Idiographic prediction of loneliness and procrastination
2022-01-21
Chapter 1 Workspace
1.1 Packages
library(knitr) # creating tables
library(kableExtra) # formatting and exporting tables
library(rio) # importing html
library(readxl) # read excel codebooks and documentation
library(psych) # biscuit / biscwit
library(glmnet) # elastic net regression
library(glmnetUtils) # extension of basic elastic net with CV
library(caret) # train and test for random forest
library(vip) # variable importance
library(Amelia) # multiple imputation (of time series)
library(lubridate) # date wrangling
library(gtable) # ggplot friendly tables
library(grid) # ggplot friendly table rendering
library(gridExtra) # more helpful ggplot friendly table updates
library(plyr) # data wranging
library(tidyverse) # data wrangling
library(ggdist) # distributional plots
library(ggridges) # more distributional plots
library(cowplot) # flexibly arrange multiple ggplot objects
library(tidymodels) # tidy model workflow and selection
# library(modeltime) # tidy models for time series
library(furrr) # mapping many models in parallel 1.2 Directory Path
res_path <- "https://github.com/emoriebeck/behavior-prediction/raw/main"
local_path <- "/Volumes/Emorie/projects/idio prediction"1.3 Codebook
Each study has a separate codebook indexing matching, covariate, personality, and outcome variables. Moreover, these codebooks contain information about the original scale of the variable, any recoding of the variable (including binarizing outcomes, changing the scale, and removing missing data), reverse coding of scale variables, categories, etc.
# list of all codebook sheets
# ipcs_codebook <- import(file = sprintf("%s/01-codebooks/codebook.xlsx", res_path), which = 2) %>%
# as_tibble()
ipcs_codebook <- sprintf("%s/01-codebooks/codebook_R1.xlsx", local_path) %>%
readxl::read_xlsx(., sheet = "codebook")
ipcs_codebook## # A tibble: 107 × 12
## category trait facet itemname old_desc scale orig_scale_num description orig_itemname reverse_code long_trait long_name
## <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr>
## 1 BFI-2 E scblty Sociabil… Was outg… Liker… 1. Is outgoin… E1 no Extravers… Sociabil…
## 2 BFI-2 E scblty Sociabil… Was talk… Liker… 46. Is talkati… E2 no Extravers… Sociabil…
## 3 BFI-2 E scblty Sociabil… Tended t… Liker… 16r. Tends to b… E3 yes Extravers… Sociabil…
## 4 BFI-2 E scblty Sociabil… Was some… Liker… 31r. Is sometim… E4 yes Extravers… Sociabil…
## 5 BFI-2 E assert Assertiv… Had an a… Liker… 6. Has an ass… E5 no Extravers… Assertiv…
## 6 BFI-2 E assert Assertiv… Was domi… Liker… 21. Is dominan… E6 no Extravers… Assertiv…
## 7 BFI-2 E assert Assertiv… Found it… Liker… 36r. Finds it h… E7 yes Extravers… Assertiv…
## 8 BFI-2 E assert Assertiv… Preferre… Liker… 51r. Prefers to… E8 yes Extravers… Assertiv…
## 9 BFI-2 E enerLev Energy L… Was full… Liker… 41. Is full of… E9 no Extravers… Energy L…
## 10 BFI-2 E enerLev Energy L… Showed a… Liker… 56. Shows a lo… E10 no Extravers… Energy L…
## # … with 97 more rowsoutcomes <- ipcs_codebook %>% filter(category == "outcome") %>% select(trait, long_name)
# ftrs <- import(file = sprintf("%s/01-codebooks/codebook.xlsx", res_path), which = 3) %>%
# as_tibble()
ftrs <- sprintf("%s/01-codebooks/codebook_R1.xlsx", local_path) %>%
readxl::read_xlsx(., sheet = "names")1.3.1 Measures
Participants responded to a large battery of trait and ESM measures as part of the larger study. The present study focuses on ESM measures whose use we preregistered. A full list of the collected measures for the study can be found in supplementary codebooks in the online materials on the OSF and GitHub. The measures collected at each wave were identical. ESM measures were used to estimate idiographic personality prediction models.
