Quora
Quora
Quora is a community question and answer website. In one of our experiments, we investigated whether performing Google web searches that are modeled after the web search behavior of individuals with a psychiatric disorder would result in advertisements and content on Quora different from a control not performing any web searches.
Methodology
- Stage 1: In stage 1, we created six different Quora accounts: 2 reflecting a subject with Mood Disorder, 2 for Schizophrenia, and 2 as the Control (that is, subject with no psychiatric illness).
- Stage 2: In stage 2, while logged into the corresponding Google accounts, we performed Google searches every 1 – 2 days. These Google searches were modeled after the internet search behavior of individuals with a mood disorder or schizophrenia. Web searches for profiles within a group were identical and performed in the same order each day, and the top web result on each search was clicked into. None of the sites clicked into belonged to Quora itself. All our experiments were done using profiles with Google “sync” on, and we did not opt out of cookies and had neither ad blockers nor strict privacy settings enabled to prevent targeted advertising. We tried to control as many factors as possible; however, there are many complex factors in play when it comes to targeted advertising.
- Stage 3: Quora claims that they offer “contextual” and “behavioral” targeting for advertisers, and we wanted to see what exactly they mean by this. After each web browsing session, we went to Quora’s homepage, and collected all the advertisements from their site’s newsfeed.
- Stage 4: At the end of our experiment, we downloaded and parsed all the Quora Digest emails each account received over the course of the experiment. Quora Digest is a daily email sent by Quora that contains a set of personalized, recommended questions.
- Stage 5: All data was analyzed using R programming.
Web Search Methodology
What DO individuals with mood disorders or schizophrenia search for online? To find out, we consulted literature on prior research that studied the most frequently used language and topics of online searches by individuals with a mood disorder or schizophrenia. Once we had a list of commonly used language and frequently searched topics, we needed to generate our simulated web searches for each profile. To accomplish this, we used something called a grammar to generate our search strings.
Tracery Grammar Code
Grammar Code Author: Aishee Mukherji
Language: Python 3.8.2
R Code Author: Aishee Mukherji
R version 4.0.4 (2021-02-15)
Results
Quora Site Advertisements
Descriptive Statistics
A chi-square test of independence shows that there is no significant difference in the distribution of Quora site advertisement topics or Quora sponsors between the three groups (P>.05).
| control (N=35) |
mood disorder (N=64) |
schizophrenia (N=63) |
P-value | |
|---|---|---|---|---|
| Quora Site Ad Type | ||||
| Games Apps & Quizzes | 8.00 (22.9%) | 11.0 (17.2%) | 14.0 (22.2%) | 0.718 |
| Healthcare | 3.00 (8.6%) | 9.00 (14.1%) | 11.0 (17.5%) | |
| News | 6.00 (17.1%) | 8.00 (12.5%) | 7.00 (11.1%) | |
| Political | 3.00 (8.6%) | 3.00 (4.7%) | 1.00 (1.6%) | |
| Selling Product(s) | 15.0 (42.9%) | 33.0 (51.6%) | 30.0 (47.6%) | |
| Quora Site Ad Sponsors | ||||
| Atlassian | 1.00 (2.9%) | 0 (0%) | 3.00 (4.8%) | 0.668 |
| CapitalOne Shopping | 7.00 (20.0%) | 14.0 (21.9%) | 16.0 (25.4%) | |
| Forge of Empires | 2.00 (5.7%) | 2.00 (3.1%) | 6.00 (9.5%) | |
| Gazillions | 1.00 (2.9%) | 3.00 (4.7%) | 2.00 (3.2%) | |
| Gundry MD | 1.00 (2.9%) | 4.00 (6.3%) | 3.00 (4.8%) | |
| Local Solar Power Programs | 2.00 (5.7%) | 2.00 (3.1%) | 2.00 (3.2%) | |
| Magellan Times | 1.00 (2.9%) | 0 (0%) | 0 (0%) | |
| ManageEngine | 1.00 (2.9%) | 0 (0%) | 0 (0%) | |
| Shopify | 1.00 (2.9%) | 2.00 (3.1%) | 3.00 (4.8%) | |
| Square Online | 1.00 (2.9%) | 0 (0%) | 1.00 (1.6%) | |
| Stansberry Research | 3.00 (8.6%) | 3.00 (4.7%) | 1.00 (1.6%) | |
| System1 | 2.00 (5.7%) | 3.00 (4.7%) | 4.00 (6.3%) | |
| The Penny Hoarder | 1.00 (2.9%) | 0 (0%) | 0 (0%) | |
| TruthFinder | 2.00 (5.7%) | 10.0 (15.6%) | 4.00 (6.3%) | |
| Upbeat News | 4.00 (11.4%) | 7.00 (10.9%) | 4.00 (6.3%) | |
| Worldlifestyle | 5.00 (14.3%) | 6.00 (9.4%) | 6.00 (9.5%) | |
| Avnet | 0 (0%) | 1.00 (1.6%) | 0 (0%) | |
| BairesDev | 0 (0%) | 1.00 (1.6%) | 0 (0%) | |
| Beverly Hills MD (BHMD) | 0 (0%) | 2.00 (3.1%) | 2.00 (3.2%) | |
| Maternity Week | 0 (0%) | 1.00 (1.6%) | 0 (0%) | |
| Network Solutions | 0 (0%) | 1.00 (1.6%) | 0 (0%) | |
| The Motley Fool | 0 (0%) | 1.00 (1.6%) | 0 (0%) | |
| Wesley Financial Group, LLC | 0 (0%) | 1.00 (1.6%) | 0 (0%) | |
| Best Gadget Advice | 0 (0%) | 0 (0%) | 2.00 (3.2%) | |
| ContentFly | 0 (0%) | 0 (0%) | 1.00 (1.6%) | |
| Eden Health | 0 (0%) | 0 (0%) | 2.00 (3.2%) | |
| History Daily | 0 (0%) | 0 (0%) | 1.00 (1.6%) |
Proportion of each type of Quora advertisement type conditioned on group
Below is a stacked bar chart of the proportion of each advertisement type of Quora newsfeed conditioned on the profile group. There does not appear to be an association in the distribution of ad types between the three different groups.
Quora Digest Emails
Descriptive Statistics
A chi-square test of independence shows that there is a significant difference in the distribution of Quora Digest topics between the three groups (P<.001).
| control (N=165) |
mood disorder (N=189) |
schizophrenia (N=207) |
P-value | |
|---|---|---|---|---|
| Quora Digest Question Topic | ||||
| Mental Health | 9.00 (5.5%) | 26.0 (13.8%) | 57.0 (27.5%) | <0.001 |
| General Health | 6.00 (3.6%) | 7.00 (3.7%) | 5.00 (2.4%) | |
| Food & Nutrition | 31.0 (18.8%) | 29.0 (15.3%) | 21.0 (10.1%) | |
| Books & Movies | 19.0 (11.5%) | 22.0 (11.6%) | 24.0 (11.6%) | |
| Music & Celebrities | 27.0 (16.4%) | 36.0 (19.0%) | 21.0 (10.1%) | |
| Society | 11.0 (6.7%) | 9.00 (4.8%) | 12.0 (5.8%) | |
| Science, Technology, Sites | 30.0 (18.2%) | 46.0 (24.3%) | 41.0 (19.8%) | |
| Education | 4.00 (2.4%) | 7.00 (3.7%) | 3.00 (1.4%) | |
| Religion, History, & Culture | 13.0 (7.9%) | 2.00 (1.1%) | 15.0 (7.2%) | |
| Business, Jobs, & Finance | 3.00 (1.8%) | 4.00 (2.1%) | 8.00 (3.9%) | |
| Law | 12.0 (7.3%) | 1.00 (0.5%) | 0 (0%) |
Proportion of each type of Quora Digest question topic conditioned on group
Below is a stacked bar chart of the proportion of each Quora Digest email question topic conditioned on the profile group. The distribution in the proportion of each topic appears to differ between the 3 study groups. Profiles from the schizophrenia group received a higher proportion of emails discussing mental health topics than the other two groups. Profiles in the mood disorder group received a larger proportion of emails relating to Science and Technology than the control received, but not more than the Schizophrenia group. Profiles in the control received a larger proportion of topics relating to food and the law than the two psychiatric disorder groups.
Word Clouds
Control
A word cloud of all nouns and adjectives with at least a frequency of 2 in the Quora Digest emails of the control.
Mood Disorder
A word cloud of all nouns and adjectives with at least a frequency of 2 in the Quora Digest emails of the Mood Disorder group.
Schizophrenia
A word cloud of all nouns and adjectives with at least a frequency of 2 in the Quora Digest emails of the Schizophrenia group.
