How to compare 2 samples based on multiple variables

Polka

New Member
#1
Hi all,

Basic stuff here but I can't wrap my head around this one. I have a table (see below) that contains different types of kin at different distances per EGO. As you can see, the table is currenntly designed as a cross-table for each ID in my data set. I want to check for the existence (or not) of a significant difference between distributions of men and women. Can't even figure what test I should use... Any comment or suggestion is most welcome. Many thanks!
ego ego_sex kin coresid d3 d10 d25 same else
16304 1 O 0 0 0 0 0 2
16304 1 C 0 0 1 0 2 19
16352 2 F 0 0 0 0 0 1
16352 2 C 0 2 0 1 4 1
16352 2 O 0 2 0 0 1 0
16352 2 P 0 2 0 0 0 0
16353 2 P 0 0 0 0 0 2
16353 2 F 0 0 0 0 0 1
16353 2 O 0 0 0 0 0 3
16353 2 C 0 0 0 1 0 7
16652 2 F 0 6 0 0 0 0
16652 2 P 0 2 0 0 0 0
31405 1 O 0 0 0 0 2 0
31405 1 P 0 0 0 0 2 0
31405 1 F 0 0 0 0 2 0
31405 1 C 0 0 0 0 5 0
 

Karabiner

TS Contributor
#2
Could you explain what this is all about? What is ego, what is kin F P O C, what kind of data did you collect, pecisely? How are your data structured in that table?

With kind regards

Karabiner
 

Polka

New Member
#3
Hi Karabiner,

Here's more information. What this is all about => I use register data linked to census data in order to measure distances between kin for different census years. Once I know the relationship (kin) I can locate them and see if they are close or not. I use different distances measures (coresid, d3,... same (as in same sub-district) and else) and count the number of kin at these distances.

Kin => I added definitions in the new table below. I restrained my analysis to rather close family: up to gr. parents and gr. children vertically, cousins, nephews/nieces and uncles/aunts horizontally.

Ego_context => the kind of environment EGO is in: rural, urban, etc.

Data structure => it is structured by EGO (ID). But since each EGO may have different kin ties in the censuses, the resulting table contains more than one line per EGO (one for each kin type). This is where I wasn't sure for the purpose of a statistical comparison.

Aim => Is there a difference between men and women regarding the spatial distributions of their kin around? Can we spot differences when origin (ego_context) is also considered? I also have the plan to do a network multilevel analysis on these data unsing a longitudinal approach. This is why that, at this stage, I was only looking to highlight with a simple test what is difficult to clearly observe with my descriptive data.

Hope this makes sense! Many thanks

Here's a clearer (I think) table:
1581081946649.png
 
Last edited: