Skip to content

Instantly share code, notes, and snippets.

Avatar

Michael Frank mcfrank

View GitHub Profile
@mcfrank
mcfrank / possessives.R
Last active Mar 3, 2020
get possessive datas from wordbank and plot
View possessives.R
library(wordbankr)
library(tidyverse)
possess_data <- get_instrument_data(language = "English (American)",
form = "WS",
items = "item_687") # note that 687 is the s-possess item
admin_data <- get_administration_data(language = "English (American)",
form = "WS")
left_join(possess_data, admin_data) %>%
@mcfrank
mcfrank / mb4_sim.R
Created Mar 2, 2020
MB4 GLM simulation
View mb4_sim.R
library(tidyverse)
n_sim <- 100
sims <- expand_grid(n_total = seq(50,500,25),
i = 1:n_sim) %>%
mutate(idx = 1:n()) %>%
split(.$idx) %>%
map_df(function (df) {
cntl_sim <- tibble(choice = c(rbinom(n = df$n_total/2, size = 1, p = .68),
rbinom(n = df$n_total/2, size = 1, p = .5)),
condition = c(rep("social", df$n_total/2),
@mcfrank
mcfrank / sim_covariates.R
Last active Oct 15, 2019
Simulate covariate effect on linear model - modified from Jan Vanhove
View sim_covariates.R
library(MASS)
library(tidyverse)
# from https://homeweb.unifr.ch/VanhoveJ/Pub/simulation_covariates.html
# Define function
# n <- 20 # observations per group
# diff <- 0 # difference between groups
# sd_y <- 3 # within-group sd of outcome (population-level)
# rho_covars <- c(0.7, 0.3, 0) # correlation of control variables with outcome
# in control group; add more coefficients to
@mcfrank
mcfrank / table_turning.R
Created Oct 10, 2019
Sample categorical regression with table turning
View table_turning.R
library(tidyverse)
library(brms)
library(langcog)
d <- tibble(subject = c(1, 1, 1, 1, 2, 2, 2, 2,
3, 3, 3, 3, 4, 4, 4, 4),
response = c("E",NA,"A","error",
"E","E","E","error",
"A","A","E","error",
"A","A","A","A"),
@mcfrank
mcfrank / covariate_sim.R
Created Oct 8, 2019
Simulate covariate adjustment
View covariate_sim.R
# based on http://egap.org/methods-guides/10-things-know-about-covariate-adjustment
library(MASS) # for mvrnorm()
library(tidyverse)
set.seed(1234567)
num.reps = 1000
# True treatment effect is 0 for every unit
adj.est = function(n, cov.matrix, treated) {
View regression_example.R
iqs <- data_frame(iq = rnorm(mean = 100, sd = 15, n= 40),
school = c(rep("Stanford",20), rep("Berkeley",20)),
year = factor(rep(c(1950,1990),20)))
summary(lm(iq ~ school * year, data = iqs))
summary(lm(scale(iq) ~ school * year, data = iqs))
@mcfrank
mcfrank / bayesian_mb1.R
Created May 3, 2019
Bayesian vs. frequentist LMMs for ManyBabies
View bayesian_mb1.R
library(brms)
library(lme4)
library(here)
library(tidyverse)
d <- read_csv(here("processed_data/03_data_trial_main.csv")) %>%
mutate(method = case_when(
method == "singlescreen" ~ "Central fixation",
method == "eyetracking" ~ "Eye tracking",
method == "hpp" ~ "HPP",
@mcfrank
mcfrank / manybabies2_sim.R
Created Jan 24, 2019
Simulation for manybabies 2 pilot
View manybabies2_sim.R
library(tidyverse)
library(ggthemes)
n_trials <- c(2,4,8)
p_anticipation <- seq(0,1,.1)
n_subs <- c(12,24,36,48)
n_sims <- 1000
d <- expand.grid(n_trials = n_trials,
p_anticipation = p_anticipation,
@mcfrank
mcfrank / negation_context.R
Created Jul 10, 2018
plot of negation in context (model results)
View negation_context.R
library(tidyverse)
# Nonexistence context, Nonexistence referent, apples? QUD
nna <- tibble(ratio = c("ratio0", "ratio1", "ratio2", "ratio3"),
probs = c(0.3323976330361966, 0.39954163105573637, 0.37453297805618113, 0.3323976330361966),
context = "nonexistence context",
referent = "nonexistence referent",
qud = "apples?")
naa <- tibble(ratio = c("ratio0", "ratio1", "ratio2", "ratio3"),
@mcfrank
mcfrank / time_transition.R
Created Jun 30, 2018
transition between states in some code
View time_transition.R
library(tidyverse)
foo <- data_frame(time = c(1:5, 1:5),
code = c(1,2,3,2,1,0,0,1,2,1),
subid = c(rep(1,5), rep(2, 5)))
foo %>%
group_by(subid) %>%
mutate(d_code = c(0,diff(code)),
one_two_transition = code == 2 & d_code == 1,
two_three_transition = code == 3 & d_code == 1) %>%
You can’t perform that action at this time.