I found your paper was a good illustration of the application of quasi-experimental approach in empirical study. Since I know little about this approach, my comment on your work will be more like an inquiry from a student of the quasi-experiment approach. For propensity score matching, you mentioned that data for matching was drawn from other survey: NRVA 2005. I am wondering whether your sample is the same as the one in NRVA 2005. If not, whether the different samples used in NRVA and yours will cause problems here? I actually can not figurer out how the sample in your future surveys can be defined and matched into the control and treatment groups by using data from other surveys. Suppose prior to your project, you have already known which population should be sampled as the control group and which as the treatment group by NRVA 2005, then you collect data for these two groups. How can the potential pre-test differences between your control and treatment groups be ruled out by matching them according to the propensity score calculated with NRVA 2005 data? Another minor problem is that some books you cited in the footnote such as “Shagdish, Cook and Campbell 2002” haven’t been included in the biography.
Very interesting topic. I had a question relating to the ownership of the land. You write that in selecting the eligible neighborhoods being on state-owned land was a requirement. Later, in your causal model, you cite as an intervening variable whether the house is owned or rented. Does this mean that the state owns the land but the house itself is owned by an individual/family? Perhaps clarification on this point would be helpful considering the importance of property rights/lack thereof in shaping the behaviors and results you are trying to evaluate.
Hi Andrea here are my comments - I very much enjoyed reading the paper and learned alot.
I think generally it would be good to include some case studies in there and what other work has been done in other countries
I don't much about the quantitative analysis you want but I think that the paper seems like it jumps into the quan analysis without explaining what the terms in the quant analysis mean. What is propensity score matching? etc. Defining these terms would be good for someone who may not have a background in the quan (this depends on your audience; if you are appealing to a larger audience you might want to define these terms more)
it would be good to have a transition between the quan section and the kabul section
what are your hypothesis? sometimes in papers i find it useful when the author says hypothesis 1: this hypothesis 2: that. By defining what your objective is it may make it easier for the reader to follow. Also maybe mention hypothesis in beginning of paper in thesis paragraph
why are you choosing reliance on housing? in the presentation you gave a sound reason but in the paper you may indicate why you choose that and why it is better than other methods of analysis
this may not work but consider whether you might reorganize things where you give the background on the slums, then talk about kabul then integrate your technical discussion with your analysis of kabul. For instance, talk about slums, then give background on kabul, then present methods, then present why choose methods you do and then set forth research design with your hypothesis.
other intervening variables: possibly ratio of workers to nonworkers (i.e. working adults to children and elderly) - I mention this because a family with 2 adults is different from a family with 2 adults and 10 children or 1 working adult 1 grandmom and 10 children; job type; location of housing??, this may be impossible to get but length of renting/ownership - this seems to be an inverse curve - people will put investment in when they first get it and when it gets old but not as much in middle
I found your paper was a good illustration of the application of quasi-experimental approach in empirical study. Since I know little about this approach, my comment on your work will be more like an inquiry from a student of the quasi-experiment approach. For propensity score matching, you mentioned that data for matching was drawn from other survey: NRVA 2005. I am wondering whether your sample is the same as the one in NRVA 2005. If not, whether the different samples used in NRVA and yours will cause problems here? I actually can not figurer out how the sample in your future surveys can be defined and matched into the control and treatment groups by using data from other surveys. Suppose prior to your project, you have already known which population should be sampled as the control group and which as the treatment group by NRVA 2005, then you collect data for these two groups. How can the potential pre-test differences between your control and treatment groups be ruled out by matching them according to the propensity score calculated with NRVA 2005 data? Another minor problem is that some books you cited in the footnote such as “Shagdish, Cook and Campbell 2002” haven’t been included in the biography.
ReplyDeleteVery interesting topic. I had a question relating to the ownership of the land. You write that in selecting the eligible neighborhoods being on state-owned land was a requirement. Later, in your causal model, you cite as an intervening variable whether the house is owned or rented. Does this mean that the state owns the land but the house itself is owned by an individual/family? Perhaps clarification on this point would be helpful considering the importance of property rights/lack thereof in shaping the behaviors and results you are trying to evaluate.
ReplyDeleteHi Andrea here are my comments - I very much enjoyed reading the paper and learned alot.
ReplyDeleteI think generally it would be good to include some case studies in there and what other work has been done in other countries
I don't much about the quantitative analysis you want but I think that the paper seems like it jumps into the quan analysis without explaining what the terms in the quant analysis mean. What is propensity score matching? etc. Defining these terms would be good for someone who may not have a background in the quan (this depends on your audience; if you are appealing to a larger audience you might want to define these terms more)
it would be good to have a transition between the quan section and the kabul section
what are your hypothesis? sometimes in papers i find it useful when the author says hypothesis 1: this hypothesis 2: that. By defining what your objective is it may make it easier for the reader to follow. Also maybe mention hypothesis in beginning of paper in thesis paragraph
why are you choosing reliance on housing? in the presentation you gave a sound reason but in the paper you may indicate why you choose that and why it is better than other methods of analysis
this may not work but consider whether you might reorganize things where you give the background on the slums, then talk about kabul then integrate your technical discussion with your analysis of kabul. For instance, talk about slums, then give background on kabul, then present methods, then present why choose methods you do and then set forth research design with your hypothesis.
other intervening variables: possibly ratio of workers to nonworkers (i.e. working adults to children and elderly) - I mention this because a family with 2 adults is different from a family with 2 adults and 10 children or 1 working adult 1 grandmom and 10 children; job type; location of housing??, this may be impossible to get but length of renting/ownership - this seems to be an inverse curve - people will put investment in when they first get it and when it gets old but not as much in middle