1.3.1.1 ESM Measures
1.3.1.1.1 Personality
Personality was assessed using the full BFI-2 (Soto & John, 2017). The scale was administered using a planned missing data design (Revelle et al., 2016). We have previously demonstrated both the between- and within-person construct validity of assessing personality using planned missing designs using the BFI-2 (https://osf.io/pj9sy/). The planned missingness was done within each Big Five trait separately, with three items from each trait included at each timepoint (75% missingness). Each item was answered relative to what a participant was just doing on a 5-point Likert-like scale from 1 “disagree strongly” to 5 “agree strongly.” Items for each person at each assessment were determined by pulling 3 numbers (1 to 12) from a uniform distribution. The order of the resulting 15 items were then randomized before being displayed to participants.
ipcs_codebook %>% filter(category == "BFI-2")## # A tibble: 60 × 12
## category trait facet itemname old_desc scale orig_scale_num description orig_itemname reverse_code long_trait long_name
## <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr>
## 1 BFI-2 E scblty Sociabil… Was outg… Liker… 1. Is outgoin… E1 no Extravers… Sociabil…
## 2 BFI-2 E scblty Sociabil… Was talk… Liker… 46. Is talkati… E2 no Extravers… Sociabil…
## 3 BFI-2 E scblty Sociabil… Tended t… Liker… 16r. Tends to b… E3 yes Extravers… Sociabil…
## 4 BFI-2 E scblty Sociabil… Was some… Liker… 31r. Is sometim… E4 yes Extravers… Sociabil…
## 5 BFI-2 E assert Assertiv… Had an a… Liker… 6. Has an ass… E5 no Extravers… Assertiv…
## 6 BFI-2 E assert Assertiv… Was domi… Liker… 21. Is dominan… E6 no Extravers… Assertiv…
## 7 BFI-2 E assert Assertiv… Found it… Liker… 36r. Finds it h… E7 yes Extravers… Assertiv…
## 8 BFI-2 E assert Assertiv… Preferre… Liker… 51r. Prefers to… E8 yes Extravers… Assertiv…
## 9 BFI-2 E enerLev Energy L… Was full… Liker… 41. Is full of… E9 no Extravers… Energy L…
## 10 BFI-2 E enerLev Energy L… Showed a… Liker… 56. Shows a lo… E10 no Extravers… Energy L…
## # … with 50 more rows1.3.1.1.2 Affect
Items capturing affect were initially pulled from the PANAS-X (Watson & Clark, 1994). In order to reduce redundancy, these were cross-referenced with the BFI-2 and duplicated items (e.g., “excited” were only asked once. Because we were not interested in scale score but in items, we further had research participants examine remaining items and asked them to indicate items that were not relevant to their experience. Finally, we added two “neutral” affect-related terms – goal-directed and purposeful. Each of these were rated on a 1 “disagree strongly” to 5 “agree strongly.”
ipcs_codebook %>% filter(category == "Affect")## # A tibble: 10 × 12
## category trait facet itemname old_desc scale orig_scale_num description orig_itemname reverse_code long_trait long_name
## <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr>
## 1 Affect angry angry angry Angry Like… <NA> <NA> A1 no Angry Negative
## 2 Affect afraid afra… afraid Afraid Like… <NA> <NA> A3 no Afraid Negative
## 3 Affect happy happy happy Happy Like… <NA> <NA> A5 no Happy Positive
## 4 Affect excited exci… excited Excited Like… <NA> <NA> A7 no Excited Positive
## 5 Affect proud proud proud Proud Like… <NA> <NA> A9 no Proud Positive
## 6 Affect guilty guil… guilty Guilty Like… <NA> <NA> A10 no Guilty Negative
## 7 Affect attentive atte… attenti… Attenti… Like… <NA> <NA> A11 no Attentive Positive
## 8 Affect content cont… content Content Like… <NA> <NA> A12 no Content Positive
## 9 Affect purposeful purp… purpose… Purpose… Like… <NA> <NA> A13 no Purposeful Neutral
## 10 Affect goaldir goal… goaldir Goal-di… Like… <NA> <NA> A14 no Goal-dire… Neutral1.3.1.1.3 Binary Situations
Binary situation indicators were derived by asking undergraduate research assistants to provide list of the common social, academic, and personal situations in which they tended to find themselves. From these, we derived a list of 19 unique situations. Separate items for arguing with or interacting with friends or relatives were composited in overall argument and interaction items. Participants checked a box for each event that occurred in the last hour (1 = occurred, 0 = did not occur).
