Social Policy Analysis II
Problem:
The effectiveness of Housing Policy for homelessness in a city of Ontario.
This policy would have the city pay for cheap apartments for all homeless people with no-strings attached.
The supporters of the policy claim that the majority of the costs in implementing the policy would be offset by the reduced of resources that the city would need to spend on homeless shelters, jail time, and emergency room visits for the chronically homeless.
Can this policy reduce other city expenditures on the homeless?
Use Differene-in-Differences method to examine whether the Housing Policy reduced jail rates.
Data source
What type of data do you need?
- You may gather data on jail rates among 10 citites at two time points, say 2005 and 2015, half of which Housing Policy in between the two time points specified.
Methodological workflow
First, run a statistical test to assess whether cities that eventually implemented the Housing Policy tended to have higher rates of jail rates in 2005 than cities that had not implemented the Housing Policy.
Then, use a t.test to assess whether cities that implemented the Housing policy tended to have higher rates of jail in 2015 than cities that had not implemented the Housing Policy.
Check whether there was an increasing/decreasing trend of jail across cities (check common trends assumption). This may suggest that the Housing policy was actually reducing/or keep at the same the rates of jail. We can model this trend with a difference-in-differences policy.
For this, simply need to construct a linear model that predicts jail rates based on implementation and year, and with an “interaction” term between implementation and year, which estimates measures whether implementating the policy altered the relationship (or trend) between time and jail rates. Use the following code for this purpose:
glm(rate ~ Implemented*Year, data=Jail)
Chapter 3 Social Policy Analysis II
Problem:
The effectiveness of Housing Policy for homelessness in a city of Ontario.
This policy would have the city pay for cheap apartments for all homeless people with no-strings attached.
The supporters of the policy claim that the majority of the costs in implementing the policy would be offset by the reduced of resources that the city would need to spend on homeless shelters, jail time, and emergency room visits for the chronically homeless.
Can this policy reduce other city expenditures on the homeless?
Use Differene-in-Differences method to examine whether the Housing Policy reduced jail rates.
3.1 Data source
What type of data do you need?
3.2 Methodological workflow
First, run a statistical test to assess whether cities that eventually implemented the Housing Policy tended to have higher rates of jail rates in 2005 than cities that had not implemented the Housing Policy.
Then, use a t.test to assess whether cities that implemented the Housing policy tended to have higher rates of jail in 2015 than cities that had not implemented the Housing Policy.
Check whether there was an increasing/decreasing trend of jail across cities (check common trends assumption). This may suggest that the Housing policy was actually reducing/or keep at the same the rates of jail. We can model this trend with a difference-in-differences policy.
For this, simply need to construct a linear model that predicts jail rates based on implementation and year, and with an “interaction” term between implementation and year, which estimates measures whether implementating the policy altered the relationship (or trend) between time and jail rates. Use the following code for this purpose:
glm(rate ~ Implemented*Year, data=Jail)