Project 2
Screen Time and Sleep
pgil9945, abri4353, fchi2838, mngu0584
University of Sydney | 23 Sep 2022
1 Executive Summary
With continuous advancements in technology, the time that one may
spend online is increasing. This increased screen time may have an
effect on sleep. Our project aims to test this possibility. We’ve
observed that increased screen-time correlates with irregular sleep
patterns, and less sleep overall.
2 Full Report
2.1 Initial Data Analysis (IDA)
In response to the research question, 44 responses were collected from students of DATA1001 as well as friends and family of team members. This was accomplished through an online questionnaire measuring qualitative (nominal and ordinal) and quantitative (discrete and continuous) factors that measure phone usage and factors affecting sleep. The questionnaire explored factors such as phone brand, weekly amount of screentime, amount of sleep and regular bedtimes.
Limitations of this survey included Android phones not providing an average weekly screen time, resulting in participants estimating their daily hours. Due to individual interpretations of a “regular” bedtime, participants may only provide an approximation, which will result in some inaccuracies. In addition, the research question explores only smartphone usage and does not account for other devices such as TVs, laptops or tablets. Gathering averages also excludes extremes from our data. Participants were assumed to provide honest and accurate accounts, as well as understand how to average their weekly screen times.
Data cleaning of the results included the translation of ‘duration’ to ‘decimal’ as well as removing impossible extremes such as “42 hours of sleep per night”. “N/A” results were placed in certain blank spaces to assist recognition of values in the R software.
head(stas)## Timestamp Age Brand Daily.Screentime
## 1 31/08/2022 11:22 56 Android (Samsung, Google)
## 2 31/08/2022 11:29 19 Android (Samsung, Google)
## 3 1/09/2022 11:50 35 Android (Samsung, Google)
## 4 1/09/2022 13:24 21 Android (Samsung, Google)
## 5 1/09/2022 13:28 18 Android (Samsung, Google)
## 6 2/09/2022 20:48 18 Android (Samsung, Google)
## Average.Screentime Screentime.decimal Average.Sleep Time.Sleep
## 1 1:50:03 1.83 7 10 - 11 pm
## 2 7:00:00 7.00 8 1 - 3 am
## 3 0:15:00 0.25 8 10 - 11 pm
## 4 4:00:00 4.00 8 12 - 1 am
## 5 3:00:00 3.00 8 10 - 11 pm
## 6 4:40:53 4.68 NA 11 - 12 pm
## Regular.Bedtime
## 1 N/A
## 2 N/A
## 3 Yes
## 4 No
## 5 Yes
## 6 Yessummary(stas)## Timestamp Age Brand Daily.Screentime
## Length:44 Min. :17.00 Length:44 Length:44
## Class :character 1st Qu.:18.00 Class :character Class :character
## Mode :character Median :19.00 Mode :character Mode :character
## Mean :21.36
## 3rd Qu.:21.00
## Max. :56.00
##
## Average.Screentime Screentime.decimal Average.Sleep Time.Sleep
## Length:44 Min. :0.250 Min. : 4.0 Length:44
## Class :character 1st Qu.:3.422 1st Qu.: 6.5 Class :character
## Mode :character Median :4.290 Median : 7.0 Mode :character
## Mean :4.601 Mean : 7.0
## 3rd Qu.:6.122 3rd Qu.: 8.0
## Max. :8.980 Max. :10.0
## NA's :1
## Regular.Bedtime
## Length:44
## Class :character
## Mode :character
##
##
##
## 2.2 Research Questions
Can phone usage have an impact on an individuals sleep routine?
