Social media’s influence on teen mental health
Ronald Assanga
2026-05-10
1. Description of the data
This study investigates whether high levels of social media use are associated with poor academic performance, reduced participation in physical activities, diminished social interaction, and increased levels of stress, anxiety, addiction, and depression.
(a) Age demographics by gender
(b) Distribution of sleep hours data
(c) Distribution of data on time spend on screen
(d) Distribution of daily social media usage data
(e) Distribution of platform usage data
(f) Distribution of academic performance data
(g) Distribution of data on time spend on physical activities
(h) Distribution of data on social interaction levels
(i) Distribution of data on stress levels
(j) Distribution of data on anxiety levels
2. Mental health scores by gender
This section examines whether mental‑health outcomes differ between male and female students by comparing their stress, anxiety, and social‑media addiction levels.
3. Assessment of Lifestyle Patterns Across Depression Categories
This section examines the associations between lifestyle habits (sleep duration, screen time, and daily social media usage)
Sleep duration by depression status
6. Analysis of academic outcomes based on platform usage
This section evaluates whether students’ academic performance differs according to the social media platforms they use. (Academics and platform)
7. Physical activity patterns by platform type
This section investigates whether students’ engagement in physical activity differs across social media platform groups. (Physical and platform)
8. Comparative analysis of mental‑health indicators across
Mental health scores, including stress, anxiety, and addiction, are compared across genders.
10. Lifestyle patterns by depression status
The analysis compares depressed and non‑depressed groups across key lifestyle variables, including daily activities, sleep duration, and screen time
Average daily social media hours by depression label
11. Mental health metrics by depression status
Average stress levels by depression label
12. Academic performance by depression label
This analysis compared the average GPA of students grouped by depression label to examine whether academic performance varies across these categories. (Average GPA by depression status)
13. Average physical activity hours by depression label
This analysis examined the relationship between depression and physical activity by assessing how activity levels vary across individuals with different depression scores. (Average physical activity hours)
14.Correlation and heatmap
This analysis explored how key behavioral and mental‑health variables relate to one another by examining correlations among screen time, age, anxiety, academic performance, depression, stress, sleep duration, and addiction levels. (Relationship between variables)
15. Conclusion
This analysis evaluated a range of behavioral, lifestyle, and platform‑related factors to determine which variables meaningfully predict depression by comparing each factor’s relationship to depression status. (prediction)
(a) Age demographics by gender
ggplot(data=social_media_data)+
geom_bar(mapping=aes(x=age, fill=gender), position="dodge")+
labs(title = "Participants’ age demographics by gender")
- The research focused on adolescents aged 13 to 19, with the majority of participants being 13 years old.
(b) Distribution of sleep hours data
ggplot(social_media_data, aes(x = sleep_hours)) +
geom_histogram(binwidth = 1, fill = "steelblue", color = "white") +
geom_freqpoly(binwidth = 1, color = "red", size = 1)+
labs(
title = "Distribution of Sleep Hours",
x = "Sleep Hours",
y = "Count"
) +
theme_minimal()## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once per session.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.
- The data indicates a symmetrical distribution, showing that most teenagers sleep between 4 and 10 hours.
- More than 200 respondents reported sleeping around 7–8 hours, while only a small number reported sleeping as little as 4 hours.
(c) Distribution of data on time spend on screen
ggplot(social_media_data, aes(x = screen_time_before_sleep)) +
geom_histogram(aes(y = after_stat(count / sum(count)) * 100),
binwidth = 1, fill = "steelblue", color = "white") +
geom_freqpoly(aes(y = after_stat(count / sum(count)) * 100),
binwidth = 1, color = "red", size = 1) +
labs(
title = "Screen Time Before Sleep",
x = "Screen time in hours",
y = "Percentage (%)"
) +
theme_minimal()
- The plot indicates that most teenagers spend 1–2 hours on social media before going to sleep.
(d) Distribution of daily social media usage data
ggplot(social_media_data, aes(x = daily_social_media_hours)) +
geom_histogram(binwidth = 1, fill = "steelblue", color = "white") +
geom_freqpoly(binwidth = 1, color = "red", size = 1)+
labs(
title = "Daily social media hours",
x = "Number of hours",
y = "Count"
) +
theme_minimal()
- The plots indicate that most users spend between 2.5 and 8 hours per day on various social media platforms.