Frequently Used Nouns and Adjectives
On this graph, only words that appeared at least 4 times in at least 1 of our three groups Quora Digest emails are shown. Notice that words directly alluding to mental health such as “schizophrenia”, “borderline”, “personality”, and “disorder”, are used more frequently in the Quora digest emails sent to schizophrenia profiles compared to mood disorder and the control, who never received emails with these words. Freddie Mercury also appears in Schizophrenia and Mood Disorder, but not as frequently in control. Book, and interestingly, the word “narcissist”, appears in similar frequency in all three groups.
Appendix: All code for this analysis.
knitr::opts_chunk$set(echo = FALSE, collapse = TRUE,
warning = FALSE, message = FALSE)
# installing packages
library(readr)
library(dplyr)
library(stringr)
library(tidyverse)
library(lubridate)
library(ggwordcloud)
library(tm)
library(SnowballC)
library(wordcloud)
library(RColorBrewer)
library(tidytext)
library(lubridate)
library(udpipe)
library(textrank)
library(lattice)
library(jcolors)
library(scales)
library(table1)
ud_model <- udpipe_download_model(language = "english")
ud_model <- udpipe_load_model(ud_model$file_model)
set.seed(122)
theme_set(theme_bw())
# Clean data frame
clean <- function(f) {
df <-
data.frame(question = str_extract_all(
read_file(f),
"To:.*.com|Date:.*\r\n|Question:.*\r*\n*.*\\?"
)) %>%
setNames(c("question")) %>%
mutate(question = str_replace_all(question, "To: |Date: |Question: |\r|\n|=", ""))
email <- df[1, 1]
Timestamp <-
dmy_hms(str_extract(df[3, 1], "[0-9]+.* (?=[\\+0000])"), tz = "America/Los_Angeles")
date <- date(Timestamp)
x <-
cbind(data.frame(email, date), question = df %>% tail(length(df) - 4))
return(x)
}
# Create word clouds
wordcloudmaker <-
function(d,
grp = NA,
words,
freq,
minfreq = 1,
maxwords = 200,
scal = c(4, .5)) {
if (!is.na(grp)) {
dt <- subset(d, group == grp)
} else {
dt <- d
}
wordcloud(
words = dt[[words]],
freq = dt[[freq]],
scale = scal,
min.freq = minfreq,
max.words = maxwords,
random.order = FALSE,
rot.per = 0.35,
colors = brewer.pal(8, "Dark2")
)
}
# Get part of speech specified
commonPOS <- function(df, grp, pos = "NOUN") {
d <- subset(df, group == grp)
d[["words"]] <- str_replace_all(d[["words"]], "e2809.", "")
stats <-
subset(as.data.frame(udpipe_annotate(ud_model, x = d[["words"]])), upos %in% pos)
stats <- txt_freq(x = stats$lemma)
stats$key <- factor(stats$key, levels = rev(stats$key))
stats
}
# Return p-value for descriptive stats
pvalue <- function(x, ...) {
# Construct vectors of data y, and groups (strata) g
y <- unlist(x)
g <- factor(rep(1:length(x), times = sapply(x, length)))
