Mental Health in Tech Workplace


In this project, our group wants to gauge the prevalence of certain mental health disorders within the tech industry, and how mental health is viewed within the tech/IT workplace. We used dataset from 2016 OSMI Mental Health in Tech Survey.


Prevalence of Mental Health

Overview

In this section, we want to analyze how different groups of people（by sex and age）perform differently in mental health conditions.
library(dplyr)
library(forcats)
library(ggplot2)
library(data.table)
library(ggpubr)
library(ggthemes)
library(tidyverse)
library(reshape2)
library(RColorBrewer)
library(readxl)
library(rgdal)
library(sp)
library(raster)
library(leaflet)
library(maps)
#library(ggmap)
library(plotly)
library(quanteda)
library(wordcloud)
library(wordcloud2)
library(readr)
library(dplyr)
library(tm)
library(tidytext)
library(patchwork)
library(SnowballC)

mental <- read.csv('mental-heath-in-tech-2016_20161114.csv', header = TRUE,encoding = "UTF-8")

Data Distribution

## data preprocessing
# gender info
temp <- mental[, grep('gender', colnames(mental))]
temp[! toupper(str_sub(temp, 1,1)) %in% c('F', 'M','W') ] <- 'Others'
temp[ toupper(str_sub(temp, 1,1)) %in% c('F', 'W') ] <- 'Female'
temp[ toupper(str_sub(temp, 1,1)) %in% c('M') ] <- 'Male'
gender <- temp
# age
age <- mental[, grep('What.is.your.age.', colnames(mental))]
breaks <- c(10*c(1:7))
age_category <-  cut(as.numeric(age), breaks,include.lowest = TRUE, right = FALSE, dig.lab=10,labels =c("10-20","20-30","30-40","40-50","50-60","60-70"))
# condition
condition <- mental[grep('condition', colnames(mental))]
#colnames(condition)
condition.diagnosed <- condition$Have.you.been.diagnosed.with.a.mental.health.condition.by.a.medical.professional.
condition.type <- condition$If.so..what.condition.s..were.you.diagnosed.with.
temp <-  str_split(condition.type, '[|]', simplify = T)

# data for plot
plotdata <- data.frame(gender = gender, age = age, age_category = as.character(age_category),
                  condition.diagnosed = condition.diagnosed,temp)
plotdata_long <- melt(plotdata, id.vars = c('gender', 'age','age_category', 'condition.diagnosed'))
#colnames(plotdata_long)

## age distribution
plotdata=plotdata[-which(plotdata$age==323),]
box_gender <- ggplot(plotdata) +
    geom_boxplot(aes(x = gender,y =age,  fill = gender, group = gender)) + 
  scale_y_continuous(breaks = seq(19, 99, 10)) + 
  scale_fill_manual(values = c("#E41A1C", "#449B75", "#AC5782")) +
    #scale_fill_manual(values=col) + 
    theme_bw()+ggtitle("Age Distribution by Sex")+xlab("Gender")+ylab("Age")+theme(plot.title = element_text(hjust = 0.5))

library(plotly) 
ggplotly(box_gender)
The age of woman mainly varies from 28 to 40, similar to that, the age of man mainly varies from 29 to 39 , both with a median age of 33.


## condition.diagnosed
library(ggpubr)
figure=ggpubr::ggarrange(
  ggplot(plotdata) +
    geom_bar(aes(x = gender, fill = condition.diagnosed), position = 'dodge') +
    scale_fill_manual(values = c("#449B75", "#E41A1C")) +
    theme_bw()+ylab("Number of People")+xlab("Gender") + 
  labs(fill="Have you been diagnosed with a mental health \n condition by a medical professional?") ,
  
  ggplot(plotdata) +
    geom_bar(aes(x = age_category,  fill = condition.diagnosed), position = 'dodge') +
    scale_fill_manual(values = c("#449B75", "#E41A1C")) +
    #scale_fill_manual(palette="Set1") +
    theme_bw()+ylab("Number of People")+xlab("Age")+ 
    theme(axis.text.x = element_text(angle = 0)), common.legend = T, legend = "bottom")

annotate_figure(figure,
               top = text_grob("Condition Diagnosed by Sex and Age",face="bold"))


For female, the number of people who have been diagnosed with some types of mental health disorders is about twice as the number of people who haven’t been diagnosed. And for male, that number tends to distribute equally. Across all age categories, the number of people who have been diagnosed with conditions is somehow more than the number of people who haven’t. In general, mental health problems are prevalent and ubiquitous within the tech industry.


Mental Disorder Type

## condition plot
col <- colorRampPalette(brewer.pal(9, 'Set1'))(6)
plotdata_long <- plotdata_long %>%
  filter(! value == '')

a=unique(plotdata_long$value);a
#  [1] "Anxiety Disorder (Generalized, Social, Phobia, etc)"                                
#  [2] "Mood Disorder (Depression, Bipolar Disorder, etc)"                                  
#  [3] "Personality Disorder (Borderline, Antisocial, Paranoid, etc)"                       
#  [4] "Attention Deficit Hyperactivity Disorder"                                           
#  [5] "Seasonal Affective Disorder"                                                        
#  [6] "Depression"                                                                         
#  [7] "Substance Use Disorder"                                                             
#  [8] "Obsessive-Compulsive Disorder"                                                      
#  [9] "Post-traumatic Stress Disorder"                                                     
# [10] "Psychotic Disorder (Schizophrenia, Schizoaffective, etc)"                           
# [11] "Eating Disorder (Anorexia, Bulimia, etc)"                                           
# [12] "Autism - while not a \"mental illness\", still greatly affects how I handle anxiety"
# [13] "Stress Response Syndromes"                                                          
# [14] "attention deficit disorder (but not the hyperactive version)"                       
# [15] "Asperger Syndrome"                                                                  
# [16] "ADD (w/o Hyperactivity)"                                                            
# [17] "Addictive Disorder"                                                                 
# [18] "Schizotypal Personality Disorder"                                                   
# [19] "PDD-NOS (see above)"                                                                
# [20] "Suicidal Ideation"                                                                  
# [21] "Attention Deficit Disorder"                                                         
# [22] "Intimate Disorder"                                                                  
# [23] "Dissociative Disorder"                                                              
# [24] "Aspergers"                                                                          
# [25] "Autism"
for(i in 1:length(a)){
  if (i != 1 & i !=2 & i!=4 & i!=8  & i!=9) {
  plotdata_long[which(plotdata_long$value==a[i]),"value"]<-"others"
    }
}

ggplot(plotdata_long) +
  geom_bar(aes(x = gender, fill = value),position = 'stack') +
  scale_fill_manual(values=col) +
  theme_bw() +
  theme(legend.title = element_blank(),
        legend.position = 'bottom',
        legend.text = element_text(size = 9))+
  guides(fill = guide_legend(ncol = 2,
                             keywidth = unit(4,'mm'),
                             keyheight = unit(4,'mm')))+ylab("Number of People")+ggtitle("Specific Mental Health Disorders Composition by Gender")+theme(plot.title = element_text(hjust = 0.5))+ xlab("Gender")

ggplot(plotdata_long) +
  geom_bar(aes(x = age_category, fill = value),position = 'stack') +
  scale_fill_manual(values=col) +
  theme_bw() +
  theme(legend.title = element_blank(),
        legend.position = 'bottom',
        legend.text = element_text(size = 9))+
  guides(fill = guide_legend(ncol = 2,
                             keywidth = unit(4,'mm'),
                             keyheight = unit(4,'mm')))+ylab("Number of People")+ggtitle("Specific Mental Health Disorders Composition by Age")+theme(plot.title = element_text(hjust = 0.5)) +xlab("Age category")

Anxiety Disorder and Mood Disorder are the two most prevalent mental health disorders across all ages for both male and female within the tech industry.


Maps

Overview

In this part, we would like to show a statistics among different states in the US. Please note that the results of states with too small sample size are not considered as reliable (e.g.South Carolina). We will not analyze those states with sample size smaller than 5.
df <- mental %>% 
  dplyr::select('Do.you.currently.have.a.mental.health.disorder.','Have.you.ever.sought.treatment.for.a.mental.health.issue.from.a.mental.health.professional.', 'What.US.state.or.territory.do.you.live.in.','Do.you.have.medical.coverage..private.insurance.or.state.provided..which.includes.treatment.of..mental.health.issues.','Does.your.employer.provide.mental.health.benefits.as.part.of.healthcare.coverage.')
names(df) <- c('disorder1','treatment','state','mental_insurance','employer_benefit1')
df <- df %>%
  mutate(disorder = case_when(
  disorder1 == 'Yes'  ~ 1,
  disorder1 == 'Maybe'  ~ 0.5,
  disorder1 == 'No'  ~ 0))
df <- df %>%
  mutate(employer_benefit = case_when(
  employer_benefit1 == 'Yes'  ~ 1,
  TRUE  ~ 0))
df <- df %>% 
  dplyr::select('state','treatment','employer_benefit','disorder')
head(df)
df1 <- df %>% 
  dplyr::select('state','treatment')
df_treatment1 <- df1 %>% group_by(state) %>% summarise(count=n())
df_treatment2 <- df1 %>% filter(treatment==1) %>% group_by(state) %>% summarise(count1=n())
df_treatment <- df_treatment1 %>% 
  left_join(df_treatment2, by = "state") %>% 
  mutate(treatment_prec = count1/count) 

#df_treatment <- df1 %>% group_by(state) %>% summarise(treatment_perc = sum('treatment'== 1))
#df1 <- group_by(df, state) %>% mutate(disorder_percent = disorder/sum(disorder))
#head(df_treatment1)
#sum(df_treatment1$count)


df2 <- df %>% 
  dplyr::select('state','employer_benefit')
df_employer_benefit1 <- df2 %>% group_by(state) %>% summarise(count=n())
df_employer_benefit2 <- df2 %>% filter(employer_benefit==1) %>% group_by(state) %>% summarise(count1=n())
df_employer_benefit <- df_employer_benefit1 %>% 
  left_join(df_employer_benefit2, by = "state") %>% 
  mutate(employer_benefit_prec = count1/count)


df_final <- merge(x = df_employer_benefit, y = df_treatment, by = "state", all = TRUE)  %>%
  dplyr::select('state','treatment_prec','employer_benefit_prec')

round_df <- function(x, digits) {
    numeric_columns <- sapply(x, mode) == 'numeric'
    x[numeric_columns] <-  round(x[numeric_columns], digits)
    x
}

df_final <- round_df(df_final, 3)

df_count <- df %>% count(state, sort = TRUE)
names(df_count) <- c('state','sample')
df_final <- merge(x = df_final, y = df_count, by = "state", all = TRUE)
# From https://www.census.gov/geo/maps-data/data/cbf/cbf_state.html

states <- shapefile("cb_2018_us_state_20m.shp")

combined <- states@data %>% 
  left_join(df_final, by = c(NAME = "state"))
states@data <- combined

Frequency of Employee Who Sought Mental Health Treatment

bins <- c(0, 0.2, 0.4, 0.6, 0.8, 1)
pal <- leaflet::colorBin("YlOrRd", domain = states$treatment_prec, bins = c(0, 0.2, 0.4, 0.6, 0.8, 1), na.color = "#C0C0C0")
labels_states <- paste( states$NAME,":",
                    "Frequency=", states$treatment_prec,",","Sample=",states$sample)

m <- leaflet() %>%
  leaflet::setView(-96, 37.8, 4) %>%
  leaflet::addProviderTiles("MapBox", options = providerTileOptions(
    id = "mapbox.light",
    accessToken = Sys.getenv('MAPBOX_ACCESS_TOKEN'))) %>% 
  leaflet::addPolygons(data=states,
    fillColor = ~pal(treatment_prec),
    weight = 2,
    opacity = 1,
    color = "white",
    dashArray = "3",
    fillOpacity = 0.7,
    highlight = highlightOptions(
      weight = 5,
      color = "white",
      dashArray = "",
      fillOpacity = 0.7,
      bringToFront = TRUE),
    label = labels_states,
    labelOptions = labelOptions(
      style = list("font-weight" = "normal", padding = "3px 8px"),
      textsize = "15px",
      direction = "auto")) %>%
  leaflet::addLegend(data=states, pal = pal, title = "Frequency of Employee Who Sought Treatment", 
                     values = ~treatment_prec, opacity = 0.7,position = "bottomright")


m
Frequency of employee who sought mental health treatment: Iowa, North Carolina, New Jersey and Connecticut have the highest scores.


Frequency of Employer with Mental Health Benefit

bins <- c(0, 0.2, 0.4, 0.6, 0.8, 1)
pal <- colorBin("YlOrRd", domain = states$employer_benefit_prec, bins = bins,na.color = "#C0C0C0")
labels_states <- paste(states$NAME,":",
                    "Frequency=", states$employer_benefit_prec,",","Sample=",states$sample)
                   

n <- leaflet() %>%
  setView(-96, 37.8, 4) %>%
  addProviderTiles("MapBox", options = providerTileOptions(
    id = "mapbox.light",
    accessToken = Sys.getenv('MAPBOX_ACCESS_TOKEN'))) %>% 
  addPolygons(data=states,
    fillColor = ~pal(states$employer_benefit_prec),
    weight = 2,
    opacity = 1,
    color = "white",
    dashArray = "3",
    fillOpacity = 0.7,
    highlight = highlightOptions(
      weight = 5,
      color = "white",
      dashArray = "",
      fillOpacity = 0.7,
      bringToFront = TRUE),
    label = labels_states,
    labelOptions = labelOptions(
      style = list("font-weight" = "normal", padding = "3px 8px"),
      textsize = "15px",
      direction = "auto")) %>%
  addLegend(data=states, pal = pal, title = "Frequency of Employer with Mental Healthcare Benefit", values = ~employer_benefit_prec, opacity = 0.7,
    position = "bottomright")


n
Frequency of employer with mental healthcare benefit: Iowa is among the highest, while New Jersey is among the lowest.


Mental Health and Company Size

Overview

Following our analysis of employees who sought mental treatment and had mental healthcare benefit, we would like to continue with relationship between company size and mental issues in the workplace.


Reactions by Gender and Company Size

We will mainly focus on the three questions below and see how gender and company size would differentiate the answer.
Q1: Would you bring up a mental health issue with a potential employer in an interview?
Q2: Do you feel that being identified as a person with a mental health issue would hurt your career?
Q3: Do you think that team members/co-workers would view you more negatively if they knew you suffered from a mental health issue?
mental1 <- mental %>% filter(gender == 'Male')%>%
          dplyr::group_by(company_size,gender, bringup_issue) %>%
          dplyr::summarise(frequency=n())%>%
          dplyr::mutate(frequency_ratio = frequency / sum(frequency)*100)

mental2 <- mental %>% filter(gender == 'Female')%>%
          dplyr::group_by(company_size,gender, bringup_issue) %>%
          dplyr::summarise(frequency=2.32*n())%>%#times male/female ratio to make sure male and female are within same sample size
          dplyr::mutate(frequency_ratio = frequency / sum(frequency)*100)

mental3 <- rbind(mental1, mental2)
fig<- ggballoonplot(mental3, y = "company_size", x = "bringup_issue", size="frequency",fill = "grey",
             facet.by = "gender", ggtheme = theme_bw())+
  scale_fill_viridis_c(option = "D",direction = -1)
fig <-  ggpar(fig,main = "Will you bring up mental issue during interviews?", submain ="Faceting by company size and gender", xlab = "Answer", ylab = "Company size")
ggplotly(fig)
From the gender perspective, it is quite uncommon for female respondents to bring up mental issues during interviews, while there are more male respondents who bring them up. From the company size perspective , employees from all company sizes tend to reject mentioning mental issues during interviews. Companies with 26-500 employees are more unlikely to receive mental issues during the interview according to the answer distribution.


mental$mental_hurt_career[mental$mental_hurt_career == "No, it has not" | mental$mental_hurt_career == "No, I don't think it would"] <- "No"
mental$mental_hurt_career[mental$mental_hurt_career == "Yes, I think it would" | mental$mental_hurt_career == "Yes, it has"] <- "Yes"
mental4 <- mental %>% filter(gender == 'Male')%>%
          dplyr::group_by(company_size,gender, mental_hurt_career) %>%
          dplyr::summarise(frequency=n())%>%
          dplyr::mutate(frequency_ratio = frequency / sum(frequency)*100)

mental5 <- mental %>% filter(gender == 'Female')%>%
          dplyr::group_by(company_size,gender, mental_hurt_career) %>%
          dplyr::summarise(frequency=2.32*n())%>%#times male/female ratio to make sure male and female are within same sample size
          dplyr::mutate(frequency_ratio = frequency / sum(frequency)*100)