(e) Distribution of platform usage data
platform_summary <- social_media_data %>%
count(platform_usage) %>%
mutate(percent = n / sum(n) * 100,
label = paste0(round(percent, 1), "%"))
ggplot(platform_summary, aes(x = "", y = percent, fill = platform_usage)) +
geom_col(width = 1) +
coord_polar(theta = "y") +
geom_text(aes(label = label),
position = position_stack(vjust = 0.5),
color = "white",
size = 4) +
labs(
title = "Platform Usage",
y = "Percentage",
x = ""
) +
theme_void()
- TikTok emerges as the most widely used platform, with Instagram following closely behind. Students who use both platforms together represent the smallest group in the dataset.
(f) Distribution of academic performance data
ggplot(social_media_data, aes(x = academic_performance)) +
geom_histogram(binwidth = 1, fill = "steelblue", color = "white") +
geom_freqpoly(binwidth = 1, color = "red", size = 1)+
labs(
title = "academic performance",
x = "GPA",
y = "# of students"
) +
theme_minimal()
- The distribution is centered around a GPA of 3, with roughly 600 students falling near this value.
- The majority of students have GPAs ranging from 2 to 4.
(g) Distribution of data on time spend on physical activities
ggplot(social_media_data, aes(x = physical_activity)) +
geom_histogram(
binwidth = 1, fill = "steelblue", color = "white") +
geom_freqpoly(binwidth = 1, color = "red", size = 1)+
labs(
title = "Physical activity",
x = "Time in hours",
y = "# of students"
) +
theme_minimal()
- Most students spend around one hour participating in physical activities, while only a small number engage in such activities for more than two hours.
(h) Distribution of data on social interaction levels
interaction_summary <- social_media_data %>%
count(social_interaction_level) %>%
mutate(percent = n / sum(n) * 100,
label = paste0(round(percent, 1), "%"))
ggplot(interaction_summary, aes(x = "", y = percent, fill = social_interaction_level)) +
geom_col(width = 1) +
coord_polar(theta = "y") +
geom_text(aes(label = label),
position = position_stack(vjust = 0.5),
color = "white",
size = 4) +
labs(
title = "Social interaction level",
y = "Percentage",
x = ""
) +
theme_void()
- Most students reported a moderate level of social interaction, which they attributed to spending significant time on social media platforms.
(i) Distribution of data on stress levels
ggplot(social_media_data, aes(x = stress_level)) +
geom_histogram(binwidth = 1, fill = "steelblue", color = "white") +
geom_freqpoly(binwidth = 1, color = "red", size = 1)+
labs(
title = "stress level",
x = "stress levels",
y = "# of students"
) +
theme_minimal()
- The distribution appears fairly uniform across the 1–10 scale, with noticeable peaks at levels 1, 5, and 10.
(j) Distribution of data on anxiety levels
ggplot(social_media_data, aes(x =anxiety_level)) +
geom_histogram(binwidth = 1, fill = "steelblue", color = "white") +
geom_freqpoly(binwidth = 1, color = "red", size = 1)+
labs(
title = "Anxiety level",
x = "anxiety levels",
y = "# of students"
) +
theme_minimal()
- The teenagers’ anxiety levels are fairly evenly distributed across the 1–10 scale, indicating that no single level dominates within the sampled group.
(k) Distribution of data on addiction levels
ggplot(social_media_data, aes(x = addiction_level)) +
geom_histogram(binwidth = 1, fill = "steelblue", color = "white") +
geom_freqpoly(binwidth = 1, color = "red", size = 1)+
labs(
title = "Addiction level",
x = "addiction levels",
y = "# of students"
) +
theme_minimal()
- The plot shows no single dominant addiction level, as the frequency remains fairly consistent across the entire range.
(l) Distribution of data on depression staus
depression_summary <- social_media_data %>%
count(depression_label) %>%
mutate(percent = n / sum(n) * 100,
label = paste0(round(percent, 1), "%"))
ggplot(depression_summary, aes(x = "", y = percent, fill = depression_label)) +
geom_col(width = 1) +
coord_polar(theta = "y") +
geom_text(aes(label = label),
position = position_stack(vjust = 0.5),
color = "white",
size = 5) +
labs(
title = "Depression Label",
x = "",
y = "Percentage"
) +
theme_void()
- Most teenagers reported that they have not experienced depression as a result of using social media.
