# Thread: Comparison of datasets of differnt sizes - Can I use ANNOVA?

1. ## Comparison of datasets of differnt sizes - Can I use ANNOVA?

I have data from an audit looking at the calculation of heart function from images. 10 unique image sets have each been processed by 63 centres. While looking at the data I have realised that only 9 different software packages are used for the calculations across all the centres and so I would like to try and find out if there are significant differences between the results calculated using each software programme. For each study, each centre has returned 2 volumes (in ml) and the % difference between the two volumes; these are theoretically continuous variables but in practice the software used to calculate them rounds the answer to the nearest integer.

So my questions: Can I use an ANOVA test to compare the results from the different software packages even though I have different numbers of data points for each (6x10 for A, 1x10 for B, 5x10 for C, 5x10 for D, 3x10 for E, 2x10 for F, 4x10 for G, 28x10 for H and 9x10 for I)? Alternatively, is there a more powerful tool I could use since the 10 datasets used are the same for all systems and so the results for study 1 will all be related to each other etc.?

Finally, as well as the 10 unique datasets I set 2 to each centre which were presented as different images but were in fact duplicates of 2 of the original image sets. I was intending to exclude these 2 from this part of my analysis but is that correct?

Many thanks!
Sarah

2. ## Re: Comparison of datasets of differnt sizes - Can I use ANNOVA?

Virtually all data is rounded, that does not make it non-interval is that was what was meant.

The answer to the question depends. If the variance is the same in the subgroups (or close) then you can. If they are not then there can be signficant issues in calculations.

Since regression has very similar assumptions to ANOVA I immagine this might be a problem with that as well, although I have never seen this issue brought up.

You seem to have two different dependent variables, the volumes and differences between them? Are you running both as dependent variables?

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts