0

In a two-way table of percentages that analyzes what will be in later models the key dependent and key independent variable, which variable is placed on the left to define rows and which variable is placed across the top to define columns?

user163367
  • 41
  • Why would it matter? Since you haven't stated how this table will be used, on the face of it it's purely a matter of personal choice how you lay out the values within it. – whuber Mar 02 '18 at 18:36

1 Answers1

1

What can bite first and most obviously is that many columns are usually much more awkward than many rows for almost any kind of display. That is true for books, journals, monitors, smart phones, and so forth. So you need to avoid the more awkward possibility, and this mundane fact may outweigh anything else.

That said, it is generally easier with tables to compare down columns than across rows -- especially if (as is good practice) numbers are aligned on the decimal point (or where it would be). So 

1
23
456
7890

is dopey and 

   1
  23
 456
7890

is recommended. So, what kind of comparisons are more interesting or useful? If reading the table both ways is equally interesting or useful, this won't settle the matter.

Perhaps too trivial to mention, yet what I do most often is try one or both possibilities and change my mind quickly if one looks horrible. That calls for changing details in a single command or function call in any software I want to take seriously. It's easier to do that than to consult some abstract principle from some expert somewhere on what you should do. You don't have to define horrible before you can recognise it.

Nick Cox
  • 56,404
  • 8
  • 127
  • 185