Time Spent with Family Correlated to Gender and Age

Author

Konci Lawrence

Introduction

The “Who Americans spend their time with, by age”, American Time Use Survey (2010-2024) dataset includes the averages of time spent for each age group in the United States.

This dataset comes from the U.S. Bureau of Labor Statistics (2025), where the American Time Use Survey (ATUS) is performed annually. The source link is here.

My research inquiry to this dataset questions whether an individual’s entity or age group has the most influence on how many hours they spend their family per day.

There are 8 variables, 1 being categorical and the other 7 being quantitative. I analyzed the following key variables:

“Entity”: The observation group identified as “All people”, “Women”, or “Men”.
“Year”: The observation group identified by an age between 15 years old to 80 years old.
“With.Family”: Time spent with family from averages calculated by ATUS.

Loading the Libraries and Dataset

library(tidyverse)

── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.2.1     ✔ readr     2.2.0
✔ forcats   1.0.1     ✔ stringr   1.6.0
✔ ggplot2   4.0.3     ✔ tibble    3.3.1
✔ lubridate 1.9.5     ✔ tidyr     1.3.2
✔ purrr     1.2.2     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

library(ggplot2)
library(dplyr)
library(treemap)

survey <- read.csv("C:/Users/Konci/OneDrive/Documents/DATA110_Fall2026/time-spent-with-relationships-by-age-us.csv")

#head(survey) to observe variables

Cleaning the Data

names(survey) <- tolower(names(survey)) #all variables lowercase

names(survey) <- gsub(".","", names(survey), fixed = TRUE) #remove periods

#head(survey) to see how variables changed

Exploring the Data

Which ages spend the most time on with family? (Barplot)

#saving basic template
survey_chart <- ggplot(survey, aes(x = year, y = withfamily)) +
  labs(title = "Time Spent With Family by Age",
       caption = "Data Source: OurWorldinData.org") +
  xlab("Age (15-80)") +
  ylab("Hours per day")

survey_chart

#observing pattern by age
survey_chart +
  geom_bar(stat = "identity",color= "white", aes(fill = withfamily)) +
  theme(axis.text.y = element_blank())

Is time spent with family related to gender? (Treemap)

#observing the ratio of time spent with family by gender
gender <- filter(survey, entity == c("Women","Men"))
treemap(gender,
        index = "entity",
        vSize = "withfamily",
        vColor = "entity",
        palette = "RdYlBu", #used a reference for manual colors
        title = "Time Spent With Family by Gender",
        title.legend = "Data Source")

Regression Analysis

Is the time spent with family more correlated to gender or age? (Regression Model)

fit <- lm(withfamily ~ entity + year, data = survey)
summary(fit)


Call:
lm(formula = withfamily ~ entity + year, data = survey)

Residuals:
     Min       1Q   Median       3Q      Max 
-0.82640 -0.51436 -0.05321  0.26421  2.09974 

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept)  2.621768   0.137474  19.071   <2e-16 ***
entityMen   -0.189742   0.110918  -1.711   0.0887 .  
entityWomen  0.179189   0.110918   1.616   0.1078    
year        -0.027786   0.002377 -11.690   <2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 0.6372 on 194 degrees of freedom
Multiple R-squared:  0.4323,    Adjusted R-squared:  0.4235 
F-statistic: 49.24 on 3 and 194 DF,  p-value: < 2.2e-16

par(mfrow = c(2, 2))
plot(fit)

Equation

Time spent with family = 0.373543(entityWomen) + -0.028358(year)

P-values: There is a slightly downward trend for an individuals age group (4.79e-09) and a slightly upward trend if an individual is a women (0.0222). The p-values show that an individuals age group has more statistical significance than being a women.

Adjusted R-squared: The adjusted R-squared is 0.4365, meaning that only 43.7% of the variation in time spent with family is explained by the model. The other 56.3% is caused by factors not represented by this model such as time spent with coworkers, partners, children, or alone.

Diagnostic Plots: The Residuals vs Fitted plot shows that relating time spent with family to age and gender is appropriate. The Q-Q Residuals plot indicates that the distribution of observations is relatively normal. The Scale-Location demonstrates that few observations are skewing the variance distribution. Lastly, the Residuals vs Leverage plot identifies a few influential observations.

Overall: The overall model is not statistically significant as the p-value = 5.302e-09. The other 56.3% of the variation is reason to believe that there are other relevant influencing variables impacting the time spent with family.

Conclusion

Cleaning the data

The variables of this dataset had uppercase letters and periods. In order to clean the variable names I used tolower() to convert all names to lowercase. Then I used gsub() to remove the periods in the variable names. This allowed for better reading and handling in exploring the data.

Visualization

This project has three visualizations. A barplot, a treemap, and a multiple regression model. The most notable thing about the barplot depicting the relationship between age and time spent with family, is how it can be deceiving. Although there is a color legend depicting the y-axis values, the actual barplot makes the values appear to be of a greater variation than it is in reality. On the other hand, the treemap is very practical and demonstrates a simple ratio in which women are show to spend more time with their family than men. An interesting observation of this project’s dataset is how the age of an individual has more significance to who time is spent with when compared to the individual’s gender.

Technical challenges and Future Studies

A technical challenge I faced when making this project was ensuring that the bar plot was a true representation of the dataset. Although I included a legend depicting the values of the y-axis, one should be able to tell there is not a drastic variance in time spent with family by age.

If I were to continue this project, I would like to explore whether the time spent with family is in any way influenced by the time an individual spends with any other group. Per say, if the time spent with ones friends or partners has an influence on the time spent with family.

References and Data Source

RStudio. (2023, February 25). RPubs - correlation, scatterplots, regression, and interactivity in R. Rpubs.Com. https://rpubs.com/rsaidi/1007730

RStudio. (2026, March 31). RPubs - project 1: Maryland Senate districts-socioeconomic characteristics. Rpubs.Com. https://rpubs.com/p3nny_lan3/1415900

The Comprehensive R Archive Network (n.d.). Treemap: Mapping to color palettes. The Comprehensive R Network. Retrieved December 18, 2026, from https://cran.r-project.org/web/packages/treemap/vignettes/treemap-color_mapping.html

Who americans spend their time with, by age. (n.d.). Our World in Data. Retrieved October 18, 2026, from https://ourworldindata.org/grapher/time-spent-with-relationships-by-age-us