors. Question is... How to deal with these interactions??, I
> mean, should I leave them in my model???, should I leave them in my
> model and included them in my results??, is it bad to have them in my
> model???. I know how to interpret them, my question is really some
> technical stuff.

What did you write in your protocol. If your protocol is vague on this
point, then you can do whatever you please. But if the protocol spelled
out a certain approach, then you need to follow that approach or report
the alternative approach in your paper as a protocol deviation.

Even if you have latitude to do what you want, you still may be at a
loss as to what to do. An interaction in many situations is effectively
the same as finding a different effect in a subgroup. So you may want to
look at some of the literature on subgroup analysis. In particular, you
need to think about the scientific plausibility of the findings. It's
plausible to believe that men have a different response to some
medications than women if the medication is sensitive to various
hormones. But it is not plausible to believe that left-handed patients
have a different response to most medications that right-handed patients.

You also did not specify how you fit the model. Did you use a stepwise
approach or something similar where you compared multiple models with
different variables and added/removed variables based on their p-values?
In this case, the interaction might be spurious. Stepwise approaches
tend to inflate p-values, and this is especially true when there are a
large number of models being considered, as is the case with
interactions. There are far more potential interactions than there are
potential main effects.

Also, look at the type of interaction you have. Is it a quantitative
interaction (the effect of A is present for one level of B and absent or
the opposite direction for another level of B)? Is it a qualitative
interaction (the effect of A is in the same direction for all levels of
B, but for some levels it is somewhat stronger and for other levels it
is somewhat weaker). Ignoring a qualitative interaction is less serious
than ignoring a quantitative interaction.

If the goal of the model is prediction rather than inference about
individual predictors, AND if you have lots of data, put in every
interaction and compare its predictive power to a model that has no
interactions (don't look at anything in between). Hold out a portion of
your sample from the model fitting and see how the predictions work on
the hold-out portion compared to the portion that was used to fit the
data. If the predictions are great for the interactions model in the
portion used in estimation, but lousy in the portion held back, that is
very good evidence that the interactions are spurious.

I had a weird interaction in one of my studies and I reported it, but
with a rather skeptical tone. It did not re-occur in a replicated study,
so if I were doing it now, I would not report it at all.

For future studies, if interactions are troublesome, don't look for
them, especially not with stepwise approaches. There's nothing wrong
with saying that you will limit your attention to a certain class of
models if previous work in the area only considered models in that same
class. One such class of models is models with no interactions. Only
look for interactions if there is a scientific reason to believe that
they may be out there. If you do look for interactions when there is no
a priori reason to believe they exist, make sure you bill by the hour
and not by the project.
--
Steve Simon, Standard Disclaimer
Sign up for The Monthly Mean, the newsletter that
dares to call itself "average" at www.pmean.com/news

=====================
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

[Granaas, Michael]

"...regretfully found 2 significant (p<0.05) interactions among my predictors"

 

Why would you regret finding interactions?  Did someone tell you that doing so is wrong or rude?

 

Interactions are potentially so much more wonderful and exciting than main effects!  You should be celebrating, not hiding your head in shame.

 

The way you deal with interactions is to interpret them. 

 

Steve Simon has already give you some general advice on how to do so, so I won't repeat it--I'll simply encourage you to start trying to interpret those potentially exciting interactions.

 

Michael 

****************************************************
Michael Granaas             [hidden email]
Assoc. Prof.                Phone: [phone]
Dept. of Psychology         FAX:  [phone]
University of South Dakota
414 E. Clark St.
Vermillion, SD 57069
*****************************************************

---

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Gerónimo Maldonado [[hidden email]]
Sent: Thursday, September 16, 2010 8:03 AM
To: [hidden email]
Subject: Pesky Statistical Interactions

Hi All,

I've constructed a binary regressional model (interactions included) and regretfully found 2 significant (p<0.05) interactions among my predictors. Question is... How to deal with these interactions??, I mean, should I leave them in my model???, should I leave them in my model and included them in my results??, is it bad to have them in my model???. I know how to interpret them, my question is really some technical stuff.

