Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Why Proc glmpower doesnot work?_ repeated time

- RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Mute
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

Posted 08-10-2018 10:53 AM
(1594 views)

I try to find out a number of sample size and effect size. however, glmpower does not work repeated time.

what is wrong in the below sas codes?

options pagesize=60 linesize=80 pageno=1 nodate;

data GLMPOWER;

input treatment $ score6 score12 score18 score24 weight @27 ;

datalines;

plac 32 36 39 41 0.5

treat 35 40 48 50 0.5

;

proc print data = GLMPOWER;

TITLE 'GLMPOWER EXAMPLE';

RUN;

proc glm data =glmpower;

class treatment;

weight weight;

model score6 score12 score18 score24 = TREATMENT;

repeated TIME;

run;

PROC GLMPOWER DATA=GLMPOWER;

CLASS TREATMENT;

WEIGHT WEIGHT;

MODEL score6 score12 score18 score24 = TREATMENT;

repeated TIME;

POWER

EFFECTS=(TREATMENT)

MTEST = HLT

ALPHA = 0.05

POWER= .9

NTOTAL = .

STDDEV= 3.2

MATRIX("WELLCORR") = LEAR(.6, 1, 4, 1 2 3 4)

CORRMAT = "WELLCORR"

RUN;

ods graphics on;

proc glmpower data =glmpower;

class treatment;

weight weight;

model score6 score12 score18 score24 = treatment;

repeated time;

power

effects = (Treatment)

mtest= hlt

alpha= 0.05

ntotal= 800

power=.

stddev= 3.2

matrix ("wellCorr") = lear( .6, 1,4, 1 2 3 4)

Corrmat= "WellCorr";

plot x= n min=6 step= 3

vary (symbol by stddev, panel by dependent source)

xopts = (ref= 15 30 crossref= yes);

run;

ods graphics off;

9 REPLIES 9

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

The is very clear about what happens:

9 proc glm data =glmpower; 10 class treatment; 11 weight weight; 12 model score6 score12 score18 score24 = TREATMENT; 13 repeated TIME; 14 run; NOTE: No multivariate tests performed for TIME due toinsufficient error degrees of freedom. NOTE: No multivariate tests performed for TIME*treatment due to insufficient error degrees of freedom.

Your example data does not have enough records to get enough degrees of freedom to do requested actions.

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

In your correction, doesnot solve the problem still if added more subject

options pagesize=60 linesize=80 pageno=1 nodate;

data GLMPOWER;

input treatment $ score6 score12 score18 score24 weight @27 ;

datalines;

plac 41 50 61 32 0.5

treat 41 39 72 80 0.5

plac 41 32 12 23 0.5

plac 40 30 12 11 0.5

treat 41 36 39 41 0.5

;

proc print data = GLMPOWER;

TITLE 'GLMPOWER EXAMPLE';

RUN;

proc glm data =glmpower;

class treatment;

weight weight;

model score6 score12 score18 score24 = TREATMENT;

repeated TIME;

run;

PROC GLMPOWER DATA=GLMPOWER;

CLASS TREATMENT;

WEIGHT WEIGHT;

MODEL score6 score12 score18 score24 = TREATMENT;

repeated TIME;

POWER

EFFECTS=(TREATMENT)

UEPSDEF=HF

ALPHA = 0.05

POWER= .9

NTOTAL = .

STDDEV= 3.2

MATRIX("WELLCORR") = LEAR(.6, 1, 4, 1 2 3 4)

CORRMAT = "WELLCORR"

RUN;

UEPSDEF=HF

ods graphics on;

proc glmpower data =glmpower;

class treatment;

weight weight;

model score6 score12 score18 score24 = treatment;

repeated time;

power

effects = (Treatment)

mtest= hlt

alpha= 0.05

ntotal= 800

power=.

stddev= 3.2

matrix ("wellCorr") = lear( .6, 1,4, 1 2 3 4)

Corrmat= "WellCorr";

plot x= n min=6 step= 3

vary (symbol by stddev, panel by dependent source)

xopts = (ref= 15 30 crossref= yes);

run;

ods graphics off;

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

What makes you think is not working:

I get this as part of the results when I run your code:

The GLM Procedure

Repeated Measures Analysis of Variance

Repeated Measures Level Information | ||||
---|---|---|---|---|

Dependent Variable | score6 | score12 | score18 | score24 |

Level of TIME | 1 | 2 | 3 | 4 |

MANOVA Test Criteria and Exact F Statistics for the Hypothesis of no TIME Effect H = Type III SSCP Matrix for TIME E = Error SSCP Matrix S=1 M=0.5 N=-0.5 |
|||||
---|---|---|---|---|---|

Statistic | Value | F Value | Num DF | Den DF | Pr > F |

Wilks' Lambda | 0.13491054 | 2.14 | 3 | 1 | 0.4569 |

Pillai's Trace | 0.86508946 | 2.14 | 3 | 1 | 0.4569 |

Hotelling-Lawley Trace | 6.41231951 | 2.14 | 3 | 1 | 0.4569 |

Roy's Greatest Root | 6.41231951 | 2.14 | 3 | 1 | 0.4569 |

MANOVA Test Criteria and Exact F Statistics for the Hypothesis of no TIME*treatment Effect H = Type III SSCP Matrix for TIME*treatment E = Error SSCP Matrix S=1 M=0.5 N=-0.5 |
|||||
---|---|---|---|---|---|

Statistic | Value | F Value | Num DF | Den DF | Pr > F |

Wilks' Lambda | 0.26090443 | 0.94 | 3 | 1 | 0.6209 |

Pillai's Trace | 0.73909557 | 0.94 | 3 | 1 | 0.6209 |

Hotelling-Lawley Trace | 2.83282115 | 0.94 | 3 | 1 | 0.6209 |

Roy's Greatest Root | 2.83282115 | 0.94 | 3 | 1 | 0.6209 |

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

You have a single observation for each treatment. You do not have enough data to do what you want as shown unless you have significantly more data. Nothing is wrong with the code per se.