#Correcting order of bed times
time.fall2 <- factor(stas$Time.Sleep, level = c('7 - 8 pm', '8 - 9 pm', '9 - 10 pm', '10 - 11 pm', '11 - 12 pm', '12 - 1 am', '1 - 3 am', '3 - 6 am', '6 am - 7 pm'))
#Average Bedtime vs Screentime (Boxplot)
ggplot(stas, aes(time.fall2, Screentime.decimal, fill = time.fall2)) +
geom_boxplot(alpha = 0.7) +
scale_fill_brewer(palette="RdBu") +
labs(title = "Average Bedtime vs Average Screentime per day", x = "Average Bedtime", y = "Average Screentime (Hours/Day)") +
theme(legend.position="none", plot.title = element_text(hjust = 0.5))
#Screen time vs Brand (Boxplot)
ggplot(stas, aes(Brand, Screentime.decimal, fill = Brand)) +
geom_boxplot(alpha=0.8) +
scale_fill_brewer(palette = "Purples") +
labs(title = "Brands and Screentime", y = "Screentime (Hours)") +
theme(plot.title = element_text(hjust = 0.5), legend.position = "none")
#Screen time vs Average Sleep (Jitter plot)
ggplot(stas, aes(Average.Sleep, Screentime.decimal, colour = Average.Sleep)) +
geom_jitter(alpha= 0.9, size = 2) +
geom_smooth(method="lm", se = F, formula = y ~ x, color = "darkslateblue") +
scale_colour_distiller(palette = "BuPu") +
labs(title = " Sleep per night vs Screen Time", y = "Screen Time (Hours/Day)", x = "Sleep (Hours/Night)") +
theme(plot.title = element_text(hjust = 0.5), legend.position="none")
#Brand and Regular Bedtime (Stacked Bar)
ggplot(stas, aes(Brand, fill = Regular.Bedtime)) + geom_bar(alpha = 0.7) +
scale_fill_brewer(palette = "Oranges") +
labs(title = "Brands and Regular Bedtimes", y = "Number of People", fill="Regular Bedtime") +
theme(plot.title = element_text(hjust = 0.5)) 
Does our data show other factors impacting sleep, like bedtime?
#Average sleep and bedtime (Boxplot)
ggplot(stas, aes(Average.Sleep, fill = time.fall2)) +
geom_boxplot(alpha = 0.7) +
coord_flip() +
scale_fill_brewer(palette="PuOr") +
labs(title = "Average sleep and Bedtimes", fill = "Average Bed Time", x = "Average Sleep (Hours/Day)") +
theme(axis.text.x=element_blank(), plot.title = element_text(hjust = 0.5))
Does age and screentime have a negative relationship? In other words, do people of an older generation spend less time on their phone?
ggplot(stas, aes(Age, Screentime.decimal, color = Screentime.decimal)) +
geom_point(alpha = 0.9, size = 2) +
geom_smooth(method="lm", se = F, formula = y ~ x, color = "steelblue3") +
labs(title = "Age vs Screentime", x = "Age (Years)", y = "Average Daily Screentime (Hours)") +
theme(plot.title = element_text(hjust = 0.5), legend.position = "none")
But is our linear model appropriate? Can we predict an outcome with it?
Our residual plot indicates an unbalanced data set. It isn’t
symmetrically distributed, and clusters around 20 mark. However, even
though the pattern is centralized around this point, it is still quite
random. In order to be able to reliably predict with this data, we would
need additional data points to fill in the age ‘gaps’, and have a more
filled-out residual plot. If the plot remained random-appearing after
the additional data points, we may be able to use the data to predict
outcomes of age and screen time.
2.3 Articles
Adolescents are 1.34 times more likely to have insufficient sleep when engaged with excessive screen times (Baiden et al. 2019) - supporting our findings that sleep negatively correlates with screen time. This is also true of older age groups, who have “significantly poorer sleep” when accumulating more than 2 hours of evening screen time (Sun et al. 2022).
2.4 References
Baiden, P., Tadeo, S., & Peters, K. (2019). The association between excessive screen-time behaviors and insufficient sleep among adolescents: Findings from the 2017 youth risk behavior surveillance system. Psychiatry Research, 281, 112586. doi: 10.1016/j.psychres.2019.112586
Sun, L., Li, K., Zhang, L., & Zhang, Y. (2022). Distinguishing the Associations Between Evening Screen Time and Sleep Quality Among Different Age Groups: A Population-Based Cross-Sectional Study. Frontiers In Psychiatry, 13. doi: 10.3389/fpsyt.2022.865688
2.5 Acknowledgments
Whole team participated in the tasks, unless otherwise stated.
31st August: Topic of ‘Sleep and Screen time’ agreed upon. Questionnaire created and tested
1st September: Finalisation of Questionnaire and Link sent out and posted - Pippa
6th September: Meeting - Initial analysis of data
File sharing established with google docs/drive
Brainstorming
answers to sections
More links sent to family/friends to gather
additional data
13th September: Sections assigned to team members via Messenger
Executive Summary - Angelika
IDA - Vania
Research Questions
+ Plots - Pippa
Articles - Angelika
Video editing/compiling -
Angelika
Script - Main insights - Maria
20th September: Meeting - Updates on section
progress
Editing to meet criteria and word count
Planning and
writing of script for video component
Filming of introductions and
script reading.