a***e
发帖数: 1627 | 1 A production engineer wishes to study the effect of cutting temperature and
cutting pressure on the surface finish of the machined component. He designs
an experiment wherein three levels of each factor are selected, and a
factorial experiment with two replicates in run. The relevant data in
certain standard units are given as follows.
Temperature
Pressure Low Medium High
Low 51.5 51.8 51.3
51.3 51.7 51.5
Medium 51.2 51.6 49.9
51.4 51.7 51.2
High 51.6 51.9 51.5
51.8 51.8 51.2
选two-way anova的model
我自己觉得应该是random，model_II但是同学说应该fixed,model_I，我就动摇了。。
这应该怎么分啊？ 这些不都是从large population里挑出来的吗？
请大家帮忙看看，谢谢 |
p*7
发帖数: 45 | 2 我刚刚开始学这个也不懂。我的感觉是FIXED，two way anova with multpile
replicants in each cell. Because there are only three levels in each factors
and all the levels are included here, it is fixed effect model. If there
are a lot of levels and we randomly selecet three, it will be random effect.
I have another question. Even we randomly select three from a population,
with only three samples, how we are going to estimate variance of the random
effect?
and
designs
【在 a***e 的大作中提到】
: A production engineer wishes to study the effect of cutting temperature and
: cutting pressure on the surface finish of the machined component. He designs
: an experiment wherein three levels of each factor are selected, and a
: factorial experiment with two replicates in run. The relevant data in
: certain standard units are given as follows.
: Temperature
: Pressure Low Medium High
: Low 51.5 51.8 51.3
: 51.3 51.7 51.5
: Medium 51.2 51.6 49.9
|
g***l
发帖数: 22 | 3 Fixed effect。这里两个variable每一个都有三个level，这些都是固定的。random
effect一般用来构造hierarchical model来处理一些lurking variables，允许对于不
同的level有不同的variance，这里不存在的。
and
designs
【在 a***e 的大作中提到】
: A production engineer wishes to study the effect of cutting temperature and
: cutting pressure on the surface finish of the machined component. He designs
: an experiment wherein three levels of each factor are selected, and a
: factorial experiment with two replicates in run. The relevant data in
: certain standard units are given as follows.
: Temperature
: Pressure Low Medium High
: Low 51.5 51.8 51.3
: 51.3 51.7 51.5
: Medium 51.2 51.6 49.9
|
g***l
发帖数: 22 | 4 How can there be lots of levels? L,M and H are predetermined before
collecting data.
And if there are only three samples it doesn't make sense to use random
effect.
factors
effect.
random
【在 p*7 的大作中提到】
: 我刚刚开始学这个也不懂。我的感觉是FIXED，two way anova with multpile
: replicants in each cell. Because there are only three levels in each factors
: and all the levels are included here, it is fixed effect model. If there
: are a lot of levels and we randomly selecet three, it will be random effect.
: I have another question. Even we randomly select three from a population,
: with only three samples, how we are going to estimate variance of the random
: effect?
:
: and
: designs
|
i**z
发帖数: 194 | 5 要看样本，fixed or random 是看如何实验设计的。
如果就这三个，那么是fixed的。
and
designs
【在 a***e 的大作中提到】
: A production engineer wishes to study the effect of cutting temperature and
: cutting pressure on the surface finish of the machined component. He designs
: an experiment wherein three levels of each factor are selected, and a
: factorial experiment with two replicates in run. The relevant data in
: certain standard units are given as follows.
: Temperature
: Pressure Low Medium High
: Low 51.5 51.8 51.3
: 51.3 51.7 51.5
: Medium 51.2 51.6 49.9
|
r***1
发帖数: 171 | 6 可以做个hausman test, 就可以检测哪个模型更适合？ |
a***e
发帖数: 1627 | 7 问了老师，说是fixed ，谢谢大家，我就是对这些很糊涂 |
a***e
发帖数: 1627 | 8 A production engineer wishes to study the effect of cutting temperature and
cutting pressure on the surface finish of the machined component. He designs
an experiment wherein three levels of each factor are selected, and a
factorial experiment with two replicates in run. The relevant data in
certain standard units are given as follows.