if (is.numeric(y)) {
# For numeric variables, perform a standard 2-sample t-test
p <- t.test(y ~ g)$p.value
} else {
# For categorical variables, perform a chi-squared test of independence
p <- chisq.test(table(y, g))$p.value
}
# Format the p-value, using an HTML entity for the less-than sign.
# The initial empty string places the output on the line below the variable label.
c("", sub("<", "<", format.pval(p, digits = 3, eps = 0.001)))
}
filenames <-
list.files("~/RStudio/Comps/healthcompsdata/emails/", full.names = TRUE)
participants <- data.frame(
email = c(
"leea5105@gmail.com",
"aj8743124@gmail.com",
"bryantcarol868@yahoo.com",
"cassidybarnes75@gmail.com",
"aubreyyates2@gmail.com",
"blakeg853@gmail.com"
),
group = c(
"mood disorder",
"schizophrenia",
"control",
"mood disorder",
"schizophrenia",
"control"
)
)
quora_ads <- read.csv("quora_ads.csv", na.strings = c("")) %>%
pivot_longer(cols = starts_with("ad"),
names_to = "ad_number",
values_to = "ad") %>%
mutate(
sponsor = str_extract(ad, ".*"),
words = str_extract_all(str_to_lower(ad), boundary("word")),
sponsor_type = case_when(
sponsor %in% c(
"BHMD",
"BHMD Dermal",
"Eden Health",
"Gundry MD",
"System1 | Bladder Cancer",
"System1 | Multiple Sclerosis",
"System1 | Spinal Muscular Atrophy"
) ~ "Healthcare",
sponsor %in% c(
"Atlassian",
"Avnet",
"BairesDev",
"CapitalOne Shopping",
"ContentFly",
"Local Solar Power Programs" ,
"ManageEngine",
"Network Solutions",
"Shopify",
"Square Online",
"The Motley Fool",
"TruthFinder",
"Wesley Financial Group, LLC"
) ~ "Selling Product(s)",
sponsor %in% c("Forge of Empires", "Gazillions", "Worldlifestyle") ~ "Games Apps & Quizzes",
sponsor %in% c("Stansberry Research") ~ "Political",
sponsor %in% c(
"Best Gadget Advice",
"History Daily",
"Magellan Times",
"Maternity Week",
"The Penny Hoarder",
"Upbeat News",
"Worldlifestyle"
) ~ "News"
),
date = mdy(time)
)
df <- lapply(filenames, clean) %>%
bind_rows() %>%
inner_join(participants, by = "email") %>%
mutate(
question = str_remove_all(question, "E28099|\""),
topic_breakdown = case_when(
str_detect(
str_to_lower(question),
"disorder|anxiety|mood|bipolar|schizophrenia|bpd|narciss|
epression|psycho|why are you sad|traumatic event|truly happy|borderline"
) ~ "mental health",
str_detect(
str_to_lower(question),
"medicine|vaccine|doctor|physician|vitamin c|hospital|health diagnosis|
pandemic|blood sugar|push up"
) ~ "general health",
str_detect(
str_to_lower(question),
"food|bread|pizza|meat|steak|vegetables|nutrition|nutrient|sugar|milk|
oysters|pasta|oats|breakfast|honey|meal|suya|tuna|onion|dish|eat less|chefs|salt|egg mcmuffin|pork|fruits"
) ~ "food",
str_detect(
str_to_lower(question),
"phone|iphone|technology|tech|computer|c\\+\\+|5g|cryptocurrency|startup|
microsoft|linux|cell phone|bot|samsung|spyware|phone|chip production|transistor|
screen recorder|batter|blockchain|dogecoin|electric car|purple bullet|cryptos"
) ~ "technology",
str_detect(
str_to_lower(question),
"snapchat|twitter|facebook|instagram|discord|youtube"
) ~ "social media",
str_detect(
str_to_lower(question),
"scientific|science|scientist|drilling|discoveries|periodic table|einstein|
inventions|atom|giant egg discovered|oldest living things|bleach|could i survive|electrons"
) ~ "science",
str_detect(
str_to_lower(question),
"music|k-pop|rock star|hard rock|song|bts|studio|rock and roll|rock band|
rock|paul mccartney|freddie mercury|john lennon|mick jagger|david crosby|billy|
beatles|peter gabriel|bobby mcgee|brian may|composer|michael jackson|singer|heavy metal|good artist"
) ~ "music",
str_detect(str_to_lower(question), "book|read|stephen king") ~ "books",
str_detect(
str_to_lower(question),
"movie|lord of the rings|mcu scene|marvel cinematic universe|oscar win"
) ~ "movie",
str_detect(str_to_lower(question), "sum of") ~ "math",
str_detect(
str_to_lower(question),
"celebrity|john wayne|prince charles|tom cruise|william shatner|princess diana"
) ~ "celebrity",
str_detect(
str_to_lower(question),
"business|job|employee|switch to product based|market|your boss|good career"
) ~ "business/marketing/jobs",
str_detect(
str_to_lower(question),
"history|henry viii|roswell|challenger astronauts|person of all time|
chemical reaction|50s and 60s|jimmy stewar|a human has survived"
) ~ "history",
str_detect(
str_to_lower(question),
