Quantcast

nested structure for Ancova

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
4 messages Options
Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

nested structure for Ancova

li li-13
Hi All,
  I have a dataset which contains 4 variables: area, group, time, y,
Area is a factor that has two levels A and B, group is a factor that is
nested within area. There are four groups within each area.
y is the response variable, and time refers to different days.

  Below is the how data looks like.

  First, I fit separate ancova for each area (area A and B), For each area,
I obtained different regression lines for each group within that area.

  mod1 <- lm(y ~ month * group, data=two[two$area=="A"]

  mod2 <- lm(y ~ month * group, data=two[two$area=="B"]

 I want to fit the model at one time by using the nested structure

 mod3 <-  lm(y ~ -1+ month * (area/group), data=two)

I get different results using mod 3 from using mod1 and mod2.

  Can someone give some suggestion on this.  How can I specify the model in
R to simultaneously fit the model to get the same results as from mod1 and
mod2.

Thanks  very much!
    Hanna

> head(two[two$area=="A",])   group time         y area
79     1    3 -1.394327    A
80     1    6 -1.435485    A
81     1    9 -1.406497    A
82     1   12 -1.265848    A
83     1    0 -1.316768    A
84     1    6 -1.431292    A> head(two[two$area=="B",])  group time
     y area
1     1    0 -2.145581    B
2     1    0 -1.910543    B
3     1    0 -2.128632    B
4     1    3 -2.079442    B
5     1    3 -2.273026    B
6     1    6 -2.312635    B

        [[alternative HTML version deleted]]

______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: nested structure for Ancova

Michael Dewey-3
Dear Hannah

You say they are different (mod3 from mod1 and mod2) but in what way are
the results different?

On 12/03/2017 15:38, li li wrote:

> Hi All,
>   I have a dataset which contains 4 variables: area, group, time, y,
> Area is a factor that has two levels A and B, group is a factor that is
> nested within area. There are four groups within each area.
> y is the response variable, and time refers to different days.
>
>   Below is the how data looks like.
>
>   First, I fit separate ancova for each area (area A and B), For each area,
> I obtained different regression lines for each group within that area.
>
>   mod1 <- lm(y ~ month * group, data=two[two$area=="A"]
>
>   mod2 <- lm(y ~ month * group, data=two[two$area=="B"]
>
>  I want to fit the model at one time by using the nested structure
>
>  mod3 <-  lm(y ~ -1+ month * (area/group), data=two)
>
> I get different results using mod 3 from using mod1 and mod2.
>
>   Can someone give some suggestion on this.  How can I specify the model in
> R to simultaneously fit the model to get the same results as from mod1 and
> mod2.
>
> Thanks  very much!
>     Hanna
>
>> head(two[two$area=="A",])   group time         y area
> 79     1    3 -1.394327    A
> 80     1    6 -1.435485    A
> 81     1    9 -1.406497    A
> 82     1   12 -1.265848    A
> 83     1    0 -1.316768    A
> 84     1    6 -1.431292    A> head(two[two$area=="B",])  group time
>      y area
> 1     1    0 -2.145581    B
> 2     1    0 -1.910543    B
> 3     1    0 -2.128632    B
> 4     1    3 -2.079442    B
> 5     1    3 -2.273026    B
> 6     1    6 -2.312635    B
>
> [[alternative HTML version deleted]]
>
> ______________________________________________
> [hidden email] mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

--
Michael
http://www.dewey.myzen.co.uk/home.html

______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: nested structure for Ancova

Richard M. Heiberger
In reply to this post by li li-13
I think you need either

mod4 <- lm( y ~ -1 + area / (month*group), data=two)

mod5 <- lm( y ~ area / (month*group), data=two)

With either of those,  area:month and area:group and Residuals add up.