2. Mental health scores by gender
Stress level per gender
ggplot(social_media_data, aes(x =gender, y = stress_level, fill = gender)) +
geom_boxplot() +
labs(
x = "gender",
y = "stress level",
title = "Stress level by gender"
) +
theme_minimal()
- Both genders show nearly identical median stress levels, each sitting at about 5 out of 10, indicating that their overall stress experience is essentially the same.
Anxiety level by gender
ggplot(social_media_data, aes(x =gender, y =anxiety_level, fill = gender)) +
geom_boxplot() +
labs(
x = "gender",
y = "anxiety level",
title = "Anxiety level by gender"
) +
theme_minimal()
- Both groups show comparable anxiety levels, each averaging around 6 out of 10, indicating no meaningful difference between them.
Addiction level by gender
ggplot(social_media_data, aes(x =gender, y = addiction_level, fill = gender)) +
geom_boxplot() +
labs(
x = "gender",
y = "addiction level",
title = "Addiction level by gender"
) +
theme_minimal()
- Both groups show nearly the same addiction score, hovering just above 5 out of 10, indicating that males and females exhibit comparable addiction levels in the dataset. In other words, there’s no meaningful difference in how addiction is distributed between the two genders.
3. Assessment of lifestyle patterns across depression categories
Sleep duration by depression status
social_media_data$depression <- factor(social_media_data$depression_label, labels = c("No Depression", "Depression"))ggplot(social_media_data, aes(x = depression, y = sleep_hours, fill = depression)) +
geom_boxplot() +
labs(
x = "Depression Status",
y = "Sleep Hours",
title = "Sleep Duration by Depression Status"
) +
theme_minimal()
The median sleep duration is 6.5 hours for the “No depression” group and 4.5 hours for the “depression” group. This suggests a potential association between depression and shorter sleep duration.
There is no overlap between the middle 50% of the two groups, indicating a statistically significant difference.
The “No depression” group has a larger interquartile range than the “depression” group, indicating greater variability in sleep duration among those without depression.
These findings support the hypothesis that individuals with depression are likely to experience significantly shorter sleep duration than those without depression. This is confirmed using the t-Test
t.test(sleep_hours ~ depression, data = social_media_data)##
## Welch Two Sample t-test
##
## data: sleep_hours by depression
## t = 15.745, df = 40.991, p-value < 2.2e-16
## alternative hypothesis: true difference in means between group No Depression and group Depression is not equal to 0
## 95 percent confidence interval:
## 1.510624 1.955162
## sample estimates:
## mean in group No Depression mean in group Depression
## 6.494183 4.761290Since the p‑value =2.2e-16 < 0.05, the two groups differ significantly. Thus, the distribution of sleep hours differs between depressed and non‑depressed participants.
Screen time before sleep by depression status
ggplot(social_media_data, aes(x = depression, y = screen_time_before_sleep, fill = depression)) +
geom_boxplot() +
labs(
x = "Depression Status",
y = "screen time before sleep",
title = "Screen time before sleep by Depression Status"
) +
theme_minimal()
- The plot indicates that individuals with depression exhibit slightly higher and more variable screen time before sleep compared to those without depression.
- The differing medians and the broader distribution for the depressed group suggest less consistent pre-sleep screen time habits.
- These findings support further investigation of sleep behavior as a potential mental health factor.
wilcox.test(screen_time_before_sleep ~ depression, data = social_media_data)##
## Wilcoxon rank sum test with continuity correction
##
## data: screen_time_before_sleep by depression
## W = 19199, p-value = 0.5707
## alternative hypothesis: true location shift is not equal to 0- Because the p‑value (0.5707) is greater than 0.05, there is no statistically significant difference between the two groups. This indicates that the amount of time spent on screens does not differ between participants with and without depression
Daily social media hours by depression status
ggplot(social_media_data, aes(x = depression, y =daily_social_media_hours, fill = depression)) +
geom_boxplot() +
labs(
x = "Depression Status",
y = "daily social media hours",
title = "Daily social media hours by Depression Status"
) +
theme_minimal()
- Individuals with depression show higher median social media use (6 hours) than those without (4.5 hours).
- The no‑depression group has a wider IQR, indicating more varied habits, while the depression group is more consistent.
- The higher median suggests a significant difference between groups.
4. Analysis of mental‑health outcomes across social media platforms
Average stress level by platform
Stress levels are analyzed in relation to the use of specific social media platforms.
stress_mean<-social_media_data %>%
group_by(platform_usage) %>%
summarise(avg_stress = mean(stress_level, na.rm = TRUE))stress_mean## # A tibble: 3 × 2
## platform_usage avg_stress
## <chr> <dbl>
## 1 Both 5.55
## 2 Instagram 5.50
## 3 TikTok 5.29ggplot(data = stress_mean) +
geom_bar(
aes(x = platform_usage, y = avg_stress, fill = platform_usage),
stat = "identity"
) +
geom_text(
aes(
x = factor(platform_usage),
y = avg_stress,
label = round(avg_stress, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "Social Media Platform",
y = "Average stress level",
title = "Average Stress level by Platform Usage"
)
- The “Both” group shows the highest average stress levels, followed by the “Instagram” group, while the “TikTok” group has the lowest average stress.
Average anxiety level by platform usage
Anxiety levels are analyzed in relation to the use of specific social media platforms.
anxiety_mean<-social_media_data %>%
group_by(platform_usage) %>%
summarise(avg_anxiety = mean(anxiety_level, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## platform_usage avg_anxiety
## <chr> <dbl>
## 1 Both 5.49
## 2 Instagram 5.67
## 3 TikTok 5.75ggplot(data = anxiety_mean) +
geom_bar(
aes(x = platform_usage, y = avg_anxiety, fill = platform_usage),
stat = "identity"
) +
geom_text(
aes(
x = factor(platform_usage),
y = avg_anxiety,
label = round(avg_anxiety, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "Social media platform",
y = "Average anxiety level",
title = "Average anxiety level by Platform Usage"
)
- “TikTok” users exhibit the highest average anxiety level, exceeding a score of 5.
- In contrast, Instagram users and those who use both platforms have average anxiety levels that are close to 5.
Average addiction level by platform usage
Addiction levels are analyzed in relation to the use of specific social media platforms.
addiction_mean<-social_media_data %>%
group_by(platform_usage) %>%
summarise(avg_addiction = mean(addiction_level, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## platform_usage avg_addiction
## <chr> <dbl>
## 1 Both 5.50
## 2 Instagram 5.58
## 3 TikTok 5.62ggplot(data = addiction_mean) +
geom_bar(
aes(x = platform_usage, y = avg_addiction, fill = platform_usage),
stat = "identity"
) +
geom_text(
aes(
x = factor(platform_usage),
y = avg_addiction,
label = round(avg_addiction, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "Social media platform",
y = "Average addiction level",
title = "Average addiction level by Platform Usage"
)
- The addiction levels are remarkably similar across all three categories.
- This suggests that the choice of social media platform does not significantly affect the average reported level of addiction.
5. Lifestyle analysis by social media platform
Average daily social media hours by platform usage
daily_usage_mean<-social_media_data %>%
group_by(platform_usage) %>%
summarise(daily_usage_mean = mean(daily_social_media_hours, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## platform_usage daily_usage_mean
## <chr> <dbl>
## 1 Both 4.52
## 2 Instagram 4.56
## 3 TikTok 4.53ggplot(data = daily_usage_mean) +
geom_bar(
aes(x = platform_usage, y = daily_usage_mean, fill = platform_usage),
stat = "identity"
) +
geom_text(
aes(
x = factor(platform_usage),
y = daily_usage_mean,
label = round(daily_usage_mean, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "Social media platform",
y = "Average daily social media hours",
title = "Average daily social media hours by Platform Usage"
)
- The average usage across all three categories is nearly identical, at approximately 4.5 to 5 hours per day.
Average sleep hours by platform usage
sleep_hours_mean<-social_media_data %>%
group_by(platform_usage) %>%
summarise(sleep_hours_mean = mean(sleep_hours, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## platform_usage sleep_hours_mean
## <chr> <dbl>
## 1 Both 6.46
## 2 Instagram 6.44
## 3 TikTok 6.45ggplot(data = sleep_hours_mean) +
geom_bar(
aes(x = platform_usage, y = sleep_hours_mean, fill = platform_usage),
stat = "identity"
) +
geom_text(
aes(
x = factor(platform_usage),
y = sleep_hours_mean,
label = round(sleep_hours_mean, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "Social media platform",
y = "Average sleep hours",
title = "Average sleep hours by Platform Usage"
)
- Users of “Both” Instagram and TikTok appear to average around 6 hours of sleep.
- Overall, there is no significant difference in average sleep duration across the different social media platform categories.
