## Problemset Title ---- ## [ PROJ ] < Problem Set 5 > ## [ FILE ] < lastname_firstname_problemset5.R > ## [ AUTH ] < Your Name / GitHub handle > ## [ INIT ] < Date you started the file > ## 31 points (+1 bonus) ## Overview ---- # In this problem set, we will be working with the student list data that University of Illinois-Chicago # purchased from College Board. Specifically, we will use the list from one specific order, where # UI-Chicago filtered for all prospects who identified as American Indian or Alaska Native and scored # within a specified test score/GPA range. Here https://anyone-can-cook.github.io/rclass1/data/prospect_list/uic_order_487927.pdf # is the order summary file containing the detailed search criteria. # To this student list data, we have also merged in Census data on zip-code characteristics and NCES data # on high school characteristics for each prospect. Thus, some variables in the data are prospect-level # variables, while others are measured at the zip-code level or school level. These include characteristics # for the zip code the prospect lives in and characteristics for the high school which the prospect # attends -- those variables do not vary across prospects within the same zip-code or school. ## Data # In this task, we are analyzing the characteristics of prospective students who identified as American # Indian or Alaska Native when they took the SAT test. We analyze the ethnicity categories and race # categories these students selected, where these students live, and their intended major. With respect to # course learning goals, these analyses will help you practice processing across observations. From a # substantive perspective, quantitative analyses seldom focus on students who identify as American Indian # or Alaska Native, so the UI-Chicago student list purchase offers an opportunity to learn a little more # about these students. # A note on terms for race and ethnicity categories: This problem set uses categories adopted by the U.S. # Census. For example, the problem set uses "American Indian or Alaska Native" rather than the terms # "Native American" or "Indigenous" and uses the term "Hispanic" rather than "Latinx." ## Question 1: Loading library and data ---- # Load the `tidyverse` library. rm(list=ls()) library(tidyverse) # Use `load()` and `url()` to load the `list_native_df` dataframe from # https://github.com/anyone-can-cook/rclass1/raw/master/data/prospect_list/list_native_df.RData. load(url('https://github.com/anyone-can-cook/rclass1/raw/master/data/prospect_list/list_native_df.RData')) # Let's investigate the `list_native_df` dataframe. First, use `head()` and `glimpse()` to preview # the data. list_native_df %>% head() list_native_df %>% glimpse() # Q1.1 ---- # 1 point # For each of the following ethnicity variables, use the `count()` function to count its unique values: # `stu_cuban` # `stu_mexican` # `stu_puerto_rican` # `stu_other_hispanic` # `stu_non_hispanic` # `stu_ethnicity_no_response` # In an object named `q1_1`, assign a numeric vector with the number of students who answered `Y` # for each of the 6 variables above (the `n` value in the `Y` row of each count). # Like this: `q1_1 <- c(1,2,3,4,5,6)` # Note: the order of each count in the vector does not matter. q1_1 <- # Q1.2 ---- # 1 point # For each of the following race variables, use the `count()` function to count its unique values: # `stu_american_indian` # `stu_asian` # `stu_black` # `stu_native_hawaiian` # `stu_white` # `stu_race_no_response` # In an object named `q1_1`, assign a numeric vector with the number of students who answered `Y` # for each of the 6 variables above (the `n` value in the `Y` row of each count). # Like this: `q1_1 <- c(1,2,3,4,5,6)` # Note: the order of each count in the vector does not matter. list_native_df %>% count(stu_american_indian) list_native_df %>% count(stu_asian) list_native_df %>% count(stu_black) list_native_df %>% count(stu_native_hawaiian) list_native_df %>% count(stu_white) list_native_df %>% count(stu_race_no_response) q1_2 <- ## Question 2: Recreating College Board's aggregate race/ethnicity variable ---- # In the questionnaire that students fill out during the College Board exams, they are allowed to select # multiple ethnicity and race categories that they identify as. For example, a student who checks the box # for "Cuban" could also check the box for "Non-hispanic." Similarly, a student who checks the box for # "American Indian or Alaska Native" could also check the box for "Black." Here https://research.collegeboard.org/about-us/changes-to-race-ethnicity-reporting # are more details on how College Board defines their race and ethnicity data. # These College Board variables are based off of the U.S. Census variables, as defined here # https://www.census.gov/topics/population/race/about.html. The specific Census variables we use in our # dataset can be found here https://www.socialexplorer.com/data/ACS2019_5yr/metadata/?ds=ACS19_5yr&table=B03002. # College Board also reports the student's aggregate race/ethnicity per U.S. Department of Education # reporting guidelines, as defined here https://anyone-can-cook.github.io/rclass1/data/prospect_list/cb_race_ethnicity_fields.pdf # (see last page). This derived category allocates each student into 1 category. Below, we will recreate # this College Board variable (`race_cb`). # To do that, we will first create 0/1 indicators for each disaggregated race and ethnicity variable. For # example, we will create the 0/1 indicator variable `stu_hispanic_01`, whose value will be `1` if the # student identifies as hispanic and `0` otherwise. Then, these 0/1 indicators, along with a couple other # variables we create, will be used as input to recreate the `race_cb` variable. # Run the following code to create the new race/ethnicity categories. All code is provided for you. # WARNING: Running this big chunk of code a second time will give you an `error in mutate()` warning. # This is because the code creates new variables and deletes the source variables, so it cannot run a second time. # You can either leave it alone and ignore it, or re-load the original data to re-run it. list_native_df <- list_native_df %>% mutate( stu_hispanic_01 = case_when( (stu_cuban == 'Y' | stu_mexican == 'Y' | stu_puerto_rican == 'Y' | stu_other_hispanic == 'Y') ~ 1, (stu_non_hispanic == 'Y' & is.na(stu_cuban) & is.na(stu_mexican) & is.na(stu_puerto_rican) & is.na(stu_other_hispanic)) ~ 0 ), stu_cuban_01 = case_when( stu_cuban == 'Y' ~ 1, is.na(stu_cuban) & is.na(stu_ethnicity_no_response) ~ 0 ), stu_mexican_01 = case_when( stu_mexican == 'Y' ~ 1, is.na(stu_mexican) & is.na(stu_ethnicity_no_response) ~ 0 ), stu_puerto_rican_01 = case_when( stu_puerto_rican == 'Y' ~ 1, is.na(stu_puerto_rican) & is.na(stu_ethnicity_no_response) ~ 0 ), stu_other_hispanic_01 = case_when( stu_other_hispanic == 'Y' ~ 1, is.na(stu_other_hispanic) & is.na(stu_ethnicity_no_response) ~ 0 ), stu_american_indian_01 = case_when( stu_american_indian == 'Y' ~ 1, is.na(stu_american_indian) & is.na(stu_race_no_response) ~ 0 ), stu_asian_01 = case_when( stu_asian == 'Y' ~ 1, is.na(stu_asian) & is.na(stu_race_no_response) ~ 0 ), stu_black_01 = case_when( stu_black == 'Y' ~ 1, is.na(stu_black) & is.na(stu_race_no_response) ~ 0 ), stu_native_hawaiian_01 = case_when( stu_native_hawaiian == 'Y' ~ 1, is.na(stu_native_hawaiian) & is.na(stu_race_no_response) ~ 0 ), stu_white_01 = case_when( stu_white == 'Y' ~ 1, is.na(stu_white) & is.na(stu_race_no_response) ~ 0 ), race_ct = rowSums(dplyr::across(c(stu_american_indian_01, stu_asian_01, stu_black_01, stu_native_hawaiian_01, stu_white_01)), na.rm = TRUE), multi_race_01 = if_else(race_ct >= 2, 1, 0, missing = NULL), race_cb = case_when( is.na(stu_hispanic_01) == 1 | (stu_hispanic_01 == 0 & stu_race_no_response == 'Y') ~ 'no_response', (stu_american_indian_01 == 1 & multi_race_01 == 0 & stu_hispanic_01 == 0) ~ 'ai_an', (stu_asian_01 == 1 & multi_race_01 == 0 & stu_hispanic_01 == 0) ~ 'asian', (stu_black_01 == 1 & multi_race_01 == 0 & stu_hispanic_01 == 0) ~ 'black', (stu_hispanic_01 == 1) ~ 'hispanic', (stu_native_hawaiian_01 == 1 & multi_race_01 == 0 & stu_hispanic_01 == 0) ~ 