Counting Defiers: Examples from Health Care Amanda

Published  . 0 views
↓ Download
Counting Defiers: Examples from Health Care Amanda
1 / 1
Counting Defiers: Examples from Health Care Amanda - slide 1 of 54 Counting Defiers: Examples from Health Care Amanda - slide 2 of 54 Counting Defiers: Examples from Health Care Amanda - slide 3 of 54 Counting Defiers: Examples from Health Care Amanda - slide 4 of 54 Counting Defiers: Examples from Health Care Amanda - slide 5 of 54 Counting Defiers: Examples from Health Care Amanda - slide 6 of 54 Counting Defiers: Examples from Health Care Amanda - slide 7 of 54 Counting Defiers: Examples from Health Care Amanda - slide 8 of 54 Counting Defiers: Examples from Health Care Amanda - slide 9 of 54 Counting Defiers: Examples from Health Care Amanda - slide 10 of 54 Counting Defiers: Examples from Health Care Amanda - slide 11 of 54 Counting Defiers: Examples from Health Care Amanda - slide 12 of 54 Counting Defiers: Examples from Health Care Amanda - slide 13 of 54 Counting Defiers: Examples from Health Care Amanda - slide 14 of 54 Counting Defiers: Examples from Health Care Amanda - slide 15 of 54 Counting Defiers: Examples from Health Care Amanda - slide 16 of 54 Counting Defiers: Examples from Health Care Amanda - slide 17 of 54 Counting Defiers: Examples from Health Care Amanda - slide 18 of 54 Counting Defiers: Examples from Health Care Amanda - slide 19 of 54 Counting Defiers: Examples from Health Care Amanda - slide 20 of 54 Counting Defiers: Examples from Health Care Amanda - slide 21 of 54 Counting Defiers: Examples from Health Care Amanda - slide 22 of 54 Counting Defiers: Examples from Health Care Amanda - slide 23 of 54 Counting Defiers: Examples from Health Care Amanda - slide 24 of 54 Counting Defiers: Examples from Health Care Amanda - slide 25 of 54 Counting Defiers: Examples from Health Care Amanda - slide 26 of 54 Counting Defiers: Examples from Health Care Amanda - slide 27 of 54 Counting Defiers: Examples from Health Care Amanda - slide 28 of 54 Counting Defiers: Examples from Health Care Amanda - slide 29 of 54 Counting Defiers: Examples from Health Care Amanda - slide 30 of 54 Counting Defiers: Examples from Health Care Amanda - slide 31 of 54 Counting Defiers: Examples from Health Care Amanda - slide 32 of 54 Counting Defiers: Examples from Health Care Amanda - slide 33 of 54 Counting Defiers: Examples from Health Care Amanda - slide 34 of 54 Counting Defiers: Examples from Health Care Amanda - slide 35 of 54 Counting Defiers: Examples from Health Care Amanda - slide 36 of 54 Counting Defiers: Examples from Health Care Amanda - slide 37 of 54 Counting Defiers: Examples from Health Care Amanda - slide 38 of 54 Counting Defiers: Examples from Health Care Amanda - slide 39 of 54 Counting Defiers: Examples from Health Care Amanda - slide 40 of 54 Counting Defiers: Examples from Health Care Amanda - slide 41 of 54 Counting Defiers: Examples from Health Care Amanda - slide 42 of 54 Counting Defiers: Examples from Health Care Amanda - slide 43 of 54 Counting Defiers: Examples from Health Care Amanda - slide 44 of 54 Counting Defiers: Examples from Health Care Amanda - slide 45 of 54 Counting Defiers: Examples from Health Care Amanda - slide 46 of 54 Counting Defiers: Examples from Health Care Amanda - slide 47 of 54 Counting Defiers: Examples from Health Care Amanda - slide 48 of 54 Counting Defiers: Examples from Health Care Amanda - slide 49 of 54 Counting Defiers: Examples from Health Care Amanda - slide 50 of 54 Counting Defiers: Examples from Health Care Amanda - slide 51 of 54 Counting Defiers: Examples from Health Care Amanda - slide 52 of 54 Counting Defiers: Examples from Health Care Amanda - slide 53 of 54 Counting Defiers: Examples from Health Care Amanda - slide 54 of 54
Description: Counting Defiers: Examples from Health Care Amanda E. Kowalski Gail Wilensky Professor of Applied Economics and Public Policy University of Michigan May 2021 1 of 51 Doing More When Youre Running LATE: Applying Marginal Treatment Effect

Related Topics

Download Presentation

"Counting Defiers: Examples from Health Care Amanda" is the property of its rightful owner. Permission is granted to download and print the materials on this website for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.

