diff --git a/docs/chapter_2.html b/docs/chapter_2.html index 6e351b4..0250e63 100644 --- a/docs/chapter_2.html +++ b/docs/chapter_2.html @@ -2,12 +2,12 @@ - + -Hacking Religion: TRS & Data Science in Action - 2  Survey Data: Spotlight Project +Hacking Religion: TRS & Data Science in Action - 2  Getting into the nitty-gritty details + @@ -110,8 +91,7 @@ div.csl-indent { "search-more-matches-text": "more matches in this document", "search-clear-button-title": "Clear", "search-detached-cancel-button-title": "Cancel", - "search-submit-button-title": "Submit", - "search-label": "Search" + "search-submit-button-title": "Submit" } } @@ -127,10 +107,10 @@ div.csl-indent { - + - @@ -161,13 +141,13 @@ div.csl-indent {
  • - 1  The 2021 UK Census + 1  Preamble
  • - 2  Survey Data: Spotlight Project + 2  Getting into the nitty-gritty details
  • @@ -210,17 +190,23 @@ div.csl-indent {

    Table of contents

    @@ -229,7 +215,7 @@ div.csl-indent {
    -

    2  Survey Data: Spotlight Project

    +

    2  Getting into the nitty-gritty details

    @@ -244,44 +230,50 @@ div.csl-indent {
    -

    In the last chapter we explored some high level data about religion in the UK. This was a census sample, which usually refers to an attempt to get as comprehensive a sample as possible. But this is actually fairly unusual in practice. Depending on how complex a subject is, and how representative we want our data to be, it’s much more common to use selective sampling, that is survey responses at n=100 or n=1000 at a maximum. The advantage of a census sample is that you can explore how a wide range of other factors - particularly demographics - intersect with your question. And this can be really valuable in the study of religion, particularly as you will see as we go along that responses to some questions are more strongly correlated to things like economic status or educational attainment than they are to religious affiliation. It can be hard to tell if this is the case unless you have enough of a sample to break down into a number of different kinds of subsets. But census samples are complex and expensive to gather, so they’re quite rare in practice.

    -

    For this chapter, I’m going to walk you through a data set that a colleague (Charles Ogunbode) and I collected in 2021. Another problem with smaller, more selective samples is that researchers can often undersample minoritised ethnic groups. This is particularly the case with climate change research. Until the time we conducted this research, there had not been a single study investigating the specific experiences of people of colour in relation to climate change in the UK. Past researchers had been content to work with large samples, and assumed that if they had done 1000 surveys and 50 of these were completed by people of colour, they could “tick” the box. But 5% is actually well below levels of representation in the UK generally, and even more sharply the case for specific communities. And if we bear in mind that non-white respondents are (of course!) a highly heterogenous group, we’re even more behind in terms of collecting data that can improve our knowledge. Up until recently researchers just haven’t been paying close enough attention to catch the significant neglect of the empirical field that this represents.

    +

    In this chapter, we’ll explore the diverse variety of ways you can frame collecting data around religion. Before we dive into that all, however, you might be wondering, why does it all really matter? Can’t we just use the census data and assume that’s a reasonably accurate approximation? I’ll explore the importance of getting the framing right, or better yet, working with data that seeks to unpack religious belonging, identity, and beliefs (or unbelief) in a variety of ways, but an example might serve to explain why this is important.

    +

    The 2016 presidential election result in the USA came as a surprise to many data analysts and pollsters. As the dust settled, a number of analysis scrambled to make sense of things and identify some hidden factor that might have tipped the balance away from the expected winner Hilary Clinton. One of the most widely circulated data points was the role of white evangelical Christians in supporting Trump. Exit polls reported that 81% of this constituency voted for Trump and many major media outlets reported this figure prominently, with public commentary from many religious leaders on the meaning this figure had the social direction of evangelical Christianity.

    +

    Far too few observers paused to ask what those exit polls were measuring and a closer look at that information reveals some interesting nuances. There is only a single firm that runs exit polling in the USA, Edison Research, who is contracted to do this work by a consortium of major media news outlets (“the National Election Pool”), which represents ABC News, Associated Press, CBS News, CNN, Fox News, and NBC News. It’s not a process driven by slow, nuanced, scholarly researchers strapped for funding, it’s a rapid high-stakes data collection exercise meant to provide data which can feed into the election week news cycle. The poll doesn’t ask respondents simply if they are “evangelical” it uses a broader proxy question to do this: “Would you describe yourself as a born-again or evangelical Christian?” This term “born-again” can be a useful proxy, but it can also prove misleading. When asked if they are “born again” people who identify with a number of non-Christian religions, and people who might describe themselves as non-religious will also often answer “yes”. This is particularly salient given the 2016 exit survey asked this question before asking specifically what a person’s religion was, so as Pew Research reported, “everyone who takes the exit poll (including religious “nones” and adherents of non-Christian faiths) has the opportunity to identify as a born-again or evangelical Christian.”

    +

    While the “born-again” Christian category tends to correlate to high levels of attendance at worship services, in this case some researchers found that white protestant Christian voters for Trump tended to have low levels of participation in activities. We don’t have access to the underlying data, and ultimately the exit polling was quite limited in scope (in some instances respondents weren’t even asked about religion), so we’ll never really have a proper understanding of what happened demographically in that election. But it’s an interesting puzzle to consider how different ways to record participation in religion might fit together, or even be in tension with one another. For this chapter, we’re going to take a look at another dataset which gives us exactly this kind of opportunity, to see how different kinds of measurement might reinforce or relate with one another.

    +
    +

    3 Survey Data: Spotlight Project

    +

    In the last chapter we explored some high level data about religion in the UK. This was a census sample, which usually refers to an attempt to get as comprehensive a sample as possible. But this is actually fairly unusual in practice. Depending on how complex a subject is and how representative we want our data to be, it’s much more common to use selective sampling, that is survey responses at n=100 or n=1000 at a maximum. The advantage of a census sample is that you can explore how a wide range of other factors - particularly demographics - intersect with your question. And this can be really valuable in the study of religion, particularly as you will see as we go along that responses to some questions are more strongly correlated to things like economic status or educational attainment than they are to religious affiliation. It can be hard to tell if this is the case unless you have enough of a sample to break down into a number of different kinds of subsets. But census samples are complex and expensive to gather, so they’re quite rare in practice.

