Which is the best way to present replicate data in a table (where Result A and Result B are two separate data sets - imagine this is gene expression data, and the results are expression levels of Gene A and Gene B)?
This way:
| Result A | Result B |
|---|
| Sample 1 | 1 | 210 |
|---|
| Sample 1 | 2 | 211 |
|---|
| Sample 1 | 3 | 212 |
|---|
| Sample 2 | 21 | 3 |
|---|
| Sample 2 | 22 | 4 |
|---|
| Sample 2 | 23 | 5 |
|---|
| Sample 3 | 333 | 45 |
|---|
| Sample 3 | 344 | 46 |
|---|
| Sample 3 | 355 | 47 |
|---|
or this way:
| Result A | Result A | Result A | Result B | Result B | Result B |
|---|
| Sample 1 | 1 | 2 | 3 | 210 | 211 | 212 |
|---|
| Sample 2 | 21 | 22 | 23 | 3 | 4 | 5 |
|---|
| Sample 3 | 333 | 344 | 355 | 45 | 46 | 47 |
|---|
or this way, with a separate table for each set of results (I think this direction is what Google Docs would prefer, but it could easily go horizontally instead):
Result A
| Sample 1 | Sample 2 | Sample 3 |
|---|
| 1 | 21 | 333 |
| 2 | 22 | 334 |
| 3 | 23 | 335 |
Result B
| Sample 1 | Sample 2 | Sample 3 |
|---|
| 210 | 3 | 45 |
| 211 | 4 | 46 |
| 212 | 5 | 47 |
or maybe even multiple results, joined by a separator, within the same cell:
| Result A | Result B |
|---|
| Sample 1 | 1;2;3 | 210;211;212 |
|---|
| Sample 2 | 21;22;23 | 3;4;5 |
|---|
| Sample 3 | 333;344;355 | 45;46;47 |
|---|
Either separate tables or with a 'rowspan="3"' on each of the 3 sample sets in your first table, or 'colspan="3"' on the second. Plus, all your table headers should be marked up with TH elements, even when they are row and not column headers (as in table 1). Add a 'scope="row"' attribute and you'll have the right semantics too.
I'd ignored the HTML semantics because it was hard enough work cleaning up the HTML to this state from OpenOffice or Google Docs, but yes, you're right as far as presentation is concerned.
Are "merged cells" a good thing for actually working with the data though?
if the point of this is to compare A and B, then the first table wins by far, although I think it could be prettied up a bit.
Good point: I think you're more likely to want to be comparing between samples within the same results set, than the same sample between distinct sets of results.
Also, the second table should probably be rotated to be vertical for a better comparison.
Basically the question is whether you should repeat the 'sample' heading, or the 'result' heading, or neither.
Since you can rotate both table 1 and 2 90 degrees and get basically the same visual result, it shouldn't matter much which option you go with. As I'm an HTML geek, I can't really ignore the semantics, so I'd merge the repeated headings no matter which direction they apply. :-)
_sample_ ______result______
A B
1 2 211
2 22 4
3 334 46
If you want to get really fancy, put the std devs in. You don't need to show the complete dataset -- that's supplementary info, where the format is not so important because anyone looking at it is happy to root around some. If the point is to compare samples within a result set, I think separate tables would be best -- although even then [mean +/- std dev] gives you that information in a rather easier-to-use format.
Ah, tags stripped I take it?
*previews, like he should have in the first place*
Yup. Oh well, you get the idea.
I thought your title said "replicants" and it was a post about the new version of Bladerunner. Ah well.