Field containing a list of sizes or a table ?

Question

Iain was kind enough to advise on using a table instead of a field containing a list of 6 comma separated items.
I did this, but can I throw a new slant on why I'd prefer to use a field ?
JEWELLERY WEB PAGE
select * from merchandise where merchandise.category = "Jewellery"
PRODUCT DETAIL PAGE
When a user clicks on any items from above page, the hyperlink sends the unique merchandise_ID in the link, to a product detail page using a new query:
select * from merchandise where merchandise.merchandise_ID = URL.merch_ID'
This jewellery comes in various sizes. These sizes were in a comma separated list in a field called 'sizes' in the merchandise table. I'd been advised to move them to a table instead. If I moved the sizes to a separate table I'd have to add a LEFT JOIN to include the sizes table or run a separate query on the sizes table (where sizes.merchandise_ID = URL.merch_ID) - which is a performance hit.
All I need to do with the sizes is produce a drop down list the user can choose from - so isn't it better I just put the sizes in a single field, separated by commas ? Then use Coldfusions inbuilt functions to parse the list into a drop-down box ?

Answer 1

A thought experiment that might convince you.
Lets compare these two data schemas.
CSV
Item
Size
Shirt
Small,Medium,Large
Coat
Medium,Small,Large
Pants
Small,Large,Medium
Shoes
Large,Small,Medium
NORMALIZED
Size Table
Key
Size
1
Small
2
Medium
3
Large
Item - Size Join Table
ItemKey
SizeKey
1
1
1
2
1
3
2
1
2
2
2
3
3
1
3
2
3
3
4
1
4
2
4
3
Item Table
Key
Item
1
Shirt
2
Coat
3
Pants
4
Shoes
Now imagine these two systems extended to multiple thousands of items.
Then imaging that some marketing type decided that "Large" was an ugly description for that size so they want to change it to "Husky"  and|or they want to change the "Small" size to "Petite".
In the Normalized version you make one or two updated to the size table.  I don't even want to try and describe the ugly update you have to try and create that involves looping and string manipulations to try and update all the records in the CSV version.