two way ANOVA test to compare three different groups

[Barkamih]

 Hi all 

I want to do an ANOVA test to compare this kind of data, I have three different groups NT as a control group, ND and PTX are the treatment groups,  each group has three different readings 24h, 48h and 72h. my data looks like below. So, what is my code should look like?

example, I want to compare 24h values of NT (control) by 24h of ND and PTX. as well as 48h and 72h. 

I hope you got my  point 

regards 

Hours	Treatment	Value
24H	NT	96.82479
24H	NT	107.7479
24H	NT	95.42732
24H	ND	100.4287
24H	ND	102.0154
24H	ND	98.723
24H	PTX 2.5nM/ml	58.0074
24H	PTX 2.5nM/ml	63.0915
24H	PTX 2.5nM/ml	70.4211
48H	NT	96.82479
48H	NT	96.82479
48H	NT	95.42732
48H	ND	102.2046
48H	ND	97.09953
48H	ND	100.4259
48H	PTX 2.5nM/ml	58.23748
48H	PTX 2.5nM/ml	58.83639
48H	PTX 2.5nM/ml	54.70102
72H	NT	96.82479
72H	NT	107.7479
72H	NT	95.42732
72H	ND	103.9728
72H	ND	98.43996
72H	ND	96.73638
72H	PTX 2.5nM/ml	34.60215
72H	PTX 2.5nM/ml	43.03356
72H	PTX 2.5nM/ml	35.74294

[sld]

Please show us the code you have tried, including code for data input.

If this is a repeated measures study, then you will want to use either MIXED or GLIMMIX, and you'll need to add a subject identifier to your data set.

I hope this helps.

[Barkamih]

I didn't use any code yet, and the importing data was done normally because I have this data in excel sheet, my

data named cancer, this is a clinical data.

regards 

[Rick_SAS · Accepted solution]

This is a balanced repeated-values ANOVA. When you don't have missing values or mistimed measurements (as seems to be the case) the traditional approach would be to use PROC GLM and a REPEATED statement, as shown in this example from the GLM documentation.  The modern approach (which sld advocates) is to use PROC MIXED, as shown in this example from the PROC MIXED documentation.

In both cases, the initial MODEL statement will be 

CLASS Hours Treatment;

MODEL Values = Hours Treatment Hours*Treatment;

[aska_ujita]

Hello Rick!

If I have the same situation, but one data for each group, what PROC can I use?

Example:

DAY	GROUP	MEAN
1	1	40
3	1	20
5	1	10
1	2	60
3	2	55
5	2	30

I would like to compare (P value) the mean between group 1 and 2 in each day (1, 3, 6).

Thank you.

Aska.

[Rick_SAS]

The data does not match your question, but you can use the BY statement to compare samples on various days to each other.

The example data that you provide does not have any variation, but If you have the raw data, then you can use

proc glm;

by day;

class group;

model Y = group;

run;

[aska_ujita]

Hello Rick, 

Yes, it haven't variation. Is just one information per day for each group.

But is still possible to make a simple analysis to compare?

Actually, I have measured temperament scores (0, 1, 2, 3, 4 and 5) in 40 animals (3 and 4 years old, distributed randomly) that is divided in two treatments: Treated group and control group.
I collected the temperament in day 1, 2, 3, 4 and 5 of experiment.

So... I would like to know if is possible to organize these scores in percentage (%) and make a table automatically by sas. I know that I can see the percentage by PROC FREQ, but it is possible to make a table with these output datas by sas? Than maybe I can use the procedure that you advise me before...??

thank you.

Aska.