@jleexx35 wrote:

I try to find out a number of sample size and effect size. however, glmpower does not work repeated time.

what is wrong in the below sas codes?

options pagesize=60 linesize=80 pageno=1 nodate;

data GLMPOWER;

@input treatment $ score6 score12 score18 score24 weight @27 ;

datalines;

plac 32 36 39 41 0.5

treat 35 40 48 50 0.5

;

proc print data = GLMPOWER;

TITLE 'GLMPOWER EXAMPLE';

RUN;proc glm data =glmpower;

class treatment;

weight weight;

model score6 score12 score18 score24 = TREATMENT;

repeated TIME;

run;

PROC GLMPOWER DATA=GLMPOWER;

CLASS TREATMENT;

WEIGHT WEIGHT;

MODEL score6 score12 score18 score24 = TREATMENT;

repeated TIME;

POWER

EFFECTS=(TREATMENT)

MTEST = HLT

ALPHA = 0.05

POWER= .9

NTOTAL = .

STDDEV= 3.2

MATRIX("WELLCORR") = LEAR(.6, 1, 4, 1 2 3 4)

CORRMAT = "WELLCORR"

RUN;ods graphics on;

proc glmpower data =glmpower;

class treatment;

weight weight;

model score6 score12 score18 score24 = treatment;

repeated time;

power

effects = (Treatment)

mtest= hlt

alpha= 0.05

ntotal= 800

power=.

stddev= 3.2

matrix ("wellCorr") = lear( .6, 1,4, 1 2 3 4)

Corrmat= "WellCorr";

plot x= n min=6 step= 3

vary (symbol by stddev, panel by dependent source)

xopts = (ref= 15 30 crossref= yes);

run;

ods graphics off;

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

I think for understanding this model, I will explain more about it.

my main focus is to calculate a number of sample size using glmpower. I directly follow the SAS glmpower reference. In terms of the reference, the limited info means by time and groups. I exactly follow the step as the reference indicated. however GLM doesnot produce f test results and proglm doesnot run with the sas codes indicate in the reference. I want to get multiple simulations using proc powerglm but it doesnot produce any output, as well as code of 'repeated time' does not work( changed a red color).

can you share your idea in terms of glmpower not glm.

thank you so much!!

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

@jleexx35 wrote:

I think for understanding this model, I will explain more about it.

my main focus is to calculate a number of sample size using glmpower. I directly follow the SAS glmpower reference. In terms of the reference, the limited info means by time and groups. I exactly follow the step as the reference indicated. however GLM doesnot produce f test results and proglm doesnot run with the sas codes indicate in the reference. I want to get multiple simulations using proc powerglm but it doesnot produce any output, as well as code of 'repeated time' does not work( changed a red color).

can you share your idea in terms of glmpower not glm.

thank you so much!!

What reference? Are you trying to calculate a sample size? If so then you would use PROC POWER most likely...

Please explain what you're trying to achieve instead and then we can help to point you in the correct direction.

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

PROC GLM will not run with only the two observations. GLMPOWER works fine. The GLMPOWER, you were missing a semicolon and it seems to require integer weights, so I changed them to 1.

```
data GLMPOWER;
input treatment $ score6 score12 score18 score24 weight @27;
datalines;
plac 32 36 39 41 1
treat 35 40 48 50 1
;
proc print data=GLMPOWER;
TITLE 'GLMPOWER EXAMPLE';
RUN;
PROC GLMPOWER DATA=GLMPOWER;
CLASS TREATMENT;
WEIGHT WEIGHT;
MODEL score6 score12 score18 score24=TREATMENT;
repeated TIME;
POWER EFFECTS=(TREATMENT) MTEST=HLT ALPHA=0.05 POWER=.9 NTOTAL=.
STDDEV=3.2 MATRIX("WELLCORR")=LEAR(.6, 1, 4, 1 2 3 4) CORRMAT="WELLCORR";
RUN;
ods graphics on;
proc glmpower data=glmpower;
class treatment;
weight weight;
model score6 score12 score18 score24=treatment;
repeated time;
power effects=(Treatment) mtest=hlt alpha=0.05 ntotal=800 power=.
stddev=3.2 matrix ("WELLCORR")=lear(.6, 1, 4, 1 2 3 4) Corrmat="WELLCORR";
plot x=n min=6 step=3 vary (symbol by stddev, panel by dependent source)
xopts=(ref=15 30 crossref=yes);
run;
ods graphics off;
```

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

If proc powerglm codes are fine, please share your output with me?

I face continue errors, not codes are not computed at all.

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

Run the code I posted above and include the errors shown in your response.

**Available on demand!**

Missed SAS Innovate Las Vegas? Watch all the action for free! View the keynotes, general sessions and 22 breakouts on demand.

What is ANOVA?

ANOVA, or Analysis Of Variance, is used to compare the averages or means of two or more populations to better understand how they differ. Watch this tutorial for more.

Find more tutorials on the SAS Users YouTube channel.