home.social

#instrumentalvariable — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #instrumentalvariable, aggregated by home.social.

  1. Asking for help/refs: I have observational data where some patients always get treatment T=A due to their diagnosis (D) while the rest get either T=A or T=B based on clinical judgement. It seems to me that it might be possible to get closer to causal claims for T by exploiting the (lack of) differences between the "T=A, D=1" and "T=A, D=0" groups. This seems akin to IV analysis with D as instrument. Any ideas? A bit more detail at: discourse.datamethods.org/t/ob #stats #instrumentalVariable #biostats

  2. Asking for help/refs: I have observational data where some patients always get treatment T=A due to their diagnosis (D) while the rest get either T=A or T=B based on clinical judgement. It seems to me that it might be possible to get closer to causal claims for T by exploiting the (lack of) differences between the "T=A, D=1" and "T=A, D=0" groups. This seems akin to IV analysis with D as instrument. Any ideas? A bit more detail at: discourse.datamethods.org/t/ob #stats #instrumentalVariable #biostats

  3. Asking for help/refs: I have observational data where some patients always get treatment T=A due to their diagnosis (D) while the rest get either T=A or T=B based on clinical judgement. It seems to me that it might be possible to get closer to causal claims for T by exploiting the (lack of) differences between the "T=A, D=1" and "T=A, D=0" groups. This seems akin to IV analysis with D as instrument. Any ideas? A bit more detail at: discourse.datamethods.org/t/ob #stats #instrumentalVariable #biostats

  4. Asking for help/refs: I have observational data where some patients always get treatment T=A due to their diagnosis (D) while the rest get either T=A or T=B based on clinical judgement. It seems to me that it might be possible to get closer to causal claims for T by exploiting the (lack of) differences between the "T=A, D=1" and "T=A, D=0" groups. This seems akin to IV analysis with D as instrument. Any ideas? A bit more detail at: discourse.datamethods.org/t/ob #stats #instrumentalVariable #biostats

  5. Asking for help/refs: I have observational data where some patients always get treatment T=A due to their diagnosis (D) while the rest get either T=A or T=B based on clinical judgement. It seems to me that it might be possible to get closer to causal claims for T by exploiting the (lack of) differences between the "T=A, D=1" and "T=A, D=0" groups. This seems akin to IV analysis with D as instrument. Any ideas? A bit more detail at: discourse.datamethods.org/t/ob #stats #instrumentalVariable #biostats