mental6 <- rbind(mental4, mental5)
fig2<- ggballoonplot(mental6, y = "company_size", x = "mental_hurt_career", fill = "grey", size="frequency",
             facet.by = "gender", ggtheme = theme_bw())+
  scale_fill_viridis_c(option = "D",direction = -1)
fig2 <-  ggpar(fig2,main = "Will identification of mental health issues hurt your career?", submain ="Faceting by company size and gender", xlab = "Answer", ylab = "Company size")
ggplotly(fig2)
From the gender perspective, female respondents tend to admit mental issues would hurt their careers, while certain males are more likely to consider mental issues would not do harm to careers. From the company size perspective , employees from all company sizes tend to connect mental issues with harm to careers. Especially, respondents from companies with 26-500 or more than 1000 employees are more likely to agree with this statement. In addition, responding with ambiguous answers is also common for respondents from all levels of companies.


mental$coworker_views[mental$coworker_views == "No, they do not" | mental$coworker_views == "No, I don't think they would"] <- "No"
mental$coworker_views[mental$coworker_views == "Yes, I think they would" | mental$coworker_views == "Yes, they do"] <- "Yes"

mental7 <- mental %>% filter(gender == 'Male')%>%
          dplyr::group_by(company_size,gender, coworker_views) %>%
          dplyr::summarise(frequency=n())%>%
          dplyr::mutate(frequency_ratio = frequency / sum(frequency)*100)

mental8 <- mental %>% filter(gender == 'Female')%>%
          dplyr::group_by(company_size,gender, coworker_views) %>%
          dplyr::summarise(frequency=2.32*n())%>%#times male/female ratio to make sure male and female are within same sample size
          dplyr::mutate(frequency_ratio = frequency / sum(frequency)*100)

mental9 <- rbind(mental7, mental8)
fig3<- ggballoonplot(mental9, y = "company_size", x = "coworker_views", fill = "grey", size="frequency",
             facet.by = "gender", ggtheme = theme_bw())+
  scale_fill_viridis_c(option = "D",direction = -1)
fig3 <-  ggpar(fig3,main = "Will coworkers view you more negatively?", submain ="Faceting by company size and gender", xlab = "Answer", ylab = "Company size")
ggplotly(fig3)
We can see male and female respondents tend to distribute equally in this question. Respondents from companies with 26-100 or more than 1000 employees are more likely to agree that coworkers will view them more negatively.


Text Analysis

In the following part, we want to further explore the relationship of people’s attitudes towards mental health in workplace and company size. We will do text analysis using participants’ answer for would you bring up a mental health issue with a potential employer in an interview, why and why not. Besides, we plan to divide company size into three categories: small(n<=100), medium-sized(100<n<500), large(n>=500).


Wordcloud for Each Answer

We plan to draw word clouds for each answer and differnt company sizes are distinguished by color: green indicates small firm, orange represents medium-scale enterprise and purple presents large company.


Wordcloud for Answer=Maybe
#answer = maybe
#create corpus
mental_health_maybe <- filter(mental_health, bringup_issue == "Maybe")
mental_health_maybe$doc_id <- as.character(c(1:nrow(mental_health_maybe)))
mental_health_maybe <- mental_health_maybe[, c(67,38,65,64)]
maybe_source <- DataframeSource(mental_health_maybe) 
maybe_corpus <- VCorpus(maybe_source)
#clean corpus
clean_corpus <- function(corpus){
  corpus <- tm_map(corpus, removePunctuation)
  corpus <- tm_map(corpus, content_transformer(tolower))
  corpus <- tm_map(corpus, removeWords, c("mental", "health", "interview", "feel", "bring", "want", "made", "get", "employer", "hire", "need", "know", "may", "affect", "job", "might", "reason", "answer", "think",  stopwords("en")))
  corpus <- tm_map(corpus, removeNumbers)
  corpus <- tm_map(corpus, stripWhitespace)
  return(corpus)
}
maybe_clean <- clean_corpus(maybe_corpus)
#stem and stem completion
stemCompletion2 <- function(x, dictionary) {
   x <- unlist(strsplit(as.character(x), " "))
   x <- x[x != ""]
   x <- stemCompletion(x, dictionary=dictionary)
   x <- paste(x, sep="", collapse=" ")
   stripWhitespace(x)
}
maybe_stemmed <- tm_map(maybe_clean, stemDocument)
maybe_compl <- lapply(maybe_stemmed, stemCompletion2, dictionary = maybe_clean) %>% VectorSource() %>% Corpus()
#word cloud
maybe_tdm <- TermDocumentMatrix(maybe_compl)
maybe_tf_idf <- tidy(maybe_tdm)  %>% left_join(.,mental_health_maybe[,c(1,3,4)],by=c("document"="doc_id")) %>% 
  bind_tf_idf(term, document, count) %>%
  arrange(desc(tf_idf))
set.seed(1000)
wordcloud2(maybe_tf_idf[, c(1,6)], color = maybe_tf_idf$color_size, shape = "diamond")


Wordcloud for Answer=Yes
#answer = yes
#create corpus
mental_health_yes <- filter(mental_health, bringup_issue == "Yes")
mental_health_yes$doc_id <- as.character(c(1:nrow(mental_health_yes)))
mental_health_yes <- mental_health_yes[, c(67,38,65,64)]
yes_source <- DataframeSource(mental_health_yes) 
yes_corpus <- VCorpus(yes_source)
#clean corpus
yes_clean <- clean_corpus(yes_corpus)
#stem and stem completion
yes_stemmed <- tm_map(yes_clean, stemDocument)
yes_compl <- lapply(yes_stemmed, stemCompletion2, dictionary = yes_clean) %>% VectorSource() %>% Corpus()
#word cloud
yes_tdm <- TermDocumentMatrix(yes_compl)
yes_tf_idf <- tidy(yes_tdm)  %>% left_join(.,mental_health_yes[,c(1,3,4)],by=c("document"="doc_id")) %>% 
  bind_tf_idf(term, document, count) %>%
  arrange(desc(tf_idf))
set.seed(1000)
wordcloud2(yes_tf_idf[, c(1,6)], color = yes_tf_idf$color_size, shape = "diamond")


Wordcloud for Answer=No
#answer = no
#create corpus
mental_health_no <- filter(mental_health, bringup_issue == "No")
mental_health_no$doc_id <- as.character(c(1:nrow(mental_health_no)))
mental_health_no <- mental_health_no[, c(67,38,65,64)]
no_source <- DataframeSource(mental_health_no) 
no_corpus <- VCorpus(no_source)
#clean corpus
no_clean <- clean_corpus(no_corpus)
#stem and stem completion
no_stemmed <- tm_map(no_clean, stemDocument)
no_compl <- lapply(no_stemmed, stemCompletion2, dictionary = no_clean) %>% VectorSource() %>% Corpus()
#word cloud
no_tdm <- TermDocumentMatrix(no_compl)
no_tf_idf <- tidy(no_tdm)  %>% left_join(.,mental_health_no[,c(1,3,4)],by=c("document"="doc_id")) %>% 
  bind_tf_idf(term, document, count) %>%
  arrange(desc(tf_idf))
set.seed(1000)
wordcloud2(no_tf_idf[, c(1,6)], color = no_tf_idf$color_size, shape = "diamond")


Word Frequency for Each Answer

We visualize the top 5 most frequent words for each answer and analyze how the words differ by company size.
#top 10 words in maybe
b1 <- filter(maybe_tf_idf, !(term %in% c("issue","issues","work","make","still"))) %>% group_by(`company size`, term) %>% 
  summarize("frequency"=sum(count)) %>% slice_max(frequency, n=5) %>% 
  ggplot(aes(reorder(term, frequency), frequency, fill=`company size`)) + 
  geom_bar(stat = "identity") + coord_flip() + 
  facet_wrap(~`company size`, scales="free_x") +
  scale_fill_brewer(palette = "Set2", direction=-1) +
  guides(fill=FALSE) +
  ggtitle("Answer = Maybe") +
  theme(axis.title.y = element_blank(),
        axis.title.x = element_blank(),
        panel.background = element_blank(), 
        plot.title = element_text(face = "bold", color = "black", size = 10))
b1

Unsure words like depends, discuss and potential are usually used by participants from different sized companies to show their attitudes, which is consistent with their answer. They also use stigma to express their worries about the issue. Besides, unique words like business from large companies, accomondation from medium_sized ones and wouldn’t as well as don’t from small firms also present different perspectives from respondents in different sized companies.
#top 10 words in yes
b2 <- yes_tf_idf %>% filter(!(term %in% c("issue","issues","work","answer","reason","hearing","generally","felt","bringing","allow","ahead","time","theyre","see","part","let","talk","point","fact","ensure"))) %>% group_by(`company size`, term) %>% 
  summarize("frequency"=sum(count)) %>% slice_max(frequency, n=4) %>% 
  ggplot(aes(reorder(term, frequency), frequency, fill=`company size`)) + 
  geom_bar(stat = "identity") + coord_flip() + 
  facet_wrap(~`company size`, scales="free_x") +
  scale_fill_brewer(palette = "Set2", direction=-1) +
  guides(fill=FALSE) +
  ggtitle("Answer = Yes") +
  theme(axis.title.y = element_blank(),
        axis.title.x = element_blank(),
        panel.background = element_blank(), 
        plot.title = element_text(face = "bold", color = "black", size = 10))
b2

Those who answer yes tend to use terms like impact, important and good to show the importance of mentioning mental health issue. In addition, words like performance from small companies indicate that they focus on work performance and may not mind such issues. People from medium-sacle firms use a lot negative words like unfortunate and problematic although they answer yes.
#top 10 words in no
b3 <- filter(no_tf_idf, !(term %in% c("issue","issues","work"))) %>% group_by(`company size`, term) %>% 
  summarize("frequency"=sum(count)) %>% slice_max(frequency, n=5) %>% 
  ggplot(aes(reorder(term, frequency), frequency, fill=`company size`)) + 
  geom_bar(stat = "identity") + coord_flip() + 
  facet_wrap(~`company size`, scales="free_x") +
  scale_fill_brewer(palette = "Set2", direction=-1) +
  guides(fill=FALSE) +
  ggtitle("Answer = No") +
  theme(axis.title.y = element_blank(),
        axis.title.x = element_blank(),
        panel.background = element_blank(), 
        plot.title = element_text(face = "bold", color = "black", size = 10))
b3

Participants who answered no tend to use negative words like stigma, don’t and wouldn’t, quite similar to words from those who answered maybe, to show their negative attitudes. Besides, respondents from medium-scale enterprises frequently use discrimination to explain why they wouldn’t talk about mental health issue in interviews.


Sentiment Analysis for Each Answer

#import Hu & Liu Dictionary
pos <- read.table("D:/QMSS 2021 SPRING/DV/course_content4/Lectures/Week09/data/dictionaries/positive-words.txt", as.is=T)
neg <- read.table("D:/QMSS 2021 SPRING/DV/course_content4/Lectures/Week09/data/dictionaries/negative-words.txt", as.is=T)
#define sentiment fuction
sentiment <- function(words){
  tok <- quanteda::tokens(words)
  pos.count <- sum(tok[[1]]%in%pos[,1])
  neg.count <- sum(tok[[1]]%in%neg[,1])
  out <- (pos.count - neg.count)/(pos.count+neg.count)
  return(out)
}
#calculate the sentiment
mental_health$sentiment <- sapply(mental_health$text, sentiment)

#plot the relationship between sentiment and answer
p1 <- ggplot(mental_health, aes(x = bringup_issue, y = sentiment)) + 
  geom_boxplot(aes(fill=`company size`))  + 
  stat_summary(mapping=aes(group=bringup_issue),fun="mean",geom="point",shape=23,size=3,fill="white") + 
  facet_wrap(~`company size`) + 
  scale_fill_manual(values = c("#8DA0CB","#FC8D62","#66C2A5")) +
  labs(title = "Distribution of Sentiment Score by Answer and Company Size", y = "Sentiment Score") + 
  theme(legend.position = 'none', 
        axis.title.x = element_blank(),
        axis.title.y = element_text(vjust = 2),
        panel.background = element_blank(), 
        panel.grid.major = element_line(color = "gray50", size = 0.5),
        panel.grid.major.x = element_blank(),
        plot.title = element_text(face = "bold", color = "black", size = 12))
ggplotly(p1, tooltip = "sentiment")
In sum, respondents who answered maybe and no show a negative sentiment towards the question, which is correspond to the similarity of word frequency for the two answers, while those answering yes present a positive sentiment, except for participants from small firms. There is a clear pattern that the larger company a respondent comes from, the more positive he/she will be on whether to bring up a mental health issue in an interview.


Mental Health and Job Type

Overview

In this section, we would like to explore the relationship between job type and mental issues in the workplace.


Reactions by Job Type

For the plot below, we can see the answers fromt the waffle chart to the question “whether mental health issues would hurt your careers” faceting by the most common job types in the survey, including back-end engineers, front-end engineers, Team Lead, DevOps and Dev Evangelist. Also, we displayed the waffle chart faceting by people who think mental health issue would/maybe/would not hurt their careers to visualize from another perspective.
library(stringr)  
library(waffle)  
library(viridis)  
setnames(mental,"Which.of.the.following.best.describes.your.work.position.","work_position_general")  
  
work_position <- mental$work_position_general  
work_position <- word(work_position,1,sep = "\\|")  
work_position <- as.data.frame(work_position)  
mental <- cbind(mental, work_position)  
arrange(plyr::count(mental, 'work_position'),desc(freq)) %>% filter(work_position!="Other") %>% top_n(6) 
mental$mental_hurt_career <- as.character(mental$mental_hurt_career)

hurt_career <- as.data.frame(table(mental$mental_hurt_career)) 
names(hurt_career) <- c("Do you feel that a mental health issue would hurt your career?", "Frequency")
hurt_career
#plyr::count(mental$mental_hurt_career)

#arrange by job types 
mental_wp_back <- mental %>% dplyr::filter(work_position== "Back-end Developer") %>% dplyr::select(mental_hurt_career, work_position) 
back <- table(mental_wp_back$mental_hurt_career) 
back_w <- waffle(back, rows = 12) + 
  labs(title = "Back-end Developer") + 
  theme(plot.title = element_text(color = "black", size = 10))+ 
  gradient_color("RdYlBu")

mental_wp_front <- mental%>% dplyr::filter(work_position== "Front-end Developer") %>% dplyr::select(mental_hurt_career, work_position)                 
front <- table(mental_wp_front$mental_hurt_career) 
front_w <- waffle(front, rows = 9)+ 
  labs(title = "Front-end Developer") + 
  theme(plot.title = element_text(color = "black", size = 10))+ 
  gradient_color("RdYlBu")


mental_wp_lead <- mental %>% dplyr::filter(work_position== "Supervisor/Team Lead")%>% dplyr::select(mental_hurt_career, work_position) 
lead <- table(mental_wp_lead$mental_hurt_career) 
lead_w <- waffle(lead, rows = 9)+ 
  labs(title = "Supervisor/Team Lead") + 
  theme(plot.title = element_text(color = "black", size = 10))+ 
  gradient_color("RdYlBu")
 

mental_wp_devops <- mental %>% dplyr::filter(work_position== "DevOps/SysAdmin")%>% dplyr::select(mental_hurt_career, work_position) 
devops<- table(mental_wp_devops$mental_hurt_career) 
devops_w <- waffle(devops, rows = 9)+ 
  labs(title = "DevOps/SysAdmin") + 
  theme(plot.title = element_text(color = "black", size = 10))+ 
  gradient_color("RdYlBu")
 
mental_wp_advocate <- mental %>% dplyr::filter(work_position== "Dev Evangelist/Advocate")%>% dplyr::select(mental_hurt_career, work_position) 
advocate<- table(mental_wp_advocate$mental_hurt_career) 
advocate_w <- waffle(advocate, rows = 6)+ 
  labs(title = "Dev Evangelist/Advocate") + 
  theme(plot.title = element_text(color = "black", size = 10))+ 
  gradient_color("RdYlBu")


mental_wp_support <- mental %>% dplyr::filter(work_position== "Support")%>% dplyr::select(mental_hurt_career, work_position) 
support<- table(mental_wp_support$mental_hurt_career) 
support_w <- waffle(support, rows = 6)+ 
  labs(title = "Support") + 
  theme(plot.title = element_text(color = "black", size = 10))+ 
  gradient_color("RdYlBu")

back_w + lead_w+ front_w + devops_w + advocate_w + support_w+ plot_annotation(title = 'Will identification of mental health issue would hurt your career?', subtitle = 'Faceting by job types') + plot_layout(nrow =3, byrow = FALSE)  

#arrange answers


top6 <-mental %>% dplyr::filter(work_position!="Other") %>%
                dplyr::select(mental_hurt_career, work_position) %>% 
                dplyr::filter(work_position == "Back-end Developer" | work_position == "Front-end Developer" | work_position == "Support"| work_position == "Supervisor/Team Lead"| work_position == "DevOps/SysAdmin"| work_position == "Dev Evangelist/Advocate")

mental_maybe <- top6 %>% dplyr::filter(mental_hurt_career== "Maybe") 
maybe <- table(mental_maybe$work_position) 
maybe_w <- waffle(maybe, rows = 16, legend_pos="bottom") + 
  labs(title = "") + 
  theme(plot.title = element_text(color = "black", size = 10))+ 
  gradient_color("RdYlBu")+ 
  labs(title = "Answer: Maybe") + 
  theme(plot.title = element_text(color = "black", size = 10))


mental_yes <- top6 %>% dplyr::filter(mental_hurt_career== "Yes") 
yes <- table(mental_yes$work_position) 
yes_w <- waffle(yes, rows = 16, legend_pos="bottom") + 
  labs(title = "") + 
  theme(plot.title = element_text(color = "black", size = 10))+ 
  gradient_color("RdYlBu")+ 
  labs(title = "Answer: Yes") + 
  theme(plot.title = element_text(color = "black", size = 10))


mental_no <- top6 %>% dplyr::filter(mental_hurt_career== "No") 
no <- table(mental_no$work_position) 
no_w <- waffle(no, rows = 8, legend_pos="bottom") + 
  labs(title = "") + 
  theme(plot.title = element_text(color = "black", size = 10))+ 
  gradient_color("RdYlBu")+ 
  labs(title = "Answer: No") + 
  theme(plot.title = element_text(color = "black", size = 10))

yes_w + no_w + maybe_w +plot_annotation(title='Will identification of mental health issue would hurt your career?', subtitle="Faceting by people who think mental health issue would/maybe/would not hurt their careers")+ plot_layout(ncol = 3, nrow=1, byrow = FALSE,guides = "collect") & theme(legend.position='bottom')

We can see most of the respondents from different job types agree with the statement that mental health issues would hurt their careers, despite the fact that around one-third of them answer “maybe”.


Text Analysis

Again, we want to further explore how people from different jobs view mental health issues in workplace. We will do text analysis using participants’ answer for would you bring up a mental health issue with a potential employer in an interview, why and why not. Besides, we divide job types into three categories: Developer, Supporter and Other.


Wordcloud for Each Answer

We plan to draw word clouds for each answer and differnt job types are distinguished by color: green indicates supporter, orange represents other and purple presents developer.


Wordcloud for Answer=Maybe
#answer = maybe
#create corpus
mental_health_maybe_job <- filter(mental_health, bringup_issue == "Maybe")
mental_health_maybe_job$doc_id <- as.character(c(1:nrow(mental_health_maybe_job)))
mental_health_maybe_job <- mental_health_maybe_job[, c(68,38,66,63)]
maybe_job_corpus <- DataframeSource(mental_health_maybe_job) %>% VCorpus(.)
#clean 
maybe_job_clean <- clean_corpus(maybe_job_corpus)
maybe_job_stemmed <- tm_map(maybe_job_clean, stemDocument)
maybe_job_compl <- lapply(maybe_job_stemmed, stemCompletion2, dictionary = maybe_job_clean) %>% VectorSource() %>% Corpus()
#word cloud
maybe_job_tdm <- TermDocumentMatrix(maybe_job_compl)
maybe_job_tf_idf <- tidy(maybe_job_tdm)  %>% left_join(.,mental_health_maybe_job[,c(1,3,4)],by=c("document"="doc_id")) %>% 
  bind_tf_idf(term, document, count) %>%
  arrange(desc(tf_idf))
set.seed(1000)
wordcloud2(maybe_job_tf_idf[, c(1,6)], color = maybe_job_tf_idf$color_job, shape = "diamond")


Wordcloud for Answer=Yes
#answer = yes
#create corpus
mental_health_yes_job <- filter(mental_health, bringup_issue == "Yes")
mental_health_yes_job$doc_id <- as.character(c(1:nrow(mental_health_yes_job)))
mental_health_yes_job <- mental_health_yes_job[, c(68,38,66,63)]
yes_job_corpus <- DataframeSource(mental_health_yes_job) %>% VCorpus(.)
#clean 
yes_job_clean <- clean_corpus(yes_job_corpus)
yes_job_stemmed <- tm_map(yes_job_clean, stemDocument)
yes_job_compl <- lapply(yes_job_stemmed, stemCompletion2, dictionary = yes_job_clean) %>% VectorSource() %>% Corpus()
#word cloud
yes_job_tdm <- TermDocumentMatrix(yes_job_compl)
yes_job_tf_idf <- tidy(yes_job_tdm)  %>% left_join(.,mental_health_yes_job[,c(1,3,4)],by=c("document"="doc_id")) %>% 
  bind_tf_idf(term, document, count) %>%
  arrange(desc(tf_idf))
set.seed(1000)
wordcloud2(yes_job_tf_idf[, c(1,6)], color = yes_job_tf_idf$color_job, shape = "diamond")