"teacher|phd|graduate|school|student|educat|community college"
) ~ "education",
str_detect(
str_to_lower(question),
"my son|parent|daughter|family|kids|child|12 year old"
) ~ "family/children",
str_detect(
str_to_lower(question),
"religion|bible|eckhart tolle|demon|angels|lucifer effect"
) ~ "religion",
str_detect(str_to_lower(question), "puppy|dog|cat|pigeon") ~ "animal",
str_detect(
str_to_lower(question),
"it correct to say between|bad handwriting"
) ~ "grammar",
str_detect(str_to_lower(question), "money|financ|wealth") ~ "finance",
str_detect(str_to_lower(question), "british|usa|uk|country|turkey") ~ "culture/country",
str_detect(
str_to_lower(question),
"military|ar-15|gun|law|cops|legal|police officer|property line|horrific crime "
) ~ "law",
str_detect(
str_to_lower(question),
"mature|romantic|compromising|inappropriate|ruined your life|
experience you had|etiquette points|spanking|walked into your home|
insane thing|dangerous person|embarrassing moment|trusted your gut|
found the one|made you feel sorry|secret life|wedding guest|parked car"
) ~ "other social"
),
topic = fct_relevel(
case_when(
topic_breakdown == "mental health" ~ "Mental Health",
topic_breakdown == "general health" ~ "General Health",
topic_breakdown == "food" ~ "Food & Nutrition",
topic_breakdown == "law" ~ "Law",
topic_breakdown %in% c("science", "technology", "social media") ~ "Science, Technology, Sites",
topic_breakdown %in% c("books", "movie") ~ "Books & Movies",
topic_breakdown %in% c("music", "celebrity") ~ "Music & Celebrities",
topic_breakdown %in% c("math", "education", "grammar") ~ "Education",
topic_breakdown %in% c("family/children", "other social", "animal") ~ "Society",
topic_breakdown %in% c("finance", "business/marketing/jobs") ~ "Business, Jobs, & Finance",
topic_breakdown %in% c("history", "religion", "culture/country") ~ "Religion, History, & Culture",
TRUE ~ topic_breakdown
),
"Mental Health",
"General Health",
"Food & Nutrition",
"Books & Movies",
"Music & Celebrities",
"Society",
"Science, Technology, Sites",
"Education",
"Religion, History, & Culture",
"Business, Jobs, & Finance",
"Law"
)
)
write.csv(participants, file = "participants.csv")
write.csv(df, file = "quora_digest_out.csv")
q <- quora_ads %>% select(-words)
write.csv(as.data.frame(q), file = "quora_ads_out.csv")
quora_ads.rmna <- quora_ads %>%
filter(!is.na(sponsor_type)) %>%
mutate(
sponsor = case_when(
str_detect(sponsor, "BHMD") ~ "Beverly Hills MD (BHMD)",
str_detect(sponsor, "System1") ~ "System1",
TRUE ~ sponsor
)
)
label(quora_ads.rmna$sponsor_type) <- "Quora Site Ad Type"
label(quora_ads.rmna$sponsor) <- "Quora Site Ad Sponsors"
table1(
~ sponsor_type + sponsor |
group,
data = quora_ads.rmna,
overall = F,
extra.col = list(`P-value` = pvalue)
)
quora_ads %>%
filter(!is.na(sponsor_type)) %>%
group_by(group, sponsor_type) %>%
summarize(n = n()) %>%
ungroup() %>%
ggplot(aes(
fill = sponsor_type,
y = n,
x = group,
label = n
)) +
geom_bar(position = "fill", stat = "identity") +
coord_flip() +
labs(x = "Group",
y = "Proportion",
fill = "Advertisement Category",
title =str_wrap("Stacked bar chart of the proportion of each Quora advertisement type conditioned on study",75),
caption = "*Number of questions in each category labeled on chart") +
theme(legend.position = "bottom") +
geom_text(position = position_fill(vjust = 0.5)) +
scale_fill_brewer(palette = "Set3") +
guides(fill = guide_legend(ncol = 2))
ggsave("quora_ads_bar.png", width = 10, height = 7, units = "in")
label(df$topic) <- "Quora Digest Question Topic"
table1(
~ topic |
group,
data = df,
overall = F,
extra.col = list(`P-value` = pvalue)
)
digest <- df %>%
mutate(words = str_extract_all(str_to_lower(question), boundary("word"))) %>%
unnest(cols = c(words)) %>%
mutate(words_rmstop = ifelse(
words %in% stopwords("english") |
removeNumbers(words) == "",
NA,
as.character(words)
))
df %>%
group_by(group, topic) %>%
summarize(n = n()) %>%
ungroup() %>%
ggplot(aes(
fill = topic,
y = n,
x = group,
label = n
)) +
geom_bar(position = "fill", stat = "identity") +
coord_flip() +
labs(x = "Group",
y = "Proportion",
title = str_wrap("Stacked bar chart of the proportion of Quora digest email question type conditioned on study group", 75),
fill = "Recommended Question Topic",
