average of one variable, uniqueness of another variable

[Judith Vick]

I am pretty inexperienced with Stata -- hoping someone can help me here, as I can't figure out how to word what I am trying to do in order to google:

I have a list of patients for whom each has an estimated value provided by a physician.

In this Example 1 below, there are 5 patients, with an average estimate of 68.75 (one missing value) made by 3 unique physicians.

Example 1

	estimate	physician
patient #1	50	1
patient #2	50	2
patient #3	100	3
patient #4	missing	3
patient #5	75	3
	68.7

In Example 2 below, there are 9 patients, with an average estimate of 67.8 (two missing values) made by 5 unique physicians.

Example 2

	estimate	physician
patient #1	50	1
patient #2	50	2
patient #3	100	3
patient #4	missing	3
patient #5	75	3
patient #6	75	4
patient #7	missing	5
patient #8	50	6
patient #9	75	6
	67.8

How can I write code to determine the number of unique physicians that contributed to the average estimate value?

Thank you in advance!

Judith

[Rich Goldstein]

your example data is not nearly as helpful as it. could be - please read the FAQ (or, at least, the part dealing with -dataex-); I think that the following user-written command will help: -distinct-; use -search- to find and download

[Andrew Musau]

Here is one approach, guessing the structure of your data.

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(group physician observation)
1 1 1
1 1 .
1 2 2
1 3 3
1 4 .
1 5 .
2 1 1
2 1 2
2 2 3
2 2 .
end

frame put group physician if !missing(observation), into(count)
frame count{
    contract group physician
    collapse (count) physician, by(group)
}
frlink m:1 group, frame(count)
frget wanted= physician, from(count)
drop count
frame drop count

Res.:

. l, sepby(group)

     +--------------------------------------+
     | group   physic~n   observ~n   wanted |
     |--------------------------------------|
  1. |     1          1          1        3 |
  2. |     1          1          .        3 |
  3. |     1          2          2        3 |
  4. |     1          3          3        3 |
  5. |     1          4          .        3 |
  6. |     1          5          .        3 |
     |--------------------------------------|
  7. |     2          1          1        2 |
  8. |     2          1          2        2 |
  9. |     2          2          3        2 |
 10. |     2          2          .        2 |
     +--------------------------------------+

[Judith Vick]

I rewrote example 2 using dataex:

* Example generated by -dataex-. For more info, type help dataex
clear
input float(patient estimate physician)
1  50 1
2  50 2
3 100 3
4   . 3
5  75 3
6  75 4
7   . 5
8  50 6
9  75 6
end

I am trying to write code that determines the number of unique physicians that contributed non-missing values to "estimate." In this example, the answer is five unique physicians: physicians 1, 2, 3, 4, and 6. Physician 5 is not counted because there is a missing value for estimate for patient 7, the only patient for whom she was the physician.

[Judith Vick]

Thank you Andrew Musau for your quick reply even before I posted appropriately with dataex. I don't think the way you posted applies to my example because I don't have "groups" of patients in my data. Rich Goldstein thank you also for your quick response, I hope my question is now clearer with the example.

[Rich Goldstein]

if you have installed -distinct-, here is one way (but I'm still not sure I understand what you want or why):

. distinct physician if estimate<.

----------------------------------
           |     total   distinct
-----------+----------------------
 physician |         7          5
----------------------------------

[Judith Vick]

Thank you!! that worked.