Average screen time hours by platform usage
screen_time_hours_mean<-social_media_data %>%
group_by(platform_usage) %>%
summarise(avg_screen_time = mean(screen_time_before_sleep, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## platform_usage avg_screen_time
## <chr> <dbl>
## 1 Both 1.75
## 2 Instagram 1.71
## 3 TikTok 1.76ggplot(data = screen_time_hours_mean) +
geom_bar(
aes(x = platform_usage, y =avg_screen_time , fill = platform_usage),
stat = "identity"
) +
geom_text(
aes(
x = factor(platform_usage),
y = avg_screen_time,
label = round(avg_screen_time, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "Social media platform",
y = "Average screen time before sleep (hours)",
title = "Average screen time before sleep by Platform Usage"
)
- All three categories (Both, Instagram, and TikTok) show very similar average screen time before sleep, ranging from approximately 1.75 to 1.8 hours.
6. Analysis of academic outcomes based on platform usage
The relationship between academic performance and platform usage is assessed.
GPA_mean<-social_media_data %>%
group_by(platform_usage) %>%
summarise(avg_GPA = mean(academic_performance, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## platform_usage avg_GPA
## <chr> <dbl>
## 1 Both 2.98
## 2 Instagram 3.00
## 3 TikTok 3.00ggplot(data =GPA_mean) +
geom_bar(
aes(x = platform_usage, y =avg_GPA, fill = platform_usage),
stat = "identity"
) +
geom_text(
aes(
x = factor(platform_usage),
y = avg_GPA,
label = round(avg_GPA, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "Social media platform",
y = "Average GPA",
title = "Average GPA by Platform Usage"
)
- The GPA values appear to be fairly similar across all three social media platform groups.
7. Physical activity patterns by platform type
The relationship between academic performance and platform usage is assessed.
physical_activity_mean<-social_media_data %>%
group_by(platform_usage) %>%
summarise(avg_physical = mean(physical_activity, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## platform_usage avg_physical
## <chr> <dbl>
## 1 Both 1.02
## 2 Instagram 1.04
## 3 TikTok 0.982ggplot(data = physical_activity_mean) +
geom_bar(
aes(x = platform_usage, y =avg_physical, fill = platform_usage),
stat = "identity"
) +
geom_text(
aes(
x = factor(platform_usage),
y = avg_physical,
label = round(avg_physical, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "Social media platform",
y = "Mean physical activity time (in hours)",
title = "Mean physical activity time by platform usage group"
)
- The plots indicate that there is little variation in physical activity among the groups.
- All three categories reported an average physical activity duration of approximately one hour.
8. Comparative analysis of mental‑health indicators across genders
Stress level by gender
stress_gender_mean<-social_media_data %>%
group_by(gender) %>%
summarise(avg_stress = mean(stress_level, na.rm = TRUE))%>%
print()## # A tibble: 2 × 2
## gender avg_stress
## <chr> <dbl>
## 1 female 5.42
## 2 male 5.47ggplot(data =stress_gender_mean ) +
geom_bar(
aes(x = gender, y =avg_stress, fill = gender),
stat = "identity"
) +
labs(
x = "gender",
y = "average sleep hours",
title = "Average sleep hours by gender"
)
- There is a slight difference in sleep patterns between the two groups.
- On average, males in the dataset sleep approximately 0.5 hours more than females.
Anxiety level by gender
anxiety_gender_mean<-social_media_data %>%
group_by(gender) %>%
summarise(avg_anxiety = mean(anxiety_level, na.rm = TRUE))%>%
print()## # A tibble: 2 × 2
## gender avg_anxiety
## <chr> <dbl>
## 1 female 5.69
## 2 male 5.59ggplot(data =anxiety_gender_mean ) +
geom_bar(
aes(x = gender, y =avg_anxiety, fill = gender),
stat = "identity"
) +
geom_text(
aes(
x = factor(gender),
y = avg_anxiety,
label = round(avg_anxiety, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "gender",
y = "average anxiety level",
title = "Average anxiety level by gender"
)
- Females show a slightly higher average anxiety level than males.
Anxiety level by gender
addiction_gender_mean<-social_media_data %>%
group_by(gender) %>%
summarise(avg_addiction = mean(addiction_level, na.rm = TRUE))%>%
print()## # A tibble: 2 × 2
## gender avg_addiction
## <chr> <dbl>
## 1 female 5.49
## 2 male 5.64ggplot(data =addiction_gender_mean ) +
geom_bar(
aes(x = gender, y =avg_addiction, fill = gender),
stat = "identity"
) +
geom_text(
aes(
x = factor(gender),
y = avg_addiction,
label = round(avg_addiction, 2)
),
vjust = -0.5,
size = 3.3
) +
labs(
x = "gender",
y = "average addiction level",
title = "Average addiction level by gender"
)
- The average addiction levels for both males and females are very similar, with both groups averaging around a value of 5.
