Data
====

1. election88.data.R
  - N            : number of observations
  - n_age        : number of age groups
  - n_edu        : number of edu groups
  - n_age_edu    : number of age-edu groups
  - n_state      : number of states
  - n_region_full: number of regions
  - age          : age category
  - age_edu      : age-edu value
  - black        : is black? 1: Yes, 0: No
  - edu          : education level
  - female       : is female? 1: Yes, 0: No
  - region_full  : region number
  - state        : state number
  - v_prev_full  : preview values
  - y            : vote outcome

2. pilots.data.R
  - N          : number of observations
  - n_groups   : number of groups
  - n_scenarios: number of scenarios
  - group_id   : group id
  - scenario_id: scenario id
  - y          : score

Models
======

1. Multilevel model with varying intercept
  election88.stan: glmer(y ~ black + female + (1 | state), family=binomial(link="logit"))

2. Multilevel model with several group level predictors
  election88_full.stan: glmer(y ~ black + female + v_prev_full + (1 | age) + (1 | age_edu) 
				  + (1 | state) + (1 | region_full), 
			      family=binomial(link="logit"))
  pilots.stan         : lmer(y ~ 1 + (1 | group) + (1 | scenario))

3. Above models with Matt trick
  pilots_chr.stan: lmer(y ~1 + (1 | group) + (1 | scenario))