Presentation Transcript

slide1. Counting Defiers: Examples from Health Care Amanda E. Kowalski
Gail Wilensky Professor of Applied Economics and Public Policy
University of Michigan

May 2021 1 of 51<br>
slide2. “Doing More When You’re Running LATE: Applying Marginal Treatment Effect Methods to Examine Treatment Effect Heterogeneity in Experiments.” NBER WP 22363.

“Reconciling Seemingly Contradictory Results from the Oregon Health Insurance Experiment and the Massachusetts Health Reform.” Accepted, Review of Economics and Statistics.

“Behavior within a Clinical Trial and Implications for Mammography Guidelines.” Revise & Resubmit, Review of Economic Studies.

“How to Examine External Validity Within an Experiment.” Revised & Resubmitted, Journal of Economics and Management Strategy.

“A Model of a Randomized Experiment with an Application to the PROWESS Clinical Trial.” NBER WP 25670.

“Counting Defiers.” NBER WP 25671.

“Counting Defiers: Examples from Health Care.” arXiv:1912.06739. 2 of 51<br>
slide3. 3 of 51<br>
slide4. 4 of 51<br>
slide5. Conduct informative finite sample inference on # defiers and analogous quantities that capture heterogeneous intervention effects, provide novel test of LATE monotonicity
Perform inference on count in finite sample, not fraction in infinite population
Model randomization process using explicit likelihood function, unlike Fisher (1935), Dwass (1957), Canay et al. (2017), etc.
Do not treat # defiers as nuisance parameter, unlike Copas (1973), Chung et al. (2013), Chiba (2015), Rigdon and Hudgens (2015), Ding, Feller, and Miratrix (2016), Li and Ding (2016), Ding and Miratrix (2019), Wu and Ding (2020)
Conduct exact inference, even when testing multiple hypotheses simultaneously
Use likelihood ratio test statistic, which also facilitates informative inference on other quantities
Engage with impossibility results based on Boole-Fréchet-Hoeffding bounds
Use more data and a stronger assumption about the randomization process than applications in the treatment effects literature: Heckman et al. (1997), Manski (1997), Tian and Pearl (2000), Zhang and Rubin (2003), Fan and Park (2010), Mullahy (2018), Ding and Miratrix (2019)
Demonstrate importance of stronger assumption: conduct informative inference on # defiers even with limited data
Engage with impossibility results and tests of LATE monotonicity based on simplifying asymptotic assumptions
Use less data than applications, which require Z, D, and Y, as opposed to Z and D or Y: Balke and Pearl (1997), Imbens and Rubin (1997), Heckman and Vytlacil (2001), Heckman and Vytlacil (2005), Richardson and Robins (2010), Huber and Mellace (2013, 2015), Kitagawa (2015), Mourifié and Wan (2017), Machado et al. (2019)
Demonstrate asymptotic assumption obscures information by adding one to model: demonstrate that resulting likelihood is an average of the finite sample likelihood and does not vary with # defiers. Also demonstrate equivalence with likelihood used by Barnard (1947), Kline and Walters (2020) for informative inference on average intervention effect
Demonstrate usefulness in hypothetical drug trial
Point estimate implies that at least 40 out of 100 individuals would be saved, but I infer with 95% confidence that at least 3 individuals would be killed 5 of 51<br>
slide6. Conduct informative finite sample inference on # defiers and analogous quantities that capture heterogeneous intervention effects, provide novel test of LATE monotonicity
Perform inference on count in finite sample, not fraction in infinite population
Model randomization process using explicit likelihood function, unlike Fisher (1935), Dwass (1957), Canay et al. (2017), etc.
Do not treat # defiers as nuisance parameter, unlike Copas (1973), Chung et al. (2013), Chiba (2015), Rigdon and Hudgens (2015), Ding, Feller, and Miratrix (2016), Li and Ding (2016), Ding and Miratrix (2019), Wu and Ding (2020)
Conduct exact inference, even when testing multiple hypotheses simultaneously
Use likelihood ratio test statistic, which also facilitates informative inference on other quantities
Engage with impossibility results based on Boole-Fréchet-Hoeffding bounds
Use more data and a stronger assumption about the randomization process than applications in the treatment effects literature: Heckman et al. (1997), Manski (1997), Tian and Pearl (2000), Zhang and Rubin (2003), Fan and Park (2010), Mullahy (2018), Ding and Miratrix (2019)
Demonstrate importance of stronger assumption: conduct informative inference on # defiers even with limited data
Engage with impossibility results and tests of LATE monotonicity based on simplifying asymptotic assumptions
Use less data than applications, which require Z, D, and Y, as opposed to Z and D or Y: Balke and Pearl (1997), Imbens and Rubin (1997), Heckman and Vytlacil (2001), Heckman and Vytlacil (2005), Richardson and Robins (2010), Huber and Mellace (2013, 2015), Kitagawa (2015), Mourifié and Wan (2017), Machado et al. (2019)
Demonstrate asymptotic assumption obscures information by adding one to model: demonstrate that resulting likelihood is an average of the finite sample likelihood and does not vary with # defiers. Also demonstrate equivalence with likelihood used by Barnard (1947), Kline and Walters (2020) for informative inference on average intervention effect
Demonstrate usefulness in hypothetical drug trial
Point estimate implies that at least 40 out of 100 individuals would be saved, but I infer with 95% confidence that at least 3 individuals would be killed 6 of 51<br>
slide7. Conduct informative finite sample inference on # defiers and analogous quantities that capture heterogeneous intervention effects, provide novel test of LATE monotonicity
Perform inference on count in finite sample, not fraction in infinite population
Model randomization process using explicit likelihood function, unlike Fisher (1935), Dwass (1957), Canay et al. (2017), etc.
Do not treat # defiers as nuisance parameter, unlike Copas (1973), Chung et al. (2013), Chiba (2015), Rigdon and Hudgens (2015), Ding, Feller, and Miratrix (2016), Li and Ding (2016), Ding and Miratrix (2019), Wu and Ding (2020)
Conduct exact inference, even when testing multiple hypotheses simultaneously
Use likelihood ratio test statistic, which also facilitates informative inference on other quantities
Engage with impossibility results based on Boole-Fréchet-Hoeffding bounds
Use more data and a stronger assumption about the randomization process than applications in the treatment effects literature: Heckman et al. (1997), Manski (1997), Tian and Pearl (2000), Zhang and Rubin (2003), Fan and Park (2010), Mullahy (2018), Ding and Miratrix (2019)
Demonstrate importance of stronger assumption: conduct informative inference on # defiers even with limited data
Engage with impossibility results and tests of LATE monotonicity based on simplifying asymptotic assumptions
Use less data than applications, which require Z, D, and Y, as opposed to Z and D or Y: Balke and Pearl (1997), Imbens and Rubin (1997), Heckman and Vytlacil (2001), Heckman and Vytlacil (2005), Richardson and Robins (2010), Huber and Mellace (2013, 2015), Kitagawa (2015), Mourifié and Wan (2017), Machado et al. (2019)
Demonstrate asymptotic assumption obscures information by adding one to model: demonstrate that resulting likelihood is an average of the finite sample likelihood and does not vary with # defiers. Also demonstrate equivalence with likelihood used by Barnard (1947), Kline and Walters (2021) for inference on average intervention effect
Demonstrate usefulness in hypothetical drug trial
Point estimate implies that at least 40 out of 100 individuals would be saved, but I infer with 95% confidence that at least 3 individuals would be killed 7 of 51<br>
slide8. Conduct informative finite sample inference on # defiers and analogous quantities that capture heterogeneous intervention effects, provide novel test of LATE monotonicity
Perform inference on count in finite sample, not fraction in infinite population
Model randomization process using explicit likelihood function, unlike Fisher (1935), Dwass (1957), Canay et al. (2017), etc.
Do not treat # defiers as nuisance parameter, unlike Copas (1973), Chung et al. (2013), Chiba (2015), Rigdon and Hudgens (2015), Ding, Feller, and Miratrix (2016), Li and Ding (2016), Ding and Miratrix (2019), Wu and Ding (2020)
Conduct exact inference, even when testing multiple hypotheses simultaneously
Use likelihood ratio test statistic, which also facilitates informative inference on other quantities
Engage with impossibility results based on Boole-Fréchet-Hoeffding bounds
Use more data and a stronger assumption about the randomization process than applications in the treatment effects literature: Heckman et al. (1997), Manski (1997), Tian and Pearl (2000), Zhang and Rubin (2003), Fan and Park (2010), Mullahy (2018), Ding and Miratrix (2019)
Demonstrate importance of stronger assumption: conduct informative inference on # defiers even with limited data
Engage with impossibility results and tests of LATE monotonicity based on simplifying asymptotic assumptions
Use less data than applications, which require Z, D, and Y, as opposed to Z and D or Y: Balke and Pearl (1997), Imbens and Rubin (1997), Heckman and Vytlacil (2001), Heckman and Vytlacil (2005), Richardson and Robins (2010), Huber and Mellace (2013, 2015), Kitagawa (2015), Mourifié and Wan (2017), Machado et al. (2019)
