2 sql queries 2 different results using UNION - sql

Question

I have 2 queries that start with same tables but filter different columns. Both queries are unioned together so I can get a single count of people without duplication.
If I run the queries with the union commented out I have the same number of rows in each 1,953. When I run with the union I get 1,816 in one and 1,922 in the other.
My data is just an account # like 123456 in the first column and a 1/0 in the second column. Help me understand how this can happen if I am starting with the same number of rows.
Here is one of the queries
select distinct acct#,
case
when (lastFilledDate is not null and lastFilledDate<>'00/00/00') or
([Last Filled DC] is not null and [Last Filled DC]<>'00/00/00') or
(vivitrol is not null and vivitrol <>'00/00/00') or
(sublocade is not null and sublocade <>'00/00/00') or
(naltrexone is not null and naltrexone <>'00/00/00') then 1
else 0 end as result
from
(
select Acct#, DOB, [COE Contact Note], [COE-INTAKA Doc], [COE-MOM
Doc], lastFilledDate, [Last Filled DC],vivitrol,sublocade,naltrexone,
ROW_NUMBER() over (partition by Acct# order by [COE-INTAKA Doc] desc)
as apptRows
from tblAppBSCImportDashCOE2279 as main
where (([COE-MOM Doc]='Yes' and [COE Contact Note] is not null) or
[COE-MOM Doc]='No') and Appt is not null
) as sub
where apptRows=1
union
select distinct acctNo,
case
when
providerMAT='The Wright Center' and [COE-MOM Doc] is not null then
1
else 0
end as result
from
(
select acctNo, [COE-MOM Doc], MAT, providerMAT,
ROW_NUMBER() over (partition by acctNo order by COEBNMOM, [COE-MOM Doc]
desc) as apptRows
from tblAppBSCImportDashCOEHM2544 as main
where [COE-MOM Doc] is not null or COEBNMOM is not null
) as sub
where apptRows=1
results look like
acct# result
123456 1
234567 0

Answer 1

There is one possibility. The records you selected may result in duplicate records WITHIN each select statement. Let me try to illustrate with an example. (you can input the following query into your session to follow along)
IF OBJECT_ID('TEMPDB..#TEMP1') IS NOT NULL
DROP TABLE #TEMP1
IF OBJECT_ID('TEMPDB..#TEMP2') IS NOT NULL
DROP TABLE #TEMP2
CREATE TABLE #TEMP1(
id INT
,account INT
,amount INT
,yes_no INT
)
INSERT INTO #TEMP1 (id,account,amount,yes_no)
VALUES(1,123456,5,0)
,(2,123456,10,0)
,(3,123456,20,0)
CREATE TABLE #TEMP2(
id INT
,account INT
,amount INT
,yes_no INT
)
INSERT INTO #TEMP2 (id,account,amount,yes_no)
VALUES(4,123456,5,0)
,(5,123456,10,0)
,(6,123456,20,0)
SELECT *
FROM #TEMP1
SELECT *
FROM #TEMP2
Output of this is 2 tables with distinct records:
Now suppose I write queries that select account and the 'yes_no' column:
SELECT account,yes_no
FROM #TEMP1
SELECT account,yes_no
FROM #TEMP2
You can see that now all of the records are the same values within each select statement. So what do you think happens when I union these queries together?
SELECT account,yes_no
FROM #TEMP1
UNION
SELECT account,yes_no
FROM #TEMP2
UNION will output the distinct values of the ENTIRE OUTPUT, which also applies within each query. This is an extreme example of what I think you are experiencing. You need to include some sort of ID for each query such that it can be distinguished from other records within the query, like;
SELECT id,account,yes_no
FROM #TEMP1
UNION
SELECT id,account,yes_no
FROM #TEMP2