Skip to content

Person Dcpst enum still returns 'PreviouslyPartnered' which isn't generated in initial populations #311

@andrewbaxter439

Description

@andrewbaxter439

Problem

  • Merging in multiple pull requests #294 merges in changes discussed in Data&estimates update 5nov #283 which makes dcpst (partnership) into a two-category variable: partnered and single
  • regression estimates derived from initial population files generate one coefficient (e.g. Dcpst_Single over baseline 'partnered')
  • getDcpst still returns PreviouslyPartnered when should return Single for newly singly persons
  • thus the Dcpst_single coefficient is estimated for this population but never applied

https://github.com/centreformicrosimulation/SimPaths/blob/bb58b85989cdd377477e7c590b3630793971fba1/src/main/java/simpaths/model/Person.java#L4648-L4659

Suggestions

Either:

  • if 'previously partnered' is not intended for use anymore anywhere then remove from getDcpst and return 'single' when relevant
  • or estimate the three-category effects on a separate dummy to restore 'previously partnered' effects.

Metadata

Metadata

Labels

No labels
No labels

Type

No type
No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions