Feeling Connected

Overview

Stilesians were asked:

On a scale of 1-10: On average, how connected have you felt to other people in the Stiles community during each time period? (1 = very disconnected, 10 = very connected.) Just answer to your best estimation/memory. First-years, it’s ok if your answers are very low for the periods before Fall semester :).

Guiding Questions:

Also note that the remainder of the year, from November 21 through December, is just assumed to be constant.

Read Data

# Load data
t <- "20201227_235052"
data.connectedness <- read.csv(paste("data/connectedness_", t, ".csv", sep="")) %>%
    mutate(Housing.F20=factor(Housing.F20, c("Stiles", "Old Campus", "Off-Campus", "Remote"))) %>%
    mutate(Enrolled.F20=case_when(
        Enrolled.F20 ~ "Enrolled",
        !Enrolled.F20 ~ "Not Enrolled"
    )) %>%
    mutate(Year=as.factor(Year))
# "Jitter" values just a little bit, so it's easier to see overlapping points
c_jitter <- data.connectedness %>%
    mutate(Day=jitter(Day, factor=1)) %>%  # x
    mutate(Connectedness=jitter(Connectedness, factor=.5))  # y

First Look: Small Multiples

g_all <- (
    ggplot(c_jitter %>% filter(!is.na(Housing.F20) & !is.na(Enrolled.F20) & !is.na(Year)),
           aes(x=Day, y=Connectedness, color=Year, group=ID)) +
    geom_line(alpha=0.5) +
    geom_point(alpha=0.6) +
    scale_y_continuous(breaks=1:10) +
    scale_x_continuous(breaks=unique(data.connectedness$Day)) +
    scale_color_colorblind() +
    facet_grid(Housing.F20 ~ Enrolled.F20) +
    theme_minimal()
) %>% ggplotly()
g_all

Figure 1: Vertical columns represent enrollment status in Fall 2020, horizontal rows represent proximity to campus in Fall 2020.

Note that you can selectively view graduation years, or groups of graduation years, by clicking on the legend above.

Focus on the Fall semester of 2020: what’s going on in Day 327?

Timeline review:

Epoch (t=value) Time frame Date range (2020) Day of Year (Starting)
t1 Before Spring Break < 3/7 13
t2 Spring Break 3/7 - 3/22 67
t3 Remainder of the Spring semester 3/23 - 5/6 83
t4 Summer 5/7 - Late August 128
t5 Fall semester, before Break 8/31 - 11/21 244
t6 Remainder of the Fall semester 11/22 - 2020 327

Deep Dives

On summary statistics (means and standard errors).

Graduation Year

data.summ.yr <- data.connectedness %>%
    filter(!is.na(Housing.F20) & !is.na(Enrolled.F20) & !is.na(Year)) %>%
    group_by(Year, Day) %>%
    summarize(mean_se(Connectedness), N..Responses=n()) %>%
    mutate(Connectedness=y)

dodge <- position_dodge(12)
g <- ggplot(data.summ.yr, aes(x=Day, y=Connectedness, color=Year, text=N..Responses)) +
    geom_line(position=dodge) +
    geom_errorbar(aes(ymin=ymin, ymax=ymax), position=dodge, width=10) +
    scale_y_continuous(breaks=1:10) +
    scale_x_continuous(breaks=unique(data.connectedness$Day)) +
    scale_color_colorblind() +
    theme_minimal()
ggplotly(g)

Note that setting tooltips with geom_line is funky.1

Proximity to Campus

data.summ.yr <- data.connectedness %>%
    filter(!is.na(Housing.F20) & !is.na(Enrolled.F20) & !is.na(Year)) %>%
    group_by(Housing.F20, Day) %>%
    summarize(mean_se(Connectedness), N..Responses=n()) %>%
    mutate(Connectedness=y)

dodge <- position_dodge(12)
g <- ggplot(data.summ.yr, aes(x=Day, y=Connectedness, color=Housing.F20, text=N..Responses)) +
    geom_line(position=dodge) +
    geom_errorbar(aes(ymin=ymin, ymax=ymax), position=dodge, width=10) +
    scale_y_continuous(breaks=1:10) +
    scale_x_continuous(breaks=unique(data.connectedness$Day)) +
    scale_color_colorblind() +
    theme_minimal()
ggplotly(g)

Enrollment Status

data.summ.yr <- data.connectedness %>%
    filter(!is.na(Housing.F20) & !is.na(Enrolled.F20) & !is.na(Year)) %>%
    group_by(Enrolled.F20, Day) %>%
    summarize(mean_se(Connectedness), N..Responses=n()) %>%
    mutate(Connectedness=y)

dodge <- position_dodge(12)
g <- ggplot(data.summ.yr, aes(x=Day, y=Connectedness, color=Enrolled.F20, text=N..Responses)) +
    geom_line(position=dodge) +
    geom_errorbar(aes(ymin=ymin, ymax=ymax), position=dodge, width=10) +
    scale_y_continuous(breaks=1:10) +
    scale_x_continuous(breaks=unique(data.connectedness$Day)) +
    scale_color_colorblind() +
    theme_minimal()
ggplotly(g)

Granular Views

The following plots offer a more granular look at Stiles-connectivity through time from the perspective of each of the following factors: graduation year, proximity to campus in the Fall, and enrollment status in the Fall.

g_year <- (
    ggplot(c_jitter %>% filter(!is.na(Year)),
           aes(x=Day, y=Connectedness, color=Year, group=ID)) +
    geom_line(alpha=0.25) +
    geom_point(alpha=0.35) +
    scale_y_continuous(breaks=1:10) +
    scale_x_continuous(breaks=unique(data.connectedness$Day)) +
    scale_color_colorblind() +
    theme_minimal()
) %>% ggplotly()

g_proximity <- (
    ggplot(c_jitter %>% filter(!is.na(Housing.F20)),
            aes(x=Day, y=Connectedness, color=Housing.F20, group=ID)) +
    geom_line(alpha=0.15) +
    geom_point(alpha=0.35) +
    scale_y_continuous(breaks=1:10) +
    scale_x_continuous(breaks=unique(data.connectedness$Day)) +
    scale_color_brewer(palette="Spectral") +
    theme_minimal()
) %>% ggplotly()

g_enrollment <- (
    ggplot(c_jitter %>% filter(!is.na(Enrolled.F20)),
           aes(x=Day, y=Connectedness, color=Enrolled.F20, shape=Enrolled.F20, group=ID)) +
    geom_line(alpha=0.25) +
    geom_point(alpha=0.35) +
    scale_y_continuous(breaks=1:10) +
    scale_x_continuous(breaks=unique(data.connectedness$Day)) +
    scale_color_colorblind() +
    theme_minimal()
) %>% ggplotly()

# https://stackoverflow.com/questions/36227482/r-ggplot-dodging-geom-lines
# https://ggplot2.tidyverse.org/reference/scale_brewer.html

# subplot(g_year, g_proximity, g_enrollment, shareX=T, nrows=3)
g_year

g_proximity

g_enrollment

  1. https://github.com/ropensci/plotly/issues/1153↩︎