count uniques anomaly
 

Here's what I do on Friday nights!!!

Can anyone explain why the result of this formula is 5:

=SUMPRODUCT((C1:C5<"")/COUNTIF(C1:C5,C1:C5&""))

screencap:

http://img382.imageshack.us/img382/107/uniques1kt6.jpg

I can understand why the result of this one is #DIV/0! (same formula,
different data):

screencap:

http://img376.imageshack.us/img376/558/uniques2nl1.jpg

It seems that COUNTIF is getting "confused" in the first example!

I know that COUNTIF has trouble with *text* true/false, but I can't figure
out what's going on in the first example. If you remove the logical FALSE
then the result is #DIV/0! which I would expect.

To count text true/false:

=COUNTIF(rng,"true*")
=COUNTIF(rng,"false*")

To count logical TRUE/FALSE:

=COUNTIF(rng,true)
=COUNTIF(rng,"true")
=COUNTIF(rng,false)
=COUNTIF(rng,"false")

--
Biff
Microsoft Excel MVP

count uniques anomaly
 

hi, Sir ! (just wild ideas)

- once you know this...
I know that COUNTIF has trouble with *text* true/false ...

- give to countif a little help and change your formula...
from: =SUMPRODUCT((C1:C5<"")/COUNTIF(C1:C5,C1:C5&""))
to: =SUMPRODUCT((C1:C5<"false")/COUNTIF(C1:C5,C1:C5&""))

(as I said... *just wild ideas*)
hth,
hector.

__ OP __
Here's what I do on Friday nights!!!
Can anyone explain why the result of this formula is 5:
=SUMPRODUCT((C1:C5<"")/COUNTIF(C1:C5,C1:C5&""))
screencap:
http://img382.imageshack.us/img382/107/uniques1kt6.jpg
I can understand why the result of this one is #DIV/0! (same formula, different data):
screencap:
http://img376.imageshack.us/img376/558/uniques2nl1.jpg
It seems that COUNTIF is getting "confused" in the first example!
I know that COUNTIF has trouble with *text* true/false, but I can't figure out what's going on in the first example.
If you remove the logical FALSE then the result is #DIV/0! which I would expect.
To count text true/false:
=COUNTIF(rng,"true*")
=COUNTIF(rng,"false*")
To count logical TRUE/FALSE:
=COUNTIF(rng,true)
=COUNTIF(rng,"true")
=COUNTIF(rng,false)
=COUNTIF(rng,"false")
--
Biff
Microsoft Excel MVP

count uniques anomaly
 

The test C1:C5<"" counts the FALSE as a value, whereas the COUNTIF ignores
it.

This works for the first dataset

=SUMPRODUCT((NOT(ISNUMBER(MATCH(C1:C5,{"FALSE","TR UE"},0))))*(C1:C5<"")/COUNTIF(C1:C5,C1:C5&""))

--
HTH

Bob

(there's no email, no snail mail, but somewhere should be gmail in my addy)

"T. Valko" wrote in message
...
Here's what I do on Friday nights!!!

Can anyone explain why the result of this formula is 5:

=SUMPRODUCT((C1:C5<"")/COUNTIF(C1:C5,C1:C5&""))

screencap:

http://img382.imageshack.us/img382/107/uniques1kt6.jpg

I can understand why the result of this one is #DIV/0! (same formula,
different data):

screencap:

http://img376.imageshack.us/img376/558/uniques2nl1.jpg

It seems that COUNTIF is getting "confused" in the first example!

I know that COUNTIF has trouble with *text* true/false, but I can't figure
out what's going on in the first example. If you remove the logical FALSE
then the result is #DIV/0! which I would expect.

To count text true/false:

=COUNTIF(rng,"true*")
=COUNTIF(rng,"false*")

To count logical TRUE/FALSE:

=COUNTIF(rng,true)
=COUNTIF(rng,"true")
=COUNTIF(rng,false)
=COUNTIF(rng,"false")

--
Biff
Microsoft Excel MVP

count uniques anomaly
 

The test C1:C5<"" counts the FALSE as a value, whereas the COUNTIF ignores
it.

No, COUNTIF is counting it. That's the problem.

Within COUNTIF:

false = FALSE and FALSE&"" = FALSE

But

false < false and FALSE&"" < false

So each element of the criteria array is being matched to the logical FALSE
in the range array and being counted.

My actual data didn't contain any logicals but it did contain text
true/false. I used this:

=SUMPRODUCT((A2:A3992<"")/COUNTIF(A2:A3992,A2:A3992&"*"))

The samples I posted were the results of me tinkering.

Another thing to consider is if the data did contain either true/TRUE and/or
false/FALSE should they be considered equal or not?

Also note that the above formula will not work on numbers


--
Biff
Microsoft Excel MVP


"Bob Phillips" wrote in message
...
The test C1:C5<"" counts the FALSE as a value, whereas the COUNTIF
ignores it.

This works for the first dataset

=SUMPRODUCT((NOT(ISNUMBER(MATCH(C1:C5,{"FALSE","TR UE"},0))))*(C1:C5<"")/COUNTIF(C1:C5,C1:C5&""))

--
HTH

Bob

(there's no email, no snail mail, but somewhere should be gmail in my
addy)

"T. Valko" wrote in message
...
Here's what I do on Friday nights!!!

Can anyone explain why the result of this formula is 5:

=SUMPRODUCT((C1:C5<"")/COUNTIF(C1:C5,C1:C5&""))

screencap:

http://img382.imageshack.us/img382/107/uniques1kt6.jpg

I can understand why the result of this one is #DIV/0! (same formula,
different data):

screencap:

http://img376.imageshack.us/img376/558/uniques2nl1.jpg

It seems that COUNTIF is getting "confused" in the first example!

I know that COUNTIF has trouble with *text* true/false, but I can't
figure out what's going on in the first example. If you remove the
logical FALSE then the result is #DIV/0! which I would expect.

To count text true/false:

=COUNTIF(rng,"true*")
=COUNTIF(rng,"false*")

To count logical TRUE/FALSE:

=COUNTIF(rng,true)
=COUNTIF(rng,"true")
=COUNTIF(rng,false)
=COUNTIF(rng,"false")

--
Biff
Microsoft Excel MVP

count uniques anomaly
 

hi, guys !

as you stated in yur first post...
... COUNTIF has trouble with *text* true/false ...

and, yes...
... COUNTIF is getting "confused" in the first example! ...

IF you change your (text) false into falso (spanish or whatever other language)
you will get the "correct" count of "non-uniques" (within countif function)

hth,
hector.

__ OP __
The test C1:C5<"" counts the FALSE as a value, whereas the COUNTIF ignores it.

No, COUNTIF is counting it. That's the problem.
Within COUNTIF:
false = FALSE and FALSE&"" = FALSE

But
false < false and FALSE&"" < false

So each element of the criteria array is being matched to the logical FALSE in the range array and being counted.

My actual data didn't contain any logicals but it did contain text true/false. I used this:

=SUMPRODUCT((A2:A3992<"")/COUNTIF(A2:A3992,A2:A3992&"*"))

The samples I posted were the results of me tinkering.

Another thing to consider is if the data did contain either true/TRUE and/or false/FALSE should they be considered equal or not?

Also note that the above formula will not work on numbers
--
Biff
Microsoft Excel MVP