I need your scientific inputs to help me design developmental stimulation study by simply matching intervention and control groups (under 5 children living in poverty) on age and sex, without randomization. I choose to design this way b/c of practical reasons. What do you think?
You have some options for an observational study that would be detailed in a good methods textbook. First of all, you will need the substantive knowledge of the treatment mechanism to ensure that you have the right variables. If your treatment is assigned based on some kind of test score, diagnostic, or means test (e.g. parent income), then you should try a Regression Discontinuity Design to estimate a local average treatment effect on the treated. If you have a control group that is assigned some other way, then you will need 1) to have all variables that might effect the selection process (selection on variables assumption) and 2) common support in the distributions of the treatment and control groups on these variables. See Morgan & Winship (2007) or Rosenbaum (2002) or Don Rubin's work.
I echo Richard's remarks. I am providing a link to a paper that I wrote about applying matching techniques (those that Richard noted) in a prospective manner. I think it is what you are looking to do.
I hope this helps
Ariel
Article Designing a Prospective Study When Randomization is Not Feasible
Wonderful resource again! Thanks Ariel!
I still need some more ideas from fellow experts!
- Badges
- Science topic
- Similar topics
- Psychology
- Developmental Psychology