    +

    For this chapter, I’m going to walk you through a data set that a colleague (Charles Ogunbode) and I collected in 2021. Another problem with smaller, more selective samples is that researchers can often undersample minoritised ethnic groups. This is particularly the case with climate change research. Until the time we conducted this research, there had not been a single study investigating the specific experiences of people of colour in relation to climate change in the UK. Past researchers had been content to work with large samples, and assumed that if they had done 1000 surveys and 50 of these were completed by people of colour, they could “tick” the box. But 5% is actually well below levels of representation in the UK generally, and even more sharply the case for specific communities and regions in the UK. And if we bear in mind that non-white respondents are (of course!) a highly heterogenous group, we’re even more behind in terms of collecting data that can improve our knowledge. Up until recently researchers just haven’t been paying close enough attention to catch the significant neglect of the empirical field that this represents.

    While I’ve framed my comments above in terms of climate change research, it is also the case that, especially in diverse societies like the USA, Canada, the UK etc., paying attention to non-majority groups and people and communities of colour automatically draws in a strongly religious sample. This is highlighted in one recent study done in the UK, the “Black British Voices Report” in which the researchers observed that “84% of respondents described themselves as religious and/or spiritual”. My comments above in terms of controlling for other factors remains important here - these same researchers also note that “despire their significant important to the lives of Black Britons, only 7% of survey respondents reported that their religion was more defining of their identity than their race”.

    We’ve decided to open up access to our data and I’m highlighting it in this book because it’s a unique opportunitiy to explore a dataset that emphasises diversity from the start, and by extension, provides some really interesting ways to use data science techniques to explore religion in the UK.

    -
    -

    2.1 Loading in some data

    +
    +
    +

    4 Loading in some data

    +

    The first thing to note here is that we’ve drawn in a different type of data file, this time from an .sav file, which is usually produced by the statistics software package SPSS. This uses a different R Library (I use haven for this, which is now included as part of the tidyverse). The upside is that in some cases where you have survey data which includes a code like “very much agree” which corresponds to a value like “1” this package will preserve both the value and the text in the R dataframe that is created. This can be useful as there will be cases where, for the sake of analysis, you want the numeric values, and in other cases, for the sake of visualisation, you want the tidy names. It’s a sort of “have your cake and eat it too” situation!

    +

    Here’s how you can load in the sample data I’ve provided for this chapter:

    # R Setup -----------------------------------------------------------------
 setwd("/Users/kidwellj/gits/hacking_religion_textbook/hacking_religion")
 library(here)  |> suppressPackageStartupMessages()
 library(tidyverse)  |> suppressPackageStartupMessages()
-# used for importing SPSS .sav files
-library(haven)   |> suppressPackageStartupMessages()
-here::i_am("chapter_2.qmd")
    +library(haven) +here::i_am("chapter_2.qmd")
    here() starts at /Users/kidwellj/gits/hacking_religion_textbook/hacking_religion
    climate_experience_data <- read_sav(here("example_data", "climate_experience_data.sav"))
    -

    The first thing to note here is that we’ve drawn in a different type of data file, this time from an .sav file, usully produced by the statistics software package SPSS. This uses a different R Library (I use haven for this). The upside is that in some cases where you have survey data with both a code and a value like “1” is eqivalent to “very much agree” this will preserve both in the R dataframe that is created. Now that you’ve loaded in data, you have a new R dataframe called “climate_experience_data” with a lot of columns with just under 1000 survey responses.

    +

    Now that you’ve loaded in data, you have a new R dataframe called “climate_experience_data” with a lot of columns with just under 1000 survey responses.

    -
    -

    2.2 How can you ask about religion?

    -

    One of the challenges we faced when running this study is how to gather responsible data from surveys regarding religious identity. We’ll dive into this in depth as we do analysis and look at some of the agreements and conflicts in terms of respondent attribution. Just to set the stage, we used the following kinds of question to ask about religion and spirituality:

    -
    -

    2.2.1 “What is your religion?”

    -
      -
    1. Question 56 asks respondents simply, “What is your religion?” and then provides a range of possible answers. We included follow-up questions regarding denomination for respondents who indicated they were “Christian” or “Muslim”. For respondents who ticked “Christian” we asked, “What is your denomination?” nad for respondents who ticked “Muslim” we asked “Which of the following would you identify with?” and then left a range of possible options which could be ticked such as “Sunni,” “Shia,” “Sufi” etc.
      2. -
    -

    This is one way of measuring religion, that is, to ask a person if they consider themselves formally affiliated with a particular group. This kind of question has some (serious) limitations, but we’ll get to that in a moment.

    +
    +

    5 How can you ask about religion?

    +

    One of the challenges we faced when running this study is how to gather responsible data from surveys regarding religious identity. As we already hinted in the last chapter, this is one of the key problems we explore in this chapter. We’ll dive into this in depth as we do analysis and look at some of the agreements and conflicts in terms of respondent attribution. Just to set the stage, for the project spotlight dataset we used the following kinds of question to ask about religion and spirituality:

    +
    +

    5.1 “What is your religion?”

    +

    The first, and perhaps most obvious question (Question 56 in the dataset) asks respondents simply, “What is your religion?” and then provides a range of possible answers. We included follow-up questions regarding denomination for respondents who indicated they were “Christian” or “Muslim”. For respondents who ticked “Christian” we asked, “What is your denomination?” and for respondents who ticked “Muslim” we asked “Which of the following would you identify with?” and then left a range of possible options which could be ticked such as “Sunni,” “Shia,” “Sufi” etc.

    +

    This is one way of measuring religion, that is, to ask a person if they consider themselves formally affiliated with a particular group. This kind of question has some limitations, but we’ll get to that in a moment.

    -
    -

    2.2.2 “How religious would you say you are?”

    -

    We also asked respondents (Q57): “Regardless of whether you belong to a particular religion, how religious would you say you are?” and then provided a slider from 0 (not religious at all) to 10 (very religious).

    +
    +

    5.2 “How religious would you say you are?”

    +

    We also asked respondents (Q57): “Regardless of whether you belong to a particular religion, how religious would you say you are?” and then provided a sliding scale from 0 (not religious at all) to 10 (very religious). Seen in this way, we had a tradition-neutral measurement of religious intensity.

    -
    -

    2.2.3 Participation in Worship

    -

    We included some classic indicators about how often respondents go to worship (Q58): Apart from weddings, funerals and other special occasions, how often do you attend religious services? and (Q59): “Q59 Apart from when you are at religious services, how often do you pray?”

    +
    +

    5.3 Social and personal participation in activity

    +

    We included another classic indicator asking how often respondents go to worship (Q58): “Apart from weddings, funerals and other special occasions, how often do you attend religious services?”. The individual counterpart to this question about social participation came next in the form of (Q59): “Apart from when you are at religious services, how often do you pray?” As with the previous question, the answers here also came in an descending scale of intensity:

    • More than once a week (1)
    • Once a week (2)
      • @@ -289,12 +281,12 @@ div.csl-indent {
    • Only on special holy days (4)
    • Never (5)
    -

    Each of these measures a particular kind of dimension, and it is interesting to note that sometimes there are stronger correlations between how often a person attends worship services (weekly versus once a year) and a particular view, than there is between their affiliation (if they are Christian or Pagan). We’ll do some exploratory work shortly to see how this is the case in our sample.

    +

    Do note the descending order here, which is different from the ascending scale for most other questions. This becomes relevant later when we explore correlations across questions. As we’ll note later on, each of these measures a particular kind of dimension, and it is interesting to note that sometimes there are stronger correlations between how often a person attends worship services (weekly versus once a year) and a particular view (in the case of our survey on environmental issues), than there is between their affiliation (if they are Christian or Pagan). We’ll do some exploratory work shortly to see how this is the case in our sample.

    -
    -

    2.2.4 Spirituality

    -

    We also included a series of questions about spirituality in Q52 and used a slightly overlapping nature relatedness scale Q51.

    -

    You’ll find that many surveys will only use one of these forms of question and ignore the rest. I think this is a really bad idea as religious belonging, identity, and spirituality are far too complex to work off a single form of response. We can also test out how these different attributions relate to other demographic features, like interest in politics, economic attainment, etc.

    +
    +

    5.4 Spirituality

    +

    We also included a series of questions about spirituality in Q52 and used a slightly overlapping nature relatedness scale Q51 which we’ll unpack a bit further below. There are many other types of question you can ask. In fact, in my teaching, one of my favourite exercises is to ask a student group to brainstorm as many ways as possible to ask a person about their religion whilst using a different word for religion in each question. We’ve managed to come up with dozens, possibly hundreds over the year, exploring faith, ritual, spirituality, transcendence, connection, belief, unbelief, sacredness and more. The key thing is that these questions are not directly interchangeable, but they will almost inevitably overlap. If you want to make constructive claims about how religion relates to some aspect of daily life, you will need to carefully consider how you can relate to this plurality in framing everyday experience. In the best case scenario, I think, you should find ways to capture a variety of dimensions and then test for correlations and clusters among your data. We’ll do some exploration further below so you can see a bit of what I mean.

    +

    You’ll find that many surveys will only use one of these forms of question and ignore the rest. I think this is a really bad idea as religious belonging, identity, and spirituality are far too complex to work off a single form of response. We can also test out how these different attributions relate to other demographic features, like interest in politics, economic attainment, etc. so it’s equally important to test for non-religion factors that may have a stronger bearing on someone’s actions or sentiments.

    @@ -306,26 +298,27 @@ So Who’s Religious?

    As I’ve already hinted in the previous chapter, measuring religiosity is complicated. I suspect some readers may be wondering something like, “what’s the right question to ask?” here. Do we get the most accurate representation by asking people to self-report their religious affiliation? Or is it more accurate to ask individuals to report on how religious they are? Is it, perhaps, better to assume that the indirect query about practice, e.g. how frequently one attends services at a place of worship may be the most reliable proxy?

    -

    Highlight challenges of various approaches pointing to literature.

    +

    In the past scholars have worked with a minimalist definition of religion, e.g. measuring only those people who participate in worship services with a high level of frequency, or demonstrate a high level of commitment to a range of pre-determined doctrinal positions or beliefs. This relates to a suspicion which was popular in the 20th century, that with the advent of modernity, religion would naturally decline. This has not proven to be the case, with a range of resurgence and transformation of “old” religions and a similar multiplication of new religious and spiritual movements. Scholars tend to refer to this awareness as relating to a post-secular study of religion, and this kind of study tends to be more maximal in orientation, finding religion, belief, and spirituality in a variety of unexpected forms and places (like football, cooking, capitalism, and popular culture). Scholars here also emphasise the ways that religion can be hidden or “tacit,” and may also be non-exclusive, with individual persons adhering to a range of religious traditions in more creative forms of appropriation. We find Christian animists and spiritual atheists, and doctrinal positions which overlap and migrate. One place that scholars have found this to be widely the case is in contemporary belief in paranormal phenomena, which can transcend particular religious identities, and be quite widespread (over 80%) even in so-called advanced scientific societies.

    -
    -

    2.3 Exploring data around religious affiliation:

    -

    Let’s dive into the data and see how this all works out. We’ll start with the question 56 data, around religious affiliation:

    +
    +

    6 Exploring data around religious affiliation:

    +

    Let’s dive into the data and do some initial exploration to map out what we have in this survey. We’ll start with the question 56 data, around religious affiliation. As usual, we’ll begin by loading in some data:

    religious_affiliation <- as_tibble(as_factor(climate_experience_data$Q56))
 names(religious_affiliation) <- c("response")
 religious_affiliation <- filter(religious_affiliation, !is.na(response))
    -

    There are few things we need to do here to get the data into initial proper shape. This might be called “cleaning” the data:

    +

    There are few things we need to do here to get the data into initial proper shape. This is often referred to as “cleaning” the data:

      -
    1. Because we imported this data from an SPSS .sav file format using the R haven() library, we need to start by adapting the data into a format that our visualation engine ggplot can handle (a dataframe).
      2. +
    2. Because we imported this data from an SPSS .sav file format using the R haven() library, we need to start by adapting the data into a format that our visualation engine ggplot can handle (a dataframe).
    3. Next we’ll rename the columns so these names are a bit more useful.
    4. We need to omit non-responses so these don’t mess with the counting (these are NA in R)
    -

    If we pause at this point to view the data, you’ll see it’s basically just a long list of survey responses. What we need is a count of each unique response (or factor). This will take a few more steps:

    +

    As in the previous chapter, I’ve provided sample data here that needs a bit of work. This gives you a chance to see what this all looks like in practice, and offers some examples you can apply later to your own datasets.

    +

    If we pause at this point to view the data, you’ll see it’s basically just a long list of survey responses. What we need is a count of each unique response (these are sometimes called a factor). This will take a few more steps:

    religious_affiliation_sums <- religious_affiliation %>%  
 1  dplyr::count(response, sort = TRUE) %>%
@@ -336,15 +329,15 @@ So Who’s Religious?
 
    
 
    1
    
 
    
-First we generate new a dataframe with sums per category and
+First we generate new a dataframe with sums per category and
 
    
 
    2
    
 
    
-…sort in descending order
+…sort in descending order
 
    
 
    3
    
 
    
-Then we add new column with percentages based on the sums you’ve just generated
+Then we add new column with percentages based on the sums you’ve just generated
    @@ -353,7 +346,7 @@ So Who’s Religious? 
    head(religious_affiliation_sums)
    -
    # A tibble: 6 x 3
+
    # A tibble: 6 × 3
   response                        n perc   
   <fct>                       <int> <chr>  
 1 Christian                     342 "33.9%"
@@ -364,6 +357,7 @@ So Who’s Religious?
 6 Spiritual but not religious    38 " 3.8%"
    +

    We can view this as a bar chart using ggplot in ways that are similar to the exercises in the last chapter:

    # make plot
 ggplot(religious_affiliation_sums, aes(x = n, y = response)) +
@@ -376,7 +370,7 @@ So Who’s Religious?
    -

    I’ve added one feature to our chart that wasn’t in the bar charts in chapter 1, text labels with the actual value on each bar.

    +

    You may notice that I’ve added one feature to our chart that wasn’t in the bar charts in chapter 1, text labels with the actual value on each bar using geom_text.

    You may be thinking about the plots we’ve just finished in chapter 1 and wondering how they compare. Let’s use the same facet approach that we’ve just used to render this data in a subsetted way.

    # First we need to add in data on ethnic self-identification from our respondents:
@@ -389,53 +383,57 @@ So Who’s Religious?
   dplyr::count(Religion, sort = TRUE) %>%
   dplyr::mutate(Religion = forcats::fct_rev(forcats::fct_inorder(Religion)))
 
-plot1 <- ggplot(religious_affiliation_ethnicity_sums, aes(x = n, y = Religion)) +
-  geom_col(colour = "white") + facet_wrap(~Ethnicity, scales="free_x")
-
-ggsave("chart.png", plot=plot1, width = 8, height = 10, units=c("in"))
    +religious_affiliation_ethnicity_plot <- ggplot(religious_affiliation_ethnicity_sums, aes(x = n, y = Religion)) + geom_col(colour = "white") + facet_wrap(~Ethnicity, scales = "free_x", labeller = label_wrap_gen(width = 24)) + theme(strip.text.x = element_text(size = 8)) + theme(strip.text.y = element_text(size = 6)) + +religious_affiliation_ethnicity_plot
    +
    +

    +
    ggsave("figures/spotlight_religious_affiliation_ethnicity.png", plot=religious_affiliation_ethnicity_plot, width = 8, height = 10, units=c("in"))
    +
    +

    You’ll notice that I’ve tweaked the display of facet titles a bit here so that the text wraps using labeller = label_wrap_gen(width = 24), since there are a lot of facets here, which are all interesting, I’ve also reduced the size of text for x- and y- axes using theme(strip.text.x = element_text().

    -
    -

    2.4 Working With a Continum: Religiosity and Spirituality

    +
    +

    7 Working With a Continum: Religiosity and Spirituality

    So far we’ve just worked with bar plots, but there are a lot of other possible visualisations and types of data which demand them.

    As I’ve mentioned above, on this survey we also asked respondents to tell us on by rating themselves on a scale of 0-10 with 0 being “not religious at all” and 10 being “very religious” in response to the question, “Regardless of whether you belong to a particular religion, how religious would you say you are?”

    We’ll recycle some code from our previous import to bring in the Q57 data:

    -
    religiosity <- as_tibble(as_factor(climate_experience_data$Q57_1))
-names(religiosity) <- c("response")
-religiosity <- filter(religiosity, !is.na(response))
-religiosity_sums <- religiosity %>%  
-1  dplyr::count(response) %>%
-  dplyr::mutate(response = forcats::fct_rev(forcats::fct_inorder(response)))
-religiosity_sums <- religiosity_sums %>% 
-  dplyr::mutate(perc = scales::percent(n / sum(n), accuracy = .1, trim = FALSE))
    +
    religiosity <- as_tibble(as_factor(climate_experience_data$Q57_1))
+names(religiosity) <- c("response")
+religiosity <- filter(religiosity, !is.na(response))
+religiosity_sums <- religiosity %>%  
+1  dplyr::count(response) %>%
+  dplyr::mutate(response = forcats::fct_rev(forcats::fct_inorder(response)))
+religiosity_sums <- religiosity_sums %>% 
+  dplyr::mutate(perc = scales::percent(n / sum(n), accuracy = .1, trim = FALSE))
    -
    1
    +
    1
    -Note: we have removed sort = TRUE in the above statement as it will enforce sorting the data by quantities rather than the factor order. It wouldn’t really make sense to plot this chart in the order of response. +Note: we have removed sort = TRUE in the above statement as it will enforce sorting the data by quantities rather than the factor order. It wouldn’t really make sense to plot this chart in the order of response.

    Now, let’s plot that data:

    -
    caption <- "Respondent Religiosity"
-ggplot(religiosity_sums, aes(x = response, y = n, color=response)) +
-1  geom_col(colour = "white", aes(fill = response)) +
-  ## get rid of all elements except y axis labels + adjust plot margin
-2  coord_flip() +
-  theme(plot.margin = margin(rep(15, 4))) +
-  labs(caption = caption)
    +
    caption <- "Respondent Religiosity"
+ggplot(religiosity_sums, aes(x = response, y = n, color=response)) +
+1  geom_col(colour = "white", aes(fill = response)) +
+  ## get rid of all elements except y axis labels + adjust plot margin
+2  coord_flip() +
+  theme(plot.margin = margin(rep(15, 4))) +
+  labs(caption = caption)
    -
    1
    +
    1
    -We’ve added colors, because colours are fun. +We’ve added colors, because colours are fun.
    -
    2
    +
    2
    -Also new here is coord_flip to rotate the chart so we have bars going horizontally +Also new here is coord_flip to rotate the chart so we have bars going horizontally
    @@ -443,519 +441,531 @@ So Who’s Religious?

    +
    +

    7.1 Quick excursus: making things pretty with themes

    Since we’re thinking about how things look just now, let’s play with themes for a minute. ggplot is a really powerful tool for visualising information, but it also has some quite nice features for making things look pretty.

    If you’d like to take a proper deep dive on all this theme stuff, R-Charts has a great set of examples showing you how a number of different theme packages look in practice, “R-Charts on Themes”.

    R has a number of built-in themes, but these are mostly driven by functional concerns, such as whether you might want to print your chart or have a less heavy look overall. So for example you might use theme_light() in the following way:

    -
    ggplot(religiosity_sums, aes(x = response, y = n, color=response)) +
-  geom_col(colour = "white", aes(fill = response)) +
-  ## get rid of all elements except y axis labels + adjust plot margin
-  coord_flip() +
-  theme(plot.margin = margin(rep(15, 4))) +
-  labs(caption = caption) +
-  theme_light()
    +
    ggplot(religiosity_sums, aes(x = response, y = n, color=response)) +
+  geom_col(colour = "white", aes(fill = response)) +
+  ## get rid of all elements except y axis labels + adjust plot margin
+  coord_flip() +
+  theme(plot.margin = margin(rep(15, 4))) +
+  labs(caption = caption) +
+  theme_light()

    You can also use additional packages like ggthemes() or hrbrthemes() so for example we might want to try the pander theme which has it’s own special (and very cheerful) colour palette.

    -
    library(ggthemes)  |> suppressPackageStartupMessages()
-ggplot(religiosity_sums, aes(x = response, y = n, color=response)) + geom_col(colour = "white", aes(fill = response)) + coord_flip() + theme(plot.margin = margin(rep(15, 4))) + labs(caption = caption) + 
-  theme_pander() +
-  scale_fill_pander()
    +
    library(ggthemes)  |> suppressPackageStartupMessages()
+ggplot(religiosity_sums, aes(x = response, y = n, color=response)) + geom_col(colour = "white", aes(fill = response)) + coord_flip() + theme(plot.margin = margin(rep(15, 4))) + labs(caption = caption) + 
+  theme_pander() +
+  scale_fill_pander()

    Or, you might try the well-crafted typgraphy from hbrthemes in the theme_ipsum_pub theme:

    +

    Note: this library will expect your system to have certain fonts installed and available for RStudio to use. You may want to run the following command to import your system fonts to R so they are available: extrafont::font_import(). This will take a bit of time to run. The package will also save fonts to a folder on your PC so you can install them if you don’t already have them, you can run import_public_sans() to get the path for these files and install in R.

    -
    library(hrbrthemes)  |> suppressPackageStartupMessages()
-ggplot(religiosity_sums, aes(x = response, y = n, color=response)) + geom_col(colour = "white", aes(fill = response)) + coord_flip() + theme(plot.margin = margin(rep(15, 4))) + labs(caption = caption) + 
-  theme_ipsum_pub() +
-  scale_fill_pander()
    +
    library(hrbrthemes)  |> suppressPackageStartupMessages()
+ggplot(religiosity_sums, aes(x = response, y = n, color=response)) + geom_col(colour = "white", aes(fill = response)) + coord_flip() + theme(plot.margin = margin(rep(15, 4))) + labs(caption = caption) + 
+  theme_ipsum_pub() +
+  scale_fill_pander()

    -

    We’re going to come back to this chart, but let’s set it to one side for a moment and build up a visualisation of the spirituality scale data. The spirituality scale questions come from research by ___ and ___. These researchers developed a series of questions which they asked respondents in a survey. The advantage here is that you’re getting at the question of spirituality from a lot of different angles, and then you combine the scores from all the questions to get a mean “spirituality score”.

    -
    -
    ### Spirituality scale  --------------------------------------------------------------
-# Calculate overall mean spirituality score based on six questions:
-climate_experience_data$Q52_score <- rowMeans(select(climate_experience_data, Q52a_1:Q52f_1))
    +
    +
    +
    +

    8 Spirituality

    +

    We’re going to come back to this data around religiosity, but let’s set it to one side for a moment and build up a visualisation of an adjacent measure we used in this study which focussed on spirituality.

    +
    +
    +
    +
    -

    Like we did in chapter 1, let’s start by exploring the data and get a bit of a sense of the character of the responses overall. One good place to start is to find out the mean response for our two continum questions:

    -
    -
    # t_testing and means
-
-# Spirituality scale
-
-# JK note to self: need to fix stat_summary plot here
-
-# stat_summary(climate_experience_data$Q52_score)
-mean(climate_experience_data$Q52_score)
    -
    -
    [1] 6.047454
    +
    +What is the difference between Spirituality and Religion?
    -
    # Q57 Regardless of whether you belong to a particular religion, how religious would you say you are?
-# 0-10, Not religious at all => Very religious; mean=5.58
-
-mean(climate_experience_data$Q57_1) # religiosity
    +
    +
    +

    Though the terms can tend to be used interchangeable in many cases, some scholars in religious studies and psychology have sought to develop the concept (and measurement of) spirituality as a counterpoint to religion. In some cases, scholars argue that religion is extrinsic (something outside us that we participate in) and spirituality is intrinsic (something inside ourselves that we engage with). Another way of contrasting the two concepts is to suggest that religion is social whereas spirituality is personal. As Hodge puts it, “spirituality refers to an individual’s relationship with God (or perceived Transcendence), while religion is defined as a particular set of beliefs, practices, and rituals that have been developed in community by people who share similar exis- tential experiences of transcendent reality.” Of course, as you’ll have noticed, there are many people who think of themselves as religious, but are opposed to participation in a formal religious tradition, or a social institution like a church, mosque, or denomination. So these differentiations can’t be sharply made in a conclusive way. And it’s likely that many respondents will have their own way to relate to these terms, whether it is affection or aversion.

    +
    +
    +

    For our study, we made use of a six-item intrinsic spirituality scale that was developed by David R. Hodge which is based on another instrument intended to measure “intrinsic religion” by Allport and Ross (1967). These researchers developed a series of questions which they asked respondents in a survey. The advantage here is that you’re getting at the question of spirituality from a lot of different angles and then you combine the scores from all the questions to get a mean “spirituality score”. There are many other ways that psychologists have developed to measure intrinsic religion or spirituality, and we’d encourage you to try them out (there are some references to get you started in Appendix B).

    +
    + +
    +
    +

    Content TBD.

    +
    +
    +
    +
    +
    ### Spirituality scale  --------------------------------------------------------------
+# Calculate overall mean spirituality score based on six questions:
+climate_experience_data$spirituality_score <- rowMeans(select(climate_experience_data, Q52a_1:Q52f_1))
+# Calculate overall mean nature relatedness score based on six questions:
+climate_experience_data$Q51_score <- rowMeans(select(climate_experience_data, Q51_heritage:Q51_remote_vacation))
    +
    +

    Like we did in chapter 1, let’s start by exploring the data and get a bit of a sense of the character of the responses overall. One good place to start is to find out the mean response for our two continum questions. We can start with religiosity:

    +
    +
    ### Calculating mean  --------------------------------------------------------------
+mean(climate_experience_data$Q57_1)
    [1] 5.581349
    -
    # Q58 Apart from weddings, funerals and other special occasions, how often do you attend religious services?
-# coded at 1-5, lower value = stronger mean=3.439484
-
-mean(climate_experience_data$Q58) # service attendance
    -
    -
    [1] 3.439484
    -
    # Q59 Apart from when you are at religious services, how often do you pray?
-# coded at 1-5, lower = stronger mean=2.50496
-
-mean(climate_experience_data$Q59)
    -
    -
    [1] 2.50496
    -
    -
    -

    Now let’s try out some visualisations:

    +

    Now let’s compare this with the overall mean score for our whole survey pool around spirituality:

    -
    ## Q52 Spirituality data ------------------------
-
-q52_data <- select(climate_experience_data, Q52a_1:Q52f_1)
-# Data is at wide format, we need to make it 'tidy' or 'long'
-q52_data <- q52_data %>% 
-  gather(key="text", value="value") %>%
-  # rename columns
-  mutate(text = gsub("Q52_", "",text, ignore.case = TRUE)) %>%
-  mutate(value = round(as.numeric(value),0))
    +
    mean(climate_experience_data$spirituality_score)
    +
    +
    [1] 6.047454
    +
    +
    +

    So they’re pretty close, but there’s a bit of a contrast between the responses these two measures, with our cohort measuring a bit higher on spirituality than religiosity.

    +

    This is quite a blunt measure, telling us how the whole average of all the responses compares in each case. But what is the relationship between these two measures for each individual? To find out more about this, we need to explore the correlation between points. We’ll talk about correlation analysis in a little bit, but I think it can be helpful to get ourselves back to thinking about our data as consisting of hundreds of tiny points all of which relate to a specific person who provided a range of responses.

    +

    Now let’s try out some visualisations, staring with the religiosity data.

    +
    +
    ### Plotting religiosity  --------------------------------------------------------------
+ggplot(climate_experience_data, aes(x = 1, y = Q57_1)) +
+  geom_point() +
+  labs(x = NULL, y = "Q57_1")
    +
    +

    +
    +
    +

    This is pretty disappointing, as ggplot doesn’t know what to do with the x-axis as our points are 1-dimensional, e.g. they only have one value. But it’s easy to fix! You can ask R to add random numbers for the x-axis so that we can see more of the dots and they aren’t overlapping. This is called jitter:

    +
    +
    ggplot(climate_experience_data, aes(x = 1, y = Q57_1)) +
+  geom_point(position = position_jitter(width = 0.1)) +
+  labs(x = NULL, y = "Q57_1") + theme(axis.text.x = element_blank())
    +
    +

    +
    +
    +

    You’ll also notice that we’ve hidden the x-axis value labels as these are just random numbers and not really something we want to draw attention to. We’ve also hidden the label for that axis.

    +

    Since this is quite a large plot, I’d recommend going one step further and making the dots a bit smaller, and a bit transparent (this is called “alpha” in R). The advantage of this is that we’ll be able to tell visually when dots are overlapping and register that there is a cluster. When they’re all the same black color, this is impossible to tell.

    +
    +
    ggplot(climate_experience_data, aes(x = 1, y = Q57_1)) +
+  geom_point(position = position_jitter(width = 1), color="black", size=0.5, alpha=0.3) +
+  labs(x = NULL, y = "Q57_1") + theme(axis.text.x = element_blank())
    +
    +

    +
    +
    +

    That’s a bit better. And we can start to see the weight of points hovering just over a value of 5, which aligns with our observation of the overall mean for this column of data a bit earlier in the exercise. But let’s say we’d like to be able to see this in an even more explicit way using a modification of the jitterplot with additional visual elements showing us where the mean is located. One example of this is called a boxplot:

    +
    +
    ggplot(climate_experience_data, aes(x = 1, y = Q57_1)) +
+  geom_boxplot(color = "black", fill = "lightblue", alpha = 0.7) +
+  labs(x = NULL, y = "Q57_1") + coord_flip() + theme(axis.text.y = element_blank())
    +
    +

    +
    +
    +

    I’ve flipped this chart on its side using coord_flip() because I just feel like these plot are easier to read from left to right. I also needed to adjust the concealment of labels to the y-axis.

    +

    The boxplot show us two things: the mean for the overall data using the black vertical line, and then the interquartile range (the boxes extend to the minimum and maximum values within 1.5 times of the IQR). This is helpful for us to see because, while the mean of all the values is a bit further to the right, the points we have to the left of the mean are more widely distributed.

    +
    + +
    +
    +

    Content TBD.

    +
    +
    +
    +

    I think it would be nice if we could see all the points on our chart with the boxes as you can really see how this is the case, and that’s not hard to do. We can also add a theme to make the points stand out a bit more:

    +
    +
    ggplot(climate_experience_data, aes(x = 1, y = Q57_1)) +
+  geom_boxplot(color = "black", fill = "lightgreen", alpha = 0.7) +
+  geom_jitter(color = "black", alpha = 0.3) +
+  labs(x = NULL, y = "Q57_1") + theme_ipsum() +
+  theme(axis.text.y = element_blank()) + coord_flip()
    +
    +

    +
    +
    +

    Let’s set the religiosity data to one side and look at the spirituality scale data. I’ve mentioned before that this dataset takes a set of six questions and then averages them out. It might be useful to start out by visualising each of these six separately, sticking with our jittered points-on-boxplot format for the sake of exploration. Let’s start by gathering our data:

    +
    +
    spirituality_combined <- select(climate_experience_data, Q52a_1:Q52f_1)
    +
    +

    Here we hit an aspect of ggplot that is really important to appreciate. This library doesn’t want to work with more than two columns, so if we are introducing a third layer of complexity (e.g. answers from different questions) we need to reformat the data for ggplot. The tools to do this are a core part of the tidyverse() library and the usual terminology here is to refer to “wide” data which needs to be converted to “tidy” (thus “tidyverse” for all these tools that love tidy data) or “long” data. This can be accomplished using a pretty quick operation using gather(). And we’ll follow that with a range of more typical data cleaning operations:

    +
    +
    spirituality_combined <- spirituality_combined %>% 
+  gather(key="text", value="value") %>%
+  mutate(text = gsub("Q52_", "",text, ignore.case = TRUE)) %>%
+  mutate(value = round(as.numeric(value),0)) # [1]
    Warning: attributes are not identical across measure variables; they will be
 dropped
    -
    # Change names of rows to question text
-q52_data <- q52_data %>% 
-  gather(key="text", value="value") %>%
-  # rename columns
-  mutate(text = gsub("Q52a_1", "In terms of questions I have about my life, my spirituality answers...",text, ignore.case = TRUE)) %>%
-  mutate(text = gsub("Q52b_1", "Growing spiritually is important...",text, ignore.case = TRUE)) %>%
-  mutate(text = gsub("Q52c_1", "When I<e2><80><99>m faced with an important decision, spirituality plays a role...",text, ignore.case = TRUE)) %>%
-  mutate(text = gsub("Q52d_1", "Spirituality is part of my life...",text, ignore.case = TRUE)) %>%
-  mutate(text = gsub("Q52e_1", "When I think of things that help me grow and mature as a person, spirituality has an effect on my personal growth...",text, ignore.case = TRUE)) %>%
-  mutate(text = gsub("Q52f_1", "My spiritual beliefs affect aspects of my life...",text, ignore.case = TRUE))
-
-# Plot
-# Used for gradient colour schemes, as with violin plots
-library(viridis)
    -
    -
    Loading required package: viridisLite
    +
    spirituality_combined <- spirituality_combined %>% 
+  mutate(text = gsub("Q52a_1", "In terms of questions I have about my life, my spirituality answers...",text, ignore.case = TRUE)) %>%
+  mutate(text = gsub("Q52b_1", "Growing spiritually is important...",text, ignore.case = TRUE)) %>%
+  mutate(text = gsub("Q52c_1", "When I’m faced with an important decision, spirituality plays a role...",text, ignore.case = TRUE)) %>%
+  mutate(text = gsub("Q52d_1", "Spirituality is part of my life...",text, ignore.case = TRUE)) %>%
+  mutate(text = gsub("Q52e_1", "When I think of things that help me grow and mature as a person, spirituality has an effect on my personal growth...",text, ignore.case = TRUE)) %>%
+  mutate(text = gsub("Q52f_1", "My spiritual beliefs affect aspects of my life...",text, ignore.case = TRUE)) # [2]
    -
    q52_plot <- q52_data %>%
-  mutate(text = fct_reorder(text, value)) %>% # Reorder data
-  ggplot( aes(x=text, y=value, fill=text, color=text)) +
-  geom_boxplot() +
-  scale_fill_viridis(discrete=TRUE, alpha=0.8) +
-  geom_jitter(color="black", size=0.2, alpha=0.2) +
-  theme_ipsum() +
-  theme(legend.position="none", axis.text.y = element_text(size = 8)) +
-  coord_flip() + # This switch X and Y axis and allows to get the horizontal version
-  xlab("") +
-  ylab("Spirituality scales") +
-  scale_x_discrete(labels = function(x) str_wrap(x, width = 45))
-
-# using gridExtra to specify explicit dimensions for printing
-q52_plot
    -
    -

    -
    -
    ggsave("figures/q52_boxplot.png", width = 20, height = 10, units = "cm")
    -
    -

    There’s an enhanced version of this plot we can use, called ggstatsplot() to get a different view:

    +
      +
    1. Gather text into long format
      2. +
    2. Change names of rows to question text
      3. +
    -
    # As an alternative trying ggstatsplot:
-library(rstantools)
    -
    -
    This is rstantools version 2.3.1.1
    -
    -
    library(ggstatsplot)
    -
    -
    You can cite this package as:
-     Patil, I. (2021). Visualizations with statistical details: The 'ggstatsplot' approach.
-     Journal of Open Source Software, 6(61), 3167, doi:10.21105/joss.03167
    -
    -
    q52_plot_alt <- ggbetweenstats(
-  data = q52_data,
-  x = text,
-  y = value,
-  outlier.tagging  = TRUE,
-  title = "Intrinsic Spirituality Scale Responses"
-) +
-  scale_x_discrete(labels = function(x) str_wrap(x, width = 30)) +
-  # Customizations
-  theme(
-    # Change fonts in the plot
-    text = element_text(family = "Helvetica", size = 8, color = "black"),
-    plot.title = element_text(
-      family = "Abril Fatface", 
-      size = 20,
-      face = "bold",
-      color = "#2a475e"
-    ),
-    # Statistical annotations below the main title
-    plot.subtitle = element_text(
-      family = "Helvetica", 
-      size = 12, 
-      face = "bold",
-      color="#1b2838"
-    ),
-    plot.title.position = "plot", # slightly different from default
-    axis.text = element_text(size = 10, color = "black"),
-    axis.text.x = element_text(size = 7),
-    axis.title = element_text(size = 12),
-    axis.line = element_line(colour = "grey50"),
-    panel.grid.minor = element_blank(),
-    panel.grid.major.x = element_blank(),
-    panel.grid = element_line(color = "#b4aea9"),
-    panel.grid.major.y = element_line(linetype = "dashed"),
-    panel.background = element_rect(fill = "#fbf9f4", color = "#fbf9f4"),
-    plot.background = element_rect(fill = "#fbf9f4", color = "#fbf9f4")
-  )
    -
    -
    Scale for x is already present.
-Adding another scale for x, which will replace the existing scale.
    -
    -
    q52_plot_alt
    +
    spirituality_combined %>%
+  mutate(text = fct_reorder(text, value)) %>% # Reorder data
+  ggplot(aes(x=text, y=value, fill=text, color=text)) +
+  geom_boxplot() +
+  geom_jitter(color="black", size=0.2, alpha=0.2) +
+  theme_ipsum() +
+  theme(legend.position="none", axis.text.y = element_text(size = 8)) +
+  coord_flip() + # This switch X and Y axis and allows to get the horizontal version
+  xlab("") +
+  ylab("Spirituality scales") +
+  scale_x_discrete(labels = function(x) str_wrap(x, width = 45))
    -

    +

    -
    ggsave("figures/q52_plot_alt.png", width = 20, height = 12, units = "cm")
    -

    One thing that might be interesting to test here is whether spirituality and religiosity are similar for our respondents.

    -
    ggplot(climate_experience_data, aes(x=Q52_score, y=Q57_1)) + labs(x="Spirituality Scale Score", y = "How Religious?") +
-  geom_point(size=1, alpha=0.3) + geom_smooth(method="auto", se=TRUE, fullrange=FALSE, level=0.95)
    +
    # using gridExtra to specify explicit dimensions for printing
+ggsave("figures/spirituality_boxplot.png", width = 20, height = 10, units = "cm")
    +
    +

    We’ve done a pretty reasonable exploration of these two questions. Now it’s time to visualise how they correlate to one another. We’ll work with the combined spirituality scale score for this example, but you could just as easily work with individual elements. What we’re wondering, in particular, is whether whether spirituality and religiosity are similar for our respondents. You’ll see that in this chart, I’ve handled the geom_point styling separately for each point so that we can tell them apart.

    +
    +
    ggplot(climate_experience_data, aes(x = spirituality_score, y = Q57_1)) +
+  geom_point(aes(color = "x"), size = 1, alpha = 0.2, shape = 15) +
+  geom_point(aes(color = "y"), size = 1, alpha = 0.2, shape = 17) +
+  geom_smooth(method = "auto", se = TRUE, fullrange = FALSE, level = 0.95) +
+  labs(x = "Spirituality Scale Score", y = "Religiosity") +
+  scale_color_manual(values = c("x" = "red", "y" = "blue"))
    `geom_smooth()` using method = 'gam' and formula = 'y ~ s(x, bs = "cs")'
    -

    +

    -
    # using http://sthda.com/english/wiki/ggplot2-scatter-plots-quick-start-guide-r-software-and-data-visualization
-
-ggplot(climate_experience_data, aes(x=Q52_score, y=Q57_1)) +
-  labs(x="Spirituality Scale Score", y = "How Religious?") +
-  geom_point(size=1, alpha=0.3) + stat_density_2d(aes(fill = ..level..), geom="polygon", alpha=0.3)+
-  scale_fill_gradient(low="blue", high="red") +
-  theme_minimal()
    +
    +

    It may be helpful to add a few more visual elements to help someone understand this data. Let’s try adding a density plot:

    +
    +
    library(ggExtra)
+p <- ggplot(climate_experience_data, aes(x = spirituality_score, y = Q57_1)) +
+  geom_point(aes(color = "x"), size = 1, alpha = 0.2) +
+  geom_point(aes(color = "y"), size = 1, alpha = 0.2, shape = 17) +
+  labs(x = "Spirituality Scale Score", y = "Religiosity") +
+  scale_color_manual(values = c("x" = "red", "y" = "blue"))
+p_with_density <- ggMarginal(p, type = "histogram")
+p_with_density
    +
    +

    +
    +
    +

    As an alternative we can view this as a heatmap:

    +
    +
    ggplot(climate_experience_data, aes(x=spirituality_score, y=Q57_1)) +
+  labs(x="Spirituality Scale Score", y = "How Religious?") +
+  geom_point(size=1, alpha=0.3) + stat_density_2d(aes(fill = ..level..), geom="polygon", alpha=0.3)+
+  scale_fill_gradient(low="blue", high="red") +
+  theme_minimal()
    Warning: The dot-dot notation (`..level..`) was deprecated in ggplot2 3.4.0.
-i Please use `after_stat(level)` instead.
    +ℹ Please use `after_stat(level)` instead.
    -

    -
    -
    -

    Because the responses to these two questions, about spirituality and religiosity are on a continuum, we can also use them, like we did in previous charts, to subset other datasets. A simple way of doing this is to separate our respondents into “high,” “medium,” and “low” bins for the two questions. Rather than working with hard values, like assigning 0-3, 4-6 and 7-10 for low medium and high, we’ll work with the range of values that respondents actually chose. This is particularly appropriate as the median answer to these questions was not “5”. So we’ll use the statistical concept of standard deviation, which R can calculate almost magically for us, in the following way:

    -
    -
    # Create low/med/high bins based on Mean and +1/-1 Standard Deviation
-climate_experience_data <- climate_experience_data %>%
-  mutate(
-    Q52_bin = case_when(
-      Q52_score > mean(Q52_score) + sd(Q52_score) ~ "high",
-      Q52_score < mean(Q52_score) - sd(Q52_score) ~ "low",
-      TRUE ~ "medium"
-    ) %>% factor(levels = c("low", "medium", "high"))
-  )
-
-
-## Q57 subsetting based on Religiosity --------------------------------------------------------------
-climate_experience_data <- climate_experience_data %>%
-  mutate(
-    Q57_bin = case_when(
-      Q57_1 > mean(Q57_1) + sd(Q57_1) ~ "high",
-      Q57_1 < mean(Q57_1) - sd(Q57_1) ~ "low",
-      TRUE ~ "medium"
-    ) %>% factor(levels = c("low", "medium", "high"))
-  )
    -
    -

    As in the previous chapter, it’s useful to explore multiple factors when possible. So I’d like us to take the data about political affiliation to visualise alongside our religion and spirituality data. this will help us to see where effects are more or less significant and give us a point of comparison.

    -
    -
    ## Q53 subsetting based on Political LR orientation --------------------------------------------------------------
-# Generate low/med/high bins based on Mean and SD
-climate_experience_data <- climate_experience_data %>%
-  mutate(
-    Q53_bin = case_when(
-      Q53_1 > mean(Q53_1) + sd(Q53_1) ~ "high",
-      Q53_1 < mean(Q53_1) - sd(Q53_1) ~ "low",
-      TRUE ~ "medium"
-    ) %>% factor(levels = c("low", "medium", "high"))
-  )
    -
    -

    Now let’s use those bins to explore some of the responses about attitudes towards climate change:

    -
    -
    # Faceted plot working with 3x3 grid
-df <- select(climate_experience_data, Q52_bin, Q53_bin, Q57_bin, Q58)
-names(df) <- c("Q52_bin", "Q53_bin", "Q57_bin", "response")
-facet_names <- c(`Q52_bin` = "Spirituality", `Q53_bin` = "Politics L/R", `Q57_bin` = "Religiosity", `low`="low", `medium`="medium", `high`="high")
-facet_labeller <- function(variable,value){return(facet_names[value])}
-df$response <- factor(df$response, ordered = TRUE, levels = c("1", "2", "3", "4", "5"))
-df$response <- fct_recode(df$response, "More than once a week" = "1", "Once a week" = "2", "At least once a month" = "3", "Only on special holy days" = "4", "Never" = "5")
-df %>% 
-  # we need to get the data including facet info in long format, so we use pivot_longer()
-  pivot_longer(!response, names_to = "bin_name", values_to = "b") %>% 
-  # add counts for plot below
-  count(response, bin_name, b) %>%
-  group_by(bin_name,b) %>%
-  mutate(perc=paste0(round(n*100/sum(n),1),"%")) %>% 
-  # run ggplot
-  ggplot(aes(x = n, y = "", fill = response)) +
-  geom_col(position=position_fill(), aes(fill=response)) +
-  geom_text(aes(label = perc), position = position_fill(vjust=.5), size=2) +
-  scale_fill_brewer(palette = "Dark2", type = "qual") +
-  scale_x_continuous(labels = scales::percent_format()) +
-  facet_grid(vars(b), vars(bin_name), labeller=as_labeller(facet_names)) + 
-  labs(caption = caption, x = "", y = "") + 
-  guides(fill = guide_legend(title = NULL))
    -
    -

    -
    -
    ggsave("figures/q58_faceted.png", width = 30, height = 10, units = "cm")
    -
    - - -
    -
    -
    - -
    -
    -What is Religion? -
    -
    -
    -

    Content tbd

    -
    -
    -
    -
    -
    - -
    -
    -Hybrid Religious Identity -
    -
    -
    -

    Content tbd

    -
    -
    -
    -
    -
    - -
    -
    -What is Secularisation? -
    -
    -
    -

    Content tbd

    +

    -
    -

    References

    -