As per your question ,
you have only one table in which locations by age group, ethnicity, sex dimensions are present.
So this should not created any problem on selection.
Does anybody know how I should load the population data so that the joins are appropriately made?
For this statement I am thinking you have more than one table.
Please clear your question and provide some details of your tables (Columns ) or sample data.
Population by region, age group, sex and ethnicity is in a single table. It's drawn from a pivot table, so I can re-shape it in any way I like. The table, in effect, shows the number of, say, European males aged between 40 and 49 who live in a particular suburb. The number is, of course, a single cell.
I want it to join to an "event" table. This table, of accidents, has a single row for each event; the accident event has columns for the age group, ethnic group, sex, and residential location of the subject. (Each event has a key field that successfully joins to several other tables.)
Does that help? I'm away from work for several days so I can't immediately provide a table. I hope this description does the trick!