how to get average of rows that have a certain relationship - mysql

Question

I have a bunch of data that is stored pertaining to county demographics in a database. I need to be able to access the average of data within in the state of a certain county.
For example, I need to be able to get the average of all counties who's state_id matches the state_id of the county with a county_id of 1. Essentially, if a county was in Virginia, I would need the average of all of the counties in Virginia. I'm having trouble setting up this query, and I was hoping that you guys could give me some help. Here's what I have written, but it only returns one row from the database because of it linking the county_id of the two tables together.
SELECT AVG(demographic_data.percent_white) as avg_percent_white
FROM demographic_data,counties, states
WHERE counties.county_id = demographic_data.county_id AND counties.state_id = states.state_id
Here's my basic database layout:
counties
------------------------
county_id | county_name
states
---------------------
state_id | state_name
demographic_data
-----------------------------------------
percent_white | percent_black | county_id

Answer 1

Your query is returning one row, because there's an aggregate and no GROUP BY. If you want an average of all counties within a state, we'd expect only one row.
To get a "statewide" average, of all counties within a state, here's one way to do it:
SELECT AVG(d.percent_white) AS avg_percent_white
FROM demographic_data d
JOIN counties a
ON a.county_id = d.county_id
JOIN counties o
ON o.state_id = a.state_id
WHERE o.county_id = 42
Note that there's no need to join to the state table. You just need all counties that have a matching state_id. The query above is using two references to the counties table. The reference aliased as "a" is for all the counties within a state, the reference aliased as "o" is to get the state_id for a particular county.
If you already had the state_id, you wouldn't need a second reference:
SELECT AVG(d.percent_white) AS avg_percent_white
FROM demographic_data d
JOIN counties a
ON a.county_id = d.county_id
WHERE a.state_id = 11
FOLLOWUP
Q What if I wanted to bring in another table.. Let's call it demographic_data_2 that was also linked via the county_id
A I made the assumption that the demographic_data table had one row per county_id. If the same holds true for the second table, then a simple JOIN operation.
JOIN demographic_data_2 c
ON c.county_id = d.county_id
With that table joined in, you could add an appropriate aggregate expression in the SELECT list (e.g. SUM, MIN, MAX, AVG).
The trouble spots are typically "missing" and "duplicate" data... when there isn't a row for every county_id in that second table, or there's more than one row for a particular county_id, that leads to rows not included in the aggregate, or getting double counted in the aggregate.
We note that the aggregate returned in the original query is an "average of averages". It's an average of the values for each county.
Consider:
bucket count_red count_blue count_total percent_red
------ --------- ---------- ----------- -----------
1 480 4 1000 48
2 60 1 200 30
Note that there's a difference between an "average of averages", and calculating an average using totals.
SELECT AVG(percent_red) AS avg_percent_red
, SUM(count_red)/SUM(count_total) AS tot_percent_red
avg_percent_red tot_percent_red
--------------- ---------------
39 45
Both values are valid, we just don't want to misinterpret or misrepresent either the value.