Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

Okay, so let's say I have a chart with N columns of data. What I want to do is compare every column to every other column, and get a count of how many values are equal in both columns. I know I can easily compare TWO columns simply by adding a third column and setting a value to 1 if the contents of both columns match, then counting the 1s in that third column. (there's many posts of people asking that question, but this is more complex than that.) I need to compare a bunch of columns, and I don't want to create a new comparison column for every possible pair of existing data columns. (N can get pretty large, and this method would basically require about N2/2* additional columns - not acceptable.) It seems like it should be possible to in a single formula count all matches between column A and column B, and then just copy that to compare A:C, A:D, B:C, B:D, C:D, etc., for however many columns there are. This would then only require N2/2 cells, rather than that many whole columns, which is way more efficient. Any help?

*(Technically N*(N-1)/2, but close enough.)

share|improve this question
Open to VBA? Do you know any VBA> –  AthomSfere Mar 22 '13 at 17:45
I do not, but how hard can it be? I have a good deal of experience with C/C++/C# and Java, and some in numerous other languages. Just haven't done any VBA. (I know exactly how I'd go about doing this in any C-type language, but not sure where to start with VBA.) It feels like this should be possible without it, though? –  Darrel Hoffman Mar 22 '13 at 17:50
Maybe with a very nasty combination of sorts, if, vlookups... I generally do not like VBA, but this seems to be decent fit for it to me. –  AthomSfere Mar 22 '13 at 18:03
@DarrelHoffman What do you expect from the comparison? This is a test for equality? Or some other condition? Why not arrange a graphical representation of the lines? Possibly suitable criterial or characteristic comparison, then the number of additional columns will be N * K where K - the criteria. –  STTR Mar 22 '13 at 20:46
Can you give an example of what data you have (say, 4 or 5 columns’ worth) and what you want as a result? –  Scott Mar 22 '13 at 21:05

1 Answer 1

up vote 1 down vote accepted

Here is a 3-column example you can scale up.

First you need to name all the source columns. Put column headers on the columns, to be used as range names. If (for example) you have hundreds but less than 1000, put labels like Col001, Col002, Col003, etc. Once you type the labels, Ctrl+A that block, then Alt+I (insert), N (name), C (create); check Top Row on and click OK.

Now you need a matrix area to make the calculations. For n columns, you will need a block nxn, plus one row and one column for labels:

Calculation grid

In the first cell, type the following formula and then Shift+Ctrl+Enter to enter it as an array formula:


(This means, sum 1 for every cell in range "Col01" that equals the corresponding cell in range "Col01"; if not equal, sum 0.)

Now just copy that formula down through the rest of the matrix column (don't include the cell you copied from to the paste selection, or you will get "You cannot change part of an array"). Once you have a whole column filled up, copy that column (calculation cells only) across to the other columns to fill up the matrix.

The cells along the diagonal will simply have the total number of rows in the source columns (because, for example, "Col01" always equals "Col01" perfectly. The cells mirrored accross the diagonal will have identical values, because (for example) "Col02" vs "Col01" has the same number of identical values as "Col01" vs "Col02". They are redundant, and the diagonal is not particularly useful, so you can clear them out to make it more readable.

Added more detail, in response to the comment...

In the picture below, A7:C16 (blue cells) contain the source data. The labels in row 6 are applied as range names, by selecting A6:C16 and choosing Alt+I (insert), N (name), C (create), then checking Top ON in the "Create Names From Selection" dialog and clicking OK. (Now, for example, =SUM(Col01) is the same as =SUM(A7:A16)).

Range B2:D4 is the counting matrix. Select B2, type or paste the formula and use Ctrl+Shift+Enter to enter it as an array. Copy B2 to B3:B4. Then copy B2:B4 to C2:D4 (it is a bit fussy that way, because it is an array formula). The green cells represent the counts you want to achieve. The diagonal is always maximum, because (for example) Col01 always equals Col01, cell for cell. The other white cells on the opposite side of the diagonal are redundant, mirror image of the green cells. Now you can scale it up to your requirements.

The INDIRECT function means, use the text in the referenced cell as a range name. So (for example) =SUM(INDIRECT(B$1)) means the same as =SUM(Col01). The $ signs are absolute references, so you can copy and paste the formula throughout the whole matrix without having to edit every single one. B$1 means, always use row 1 in the formula, even when you copy it down. $A2 means, always use column A, even when you copy it across.

enter image description here

Yet more detail... :^)

Please make sure:

  • The range names are applied to the source data in the flat table (that is, the original table of columns you want to compare to each other)
  • The array formula is applied to each cell in the matrix where the match counts will be calculated

My guess: the range names didn't apply properly to the source data in the flat table. In the example, "Col01" should "Refer To" =Sheet1!$A7$16, and the value(s) should be something like {"5";"7";"5";"9";...

enter image description here

Once the range names are applied properly, the array formula in the matrix cells should be applied as follows:

enter image description here

Now... since the permutations multiply fast (3 columns --> 3 comparisons, 4 --> 6, 5 --> 10, 6 --> 15, etc. etc.), the INDIRECT really comes in handy - you can type the formula once in B2, then paste it to all the other cells. (If you get the "Can't change part of array", look back at previous answer content, you'll figure it out.)

Without the INDIRECT, B2 could be:


but that means, for each of the other cells in the matrix, you would have to manually change "Col01" to "Col02" etc. etc., very tedious...

Without the range names, B2 could be:


but that editing would be even more and more and more tedious...

share|improve this answer
I think I'm missing something here. I've tried this a bunch of different ways, and I either end up with just a matrix filled with all 1's, or #REF!'s. How are you arranging this? Is the matrix on a separate sheet from the data? Do B$1 and $A2 refer to the first cell in the matrix or the first cell in the data? I've never used the INDIRECT() function before, I'm not quite sure what it does... –  Darrel Hoffman Mar 26 '13 at 12:32
See edits above, with better picture –  KnomDeGuerre Mar 27 '13 at 14:03
Okay, so I have a similar setup, though my matrix is beside my data columns instead of above them - mainly so I can freeze-panes and get the headers to always be on top. But what I'm getting is the matrix cells are only comparing the data in the same row, rather than the whole column. Thus I only get 1 and 0 for the counts - 1 if the target column equals the source column at that row and 0 if it doesn't. Can't figure out why this doesn't work, as I copied your formula verbatim (changing B$1 and $A2 to in this case T$1 and $S2 for the first cell and nothing else). –  Darrel Hoffman Mar 27 '13 at 17:28
Okay, so I got rid of the freeze-panes and moved the matrix above the data like yours. What I get now is every cell in the matrix displays "#VALUE!". I stepped through the formula, and I found that it's the INDIRECT call that always returns #VALUE!. When I look at the ranges in the Name Manager, it shows each cell containing "0" or "1" instead of just 0 or 1. Not sure how this happened, nor how to fix this. (A global replace of all "0" with 0 found no matches.) The cells are all set to be "Number" format, so I'm not sure why they seem to be showing up as strings... –  Darrel Hoffman Mar 27 '13 at 18:02
More edits above, with pictures to support explanations... –  KnomDeGuerre Mar 28 '13 at 12:40

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.