Below are two examples of a table, with different orientations of the trial header. The number of the trial represents the chronological order, but there is no relationship between all the Trial 1's, for example.

enter image description here

enter image description here

Which variation would be more preferable, and why?

If you have another, more preferable variation, please suggest it.

  • 2
    Not your question, but: You have far too many lines. In your first table you can probably throw away all lines except the one separating header and body. In the second table, you may additionally need the horizontal lines separating the different lengths. To enhance readability, add vertical spacing.
    – Wrzlprmft
    Commented Sep 20, 2017 at 11:11
  • @Wrzlprmft thank you for the suggestion, will do. Commented Sep 20, 2017 at 12:25
  • 1
    Another sidenote: Are you sure that your time measurements are accurate up to 0.01 s? The human reaction time is about 0.2 s and I don’t think you measure by other means in the presumed lab-course setting.
    – Wrzlprmft
    Commented Sep 20, 2017 at 12:35
  • @Wrzlprmft human error is considered in the variables, and is not considered here (I would, but teacher said stick to 0.01) Commented Sep 21, 2017 at 0:23

3 Answers 3


This is how I would do it (essentially your first example):


enter image description here


My rationale for this is as follows:

  • It allows for including the average over your trials – which you have to compute and probably show anyway, and here is the best place to do it, as it can be compared with the data from the individual trials.

  • Your reader usually wants to quickly assess the following:

    • Do the individual trials for a given length agree with each other or with the average?

    • Do the average values agree with some information outside this table (theoretical expectation, figure, official solution)?

    Both are not very comfortable in your second example: You either have to check groups of three that are not well separated visually or you have to go over the data in steps of three.

  • The structure of the table is more easy to grasp. Each direction is only used for one purpose – in your second example, going vertical could mean increasing the length but also going from one trial to the next.

  • The least import information you have is the trial number; in fact it shouldn’t matter at all. The suggested layout avoids it dominating the information and repeating it – if you show it at all.

  • Thank for for the suggestion. In regards to the average, my teacher asked us to explicitly show how we got the average by including sample calculations, so the average will be left to a table further down the path. Commented Sep 22, 2017 at 4:58

I would prefer the first table. It would be natural to add a column with average±stddev, which you would generally need to draw some conclusion from your experiment. In the second table, it would be more awkward to include this, due to having to span multiple rows.


Anything as complex as this I'd prefer to see in graphical form. Do you really need to show 4 significant digits? Can you create a barplot or something with these data and still get the idea across?

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .