Timeline for Can one usefully specify a multilevel-model with a partially-nested, partially non-nested structure?
Current License: CC BY-SA 4.0
15 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 17, 2020 at 12:14 | history | edited | Robert Long |
edited tags
|
|
| Jan 23, 2019 at 11:50 | history | edited | Robert Long |
Added Mixed model tag
|
|
| Jan 23, 2019 at 1:12 | vote | accept | Lepidopterist | ||
| S Jan 23, 2019 at 1:12 | history | bounty ended | Lepidopterist | ||
| S Jan 23, 2019 at 1:12 | history | notice removed | Lepidopterist | ||
| Jan 22, 2019 at 17:34 | answer | added | Robert Long | timeline score: 6 | |
| Jan 18, 2019 at 17:44 | comment | added | Lepidopterist | What lmer is doing can be specified in a fully Bayesian way if you want. Indeed that is how I am able to understand what lmer is doing given Gelman and Hill's exposition. I'm indicating that if lmer cannot do this, I am open to a Bayesian specification that makes sense. With regard to "where is the problem with this", it's that I don't know how to specify a model with partially nested and non-nested random effects. For example, naively I would guess lmer (y ~ x.centered + (1 + x.centered | eth1/eth2/eth3) + (1 + x.centered | age) + (1 + x.centered | eth1/eth2/eth3:age)). | |
| Jan 18, 2019 at 17:36 | comment | added | Sextus Empiricus | What sort of Bayesian regression/modelling are you doing? This is unclear because you refer to lmer. | |
| Jan 18, 2019 at 17:34 | comment | added | Sextus Empiricus | "nested effects + an effect which is not-nested in the nested-variables." where is the problem with this? | |
| Jan 18, 2019 at 14:30 | comment | added | Lepidopterist | That captures the nested random effects, which I was already aware of how to do. I am interested in modeling, using random effects or a more Bayesian setting, nested effects + an effect which is not-nested in the nested-variables. I gave the example of trying to model the random effects of geographical location (nested as country, state, county and city effects) plus some other random effect, like age. I'm looking first and foremost for an intuitive and practical approach in this setting. | |
| Jan 18, 2019 at 11:52 | comment | added | Sextus Empiricus | Do you want something like this: stats.stackexchange.com/questions/134995/… ? Can explain a bit more the issue that you have with this? | |
| S Jan 16, 2019 at 2:33 | history | bounty started | Lepidopterist | ||
| S Jan 16, 2019 at 2:33 | history | notice added | Lepidopterist | Draw attention | |
| Jan 16, 2019 at 2:28 | history | edited | Lepidopterist | CC BY-SA 4.0 |
added 1684 characters in body; edited title
|
| Jan 7, 2019 at 22:31 | history | asked | Lepidopterist | CC BY-SA 4.0 |