'nh_pi', (stu_white_01 == 1 & multi_race_01 == 0 & stu_hispanic_01 == 0) ~ 'white', (multi_race_01 == 1 & stu_hispanic_01 == 0) ~ 'multi_race' ) ) %>% select(-stu_cuban, -stu_mexican, -stu_puerto_rican, -stu_other_hispanic, -stu_non_hispanic, -stu_american_indian, -stu_asian, -stu_black, -stu_native_hawaiian, -stu_white, -stu_ethnicity_no_response, -stu_race_no_response) # After adding the new variables, let's investigate the `list_native_df` dataframe again. Use `head()` # and `glimpse()` to preview the data. list_native_df %>% head() list_native_df %>% glimpse() # Q2.1 ---- # 1 point # Now let's look at the `race_cb` variable we just created, which puts each student into one # race/ethnicity category. Create a new object `race_cb_freq` that shows how many students # are in each `race_cb` category: # Use `count()` to get the count for each `race_cb` category # Use `arrange()` to sort by the count in descending order race_cb_freq <- # Investigate the `race_cb_freq` object you created in the previous question by using the `typeof()` # and `str()` functions. Run your code below and answer the following questions: typeof(race_cb_freq) str(race_cb_freq) # Q2.2 ---- # 1 point # What type of object is this, and how many elements does it have? # Assign your answers to an object named `q2_2` and give it a one word character answer in quotes to the first element, # and a single numeric value to the second element. Like so: `q2_2 <- c("answer", 123)` # HINT: A dataframe is a list where each column is one element, so the number of elements # is the number of columns. You can check with length(race_cb_freq). q2_2 <- # Q2.3 ---- # 1 point # Is this object a dataframe? # How many observations does race_cb_freq have? # What are the names of the variables? # Assign your answer to an object named `q2_3` with four character elements like this: # `q2_3 <- c("answer", 123, "name1", "name2")` # HINT: typeof() shows how R stores the object. To check whether it is a dataframe, # use class(race_cb_freq) or is.data.frame(race_cb_freq). q2_3 <- # Q2.4 ---- # 2 points # Now, using `race_cb_freq`, add a column for the percentage of students in each `race_cb` category. # Use `mutate()` to create a new variable that is the percent of students in each category. # Name this new column `pct`, and remember to assign this new variable back to the `race_cb_freq` object. # HINT: Calculate the percent by dividing the count (`n`) by the sum of all counts, then multiply by 100. # Your code should look like: race_cb_freq <- race_cb_freq %>% mutate(pct = ...) race_cb_freq <- ## Question 3: Summarizing across rows ---- # Q3.1 ---- # 3 points # Now, let's investigate the 0/1 indicator variables we created earlier for each race/ethnicity # variable. First, we'll take a look at `stu_hispanic_01`. # Use `summarise()` on `list_native_df` to create three summary variables: # The total number of students (name this `n_obs`) # The total number of students where `stu_hispanic_01` is missing (name this `n_miss_hispanic`) # The percentage of students who identify as hispanic (name this `pct_hispanic`) # Assign the new summary table to an object named `q3_1`. # HINT: Your code can follow this structure: # summarise(n_obs = n(), # n_miss_hispanic = sum(is.na(...)), # pct_hispanic = mean(..., na.rm = TRUE) * 100) # Because stu_hispanic_01 is 0/1, its mean is the share of students who identify as hispanic. # na.rm = TRUE leaves out students with a missing value. q3_1 <- # Q3.2 ---- # 2 points # Next, use `summarise()` on `list_native_df` to calculate the total number of students and the # percentage of students who identify as each of the following categories of race and ethnicity. # Assign the result to an object named `race_ethnicity_pct`: # The total number of students (name this `n`) # The percentage of students for each of these variables: # `stu_cuban_01` # `stu_mexican_01` # `stu_puerto_rican_01` # `stu_other_hispanic_01` # `stu_hispanic_01` # `stu_american_indian_01` # `stu_black_01` # `stu_native_hawaiian_01` # `stu_white_01` # Please name each new race and ethnicity