caption = "*Number of questions in each category labeled on chart") +
theme(legend.position = "bottom") +
geom_text(position = position_fill(vjust = 0.5)) +
scale_fill_brewer(palette = "Set3") +
guides(fill = guide_legend(ncol = 3))
ggsave("quora_digest_bar.png", width = 11, height = 7, units = "in")
wordcloudmaker(
d = commonPOS(digest, "control", c("NOUN", "ADJ")),
words = "key",
freq = "freq",
scal = c(5, .5),
minfreq = 2
)
wordcloudmaker(
d = commonPOS(digest, "mood disorder", c("NOUN", "ADJ")),
words = "key",
freq = "freq",
scal = c(3.5, .5),
minfreq = 2
)
wordcloudmaker(
d = commonPOS(digest, "schizophrenia", c("NOUN", "ADJ")),
words = "key",
freq = "freq",
scal = c(3, .5),
minfreq = 2
)
complete_df <- commonPOS(digest, "control", c("NOUN", "ADJ")) %>%
full_join(commonPOS(digest, "mood disorder", c("NOUN", "ADJ")), by = "key") %>%
full_join(commonPOS(digest, "schizophrenia", c("NOUN", "ADJ")), by = "key") %>%
rename(
freq.control = freq.x,
freq.md = freq.y,
freq.schiz = freq,
freq_pct.control = freq_pct.x,
freq_pct.md = freq_pct.y,
freq_pct.schiz = freq_pct,
) %>%
filter(freq.control > 4 | freq.md > 4 | freq.schiz > 4) %>%
arrange(desc(freq.control), desc(freq.md), desc(freq.schiz)) %>%
pivot_longer(cols = starts_with("freq."),
names_to = "group",
values_to = "freq")
complete_df %>%
group_by(key) %>%
mutate(total = sum(freq),
group = factor(case_when(group == "freq.md" ~ "Mood Disorder",
group == "freq.schiz" ~ "Schizophrenia",
group == "freq.control" ~ "Control"))) %>%
ggplot(aes(x = reorder(key, total), y = freq, fill = group))+
geom_bar(stat="identity") +
coord_flip() +
facet_grid(~group) +
labs(x = "Nouns and Adjectives",
y = "Frequency",
fill = "Group",
title = str_wrap("Frequency of nouns and adjectives used in Quora Digest emails by study group.",65),
caption = "*Only nouns and adjectives with a frequency of at least 4 in one study group are shown.")
ggsave("nounadj.png")Data
participants.csv 🔗
| FIELD1 | group | |
|---|---|---|
| 1 | leea5105@gmail.com | mood disorder |
| 2 | aj8743124@gmail.com | schizophrenia |
| 3 | bryantcarol868@yahoo.com | control |
| 4 | cassidybarnes75@gmail.com | mood disorder |
| 5 | aubreyyates2@gmail.com | schizophrenia |
| 6 | blakeg853@gmail.com | control |
quora_ads_out.csv 🔗
| FIELD1 | id | group | name | time | ad_number | ad | sponsor | sponsor_type | date | |
|---|---|---|---|---|---|---|---|---|---|---|
| 32 | 1 | mood disorder | Ali Lee | leea5105@gmail.com | 4/20/21 | ad2 | Upbeat News How much of a psychopath are you? You may be more psychopathic than you think you are. Answer truthfully for the most accurate results. |
Upbeat News | News | 2021-04-20 |
| 33 | 1 | mood disorder | Ali Lee | leea5105@gmail.com | 4/20/21 | ad3 | TruthFinder Have you ever googled yourself? Run a "deep search" instead. This new search engine reveals so much more. Type in you name, wait 107 seconds, brace yourself. |
TruthFinder | Selling Product(s) | 2021-04-20 |
| 34 | 1 | mood disorder | Ali Lee | leea5105@gmail.com | 4/20/21 | ad4 | System1 | Bladder Cancer Symptoms of bladder cancer you may wish you knew sooner. Catching bladder cancer early may make a big difference. Look for signs and symptoms here. |
System1 | Bladder Cancer | Healthcare | 2021-04-20 |
| 35 | 1 | mood disorder | Ali Lee | leea5105@gmail.com | 4/20/21 | ad5 | Worldlifestyle A school expelled a teen without realizing who her dad is. He listened to them as they berated his daughter, but enough was enough. |
Worldlifestyle | Games Apps & Quizzes | 2021-04-20 |
| 36 | 2 | schizophrenia | Ashley Jackson | aj8743124@gmail.com | 4/20/21 | ad1 | CapitalOne Shopping The genius hack every Home Depot shopper should know. Before you shop at The Home Depot, read this. |
CapitalOne Shopping | Selling Product(s) | 2021-04-20 |
| 37 | 2 | schizophrenia | Ashley Jackson | aj8743124@gmail.com | 4/20/21 | ad2 | Upbeat News How much of a psychopath are you? You may be more psychopathic than you think you are. Answer truthfully for the most accurate results. |
Upbeat News | News | 2021-04-20 |
| 38 | 2 | schizophrenia | Ashley Jackson | aj8743124@gmail.com | 4/20/21 | ad3 | Forge of Empires Play this game for 3 minutes and see why everyone is addicted. Build, barter, and battle through historical ages in this award-winning strategy game. |
Forge of Empires | Games Apps & Quizzes | 2021-04-20 |
| 39 | 2 | schizophrenia | Ashley Jackson | aj8743124@gmail.com | 4/20/21 | ad4 | BHMD Dermal 1 simple thing you can do for a natural face lift. Beverly Hills surgeon reveals at home fix (no creams needed). For illustrative purposes only. |
BHMD Dermal | Healthcare | 2021-04-20 |
| 40 | 2 | schizophrenia | Ashley Jackson | aj8743124@gmail.com | 4/20/21 | ad5 | System1 | Bladder Cancer Symptoms of bladder cancer you may wish you knew sooner. Catching bladder cancer early may make a big difference. Look for signs and symptoms here. |
System1 | Bladder Cancer | Healthcare | 2021-04-20 |
| 41 | 3 | control | Carol Bryant | bryantcarol868@yahoo.com | 4/20/21 | ad1 | CapitalOne Shopping The genius hack every Home Depot shopper should know. Before you shop at The Home Depot, read this. |
CapitalOne Shopping | Selling Product(s) | 2021-04-20 |
| 42 | 3 | control | Carol Bryant | bryantcarol868@yahoo.com | 4/20/21 | ad2 | Forge of Empires Play this game for 3 minutes and see why everyone is addicted. Build, barter, and battle through historical ages in this award-winning strategy game. |
Forge of Empires | Games Apps & Quizzes | 2021-04-20 |
| 43 | 3 | control | Carol Bryant | bryantcarol868@yahoo.com | 4/20/21 | ad3 | Upbeat News Quiz: are you vintage enough to name these 24 old school items. Test your knowledge and be the first to score over 90%. |
Upbeat News | News | 2021-04-20 |
| 44 | 3 | control | Carol Bryant | bryantcarol868@yahoo.com | 4/20/21 | ad4 | System1 | Bladder Cancer Symptoms of bladder cancer you may wish you knew sooner. Catching bladder cancer early may make a big difference. Look for signs and symptoms here. |
System1 | Bladder Cancer | Healthcare | 2021-04-20 |
| 45 | 3 | control | Carol Bryant | bryantcarol868@yahoo.com | 4/20/21 | ad5 | NA | NA | NA | 2021-04-20 |
| 46 | 4 | mood disorder | Cassidy Barnes | cassidybarnes75@gmail.com | 4/20/21 | ad1 | Maternity Week What's the truth about Céline Dion's major weight loss? She recently confirmed what we suspected all along. |
Maternity Week | News | 2021-04-20 |
| 47 | 4 | mood disorder | Cassidy Barnes | cassidybarnes75@gmail.com | 4/20/21 | ad2 | Upbeat News How much of a psychopath are you? You may be more psychopathic than you think you are. Answer truthfully for the most accurate results. |
Upbeat News | News | 2021-04-20 |
| 48 | 4 | mood disorder | Cassidy Barnes | cassidybarnes75@gmail.com | 4/20/21 | ad3 | Upbeat News Quiz: are you vintage enough to name these 24 old school items. Test your knowledge and be the first to score over 90%. |
Upbeat News | News | 2021-04-20 |
| 49 | 4 | mood disorder | Cassidy Barnes | cassidybarnes75@gmail.com | 4/20/21 | ad4 | Worldlifestyle A school expelled a teen without realizing who her dad is. He listened to them as they berated his daughter, but enough was enough. |
Worldlifestyle | Games Apps & Quizzes | 2021-04-20 |
| 50 | 4 | mood disorder | Cassidy Barnes | cassidybarnes75@gmail.com | 4/20/21 | ad5 | Gazillions Most adults fail this United States geography quiz. How much do you remember from geography class? |
Gazillions | Games Apps & Quizzes | 2021-04-20 |
| 61 | ... | ... | ... | ... | ... | ... |
quora_digest.csv 🔗
| 59 | leea5105@gmail.com | 2021-05-12 | Do narcissists like to check their exs social media? And if so whats the reason? | mood disorder | mental health | Mental Health |
| 60 | leea5105@gmail.com | 2021-05-12 | Why didnt Jean-Claude Van Damme find more success in the movie industry? | mood disorder | movie | Books & Movies |
| 61 | leea5105@gmail.com | 2021-05-12 | Why didnt Jean-Claude Van Damme find more success in the movie industry? | mood disorder | movie | Books & Movies |
| 61 | ... | ... | ... | ... | ... | ... |
YOUTUBE
This webpage contains the mental health experiments pertaining to YouTube and targeted advertising. Under “data”, we have data collected during the experiments and the visualizations relevant to the experiment for each persona. The “methodology” page has visual explanations of the methods used in this experiment as well as a detailed description of the methods of the experiments. Finally, “Video” has a 4 minute video explaining the experiment, results, and interpretations of the results. We hope you enjoy this. Thank you!
Experiment Video
Excerpt from final presentation about YouTube
Methodology 🔬
Results (YouTube)
References
Birnbaum, Michael L., Raquel Norel, Anna Van Meter, Asra F. Ali, Elizabeth Arenare, Elif Eyigoz, Carla Agurto, Nicole Germano, John M. Kane, and Guillermo A. Cecchi. “Identifying Signals Associated with Psychiatric Illness Utilizing Language and Images Posted to Facebook.” Npj Schizophrenia 6, no. 1 (December 2020): 38. https://doi.org/10.1038/s41537-020-00125-0.
Conell, Jörn, Rita Bauer, Tasha Glenn, Martin Alda, Raffaella Ardau, Bernhard T. Baune, Michael Berk, et al. “Online Information Seeking by Patients with Bipolar Disorder: Results from an International Multisite Survey.” International Journal of Bipolar Disorders 4, no. 1 (December 2016): 17. https://doi.org/10.1186/s40345-016-0058-0.
Eichstaedt, Johannes C., Robert J. Smith, Raina M. Merchant, Lyle H. Ungar, Patrick Crutchley, Daniel Preoţiuc-Pietro, David A. Asch, and H. Andrew Schwartz. “Facebook Language Predicts Depression in Medical Records.” Proceedings of the National Academy of Sciences 115, no. 44 (October 30, 2018): 11203–8. https://doi.org/10.1073/pnas.1802331115.
“Facebook Posts Help Detect Psychiatric Illness, Study Finds.” Accessed May 12, 2021. https://nypost.com/2020/12/06/facebook-posts-help-detect-psychiatric-illness-study-finds/.
Futurity. “Facebook Posts with These Words Can Predict Depression,” October 22, 2018. https://www.futurity.org/depression-facebook-prediction-1893812/.
“How Facebook Can Be Used to Predict a Manic Episode.” Accessed January 13, 2021. https://www.vice.com/en/article/kbznv9/how-facebook-can-be-used-to-predict-a-manic-episode.
“Inside Facebook’s Suicide Algorithm.” Accessed May 12, 2021. https://www.businessinsider.com/facebook-is-using-ai-to-try-to-predict-if-youre-suicidal-2018-12.
Money and Mental Health Policy Institute. “Is Facebook Tipping-off Advertisers about Your Mental Health?,” May 3, 2017. https://www.moneyandmentalhealth.org/facebook-mental-health/.
Khazan, Olga. “What Your Facebook Posts Say About Your Mental Health.” The Atlantic, November 6, 2019. https://www.theatlantic.com/health/archive/2019/11/what-your-tweets-say-about-your-mood/601510/.
Schrank, Beate, Ingrid Sibitz, Annemarie Unger, and Michaela Amering. “How Patients with Schizophrenia Use the Internet: Qualitative Study.” Journal of Medical Internet Research 12, no. 5 (December 19, 2010): e70. https://doi.org/10.2196/jmir.1550.
Uprichard, Lucy. “The Suicide Algorithm: Facebook’s Self-Appointed Role in Mental Health.” The Technoskeptic (blog), June 28, 2017. https://thetechnoskeptic.com/suicide-algorithm-facebook-mental-health/.
“When Facebook and Instagram Think You’re Depressed.” Accessed May 12, 2021. https://www.vice.com/en/article/pg7d59/when-facebook-and-instagram-thinks-youre-depressed.