Rachel_CreateDescData&RegressionTables.Rmd

---
title: 'Rachel's Replication: Create Desc data and Regression tables'
author: "Rachel Ganly"
date: "9/28/2021"
output: html_document
---

Load packages

```{r}

#1. Load required packages for analyses

# Clear dataspace
rm(list = ls())

library(tidyverse)
library(haven)
library(skimr)
library(naniar)
library(Hmisc)
library(sjlabelled)
library(gt)
library(gtsummary)
# Rachel's comments: Found these packages in the Readme list on GitHub

library(vtable) #Rachel's comment: added package for descriptive stats


```

Create Table 1 (Rachel's version)
```{r}
#The sample is now ready for producing descriptive and inferential statistics. Run
# **"Table 1.R"** to create a descriptive table. This table provides descriptive statistics
# across respondents' sex.

# Load file

load("GSS_NoMissing_Rachel.csv")

descriptive_data=gss_rachel%>%select(srh, mental_srh, sex, education, age, marital_status, family_income, number_children, visible_minority)

st(descriptive_data, group="sex",title="Summary Statistics",  digits=2, summ=c('notNA(x)','mean(x)'), summ.names = c("Number of Individuals", "Mean"))


  
```

Create Table 1 (Version using Tyler's code)
```{r}
tbl_1 <- gss_rachel %>% 
  tbl_summary(include   = c(srh, 
                            mental_srh, 
                            education, 
                            family_income, 
                            age, 
                            marital_status, 
                            number_children, 
                            visible_minority),
              type = list(family_income   ~ 'continuous',
                          number_children ~ 'continuous'),
              
              by        = sex,
              
              statistic = list(all_continuous()  ~ "{mean} ({sd})",
                               all_dichotomous() ~ "0.{p}",
                               all_categorical() ~ "0.{p}"),
              
              digits    = list(all_continuous()  ~ 2,
                               all_categorical() ~ 0)) %>% 
  
  modify_header(update = list(label  ~ '**Variables**',
                              stat_1 ~ '**Male ({n})**', 
                              stat_2 ~ '**Female({n})**')) %>% 
  
  modify_footnote(update = c(stat_1, stat_2) ~ 
                  'Mean and standard deviations provided for continuous 
                  variables. Proportions provided for categorical variables') %>% 
  
  modify_caption('**TABLE 1. DESCRIPTIVE STATISTICS (N = 19755)**') %>% 
  
  bold_labels()


  # Prints Table 1 as HTML
tbl_1 %>% as_gt()

```

Create Table  (Version using Tyler's code)
```{r}

#Finally, you should run **"Table 2.R"** to create a table featuring two  multivariable regressions.

# Rachel's comment: I could not tell how to do this without looking at the code as there were no instructions or comments about which variables to use in the regression or which standard errors to use

# CREATING TABLE 2: OLS REGRESSIONS OF SRH AND SRMH
  # Linear Model of Self-Rated Health
    # y ~ x1 + x2 + x3 ... , data = )
# Creating SRH Table based on Linear Model
  # Linear Model of Self-Rated Mental Health
    # y ~ x1 + x2 + x3 ... , data = )
 # Creating SRMH Table based on Linear Model
  # Merging SRH and SRMH Tables to Create and Store Table 2: tbl_2
 # Prints Table 2 in HTML


  # Linear Model of Self-Rated Health
    # y ~ x1 + x2 + x3 ... , data = )
ols_srh <- lm(srh ~ education * sex + 
                    age +
                    marital_status +         
                    family_income +        
                    number_children +        
                    visible_minority, 
                data = gss_rachel,
                weights = individual_survey_weight)

  # Creating SRH Table based on Linear Model
model_srh <- 
tbl_regression(ols_srh) %>% 
  modify_header(update = list(label     ~ '**Variables**',
                              estimate  ~ '***b***',
                              std.error ~ '***se***',
                              p.value   ~ '**p-value**')) %>% 
  add_significance_stars()


  # Linear Model of Self-Rated Mental Health
    # y ~ x1 + x2 + x3 ... , data = )
ols_srmh <- lm(mental_srh ~ education * sex + 
                       age +
                       marital_status +         
                       family_income +        
                       number_children +        
                       visible_minority, 
                 data = gss_rachel,
                 weights = individual_survey_weight)

  # Creating SRMH Table based on Linear Model
model_srmh <- 
  tbl_regression(ols_srmh) %>% 
  modify_header(update = list(label     ~ '**Variables**',
                              estimate  ~ '***b***',
                              std.error ~ '***se***',
                              p.value   ~ '**p-value**')) %>% 
  add_significance_stars()


  # Merging SRH and SRMH Tables to Create and Store Table 2: tbl_2
tbl_2 <- 
tbl_merge(list(model_srh, model_srmh),
          tab_spanner = c('**Self-Rated Health**', 
                          '**Self-Rated Mental Health**')) %>% 
          modify_caption('**TABLE 2. SEX MODERATES THE HEALTH-BENEFITS OF 
                 POSTSECONDARY EDUCATIONAL ATTAINMENT, NET OF CONTROLS (N = 19755)**') %>% 
          modify_footnote(update = c('p.value_1', 'p.value_2') ~ 'Significant p-values in bold.')

    # Prints Table 2 in HTML
tbl_2 %>% as_gt()




```