- This suggests that gender is not a significant factor influencing addiction levels, as the difference between the two groups is minimal.
9. Analysis of mental health metrics by Social Interaction Levels
Average stress level by social interaction levels
stress_interaction_mean<-social_media_data %>%
group_by(social_interaction_level) %>%
summarise(avg_stress = mean(stress_level, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## social_interaction_level avg_stress
## <chr> <dbl>
## 1 high 5.51
## 2 low 5.35
## 3 medium 5.48ggplot(data = stress_interaction_mean) +
geom_bar(
aes(x =social_interaction_level, y =avg_stress, fill =social_interaction_level ),
stat = "identity"
) +
labs(
x = "social interaction level",
y = "average stress level",
title = "Average stress level by social interaction levels"
)
- The average stress level remains relatively consistent at around 5 across all three social interaction categories: high, medium, and low.
Average anxiety level by social interaction levels
anxiety_interaction_mean<-social_media_data %>%
group_by(social_interaction_level) %>%
summarise(avg_anxiety = mean(anxiety_level, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## social_interaction_level avg_anxiety
## <chr> <dbl>
## 1 high 5.70
## 2 low 5.53
## 3 medium 5.69ggplot(data = anxiety_interaction_mean) +
geom_bar(
aes(x =social_interaction_level, y =avg_anxiety, fill =social_interaction_level ),
stat = "identity"
) +
labs(
x = "social interaction level",
y = "average anxiety level",
title = "Average anxiety level by social interaction levels"
)
- The data shows no significant relationship between an individual’s level of social interaction and their average anxiety level.
- Therefore, social interaction level cannot be considered a meaningful predictor of anxiety, as the results remain consistent across all categories.
Average addiction level by social interaction levels
addiction_interaction_mean<-social_media_data %>%
group_by(social_interaction_level) %>%
summarise(avg_addiction = mean(addiction_level, na.rm = TRUE))%>%
print()## # A tibble: 3 × 2
## social_interaction_level avg_addiction
## <chr> <dbl>
## 1 high 5.59
## 2 low 5.47
## 3 medium 5.64ggplot(data = addiction_interaction_mean) +
geom_bar(
aes(x =social_interaction_level, y =avg_addiction, fill =social_interaction_level ),
stat = "identity"
) +
labs(
x = "social interaction level",
y = "average addiction level",
title = "Average addiction level by social interaction levels"
)
- The plots indicate that the average addiction levels remain almost the same across all three categories of social interaction.
- This suggests that social interaction levels do not have a significant effect on average addiction levels within this dataset.
10. Lifestyle patterns by depression status
Average daily social media hours by depression label
daily_depression_mean<-social_media_data %>%
group_by(depression_label) %>%
summarise(avg_daily_hours= mean(daily_social_media_hours, na.rm = TRUE))ggplot(data = daily_depression_mean) +
geom_bar(
aes(
x = factor(depression_label),
y = avg_daily_hours,
fill = factor(depression_label)
),
stat = "identity"
) +
geom_text(
aes(
x = factor(depression_label),
y = avg_daily_hours,
label = round(avg_daily_hours, 2)
),
vjust = -0.5,
size = 3.3
) +
scale_fill_manual(
values = c("0" = "steelblue", "1" = "tomato"),
name = "Depression label"
) +
labs(
x = "Depression label",
y = "Average daily Social media hours",
title = "Average daily social media hours by depression Label"
) +
theme_minimal()
- There is a noticeable positive link between greater social media use and having a depression label. Individuals who spend more than three to five hours per day on social media show a significantly higher likelihood of experiencing mental‑health symptoms.
Average sleep hours by depression label
sleep_depression_mean<-social_media_data %>%
group_by(depression_label) %>%
summarise(avg_sleep_hours= mean(sleep_hours, na.rm = TRUE))ggplot(data = sleep_depression_mean) +
geom_bar(
aes(
x = factor(depression_label),
y = avg_sleep_hours,
fill = factor(depression_label)
),
stat = "identity"
) +
geom_text(
aes(
x = factor(depression_label),
y = avg_sleep_hours,
label = round(avg_sleep_hours, 2)
),
vjust = -0.5,
size = 3.3
) +
scale_fill_manual(
values = c("0" = "steelblue", "1" = "tomato"),
name = "Depression label"
) +
labs(
x = "Depression label",
y = "Average sleep hours",
title = "Average sleep hours by depression Label"
) +
theme_minimal()
The data indicates a negative association between sleep duration and a depression diagnosis, meaning that as individuals sleep fewer hours, their likelihood of experiencing depression increases.
Notably, both groups average less than the recommended 7–9 hours of sleep for adults.
Average screen time before sleep by depression
screen_depression_mean<-social_media_data %>%
group_by(depression_label) %>%
summarise(avg_screen_hours= mean(screen_time_before_sleep, na.rm = TRUE))ggplot(data = screen_depression_mean) +
geom_bar(
aes(
x = factor(depression_label),
y = avg_screen_hours,
fill = factor(depression_label)
),
stat = "identity"
) +
geom_text(
aes(
x = factor(depression_label),
y = avg_screen_hours,
label = round(avg_screen_hours, 2)
),
vjust = -0.5,
size = 3.3
) +
scale_fill_manual(
values = c("0" = "steelblue", "1" = "tomato"),
name = "Depression label"
) +
labs(
x = "Depression label",
y = "Average screen time before sleep",
title = "Average screen time before sleep by depression Label"
) +
theme_minimal()
- The average amount of screen time before sleep is nearly identical for both the “Depressed” and “No Depression” groups. Based on the data, increased screen use before bedtime does not appear to be positively associated with having a depression label.
11. Mental health metrics by depression status
Average stress levels by depression label
stress_depression_mean<-social_media_data %>%
group_by(depression_label) %>%
summarise(avg_stress_hours= mean(stress_level, na.rm = TRUE))ggplot(data = stress_depression_mean) +
geom_bar(
aes(
x = factor(depression_label),
y = avg_stress_hours,
fill = factor(depression_label)
),
stat = "identity"
) +
geom_text(
aes(
x = factor(depression_label),
y = avg_stress_hours,
label = round(avg_stress_hours, 2)
),
vjust = -0.5,
size = 3.3
) +
scale_fill_manual(
values = c("0" = "steelblue", "1" = "tomato"),
name = "Depression label"
) +
labs(
x = "Depression label",
y = "Average stress levels",
title = "Average stress levels by depression Label"
) +
theme_minimal()
- There is a clear positive correlation between depression and stress levels in this dataset.
- Individuals identified as depressed report stress scores that are, on average, about 58% higher than those without a depression label.
Average anxiety levels by depression label
anxiety_depression_mean<-social_media_data %>%
group_by(depression_label) %>%
summarise(avg_anxiety_hours= mean(anxiety_level, na.rm = TRUE))ggplot(data = anxiety_depression_mean) +
geom_bar(
aes(
x = factor(depression_label),
y = avg_anxiety_hours,
fill = factor(depression_label)
),
stat = "identity"
) +
geom_text(
aes(
x = factor(depression_label),
y = avg_anxiety_hours,
label = round(avg_anxiety_hours, 2)
),
vjust = -0.5,
size = 3.3
) +
scale_fill_manual(
values = c("0" = "steelblue", "1" = "tomato"),
name = "Depression label"
) +
labs(
x = "Depression label",
y = "Average anxiety levels",
title = "Average anxiety levels by depression Label"
) +
theme_minimal()
There is a strong positive correlation between depression and anxiety in this dataset.
The pattern aligns with established psychological research showing that these conditions frequently co‑occur.
Individuals with higher depression scores also exhibit elevated levels of anxiety.
Average addiction levels by depression label
addiction_depression_mean<-social_media_data %>%
group_by(depression_label) %>%
summarise(avg_addiction_hours= mean(addiction_level, na.rm = TRUE))ggplot(data = addiction_depression_mean) +
geom_bar(
aes(
x = factor(depression_label),
y = avg_addiction_hours,
fill = factor(depression_label)
),
stat = "identity"
) +
geom_text(
aes(
x = factor(depression_label),
y = avg_addiction_hours,
label = round(avg_addiction_hours, 2)
),
vjust = -0.5,
size = 3.3
) +
scale_fill_manual(
values = c("0" = "steelblue", "1" = "tomato"),
name = "Depression label"
) +
labs(
x = "Depression label",
y = "Average addiction levels",
title = "Average addiction levels by depression Label"
) +
theme_minimal()
- The difference in average addiction scores between the two groups is minimal, indicating that, within this dataset, depression status does not meaningfully predict addiction levels.