percentage variable as `pct_ethnicity`; basically replacing the `stu` # with `pct` and removing the `_01` from the variable names above. For example, the percentage drawn from # `stu_native_hawaiian_01` would be `pct_native_hawaiian`. # HINT: You can use the same approach you used for `pct_hispanic` in Q3.1: mean(..., na.rm = TRUE) * 100 # You can also use across() for this question race_ethnicity_pct <- # REFLECTION: How do these percentages differ from the aggregated `race_cb` variable # in which each student can only be in one group? Write your answer as a comment below. # ANSWER in the commented lines below: # # # # # Next, run typeof() and str() below to help you answer Q3.3 and Q3.4: race_ethnicity_pct %>% typeof() race_ethnicity_pct %>% str() # Q3.3 ---- # 1 point # What type of object is `race_ethnicity_pct`, and how many elements does it have? # Assign your answer to an object named `q3_3` as a character vector with a single word and a single number. # Like this: `q3_3 <- c("answer", 12345)` q3_3 <- # Q3.4 ---- # 2 points # Is this object a dataframe? If so, how many observations does it have, and what are the names of the # variables? # Assign your answer to an object named `q3_4` as a character vector with one word/number per item. # Like this: `q3_4 <- c("answer", 12345, "name1", "name2", ... )` # Note: Include the names of ALL variables in race_ethnicity_pct (including `n`). q3_4 <- ## Question 4: Grouping and summarizing ---- # Q4.1 ---- # 3 points # Now using the `list_native_df`, use `group_by()` in conjunction with `summarise()` to # calculate summary results for each group. # Assign the new dataframe to an object named `zip_cbsa_group` # First, group by core-based statistical area (`zip_cbsatitle`) # and calculate the following statistics for each CBSA: # The total number of students (name this `n_obs`) # The percentage of students who identify as each of the following race/ethnicity category: # `stu_cuban_01` # `stu_mexican_01` # `stu_puerto_rican_01` # `stu_other_hispanic_01` # `stu_hispanic_01` # `stu_american_indian_01` # `stu_black_01` # `stu_native_hawaiian_01` # `stu_white_01` # Remember to name your percentage variables as `pct_ethnicity`, the same as Q3.2, and to use the # same approach: mean(..., na.rm = TRUE) * 100 # Lastly, in the same line of code, sort by the number of students per CBSA in descending order. # Note that a core-based statistical area by definition only includes urban areas. Observations where # `zip_cbsatitle` is `NA` indicates that the student does not live in a CBSA (i.e., rural location). # HINT: This is similar to the Q3.2 summarise() code with group_by() added before it and arrange() added after it. # You can also use across() for this question zip_cbsa_group <- # Q4.2 ---- # 3 points # Next, we will look at the students' zip-code level median household income (`zip_median_household_income`) by state. # Assign the new dataframe to an object named `stu_state_group`. # Group by state (`stu_state`) and calculate the following statistics for each state: # Note: We provide the name you need to give each new summary statistic in the list below # The total number of students (`n_student`) # The total number of students where `zip_median_household_income` is missing (`n_miss_inc`) # The average median household income of students (`ave_inc`) # The maximum median household income of students (`max_inc`) # The minimum median household income of students (`min_inc`) # Lastly, sort by the number of students per state in descending order. # HINT: Use na.rm = TRUE inside mean(), max(), and min(). Otherwise, any state with a # missing income will show NA. # Note: R WILL GIVE YOU WARNINGS, and a few rows (GU, VI, and the NA state row) will show NaN, Inf, # and -Inf. This is expected: every student in those rows is missing income, so there is nothing # to average. Your `stu_state_group` object will still be created correctly. stu_state_group <- # Q4.3 ---- # 1 point # In the next few questions, we'll take a look at the students' intended major choice. First, group by # major choice (`stu_major_1_group_text`) and summarize the number of students per major (`n_obs`). # Sort by the number of students in descending order and assign the result to an object named `major_group_freq`. major_group_freq <- # Q4.4 ---- # 1 point # Use `major_group_freq` to create a variable for the percentage of students in each # `stu_major_1_group_text` category. Name the new percentage variable `pct` # Assign this new dataframe to an object named `q4_4`. # Hint: use the variable you created in the previous question for the # number of students per major. This works like Q2.4. q4_4 <- # Q4.5 ---- # 1 point # Now, create the same table as the previous question that shows the count and percentage of students # for each major choice, but instead of using `group_by()` and `summarise()`, use `count()` to get the # counts from the original `list_native_df` dataframe. Make sure to sort by descending student count. # Assign the new table to an object named `q4_5`, and be sure to name your percentage variable `pct` # Hint: See Lecture Section 3.4 for usage of `count()`; the number of observations will automatically be named `n`. q4_5 <- # Q4.6 ---- # 1 point # We can also group by multiple variables. In this question, using the `list_native_df` dataframe, # group by both state (`stu_state`) and the student's intended major (`stu_major_1_group_text`), then # summarize to get the number of students per state and major (name this `n_obs`). # Sort ascending by state, then the number of students in descending order. # Assign the result to an object named `major_by_state_freq`. # HINT: You can sort by two variables in one arrange(), e.g. arrange(var1, desc(var2)) major_by_state_freq <- # Looking at the `major_by_state_freq` dataframe from the previous question, answer the following # questions: # Q4.7 ---- # 1 point # How many observations are there? Assign the numeric value to an object named `q4_7` q4_7 <- # Q4.8 ---- # 1 point # What two variables does each row/observation represent? # Assign two single, unique, words as your answers (each one in quotation marks!) to an object named `q4_8`. # Like this: `q4_8 <- c("answer1","answer2")` q4_8 <- # Q4.9 ---- # 1 point # If we were to group/summarize only by state, how many observations would the resulting object have? # Assign the numeric value to an object named `q4_9` q4_9 <- # Q4.10 ---- # 1 point # Finally, we will look at the top 3 intended major choices by students from each state. Assign a # new object named `top_3_majors` using `major_by_state_freq`. # Group by state and create the following variables (we tell you what to name them): # The top choice major by students per state (`top_major`) # The second choice major by students per state (`second_major`) # The third choice major by students per state (`third_major`) # HINT: You can use [ ] to pick the 1st, 2nd, and 3rd values within each state # OR use first() and nth() inside summarise() top_3_majors <- ## Question 5: Bonus (1 point extra credit) ---- # Perform an analysis of your choosing using `list_native_df`. Feel free to be creative! # Assign your result to an object named `q5`, and write 1-2 sentences as a comment # explaining what your analysis shows. q5 <- # ANSWER: # # # # ## Create a GitHub issue ---- # 2 points # Go to the class repository https://github.com/anyone-can-cook/rclass1_student_issues_f26/issues and create a new issue. # Refer to rclass1 student issues readme https://github.com/anyone-can-cook/rclass1_student_issues_f26/blob/main/README.md # for instructions on how to post questions or reflections. # You are also required to respond to at least one issue posted by another student. # Create an object named `issue` below and assign it the URL to your new issue, as a character object. # Make sure the URL is in quotes. issue <- # Create an object named `reply` below and assign it the URL to your response to another issue, as a character object. # Make sure the URL is in quotes. reply <- ## Submit problem set on Canvas ---- # Use this naming convention "lastname_firstname_ps#" for your R script (e.g. jaquette_ozan_problemset5.R).