repeated measures ANOVA

[versace]

Hi,

I want to run a 2 way repeated mesures ANOVA with Labview 7.0

I am actually confused about how to use the anova VI included in the package.

As an example: in my experiment I have 10 subjects. The dependent variable is the performance to a test.

 The test session is run two times (Factor 1: "Time" with two levels)

The test is  "easy" or "difficult" (Factor 2: "difficulty" with two levels). 

Each subjects is put in all the 4 conditions, hence, for each subject, I have 4 scores: time 1 easy test, time 1 difficult test, time 2 easy test, time 2 difficult test.

I don't understand from where the VI gets the information if the experimental design is within subjects (like in this case) or between subjects.

Is the terminal "observations per cell" that is telling to the VI if the experimental design is within subjects or between subjects?

After solving the problem about the within or repeated measures design, I have another question:

in this case, should I actually use the 3 D ANOVA vi and put the subjects as a third random factor?

Thank you very much

[Megan B.]

Hi versace,

The Help files for the ANOVA VIs were really beefed up in later versions of LabVIEW and even include some examples.  I think that they might contain just the information you need. 

Check out the 1D ANOVA, 2D ANOVA, and 3D ANOVA Help files.

I hope this helps!  Let us know if you have other questions.
Megan B.
National Instruments

[versace]

Hi Megan,
thanks for the reply, but the Help file does not help too much in this case.
I am not a real expert in statistics, but to me it seems that is not possible to compute with labview a repeated mesures ANOVA, only a between subject anova is implemented. Trying several "solutions" I was able to obtain correct results only for between subject designs (I compared the results of labview to the output of a statistical package).
Moreover the only example I was able to find in the help file is about 2 d anova, but that example is confusing me too: in that example there is one observation per cell (one subject for each level of the two factors in the experimental design)...I don't understand how labview is able to extract an F value from that design. With only one subject per condition there is no variance with which an F value can be computed (empirically this is confirmed by the fact that using the numbers of the example in a statistical package, the result is that no F value can be computed).
I guess that the next help file can be beefed up a bit more....
Anyway, do you think is possible to know if it is possible to run a repeated measures anova with labview? If the answer is yes, is it possible to have an example?
Thanks, francesco

[Hapenciuc]

Hi all,

I have a problem with 2D ANOVA.

I have 45 measurments of something, for 5 parts, 3 repeats of measurements, by 3 dudes.

I keep having an error -20043 (and anther time -20045)  (this mean "The level of factors is outside the allowable range of some data" and "The total number of data points must be equal to the product of levels/each factor * observations/cell. ")

I am not to sure what values to put for "observation for cell" "B level" and "A levels"

the out put array with info is void!!!

Please help it's really important.

I use LV 8.2 and the Help files didn't helped me 😞

Thanks in anticipation.

[Brian_K.]

Hi Hapenciuc,

So I read through the help file as well.  They didn't help me either, but I don't know much about ANOVA much less 2D ANOVA.  Do you think this is an error in LabVIEW or just need to know how to implement it in LabVIEW.  Have you tried your data in Excel (pretty sure it has an ANOVA function), and gotten useful results?

Also I put an indicator on the the info output and got a 2D array.  Did you get something different?

[Hapenciuc]

Hi Brian,

In fact I have fix the ANOVA issue.

The problem at me was that the numbers for A and B inputs where from 1to n. And aparentrly ANOVA needs then from 0 to n-1.

You might laugh, as I did, but this was all.

Also for the A levels and B levels the value where m respectively n. In fact I compute those numbers by extractin the maximum value of the A and B array and adding 1 (the same it is done in the ANOVA example). For the "observation per cell" input the value is the total numer of recordings divided to n*m.

Total number of recordings should be T=n*m*p ; p numer of parts involved in measuring process.

About Exel, I have neved done programing in it and I belive I will not do anything like this soon :). But, a work mate tried to do the very same thing in Exel (after he see I did it, man I am proud 🙂 ) and he didn't found no AOVA function in Exel. He tried to compute the ANOVA stuff himself, but I didn't heard from him that he succeded.

Best Regards

[versace]

Hi,
I would compare the results from Labview with those outputted by a reputable stat pack. Especially if, as you wrote, you are dealing with a repeated measures design. The VI implemented in labview performs a between subjects ANOVA. The results, obviously, will differ. From what you write, your are dealing with a one way ANOVA. It will be easy to implement the corrections looking in a stat book how to calculate degrees of freedom and sum of squares for a repeated measures design.
Cheers, Francesco