Inserting CTE results into temp table generates duplicate rows - sql

Question

I have a long series of CTEs, and I want to insert the results into a temporary table at the end. However, I am ending up with 16 rows of duplicate data instead of just the one. Here is my code:
INSERT INTO #lvl_1_results
(
[Level_1]
, [Level_0]
, [P]
, [Sim (%)]
, [V]
, [StAV (%)]
, [CV]
, [CV (%)]
, [Sim (Mean)]
, [SD]
, [IAV]
, [MAV]
, [BMP]
, [BMP (%)]
, [BMV]
, [BMV (%)]
, [BMCV]
, [BMCV (%)]
, [Sim BMP (Mean)]
, [BM SD]
, [SAP]
, [SAP (%)]
, [ActV]
, [ActV (%)]
, [ACV]
, [ACV (%)]
, [Act (Mean)]
, [Act SD]
, [MV]
, [MV (%)]
)
SELECT sc.[Level_1]
, sc.[Level_0]
, [P]
, [P] / NULLIF([MV], 0)
, [V]
, [V] / NULLIF([MV], 0)
, [CV]
, [CV] / NULLIF([upper VaR], 0)
, [Sim (Mean)]
, [SD]
, [IAV]
, [MAV]
, [BMP]
, [BMP] / NULLIF([MV], 0)
, [BMV]
, [BMV] / NULLIF([MV], 0)
, [BMCV]
, [BMCV] / NULLIF([upper bm VaR], 0)
, [Sim BMP (Mean)]
, [BM SD]
, [SAP]
, [SAP] / NULLIF([MV], 0)
, [ActV]
, [ActV] / NULLIF([MV], 0)
, [ACV]
, [ACV] / NULLIF([upper active VaR], 0)
, [Act (Mean)]
, [Act SD]
, [MV]
, [MV] / NULLIF([parent_MV], 0)
FROM sc
, s_p
, s_v
, bm_c
, bm_v
, bm_p
, a_c
, a_v
, a_p
, MV
, upper_mv
, c_l_v
, bm_c_l_v
, a_c_l_v
, s_m
, bm_m
, a_m
, sim_sd
, bm_sd
, a_sd
, s_i
, s_mar
How can I get this query to produce just one row? For performance reasons, I think I should opt for a solution that doesn't just select the distinct rows.

Answer 1

As others have indicated in their comments, this query doesn't use the ansi JOIN syntax:
TableA
[INNER/LEFT] JOIN
TableB
ON
TableA.id=tableb.id
The advantage of using this style is that you Must supply an ON clause which forces you to think about how the data in your tables is actually related
You havent told the database how this data is related at all, so all it has done is join every row it found, to every other row, giving every possible combination of rows
Because you're getting 16 rows, it implies that some of your tables have 2 or more rows. If 2 tables have 2 rows, then the result is 2x2x2x2 rows. (Or maybe two tables have 2 rows and one has 4, or maybe two tables have 4 rows.. or a 2 and an 8, either way, when multiplied so every row is joined to every other row, you're getting 16 rows)
If you have 2 tables that have values a,b and 1,2 and they're joined like you have done here the database will give you
a,1
a,2
b,1
b,2
These rows are all unique in themselves, but if you just selected the letter you'd see a a b b, and then complain there were duplicate letters. If you just look at the numbers the complaint would be the same
Add more tables in without any relationship between them, and it will get worse. Hopefully easy to see that adding another 2 row table with x and y would increase your results to 8 rows..
You have to tell your database which columns in which tables link the tables together. If you want to see which tables cause your duplicates, do a select *