12. Average GPA by depression status
gpa_depression_mean<-social_media_data %>%
group_by(depression_label) %>%
summarise(avg_gpa_hours= mean(academic_performance, na.rm = TRUE))ggplot(data =gpa_depression_mean) +
geom_bar(
aes(
x = factor(depression_label),
y = avg_gpa_hours,
fill = factor(depression_label)
),
stat = "identity"
) +
geom_text(
aes(
x = factor(depression_label),
y = avg_gpa_hours,
label = round(avg_gpa_hours, 2)
),
vjust = -0.5,
size = 3.3
) +
scale_fill_manual(
values = c("0" = "steelblue", "1" = "tomato"),
name = "Depression label"
) +
labs(
x = "Depression label",
y = "Average GPA",
title = "Average GPA by depression Label"
) +
theme_minimal()
- The average GPA is nearly identical across the two groups, indicating no meaningful difference in academic performance between students with and without a depression label.
13. Average physical activity hours by depression label
physical_depression_mean<-social_media_data %>%
group_by(depression_label) %>%
summarise(avg_physical_hours= mean(physical_activity, na.rm = TRUE))ggplot(data =physical_depression_mean) +
geom_bar(
aes(
x = factor(depression_label),
y = avg_physical_hours,
fill = factor(depression_label)
),
stat = "identity"
) +
geom_text(
aes(
x = factor(depression_label),
y = avg_physical_hours,
label = round(avg_physical_hours, 2)
),
vjust = -0.5,
size = 3.3
) +
scale_fill_manual(
values = c("0" = "steelblue", "1" = "tomato"),
name = "Depression label"
) +
labs(
x = "Depression label",
y = "Average physical activity hours",
title = "Average physical activity hours by depression Label"
) +
theme_minimal()
There is a negative correlation between depression and physical activity in this dataset, indicating that higher levels of depression are associated with reduced engagement in physical exercise.
This pattern is consistent with psychological research showing that depressive symptoms often coincide with lower levels of physical activity.
14.Correlation and Heatmap
corr_matrix<-social_media_data%>%
select(1,3,5,6,7,8,10,11,12,13)%>%
cor()corrplot(corr_matrix,
method = "color",
tl.col = "black"
)
Screen time before sleep vs age: There is a slight positive correlation between age and screen time before sleep, indicating that older participants spend somewhat more time on screens before bed, though the relationship is weak.
Anxiety level vs Academic performance: The heatmap shows a weak negative correlation, suggesting that higher anxiety may be linked to slightly lower academic performance.
Depression vs. daily social media usage: There is a mild positive correlation between depression and daily social media use, indicating that increased social media time may be linked to higher reports of depression.
Depression vs Sleep hours: There is a clear negative relationship, with participants experiencing depression reporting fewer hours of sleep on average.
Depression vs Stress Level: Depression shows a moderate positive correlation with stress, indicating that higher stress is linked to a greater likelihood of depression.
Depression vs Anxiety level: There is a positive relationship between depression and anxiety, with higher anxiety levels associated with increased reports of depression.
Addiction level vs Sleep hours: The heatmap indicates a slight negative correlation, suggesting that higher addiction levels are linked to reduced sleep duration, though the effect is small.
15. Conclusion
| Factor | Predictor of Depression | Explanation |
|---|---|---|
| Sleep duration | Yes | Depressed students sleep ~2 hours less; no overlap in IQR; highly significant difference. |
| Daily social media usage | Yes | Depressed students use social media more (6 hrs vs. 4.5 hrs). |
| Stress levels | Yes | Depressed students report ~58% higher stress. |
| Anxiety levels | Yes | Strong positive correlation; anxiety and depression co‑occur |
| Screen Time Before Sleep | No | No significant difference; p = 0.57. |
| Social Media Addiction | No | Average addiction levels nearly identical across groups |
| Academic Performance (GPA) | No | GPA is almost the same for depressed and non‑depressed students |
| Physical Activity | No | Slight negative correlation, but not strong enough to predict depression. |
| Platform Used (TikTok/Instagram/Both) | No | Stress/anxiety vary slightly by platform, but depression rates do not. |