Demonstrate asymptotic assumption obscures information by adding one to model: demonstrate that resulting likelihood is an average of the finite sample likelihood and does not vary with # defiers. Also demonstrate equivalence with likelihood used by Barnard (1947), Kline and Walters (2021) for inference on average intervention effect
Demonstrate usefulness in hypothetical drug trial
Point estimate implies that at least 40 out of 100 individuals would be saved, but I infer with 95% confidence that at least 3 individuals would be killed 8 of 51<br>
slide9. Model
Hypothesis Testing
Examples from Health Care
Impossibility Results Outline 9 of 51<br>
slide10. A.1. (No interference) Each individual has one potential outcome in the control arm and one potential outcome in the intervention arm. 10 of 51<br>
slide11. 11 of 51<br>
slide12. 12 of 51<br>
slide13. 13 of 51<br>
slide14. 14 of 51<br>
slide15. 15 of 51<br>
slide16. 16 of 51<br>
slide17. 17 of 51<br>
slide18. 18 of 51<br>
slide19. 19 of 51<br>
slide20. 20 of 51<br>
slide21. 21 of 51<br>
slide22. 22 of 51<br>
slide23. 23 of 51<br>
slide24. Model
Hypothesis Testing
Examples from Health Care
Impossibility Results Outline 24 of 51<br>
slide25. 25 of 51<br>
slide26. Model
Hypothesis Testing
Examples from Health Care
Impossibility Results Outline 26 of 51<br>
slide27. 27 of 51<br>
slide28. 28 of 51<br>
slide29. 29 of 51<br>
slide30. 30 of 51<br>
slide31. 31 of 51<br>
slide32. 32 of 51<br>
slide33. 33 of 51<br>
slide34. 34 of 51<br>
slide35. 35 of 51<br>
slide36. 36 of 51<br>
slide37. 37 of 51<br>
slide38. 38 of 51<br>
slide39. 39 of 51<br>
slide40. 40 of 51<br>
slide41. 41 of 51<br>
slide42. 42 of 51<br>
slide43. 43 of 51<br>
slide44. 44 of 51<br>
slide45. 45 of 51<br>
slide46. 46 of 51<br>
slide47. Example: Hypothetical Clinical Trial Results 47 of 51<br>
slide48. Model
Hypothesis Testing
Examples from Health Care
Impossibility Results Outline 48 of 51<br>
slide49. Impossibility Results Based on Boole-Fréchet-Hoeffding Bounds I use more data: full data configuration vs shares treated in intervention & control
I make a stronger assumption: A.2(iid) vs independence of intervention and potential outcomes
I can still conduct informative inference on # defiers even with limited data
Therefore, the stronger assumption is important 49 of 51<br>
slide50. Impossibility Results Based on Simplifying Asymptotic Assumptions 50 of 51<br>
slide51. Conduct informative finite sample inference on # defiers and analogous quantities that capture heterogeneous intervention effects, provide novel test of LATE monotonicity
Model randomization process using explicit likelihood function, unlike Fisher (1935), Dwass (1957), Canay et al. (2017), etc.
Do not treat # defiers as nuisance parameter, unlike Copas (1973), Chung et al. (2013), Chiba (2015), Rigdon and Hudgens (2015), Ding, Feller, and Miratrix (2016), Li and Ding (2016), Ding and Miratrix (2019), Wu and Ding (2020)
Perform inference on count in finite sample, not fraction in infinite population
Conduct exact inference, even when testing multiple hypotheses simultaneously
Use likelihood ratio test statistic, which also facilitates informative inference on other quantities
Engage with impossibility results based on Boole-Fréchet-Hoeffding bounds
Use more data and a stronger assumption about the randomization process than applications in the treatment effects literature: Heckman et al. (1997), Manski (1997), Tian and Pearl (2000), Zhang and Rubin (2003), Fan and Park (2010), Mullahy (2018), Ding and Miratrix (2019)
Demonstrate importance of stronger assumption: conduct informative inference on # defiers even with limited data
Engage with impossibility results and tests of LATE monotonicity based on simplifying asymptotic assumptions
Use less data than applications, which require Z, D, and Y, as opposed to Z and D or Y: Balke and Pearl (1997), Imbens and Rubin (1997), Heckman and Vytlacil (2001), Heckman and Vytlacil (2005), Richardson and Robins (2010), Huber and Mellace (2013, 2015), Kitagawa (2015), Mourifié and Wan (2017), Machado et al. (2019)
Demonstrate asymptotic assumption obscures information by adding one to model: demonstrate that resulting likelihood is an average of the finite sample likelihood and does not vary with # defiers. Also demonstrate equivalence with likelihood used by Barnard (1947), Kline and Walters (2021) for informative inference on average intervention effect
Demonstrate usefulness in hypothetical drug trial
Point estimate implies that at least 40 out of 100 individuals would be saved, but I infer with 95% confidence that at least 3 individuals would be killed 51 of 51<br>
slide52. Appendix 52 of 51<br>
slide53. 53 of 51<br>
slide54. 54 of 51<br>