[jordan.trachtenberg]

In this cross-tab below, how can we order the Likert scale (Excellent, Vgood, Good, Fair, Poor) instead of having it in ascending alphabetical order?

nhanes %>% 
    drop_na(gender, health_gen) %>% 
    tabyl(gender, health_gen) %>% 
    adorn_totals(where = c("row", "col")) %>% 
    adorn_percentages() %>% 
    adorn_pct_formatting(digits = 1) %>% 
    adorn_ns()

[diego.catalan molina]

Have you tried fct_reorder? See here some examples. In general, I tend to reorder factors (e.g., dichotomous, categorical vars) before pushing them into a table or plot to avoid cleaning variables in the table/plot.

The most basic way of formatting factors is through the factor() function. E.g., factor(health_gen, levels = c("Poor", "Fair", "Good", "Vgood", "Excellent")).

[jordan.trachtenberg]

Thanks, @diego-catalan-molina for your suggestions. I’m not sure I understand the concept of factors to apply these solutions correctly. When I try

nhanes %>% 
    drop_na(gender, health_gen) %>% 
    factor(health_gen, levels = c("Poor", "Fair", "Good", "Vgood", "Excellent")) %>% 
    tabyl(gender, health_gen) %>% 
    adorn_totals(where = c("row", "col")) %>% 
    adorn_percentages() %>% 
    adorn_pct_formatting(digits = 1) %>% 
    adorn_ns()

I get this error:

Error in factor(., health_gen, levels = c(“Poor”, “Fair”, “Good”, “Vgood”, : object ‘health_gen’ not found

When I try

nhanes %>% 
    drop_na(gender, health_gen) %>% 
    fct_reorder(health_gen, levels = c("Poor", "Fair", "Good", "Vgood", "Excellent")) %>% 
    tabyl(gender, health_gen) %>% 
    adorn_totals(where = c("row", "col")) %>% 
    adorn_percentages() %>% 
    adorn_pct_formatting(digits = 1) %>% 
    adorn_ns()

I get this error:

Error: f must be a factor (or character vector).

Is it because I’m starting with the nhanes dataframe?

[diego.catalan molina]

Sorry @jordan-trachtenberg ! I should have been more clear. This is what you want:

nhanes %>%

mutate(health_gen_asfactor = #name of your new var

factor(health_gen, #original var

levels = c(“Poor”, “Fair”, “Good”, “Vgood”, “Excellent”)) #order you want to impose on categories

) %>% # add the rest of your pipe

• This reply was modified 6 months ago by  diego.catalan molina.

[jordan.trachtenberg]

Thank you, @diego-catalan-molina .

I got this to work for the cross-tab table:

nhanes %>%
    mutate(health_gen_asfactor = factor(health_gen, 
                                        levels = c("Poor", "Fair", "Good", "Vgood", "Excellent"))
           ) %>%
    drop_na(gender, health_gen) %>% 
    tabyl(gender, health_gen_asfactor) %>% 
    adorn_totals(where = c("row", "col")) %>% 
    adorn_percentages() %>% 
    adorn_pct_formatting(digits = 1) %>% 
    adorn_ns()

[David]

Thanks for jumping in, @diego-catalan-molina !

@jordan-trachtenberg just to clarify, are you trying to make a plot or reorder in a table (or table-like output)? I ask because the answer differs depending on what you want to do. Factors are for reordering in plots. The arrange() function works well when reordering in tables.

[jordan.trachtenberg]

For now, I’m trying to order the Likert scale in a table, but I could imagine that I would also need it for a plot in the future. I guess it’s good to know that those preferences require separate functions!

[David]

I’d love to go over this in the office hours this Friday at 10am Pacific. Any chance you can join so we can discuss there?

[jordan.trachtenberg]

I will try to make it. Thanks!

[David]

Awesome, thanks!

[jordan.trachtenberg]

@dgkeyes I think I figured out the plot as well:

# Set up new data frame with desired Likert scale order
sleep_by_gender_andhealth <- nhanes %>% 
  mutate(ordered_healthgen = factor(health_gen, levels = c("Poor", "Fair", "Good", "Vgood", "Excellent"))) %>%
  group_by(gender, ordered_healthgen) %>% 
  drop_na(gender, health_gen) %>%
  summarize(mean_hrs_sleep = round(mean(sleep_hrs_night, 
                                        na.rm = TRUE), 
                                   1))
# Create visualization with new data frame
sleep_by_gender_andhealth %>%
  ggplot(mapping = aes(x = ordered_healthgen,
                       y = mean_hrs_sleep,
                       fill = gender)) +
  geom_bar(stat = "identity") +
  coord_flip()

[David]

Great!