On Sun, Mar 12, 2017 at 10:39 li li <[hidden email]> wrote:

> Hi All,
>   I have a dataset which contains 4 variables: area, group, time, y,
> Area is a factor that has two levels A and B, group is a factor that is
> nested within area. There are four groups within each area.
> y is the response variable, and time refers to different days.
>
>   Below is the how data looks like.
>
>   First, I fit separate ancova for each area (area A and B), For each area,
> I obtained different regression lines for each group within that area.
>
>   mod1 <- lm(y ~ month * group, data=two[two$area=="A"]
>
>   mod2 <- lm(y ~ month * group, data=two[two$area=="B"]
>
>  I want to fit the model at one time by using the nested structure
>
>  mod3 <-  lm(y ~ -1+ month * (area/group), data=two)
>
> I get different results using mod 3 from using mod1 and mod2.
>
>   Can someone give some suggestion on this.  How can I specify the model in
> R to simultaneously fit the model to get the same results as from mod1 and
> mod2.
>
> Thanks  very much!
>     Hanna
>
> > head(two[two$area=="A",])   group time         y area
> 79     1    3 -1.394327    A
> 80     1    6 -1.435485    A
> 81     1    9 -1.406497    A
> 82     1   12 -1.265848    A
> 83     1    0 -1.316768    A
> 84     1    6 -1.431292    A> head(two[two$area=="B",])  group time
>      y area
> 1     1    0 -2.145581    B
> 2     1    0 -1.910543    B
> 3     1    0 -2.128632    B
> 4     1    3 -2.079442    B
> 5     1    3 -2.273026    B
> 6     1    6 -2.312635    B
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> [hidden email] mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

        [[alternative HTML version deleted]]

______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: nested structure for Ancova

li li-13
Thanks so much, Richard. That  works.
 Hanna

2017-03-12 14:31 GMT-04:00 Richard M. Heiberger <[hidden email]>:

> I think you need either
>
> mod4 <- lm( y ~ -1 + area / (month*group), data=two)
>
> mod5 <- lm( y ~ area / (month*group), data=two)
>
> With either of those,  area:month and area:group and Residuals add up.
>
>
> On Sun, Mar 12, 2017 at 10:39 li li <[hidden email]> wrote:
>
>> Hi All,
>>   I have a dataset which contains 4 variables: area, group, time, y,
>> Area is a factor that has two levels A and B, group is a factor that is
>> nested within area. There are four groups within each area.
>> y is the response variable, and time refers to different days.
>>
>>   Below is the how data looks like.
>>
>>   First, I fit separate ancova for each area (area A and B), For each
>> area,
>> I obtained different regression lines for each group within that area.
>>
>>   mod1 <- lm(y ~ month * group, data=two[two$area=="A"]
>>
>>   mod2 <- lm(y ~ month * group, data=two[two$area=="B"]
>>
>>  I want to fit the model at one time by using the nested structure
>>
>>  mod3 <-  lm(y ~ -1+ month * (area/group), data=two)
>>
>> I get different results using mod 3 from using mod1 and mod2.
>>
>>   Can someone give some suggestion on this.  How can I specify the model
>> in
>> R to simultaneously fit the model to get the same results as from mod1 and
>> mod2.
>>
>> Thanks  very much!
>>     Hanna
>>
>> > head(two[two$area=="A",])   group time         y area
>> 79     1    3 -1.394327    A
>> 80     1    6 -1.435485    A
>> 81     1    9 -1.406497    A
>> 82     1   12 -1.265848    A
>> 83     1    0 -1.316768    A
>> 84     1    6 -1.431292    A> head(two[two$area=="B",])  group time
>>      y area
>> 1     1    0 -2.145581    B
>> 2     1    0 -1.910543    B
>> 3     1    0 -2.128632    B
>> 4     1    3 -2.079442    B
>> 5     1    3 -2.273026    B
>> 6     1    6 -2.312635    B
>>
>>         [[alternative HTML version deleted]]
>>
>> ______________________________________________
>> [hidden email] mailing list -- To UNSUBSCRIBE and more, see
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide http://www.R-project.org/
>> posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
>

        [[alternative HTML version deleted]]

______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Loading...