hdiffr

hdiffr performs an event history diffusion model that is modified in order to account for heterogeneity in the diffusion process. In addition to modeling characteristics that influnece the direct propensity of a focal actor to adopt, hdiffr can also model characteristics of the focal actor that make it more susceptible to the prior adoptions of other actors, as well as characteristics of prior-adopting actors that make them more or less influential on others, and characteristics of the social structure among actors that influence the transmission of influence between actors.

Author

Sang Won Han, Ph.D. candidate in Sociology at Columbia University

References

Strang, David and Nancy Brandon Tuma. 1993. “Spatial and Temporal Heterogeneity in Diffusion.” American Journal of Sociology 99(3): 614-639.

Briscoe, Forrest, Abhinav Gupta, and Mark S. Anner. 2015. “Social Activism and Practice Diffusion: How Activist Tactics Affect Non-Targeted Organizations.” Administrative Science Quarterly 60(2): 300-332.

Installation

The experimental version can be obtained via:

devtools::install_github("petershan1119/hdiffr")

Example

This is a basic example which performs heterogeneous diffusion model with multispell data:

library(hdiffr)
data(exampleData)
result <- hdiffr(data = exampleData, xvars = c('lnsale', 'roa', 'vote'), vvars = 'activistgrp', 
                 wvars = 'lnsale', zgvars = 'sic', idvar = 'gvkey',
                 multispell = 1, vintercept = 1, hrno = 1)
#> Warning in hdiffr(data = exampleData, xvars = c("lnsale", "roa", "vote"), :
#> Warning: More than one observation per ID detected.
#> Loading required package: dplyr
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

If you want to see the results, use summary:

summary(result)
#> 
#> Call:
#> survival::survreg(formula = as.formula(formula), data = data_wf_gi, 
#>     dist = dist)
#>                 Value Std. Error     z       p
#> (Intercept)    5.6198     1.3413  4.19 2.8e-05
#> lnsale        -0.3842     0.1492 -2.58   0.010
#> roa           -2.2903     2.1201 -1.08   0.280
#> vote          -0.0241     0.0230 -1.05   0.296
#> v_activistgrp -0.1493     0.0571 -2.61   0.009
#> v_intercept    0.1355     0.0826  1.64   0.101
#> w_lnsale      -0.0417     0.0377 -1.11   0.268
#> zg_sic        -0.3177     0.1340 -2.37   0.018
#> 
#> Scale fixed at 1 
#> 
#> Exponential distribution
#> Loglik(model)= -225.4   Loglik(intercept only)= -245.3
#>  Chisq= 39.72 on 7 degrees of freedom, p= 1.4e-06 
#> Number of Newton-Raphson Iterations: 6 
#> n= 575