README.md

In openmindplatform/openmindR:

openmindR

Overview

• openmindR Cleaning Functions
• openmindR Analysis Functions
• openmindR ggplot2 theme

Install package like this:

devtools::install_github("openmindplatform/openmindR")

Load package(s):

library(openmindR)
library(tidyverse)

Create Master dataset

This code merges Assessment data with Participant Progress and Access Codes tables (to add UserType, OpenMindVersion and other variables).

pacman::p_load(tidyverse, openmindR)


assessmentv6 <- om_download_at(tables = "AssessmentV6",
                               clean = T, v6.1 = T)

## ParticipantProgress1
pp1 <- om_download_at(tables = "ParticipantProgress") %>% 
  select(OMID, AccessCode, OpenMindVersion, Country, 
         Research, DateStarted, DateFinished)

## ParticipantProgress2
pp2 <- om_download_at(tables = "ParticipantProgress2") %>% 
  select(OMID, AccessCode, OpenMindVersion, Country, 
         Research, DateStarted, DateFinished)

## AccessCodes
acs <- om_download_at(tables = "AccessCodes") %>%
  select(AccessCode, UserType, GroupName)

## bring it all together
cleaned_dat <- pp1  %>%
  bind_rows(pp2) %>%
  ## turn dates into date format
  mutate(DateFinished = lubridate::as_date(DateFinished)) %>%
  mutate(DateStarted = lubridate::as_date(DateStarted)) %>%
  ## compute WithinADay
  mutate(WithinADay = as.numeric(DateStarted==DateFinished)) %>%
  ## join in Assessment data
  coalesce_join(assessmentv6, join = dplyr::right_join) %>% 
  ## join in Access Code Data
  coalesce_join(acs, join = dplyr::left_join) %>%
  ## if AccessCode is IndividualUser, then UserType is IndividualUser
  mutate(UserType = ifelse(AccessCode == "IndividualUser",
                           "IndividualUser", UserType)) %>%
  drop_na(UserType) %>%
  drop_na(DateStarted)

openmindR Cleaning Functions

The following functions are meant to turn AirTable data into a single clean file that can be analyzed.

• om_download_at

• om_filter_data

• om_clean_ppol

• om_dummy_nonwhite

• om_dummy_nonstraight

• om_gather

• om_parse_lifehacks

• merge_assessments

om_download_at

openmindR can download and clean data directly from Airtable.

• clean: Whether to return “clean” data (i.e. numeric data is numeric and minors or opt-outs are NA. Also constructs measures and computes dummy variables. (TRUE/FALSE)
• file: give a file path to where the downloaded data should be stored
• v6.1: filter down the dataset to only include Assessment V6.1 (TRUE/FALSE)

Here is a code example that will download Assessment V6, clean it, save it into a folder called “Data” under Research and filter down to only include V6.1 data.

assessmentv6 <- om_download_at(
                            tables = "AssessmentV6", 
                            clean = TRUE, 
                            file = "../../../Data/assessmentv6.1.csv",
                            v6.1 = TRUE)

Here is another example code for downloading a clean version of Assessment v7:

assessmentv7 <- om_download_at(
                            tables = "AssessmentV7", 
                            clean = TRUE)

## Seting up OM key
## Seting up key
## Download AssessmentV7 Data
## Done. AssessmentV7 Data has 1585 rows

Get Assessment V7.2 data and merge it with P2P data

## filter assessment v7 to only include 7.2
assessmentv72 <- assessmentv7 %>% 
  dplyr::filter(AssessmentVersion == 7.2) %>%
  ## remove all columns that are empty (those are from older verrsions)
  dplyr::select_if(~!all(is.na(.))) 

## download p2p dat
p2p_dat <- om_download_at(tables = "P2P", 
                          clean = TRUE)

## merge v7.2 with p2p dat
p2p_assessment <- assessmentv72 %>% 
  left_join(p2p_dat)

om_filter_data

Filter down Assessment data from AirTable by AssessmentsDone, AssessmentVersion and AccessCodes.

assessmentv7 %>% 
  # specify which number of assessment you want to have
  om_filter_data(n_assessments = 1:2,
             # assessment version?
             version = 7,
             # select Accesscode(s) 
             accesscode = "TuttlePen"
             # "TuttlePen" #try this out :)
  )

The accesscode argument is not case-sensitive and can both be used with vectors:

assessmentv7 %>% 
  # specify which number of assessment you want to have
  om_filter_data(n_assessments = 1:2,
             # assessment version?
             version = 7,
             # select Accesscode(s) 
             accesscode = c("TuttlePennStateS20", "SpriggsBTHS201S20")
  )

And individual strings:

assessmentv7 %>% 
  # specify which number of assessment you want to have
  om_filter_data(n_assessments = 1:2,
             # assessment version?
             version = 7,
             # select Accesscode(s) to produce report for
             accesscode = c("tuttle|spriggs")
  )

om_clean_ppol

Creates the following measures of Political Orientation

• ppol_raw: a variable that has all political orientation variables across all Assessment Versions (corresponds to D4)
• ppol: a factor variable ordered from “Very Progressive/left” to “Very Conservative/right”. Excludes all other categories as NA (classical liberal etc.)
• ppol_num: numeric variable ranging from 1 “Very Progressive/left” to 7 “Very Conservative/right”
• ppol_extreme: numeric variable ranging from 0 “Moderate/Middle-of-the-road” to 3 “Very Conservative/right” or “Very Progressive/left”
• ppol_cat: a factor variable which has two categories “Progressive” and “Conservative”. The rest is NA.
• ppol_catmod: a factor variable which has three categories “Progressive”, “Conservative” and “Moderates”. The rest is NA.

assessmentv7 %>% 
  om_clean_ppol()

om_dummy_nonwhite

This function creates a dummy variable from D3 (Race) called race_nonwhite and codes people who identify only as white as 0 and everyone else as 1.

assessmentv7 %>% 
  om_dummy_nonwhite()

om_dummy_nonstraight

This function creates a dummy variable from D5 (Sexuality) called sex_nonstraight and codes people who identify as heterosexual as 0 and everyone else as 1.

assessmentv7 %>% 
  om_dummy_nonstraight()

om_dummy_ut

This function creates the dummy variables from UserType

• ut_college_individ: College student (0) v. individual user (1)
• ut_corp_individ: Corp (0) v. individual user (1)
• ut_college_corp: College student (0) v. corp (1)

assessmentv7 %>% 
  om_dummy_ut()

om_gather

This function will turn Assessment data into long format.

Creates the following variables:

• Question: Q1Pre, Q2Pre, Q3Pre etc.
• Type: Pre, Post, or FollowUp
• Response: Values of the Question
• variable_code: Q1, Q2, Q3 etc.

Takes the following arguments:

• .data Assessment data
• which_strings a string indicating which variables should be parsed out (v7_var_strings has all variables for v7). The format looks as follows: “AffPol1|AffPol2|GM1” (so each variable without the “Pre”, “Post” or “FollowUp” suffix)

assessmentv7 %>% 
  om_gather(v7_var_strings) %>% 
  ## select just the relevant vars as showcase
  select(Question, Response, Type, variable_code)

om_parse_lifehacks

This function parses and cleans the list variables that contain the lifehack data. The input dataset needs to have the four following columns: LifeHacksChosen, LifeHacksComplete, LifeHacksUseful and LifeHacksReason. You can find them in the ParticipantProgress table in Airtable. The function will parse these four columns into 4 x 5 Steps variables (so 20 in total):

• LifeHack1 - LifeHack5: Which Lifehack was chosen
• LifeHacksComplete1 - LifeHacksComplete5: Did the user complete the lifehack
• LifeHacksUseful1 - LifeHacksUseful5: Did the user find completing the lifehack useful
• LifeHacksReason1 - LifeHacksReason5: Why didn’t the user complete the lifehack

## get key
key <- read_lines("../../Keys/airtabler.txt")

## download participant progress data
pp_dat <- om_download_at(tables = "ParticipantProgress")


## this parses all Lifehack data so far
parsed_lh <-  om_parse_lifehacks(pp_dat)


## if you just want the newest life hack data you can filter by OpenMind version
parsed_lh <-  pp_dat %>%
  ## making sure OpenMindVersion is numeric
  mutate(OpenMindVersion = as.numeric(OpenMindVersion)) %>% 
  ## Only include OM Version 3 and above
  filter(OpenMindVersion >= 3) %>% 
  ## parse Life hacks
  om_parse_lifehacks()

## just select the OMID and all LifeHack data (remove all other PP variables)
parsed_lh %>% 
  select(OMID, LifeHack1:LifeHacksReason5)

merge_assessments

This function merges assessment v4, v5 and v6 data and only keeps common variables.

## get previous assessment dat
v4 <- read.csv("../../../Data/2019-10-29_assessmentv4.csv") 
v5 <- read.csv("../../../Data/2019-10-29_assessmentv5.csv") 


## get (clean) assessment v6 data
v6 <- om_download_at(tables = "AssessmentV6", clean = T)


## merge all three datasets and only keep common variables
merge_assessments(v4, v5, v6)

om_reverse_code

Reverse codes items and adds them at the end of the dataset with a “*_Rev” at the end.

assessmentv7 %>% 
  om_reverse_code()

openmindR Analysis Functions

This section introduces the openmindR analysis functions.

• om_ttest
• om_lm
• om_lmer
• om_textscore

om_ttest

This function performs paired t-tests on long format Pre-Post-FollowUp data and returns stats on the model (including p-values, t-statistics and Cohen’s D effect size).

om_ttest takes two arguments:

• gathered_dat OpenMind data in long format (must be created with om_gather)
• comparison Three possible comparisons “PrePost”, “PreFollowUpT1T2” or “PreFollowUpT1T3”

PreFollowUpT1T2 performs Pre-Post comparison only for people who completed the FollowUp. PreFollowUpT1T3 is the same sample of people (only those who completed the FollowUp) but compares them Pre-FollowUp.

First we download v7 data:

## Get Key (may differ in your code)


## get v7 data
assessmentv7 <- om_download_at(key,
                            tables = "AssessmentV7",
                            clean = TRUE)

You can perform t-tests on a single variable by only gathering one variable and specifying the comparison (here: "PrePost").

## Perform t-test on a single variable
assessmentv7 %>% 
  om_gather("AffPol1") %>% 
  om_ttest(comparison = "PrePost")

However, this is not how om_ttest is intended to work. Rather it should be used on a dataset in long format with all variables that you want to perform analysis on. The next lines of code show that process.

## get results for all variables
assessmentv7  %>%
  ## select only relevant variables and composite scores
  select(OMID, AffPol1Pre:IHCultureSub3FollowUp) %>% 
  ## turn data into long format
  om_gather(which_strings = v7_var_strings)  %>%
  ## perform t-test on each variable (for Pre and Post)
  om_ttest("PrePost") %>% 
  ## arrange by cohens D
  arrange(desc(cohend)) 

Same process for Assessment v6:

## get results for all variables
assessmentv6  %>%
  ## select only relevant variables and composite scores
  select(OMID, AffPol1Pre:IntellectualHumilityFollowUp, 
         GrowthMindsetPre, GrowthMindsetPost, GrowthMindsetFollowUp,
         C1Pre, C5Pre, C6Pre, 
         C1Post, C5Post, C6Post,
         C1FollowUp, C5FollowUp, C6FollowUp,
         -contains("Preparedness3")) %>% 
  ## turn data into long format
  om_gather(which_strings = v6_var_strings)  %>%
  ## perform t-test on each variable (for Pre and Post)
  om_ttest("PrePost") %>% 
  ## arrange by cohens D
  arrange(desc(cohend)) 

perc_improved

om_ttest uses a helper function to determine the percentage of people who improved: perc_improved from T1 to T2 (or T3). For this, we need to decide which variables are coded so that “improvement” means higher values and which ones coded so that improving means lower values on the respective scales.

Here is the list of variables that should have higher values to show improvement:

##  [1] "GrowthMindset"        "CIHS_LIO"             "OutgroupLiking"      
##  [4] "OutgroupMotivation"   "Preparedness"         "C1"                  
##  [7] "C6"                   "IntellectualHumility" "GM"                  
## [10] "IHSub1"               "IHSub2"               "IHSub3"              
## [13] "IHCultureSub1"        "IHCultureSub2"        "IHCultureSub3"       
## [16] "SE"                   "Belong"               "Dissent"             
## [19] "Tolerance"            "IngroupLiking"        "IngroupMotivation"   
## [22] "OutgroupLiking"       "OutgroupMotivation"   "MotivationCon"       
## [25] "MotivationProg"

Here is the list of variables that should have lower values to show improvement:

##  [1] "AffPol1"        "AffPol2"        "GBSS"           "MAA"           
##  [5] "C5"             "Anxiety"        "Attribution"    "IntAnx"        
##  [9] "SocialDistance" "Avoidance"

om_lm

Run linear regression model (with interactions)

This function performs linear regression and gives back some neat info

Four inputs:

• .data data that was used to fit the model

• lin_mod a fitted model

• type what kind of model (currently only accepts "int" for interactions)

• switch logical. Switch variables in interaction plot. Default is FALSE

mod1 <- lm(ppol_extreme ~ ppol_cat*gender, 
           data = assessmentv7)


results <- assessmentv7 %>% 
  om_lm(lin_mod = mod1, 
        type = "int",
        switch = T)

The function returns 5 outputs (in list format).

results$model
results$table
results$estimated_means
results$plot
results$report

Show the model

results$model

## 
## Call:
## lm(formula = ppol_extreme ~ ppol_cat * gender, data = assessmentv7)
## 
## Coefficients:
##                        (Intercept)               ppol_catConservatives  
##                            1.74286                             0.05360  
##                       genderFemale  ppol_catConservatives:genderFemale  
##                            0.06101                            -0.17747

Show a regression table

results$table

## 
## ==============================================
##                                     Model 1   
## ----------------------------------------------
## (Intercept)                           1.74 ***
##                                      (0.05)   
## ppol_catConservatives                 0.05    
##                                      (0.08)   
## genderFemale                          0.06    
##                                      (0.06)   
## ppol_catConservatives:genderFemale   -0.18    
##                                      (0.11)   
## ----------------------------------------------
## R^2                                   0.00    
## Adj. R^2                              0.00    
## Num. obs.                           750       
## RMSE                                  0.69    
## ==============================================
## *** p < 0.001, ** p < 0.01, * p < 0.05

Show an interpretation of the model

results$report %>% 
  cat()

We fitted a linear model (estimated using OLS) to predict ppol_extreme with ppol_cat and gender (formula = ppol_extreme \~ ppol_cat * gender). Standardized parameters were obtained by fitting the model on a standardized version of the dataset. Effect sizes were labelled following Funder’s (2019) recommendations.

The model explains a not significant and very weak proportion of variance (R2 = 0.00, F(3, 746) = 1.01, p = 0.388, adj. R2 = 0.00). The model’s intercept, corresponding to ppol_extreme = 0, ppol_cat = Progressives and gender = Male, is at 1.74 (SE = 0.05, 95% CI [1.64, 1.84], p \< .001). Within this model:

• The effect of ppol_catConservatives is positive and can be considered as tiny and not significant (beta = 0.05, SE = 0.08, 95% CI [-0.11, 0.22], std. beta = 0.08, p = 0.518).
• The effect of genderFemale is positive and can be considered as tiny and not significant (beta = 0.06, SE = 0.06, 95% CI [-0.06, 0.19], std. beta = 0.09, p = 0.335).
• The effect of ppol_catConservatives:genderFemale is negative and can be considered as small and not significant (beta = -0.18, SE = 0.11, 95% CI [-0.40, 0.05], std. beta = -0.26, p = 0.119).

Show estimated means

results$estimated_means %>% 
  knitr::kable()

| gender | ppol_cat | Mean | SE | CI_low | CI_high | | :----- | :------------ | -------: | --------: | -------: | -------: | | Male | Progressives | 1.742857 | 0.0519687 | 1.640835 | 1.844879 | | Male | Conservatives | 1.796460 | 0.0646727 | 1.669498 | 1.923422 | | Female | Progressives | 1.803867 | 0.0361332 | 1.732933 | 1.874802 | | Female | Conservatives | 1.680000 | 0.0687481 | 1.545037 | 1.814963 |

Show a plot of the means

results$plot  +
  ylab("Political Extremity")

Interaction with Numeric variables

When interacting a numeric with a factor variable, the function will return the mean as well as one SD below/above the numeric response.

mod1 <- lm(ppol_extreme ~ ppol_cat*AffPol1Pre, 
           data = assessmentv7)


results <- assessmentv7 %>% 
  om_lm(lin_mod = mod1, 
        type = "int",
        switch = F)

results$plot +
  ylab("Political Extremity")

When both interaction variables are numeric, the function will return the mean as well as one SD below/above for both numeric responsse.

mod1 <- lm(ppol_extreme ~ GMPre*AffPol1Pre, 
           data = assessmentv7)


results <- assessmentv7 %>% 
  om_lm(lin_mod = mod1, 
        type = "int",
        switch = T)

results$plot +
  ylab("Political Extremity")

om_lmer

Run mixed models (with interactions)

This function performs a mixed model and gives back some neat info

Three inputs:

• mod a fitted model

• type what kind of model (currently only accepts "int" for interactions)

• mod_transform Indicates which values of the moderator variable should be used when plotting interaction terms (i.e. type = "int").

