make_dilated_out: Create potential dilated outcomes.
In szonszein/interference: Design-Based Estimation of Spillover Effects

Description Usage Arguments Details Value Functions References Examples

Create potential dilated outcomes for simulation.

make_dilated_out(adj_matrix, make_corr_out, seed, hop, multipliers = NULL)

make_dilated_out_full_neighborhood(
  adj_matrix,
  make_corr_out,
  multipliers = NULL,
  seed = NULL
)

`adj_matrix`	an `N` * `N` numeric matrix of 0, 1 entries such as those retuned by `make_adj_matrix`, or a `N` * `N` matrix, where `N` is the number of units.
`make_corr_out`	unused.
`seed`	random number for result replicability.
`hop`	number; either `1` or `2`. If `1` assumes first-degree interference and produces four exposure conditions, if `2` assumes second-degree interference and produces eight exposure conditions.
`multipliers`	numeric vector with dilated effects multipliers. Must be of length 3 if `hop` is 1, and length 7 if `hop` is 2. Default are `c(2,1.5,1.25)` and `c(2.25,2,1.75,1.5,1.375,1.25,1.125)`.

make_dilated_out produces potential dilated outcomes for simulation according to two different exposure mappings. An exposure mapping that assumes that interference happens only through direct peer connections (first-degree interference) which produces four exposure conditions: Direct + Indirect Exposure, Isolated Direct Exposure, Indirect Exposure, No Exposure. And an exposure mapping that assumes second-degree interference which produces eight exposure conditions. The values for the baseline No Exposure condition are drawn from an absolute standard normal distribution which is correlated with the unit's first and second order degree. The values for the other exposure conditions are obtained by multiplying the vector of multipliers by the baseline No Exposure value.

An K * N named numeric matrix, where K corresponds to the number of exposure conditions and N number of units.

make_dilated_out_full_neighborhood: Produces potential dilated outcomes for simulation according to an exposure mapping that assumes first-degree interference and that produces two exposure conditions: All Treat, a condition in which a unit and 100% of its first-degree neighbors are treated, and All Control, a unit and 100% of its first-degree neighbors are in the control condition.

Aronow, P.M. et al. (2020). Spillover effects in experimental data. arXiv preprint, arXiv:2001.05444.

adj_matrix <- make_adj_matrix(N = 9, model = 'sq_lattice')
multipliers <- c(2, 1.5, 1.25)
make_dilated_out(adj_matrix, make_corr_out, seed = 357,
multipliers = multipliers, hop = 1)
multipliers <- c(2.25,2,1.75,1.5,1.375,1.25,1.125)
make_dilated_out(adj_matrix, make_corr_out, seed = 357,
multipliers = multipliers, hop = 2)
multiplier <- 4
make_dilated_out_full_neighborhood(adj_matrix, make_corr_out,
                                   multipliers=multiplier,
                                   seed=357)