Wordcloud for Answer=No
#answer = no
#create corpus
mental_health_no_job <- filter(mental_health, bringup_issue == "No")
mental_health_no_job$doc_id <- as.character(c(1:nrow(mental_health_no_job)))
mental_health_no_job <- mental_health_no_job[, c(68,38,66,63)]
no_job_corpus <- DataframeSource(mental_health_no_job) %>% VCorpus(.)
#clean 
no_job_clean <- clean_corpus(no_job_corpus)
no_job_stemmed <- tm_map(no_job_clean, stemDocument)
no_job_compl <- lapply(no_job_stemmed, stemCompletion2, dictionary = no_job_clean) %>% VectorSource() %>% Corpus()
#word cloud
no_job_tdm <- TermDocumentMatrix(no_job_compl)
no_job_tf_idf <- tidy(no_job_tdm)  %>% left_join(.,mental_health_no_job[,c(1,3,4)],by=c("document"="doc_id")) %>% 
  bind_tf_idf(term, document, count) %>%
  arrange(desc(tf_idf))
set.seed(1000)
wordcloud2(no_job_tf_idf[, c(1,6)], color = no_job_tf_idf$color_job, shape = "diamond")


Word Frequency for Each Answer

We visualize the top 5 most frequent words for each answer and analyze how the words differ by job type.
#top 10 words in maybe
df <- filter(maybe_job_tf_idf, !(term %in% c("someone","issues","work","etc","just","felt"))) %>% group_by(job_type, term) %>% 
  summarize("frequency"=sum(count)) %>% slice_max(frequency, n=5)
b4 <- filter(maybe_job_tf_idf, !(term %in% c("someone","issues","work","etc","just","felt","issue"))) %>% group_by(job_type, term) %>% 
  summarize("frequency"=sum(count)) %>% slice_max(frequency, n=5) %>% 
  ggplot(aes(reorder(term, frequency), frequency, fill=job_type)) + 
  geom_bar(stat = "identity") + coord_flip() + 
  facet_wrap(~job_type, scales="free_x") +
  scale_fill_brewer(palette = "Set2", direction=-1) +
  guides(fill=FALSE) +
  ggtitle("Answer = Maybe") +
  theme(axis.title.y = element_blank(),
        axis.title.x = element_blank(),
        panel.background = element_blank(), 
        plot.title = element_text(face = "bold", color = "black", size = 10))
b4

Similar to word frequency for maybe by company size, uncertain words like depends, discuss and potential are usually used by participants to show their attitudes. They also use stigma to express their worries about the issue. Besides, words like accommodation from developer indicate that it depends on whether they require accommodations. While the term comfortable from supporters shows that it depends on whether they feel comfortable with the interviewer.
#top 10 words in yes

b5 <- filter(yes_job_tf_idf, !(term %in% c("issues","work","just","felt","though","keep","front","can","time","table","part","looking","ahead","employers","either","aware","exactly","end","fair","even","cards"))) %>% group_by(job_type, term) %>% 
  summarize("frequency"=sum(count)) %>% slice_max(frequency, n=5) %>% 
  ggplot(aes(reorder(term, frequency), frequency, fill=job_type)) + 
  geom_bar(stat = "identity") + coord_flip() + 
  facet_wrap(~job_type, scales="free_x") +
  scale_fill_brewer(palette = "Set2", direction=-1) +
  guides(fill=FALSE) +
  ggtitle("Answer = Yes") +
  theme(axis.title.y = element_blank(),
        axis.title.x = element_blank(),
        panel.background = element_blank(), 
        plot.title = element_text(face = "bold", color = "black", size = 10))
b5

Those who answer yes tend to use terms like impact, needed and good to show the importance of mentioning mental health issue. In addition, supporters and workers from other job types use performance to explain that they think work performance is more important than mental health issue.
#top 10 words in no
b6 <- filter(no_job_tf_idf, !(term %in% c("issues","work","etc","just","felt","issue"))) %>% group_by(job_type, term) %>% 
  summarize("frequency"=sum(count)) %>% slice_max(frequency, n=5) %>% 
  ggplot(aes(reorder(term, frequency), frequency, fill=job_type)) + 
  geom_bar(stat = "identity") + coord_flip() + 
  facet_wrap(~job_type, scales="free_x") +
  scale_fill_brewer(palette = "Set2", direction=-1) +
  guides(fill=FALSE) +
  ggtitle("Answer = No") +
  theme(axis.title.y = element_blank(),
        axis.title.x = element_blank(),
        panel.background = element_blank(), 
        plot.title = element_text(face = "bold", color = "black", size = 10))
b6

Similar to word frequency for no by company size, participants who answered no tend to use negative words like stigma, don’t and wouldn’t to show their negative attitudes. Besides, supporters frequently use discrimination to explain why they wouldn’t talk about mental health issue in interviews.


Sentiment Analysis for Each Answer

#plot the relationship between sentiment and answer
p2 <- ggplot(mental_health, aes(x = bringup_issue, y = sentiment)) + 
  geom_boxplot(aes(fill=job_type))  + 
  stat_summary(mapping=aes(group=bringup_issue),fun="mean",geom="point",shape=23,size=3,fill="white") + 
  facet_wrap(~job_type) + 
  scale_fill_manual(values = c("#8DA0CB","#FC8D62","#66C2A5")) +
  labs(title = "Distribution of Sentiment Score by Answer and Job Type", y = "Sentiment Score") + 
  theme(legend.position = 'none', 
        axis.title.x = element_blank(),
        axis.title.y = element_text(vjust = 2),
        panel.background = element_blank(), 
        panel.grid.major = element_line(color = "gray50", size = 0.5),
        panel.grid.major.x = element_blank(),
        plot.title = element_text(face = "bold", color = "black", size = 12))
ggplotly(p2 , tooltip = "sentiment")
In sum, respondents who answered maybe and no show a negative sentiment towards the question while those answering yes present a positive sentiment. Supporters get highest average sentiment score in each answer, followed by developers and then participants from other job types.