The goal of ssrhom is to analyze data from single subject designs using hierarchical ordinal regression models.
You can install ssrhom from R-universe with:
install.packages(
"ssrhom",
repos = c("https://jamesuanhoro.r-universe.dev", "https://cloud.r-project.org")
)Using the tasky dataset which comes with the package:
library(ssrhom)
head(tasky)
#> person phase count proportion time
#> 1 rebeccah A 4 0.6666667 1
#> 2 rebeccah A 2 0.3333333 2
#> 3 rebeccah A 2 0.3333333 3
#> 4 rebeccah A 0 0.0000000 4
#> 5 rebeccah A 2 0.3333333 5
#> 6 rebeccah A 0 0.0000000 6# all arguments below are required for a given dataset
tasky_model <- ssrhom_model_ab(
data = tasky,
grouping = "phase", condition = "B",
time = "time", outcome = "count", case = "person"
)
#> Warning: There were 5 divergent transitions after warmup. See
#> https://mc-stan.org/misc/warnings.html#divergent-transitions-after-warmup
#> to find out why this is a problem and how to eliminate them.
#> Warning: Examine the pairs() plot to diagnose sampling problems
#> Warning: Bulk Effective Samples Size (ESS) is too low, indicating posterior means and medians may be unreliable.
#> Running the chains for more iterations may help. See
#> https://mc-stan.org/misc/warnings.html#bulk-ess# how much autocorrelation?
print(tasky_model$model, "ac")
#> Inference for Stan model: model_ab.
#> 3 chains, each with iter=1500; warmup=750; thin=1;
#> post-warmup draws per chain=750, total post-warmup draws=2250.
#>
#> mean se_mean sd 2.5% 25% 50% 75% 97.5% n_eff Rhat
#> ac -0.05 0.01 0.18 -0.39 -0.17 -0.06 0.07 0.32 1166 1
#>
#> Samples were drawn using NUTS(diag_e) at Tue Sep 17 08:40:58 2024.
#> For each parameter, n_eff is a crude measure of effective sample size,
#> and Rhat is the potential scale reduction factor on split chains (at
#> convergence, Rhat=1).# for a list of available effect sizes:
ssrhom_list_stats()
#> mean : mean of each case in both phases
#> median : median of each case in both phases
#> mean-diff : mean difference between phases by case
#> median-diff : median difference between phases by case
#> lrr : log-ratio of means by case when data are never negative
#> lor : log-ratio of odds by case when data fall between 0 and 1 inclusive
#> nap : non-overlap of all pairs by case
#> tau : A linear transformation of NAP
#> pem : Proportion of treatment cases exceeding control cases by case
#> smd-c : Standardized mean difference using control SD as standardizer by case
#> smd-p : Standardized mean difference using pooled SD as standardizer by case
#>
#> If model was called with `increase = FALSE`, then effects are reversed.# non-overlap of all pairs
ssrhom_get_effect(tasky_model, stat = "nap")
#> # A tibble: 3 × 8
#> variable median sd q2.5 q97.5 rhat ess_bulk ess_tail
#> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 nap[amber] 0.896 0.0425 0.797 0.961 1.00 2269. 2049.
#> 2 nap[cara] 0.691 0.0882 0.504 0.843 1.00 2532. 2092.
#> 3 nap[rebeccah] 0.910 0.0442 0.801 0.976 1.00 1789. 1847.# within subject standardized mean difference using pooled SD
ssrhom_get_effect(tasky_model, stat = "smd-p")
#> # A tibble: 3 × 8
#> variable median sd q2.5 q97.5 rhat ess_bulk ess_tail
#> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 smd-p[amber] 1.87 0.372 1.25 2.71 0.999 2349. 2124.
#> 2 smd-p[cara] 0.748 0.391 0.0386 1.57 1.00 2464. 2129.
#> 3 smd-p[rebeccah] 2.15 0.542 1.33 3.47 1.00 1970. 1797.