Thanks in advance.
--

[Zdaniuk, Bozena-3]

in fact one may argue there is nothing but interactions in this world. Don't we want to start the answer to almost any question with "Well, it depends..."?
Bozena Zdaniuk
----- Original Message -----
From: "Michael Granaas" <[hidden email]>
To: [hidden email]
Sent: Thursday, September 16, 2010 6:43:26 AM GMT -08:00 US/Canada Pacific
Subject: Re: Pesky Statistical Interactions

"...regretfully found 2 significant (p<0.05) interactions among my predictors"

 

Why would you regret finding interactions?  Did someone tell you that doing so is wrong or rude?

 

Interactions are potentially so much more wonderful and exciting than main effects!  You should be celebrating, not hiding your head in shame.

 

The way you deal with interactions is to interpret them. 

 

Steve Simon has already give you some general advice on how to do so, so I won't repeat it--I'll simply encourage you to start trying to interpret those potentially exciting interactions.

 

Michael 

****************************************************
Michael Granaas             [hidden email]
Assoc. Prof.                Phone: [phone]
Dept. of Psychology         FAX:  [phone]
University of South Dakota
414 E. Clark St.
Vermillion, SD 57069
*****************************************************

---

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Gerónimo Maldonado [[hidden email]]
Sent: Thursday, September 16, 2010 8:03 AM
To: [hidden email]
Subject: Pesky Statistical Interactions

Hi All,

I've constructed a binary regressional model (interactions included) and regretfully found 2 significant (p<0.05) interactions among my predictors. Question is... How to deal with these interactions??, I mean, should I leave them in my model???, should I leave them in my model and included them in my results??, is it bad to have them in my model???. I know how to interpret them, my question is really some technical stuff.

Thanks in advance.
--

[Maurice Vergeer]

Hi,

I suggest you let theory guide you. If you don´t, you are bound to
come across interactions that may mean nothing. At least a theory
predicting what kind of interactions you could expect helps to
distinguish the nonsensical interactions from the theoretically
interesting ones.

Maurice


2010/9/16 Gerónimo Maldonado <[hidden email]>:

>
> Hi All,
>
> I've constructed a binary regressional model (interactions included) and
> regretfully found 2 significant (p<0.05) interactions among my predictors.
> Question is... How to deal with these interactions??, I mean, should I leave
> them in my model???, should I leave them in my model and included them in my
> results??, is it bad to have them in my model???. I know how to interpret
> them, my question is really some technical stuff.
>
> Thanks in advance.
> --
>
>

--
___________________________________________________________________
Maurice Vergeer
Department of communication
Radboud University�  (www.ru.nl)
PO Box 9104
NL-6500 HE Nijmegen
The Netherlands

Visiting Professor Yeungnam University, Gyeongsan, South Korea

contact:
E: [hidden email]
T: [phone] (direct)/ 3612372 (secretary) / maurice.vergeer (skype)
personal webpage: www.mauricevergeer.nl
blog:�  http://blog.mauricevergeer.nl/
Journalism: www.journalisteninhetdigitaletijdperk.nl
CENMEP New Media and European Parliament Elections 2009
http://mauricevergeer.ruhosting.nl/cenmep

Recent publications:
- Vergeer, M. & Pelzer, B. (2009). Consequences of media and Internet
use for offline and online network capital and well-being. A causal
model approach. Journal of Computer-Mediated Communication, 15,
189-210.
-Vergeer, M., Coenders, M. & Scheepers, P. (2009). Time spent on
television in European countries. In R.P. Konig, P.W.M. Nelissen, &
F.J.M. Huysmans (Eds.), Meaningful media: Communication Research on
the Social Construction of Reality (54-73). Nijmegen, The Netherlands:
Tandem Felix.
- Hermans, L., Vergeer, M., &�  d’Haenens, L. (2009). Internet in the
daily life of journalists. Explaining the use of the Internet through
work-related characteristics and professional opinions. Journal of
Computer-Mediated Communication, 15, 138-157.
___________________________________________________________________

=====================
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

[SR Millis-3]

I tend to use interactions sparingly.  I work in the biomedical field and have found the following interactions often helpful in predicting outcomes.

--Treatment and severity of disease being treated

--Age and risk factors

--Age and type of disease

--Race and disease

--Study center and treatment

--Quality and quantity of symptom

Scott
~~~~~~~~~~~
Scott R Millis, PhD, ABPP, CStat, CSci
Professor & Director of Research
Dept of Physical Medicine & Rehabilitation
Dept of Emergency Medicine
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email:  [hidden email]
Email:  [hidden email]
Tel: [phone]
Fax: [phone]

=====================
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