In the model, the dependent variable is a dichotomous data (Yes or No), the influencing factors includes two types: individual Sociodemographic data and bilateral hand data.
However, problem comes:
The unit for Dependent variable, sociodemographic data is person.
Whereas the unit for hand data is hand (each individual has two hand).
So when the model "DV=Sociodemographic + Hands" is created, I have problems as follow:
1. Do I need to stratify the sample. Because sociodemographic data is always accompanied with two hand data. I can't put the two hand data at the same row/
2. If I need to stratify, how to do?
Thanks.
You can do this with a mixed-effects model. Each datapoint has a DV (yes/no), two fixed IVs (the sociodemographic value and Hand [right/left]) and a random variable (person). Then your model (in the syntax of the {lme4} package in R) would be
DV ~ Socio*Hand + (Socio*Hand|Person)
I don't know SPSS syntax. But conceptually it's the same thing: a generalized [binomial] linear mixed-effects model with your DV, your two fixed effects, and the random effect for person. The details of how to implement it vary from software to software (there are tutorials online for this) but this concept remains the same.
Person is another factor. Each person has sociodemographic data associated with her.
Presumably each person supplied two data points, one from each hand. A given datapoint will have a DV value, a sociodemographic value, a Hand value (left or right), and a Person value ("subject5" or whatever you coded the people as).
- Badges
- Science topic