A colleague from Bates College (with whom I happened to share an advisor in grad school) visited last week to give a seminar at Lafayette and we started talking about writing a paper together. Working off each of our comparative advantages, it’s going to be about domestic violence in India.
As a result, this morning I was thinking about how to code up domestic violence to put into regression analysis and how defining gender-based or domestic violence is part and parcel to the type of question you’re trying to answer.
For example, many surveys include violence by a partner, a husband, a boyfriend, a father, an in-law, and any number of other actors. My quick response to SD this morning was to divide the categories (not mutually exclusive, perhaps) like this.
1. By a romantic partner
2. By a husband (romantic partner with legal implications)
3. By a member of her husband’s family
4. By a member of her own family.
5. By anyone when it’s gender-motivated.
2 and 3 (and possibly 1 depending on societal structures) have implications for bargaining power-type questions and investments in children. 1, 4, and 5 have greater implications for society at large.
Thoughts?
Code ’em all up, I say.
ERIN K. FLETCHER 
Way out of my depth here, but would a consideration of elder abuse be appropriate in this context? While by definition it’s not solely gender-motivated, that may be a factor.
Absolutely, we’ll likely confine the sample to people with young children, but that’s another angle that I hadn’t considered. Low life I don’t know what elder abuse looks like in India, except the tradition of widows being forced to wear white and live away from society (less common now in cities, but still present in rural areas). I’d still call that gender-based, though. I’ll have to think on it.