  • "minmax" (default) minimum and maximum values (lower and upper bounds) of the moderator are used to plot the interaction between independent variable and moderator(s).}
  • "meansd" uses the mean value of the moderator as well as one standard deviation below and above mean value to plot the effect of the moderator on the independent variable (following the convention suggested by Cohen and Cohen and popularized by Aiken and West (1991), i.e. using the mean, the value one standard deviation above, and the value one standard deviation below the mean as values of the moderator.
  • "zeromax" is similar to the "minmax" option, however, 0 is always used as minimum value for the moderator. This may be useful for predictors that don’t have an empirical zero-value, but absence of moderation should be simulated by using 0 as minimum. "quart" calculates and uses the quartiles (lower, median and upper) of the moderator value.
  • "all" uses all values of the moderator variable.

First you need to fit to bring the data into long format and fit the model.

cleaned_dat_long <- openmindR::om_gather(assessmentv7, which_strings = "AffPol1")

mixed <- lme4::lmer(Response~Time * gender + (1|OMID), data = cleaned_dat_long)

Next you can use om_lmer.

results <- om_lmer(mixed,
                   type = "int")

Show the model

results$model

## Linear mixed model fit by REML ['lmerMod']
## Formula: Response ~ Time * gender + (1 | OMID)
##    Data: cleaned_dat_long
## REML criterion at convergence: 8643.785
## Random effects:
##  Groups   Name        Std.Dev.
##  OMID     (Intercept) 24.79   
##  Residual             13.57   
## Number of obs: 945, groups:  OMID, 586
## Fixed Effects:
##               (Intercept)                   TimePost  
##                    25.642                     -3.873  
##              TimeFollowUp               genderFemale  
##                   -14.608                     10.507  
##     TimePost:genderFemale  TimeFollowUp:genderFemale  
##                    -6.358                     11.857

Show a regression table

results$table

## 
## =======================================
##                            Model 1     
## ---------------------------------------
## (Intercept)                   25.64 ***
##                               (1.83)   
## TimePost                      -3.87 *  
##                               (1.66)   
## TimeFollowUp                 -14.61    
##                              (16.27)   
## genderFemale                  10.51 ***
##                               (2.38)   
## TimePost:genderFemale         -6.36 ** 
##                               (2.07)   
## TimeFollowUp:genderFemale     11.86    
##                              (19.93)   
## ---------------------------------------
## AIC                         8659.79    
## BIC                         8698.59    
## Log Likelihood             -4321.89    
## Num. obs.                    945       
## Num. groups: OMID            586       
## Var: OMID (Intercept)        614.79    
## Var: Residual                184.03    
## =======================================
## *** p < 0.001, ** p < 0.01, * p < 0.05

Show an interpretation of the model

results$report %>% 
  cat()

We fitted a linear mixed model (estimated using REML and nloptwrap optimizer) to predict Response with Time and gender (formula = Response \~ Time * gender). The model included OMID as random effects (formula = \~1 | OMID). Standardized parameters were obtained by fitting the model on a standardized version of the dataset. Effect sizes were labelled following Funder’s (2019) recommendations.The model’s total explanatory power is substantial (conditional R2 = 0.78) and the part related to the fixed effects alone (marginal R2) is of 0.04. The model’s intercept, corresponding to Response = 0, Time = Pre, gender = Male and OMID = 1002059522913, is at 25.64 (SE = 1.83, 95% CI [22.05, 29.23], p \< .001). Within this model:

• The effect of TimePost is negative and can be considered as very small and significant (beta = -3.87, SE = 1.66, 95% CI [-7.12, -0.63], std. beta = -0.13, p \< .05).
• The effect of TimeFollowUp is negative and can be considered as medium and not significant (beta = -14.61, SE = 16.27, 95% CI [-46.50, 17.28], std. beta = -0.51, p = 0.369).
• The effect of genderFemale is positive and can be considered as small and significant (beta = 10.51, SE = 2.38, 95% CI [5.85, 15.17], std. beta = 0.36, p \< .001).
• The effect of TimePost:genderFemale is negative and can be considered as small and significant (beta = -6.36, SE = 2.07, 95% CI [-10.42, -2.30], std. beta = -0.22, p \< .01).
• The effect of TimeFollowUp:genderFemale is positive and can be considered as medium and not significant (beta = 11.86, SE = 19.93, 95% CI [-27.20, 50.92], std. beta = 0.41, p = 0.552).

Show estimated means

results$estimated_means %>% 
  knitr::kable()

| Time | predicted | std.error | conf.low | conf.high | group | group_col | | ---: | --------: | --------: | ---------: | --------: | :----- | :--------- | | 1 | 25.64231 | 1.831546 | 22.05254 | 29.23207 | Male | Male | | 1 | 36.14943 | 1.515080 | 33.17992 | 39.11893 | Female | Female | | 2 | 21.76919 | 2.133368 | 17.58786 | 25.95051 | Male | Male | | 2 | 25.91874 | 1.669731 | 22.64613 | 29.19136 | Female | Female | | 3 | 11.03439 | 16.326028 | -20.96403 | 43.03282 | Male | Male | | 3 | 33.39802 | 11.561290 | 10.73831 | 56.05773 | Female | Female |

Show a plot of the means

results$plot  +
  ggplot2::ylab("Affective Polarization") +
  ggplot2::ylim(0, 50)

om_textscore

This function performs sentiment analysis and other methods from different text tools on any dataset with a text colum that you supply.

Note: openmindR will not automatically install all the packages necessary for each text tool because it is quite a lot and there are many dependencies. Instead, it will check whether the specified method is installed on your computer before it executes the code. If it’s not, it asks you to install the relevant package. That way you only need to install what you actually use.

Arguments:

• .data dataset including text and a personal identifier (must be named OMID)
• method which method(s) should be applied? Currently supports: "nrc", "bing", "afinn", "textblob", "vader", "sentiment_stanford", "sentimentr_polarity", "affect3d", "receptiveness", and "toxicity".
• text specify the text variable which should be scored

Here an example code that scores the free text from v7 with bing, nrc and afinn sentiment dictionaries.

assessmentv7 %>% 
 om_textscore(method = c("bing", "nrc", "afinn"), 
              text = IssueFreeTextPre) 

openmindR ggplot2 theme

There are three functions for the ggplot2 theme:

• theme_om
• scale_fill_om
• scale_color_om

Make sure you have the Poppins font installed!

extrafont::font_import()

windowsFonts(`Poppins` = windowsFont("Poppins"))

Good tutorial on how to install custom fonts in R

Example

## Load tidyverse
library(tidyverse)

titanic_dat <- Titanic %>% as_tibble()

titanic_dat %>% 
  ggplot(aes(Sex, n)) +
  geom_col(aes(fill = Class), position = position_dodge()) +
  theme_om(legend_position = c(0.9, 0.75)) +
  scale_fill_om("Class") +
  facet_wrap(~Survived) +
  labs(title = "Titanic Survival by Age and Class") 

Adapt theme_om

• legend_position
• axis_text_size
• axis_title_size
• legend_text_size
• title_size

titanic_dat %>% 
  ggplot(aes(Class, n, fill = Class)) +
  geom_col() +
  theme_om(legend_position = "bottom",
           axis_text_size = 10,
           axis_title_size = 15, 
           legend_text_size = 10,
           title_size = 20) +
  scale_fill_om() +
  facet_wrap(~Survived) +
  labs(title = "Titanic Survival by Class") 

Or all text sizes at once

• overall_text_size

titanic_dat %>% 
  ggplot(aes(Class, n, fill = Class)) +
  geom_col() +
  theme_om(legend_position = "top",
           overall_text_size = 15) +
  scale_fill_om() +
  facet_wrap(~Survived) +
  labs(title = "Titanic Survival by Class") 

In case your pandoc is having problems check out this very neat fix: https://github.com/rstudio/rstudio/issues/3661#issuecomment-475705806

openmindplatform/openmindR documentation built on Nov. 13, 2021, 2:13 p.m.