Temperature
Pressure Low Medium High
Low 51.5 51.8 51.3
51.3 51.7 51.5
Medium 51.2 51.6 49.9
51.4 51.7 51.2
High 51.6 51.9 51.5
51.8 51.8 51.2
选two-way anova的model
我自己觉得应该是random，model_II但是同学说应该fixed,model_I，我就动摇了。。
这应该怎么分啊？ 这些不都是从large population里挑出来的吗？
请大家帮忙看看，谢谢 |
p*7
发帖数: 45 | 9 我刚刚开始学这个也不懂。我的感觉是FIXED，two way anova with multpile
replicants in each cell. Because there are only three levels in each factors
and all the levels are included here, it is fixed effect model. If there
are a lot of levels and we randomly selecet three, it will be random effect.
I have another question. Even we randomly select three from a population,
with only three samples, how we are going to estimate variance of the random
effect?
and
designs
【在 a***e 的大作中提到】
: A production engineer wishes to study the effect of cutting temperature and
: cutting pressure on the surface finish of the machined component. He designs
: an experiment wherein three levels of each factor are selected, and a
: factorial experiment with two replicates in run. The relevant data in
: certain standard units are given as follows.
: Temperature
: Pressure Low Medium High
: Low 51.5 51.8 51.3
: 51.3 51.7 51.5
: Medium 51.2 51.6 49.9
|
g***l
发帖数: 22 | 10 Fixed effect。这里两个variable每一个都有三个level，这些都是固定的。random
effect一般用来构造hierarchical model来处理一些lurking variables，允许对于不
同的level有不同的variance，这里不存在的。
and
designs
【在 a***e 的大作中提到】
: A production engineer wishes to study the effect of cutting temperature and
: cutting pressure on the surface finish of the machined component. He designs
: an experiment wherein three levels of each factor are selected, and a
: factorial experiment with two replicates in run. The relevant data in
: certain standard units are given as follows.
: Temperature
: Pressure Low Medium High
: Low 51.5 51.8 51.3
: 51.3 51.7 51.5
: Medium 51.2 51.6 49.9
|
|
|
g***l
发帖数: 22 | 11 How can there be lots of levels? L,M and H are predetermined before
collecting data.
And if there are only three samples it doesn't make sense to use random
effect.
factors
effect.
random
【在 p*7 的大作中提到】
: 我刚刚开始学这个也不懂。我的感觉是FIXED，two way anova with multpile
: replicants in each cell. Because there are only three levels in each factors
: and all the levels are included here, it is fixed effect model. If there
: are a lot of levels and we randomly selecet three, it will be random effect.
: I have another question. Even we randomly select three from a population,
: with only three samples, how we are going to estimate variance of the random
: effect?
:
: and
: designs
|
i**z
发帖数: 194 | 12 要看样本，fixed or random 是看如何实验设计的。
如果就这三个，那么是fixed的。
and
designs
【在 a***e 的大作中提到】
: A production engineer wishes to study the effect of cutting temperature and
: cutting pressure on the surface finish of the machined component. He designs
: an experiment wherein three levels of each factor are selected, and a
: factorial experiment with two replicates in run. The relevant data in
: certain standard units are given as follows.
: Temperature
: Pressure Low Medium High
: Low 51.5 51.8 51.3
: 51.3 51.7 51.5
: Medium 51.2 51.6 49.9
|
r***1
发帖数: 171 | 13 可以做个hausman test, 就可以检测哪个模型更适合？ |
a***e
发帖数: 1627 | 14 问了老师，说是fixed ，谢谢大家，我就是对这些很糊涂 |
t******y
发帖数: 100 | 15 我认为如果整个实验用的只有同样两个replicates，那这两个replicates就可以算作
random effect,相当于在每个replicate上采集9个数据。一共18个数据。每个cell里一
个数据。如果replicates是九对不同的，那就是fixed effect model, 3*3 two way
anova |