ipcs_codebook %>% filter(category == "sit")## # A tibble: 20 × 12
## category trait facet itemname old_desc scale orig_scale_num description orig_itemname reverse_code long_trait long_name
## <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr>
## 1 sit study study study Was stu… 0 = … <NA> <NA> sit_01 no Studying <NA>
## 2 sit argument argu… argFrnd Had an … 0 = … <NA> <NA> sit_02 no Argument Argument
## 3 sit argument argu… argFam Had an … 0 = … <NA> <NA> sit_03 no Argument Argument
## 4 sit interacted inte… IntFrnd Interac… 0 = … <NA> <NA> sit_04 no Interacted Interact…
## 5 sit interacted inte… IntFam Interac… 0 = … <NA> <NA> sit_05 no Interacted Interact…
## 6 sit lostSmthng lost… lostSmt… Lost so… 0 = … <NA> <NA> sit_06 no Lost some… Lost som…
## 7 sit late late late Was lat… 0 = … <NA> <NA> sit_07 no Late Late
## 8 sit frgtSmthng frgt… frgtSmt… Forgot … 0 = … <NA> <NA> sit_08 no Forgot so… Forgot s…
## 9 sit brdSWk brdS… brdSWk Was bor… 0 = … <NA> <NA> sit_09 no Bored wit… Bored wi…
## 10 sit excSWk excS… excSWk Was exc… 0 = … <NA> <NA> sit_10 no Excited a… <NA>
## 11 sit AnxSWk AnxS… AnxSWk Was anx… 0 = … <NA> <NA> sit_11 no Anxious a… <NA>
## 12 sit tired tired tired Felt ti… 0 = … <NA> <NA> sit_12 no Tired Tired
## 13 sit sick sick sick Felt si… 0 = … <NA> <NA> sit_13 no Sick Sick
## 14 sit sleeping slee… sleeping Was sle… 0 = … <NA> <NA> sit_15 no Sleeping Sleeping
## 15 sit class class class Was in … 0 = … <NA> <NA> sit_16 no In Class In Class
## 16 sit music music music Was lis… 0 = … <NA> <NA> sit_17 no Listening… Listenin…
## 17 sit internet inte… internet Was on … 0 = … <NA> <NA> sit_18 no On the in… On the i…
## 18 sit TV TV TV Was wat… 0 = … <NA> <NA> sit_19 no Watching … Watching…
## 19 sit prcrst prcr… prcrst Procras… 0 = … <NA> <NA> sit_20 no Procrasti… Procrast…
## 20 sit lonely lone… lonely Felt lo… 0 = … <NA> <NA> sit_21 no Lonely Lonely1.3.1.1.4 DIAMONDS Situation Features
Psychological features of situations were measured using the ultra brief version of the “Situational Eight” DIAMONDS (Duty, Intellect, Adversity, Mating, pOsitivity, Negativity, Deception, and Sociality) scale (S8-I; Rauthmann & Sherman, 2015). Items were measured on a 3-point scale from 1 “not at all” to 3 “totally.”
ipcs_codebook %>% filter(category == "S8-I")## # A tibble: 8 × 12
## category trait facet itemname old_desc scale orig_scale_num description orig_itemname reverse_code long_trait long_name
## <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr>
## 1 S8-I Duty Duty Duty Work ha… 1 (=… <NA> <NA> D1 no Duty Duty
## 2 S8-I Intellect Inte… Intelle… Deep th… 1 (=… <NA> <NA> D2 no Intellect Intellect
## 3 S8-I Adversity Adve… Adversi… Somebod… 1 (=… <NA> <NA> D3 no Adversity Adversity
## 4 S8-I Mating Mati… Mating Potenti… 1 (=… <NA> <NA> D4 no Mating Mating
## 5 S8-I pOsitivity pOsi… pOsitiv… The sit… 1 (=… <NA> <NA> D5 no pOsitivity pOsitivi…
## 6 S8-I Negativity Nega… Negativ… The sit… 1 (=… <NA> <NA> D6 no Negativity Negativi…
## 7 S8-I Deception Dece… Decepti… Somebod… 1 (=… <NA> <NA> D7 no Deception Deception
## 8 S8-I Sociability Soci… Sociabi… Social … 1 (=… <NA> <NA> D8 no Sociality Sociality1.3.1.1.5 Timing Features
The final set of features were created from the time stamps collected with each survey based on approaches used in other studies of idiographic prediction (Fisher & Soyster, 2019; . To create these, we created time of day (4; morning, midday, evening, night) and day of the week dummy codes (7). Next, we create a cumulative time variable (in hours) from first beep (not used in analyses) that we used to create linear, quadratic, and cubic time trends (3) as well as 1 and 2 period sine and cosine functions across each 24 period (e.g., 2 period sine = {cumulative time}_t and 1 period sine = {cumulative time}_t).
1.3.2 Procedure
Participants in this study were drawn from a larger personality study. All responded to two types of surveys: trait and state (Experience Sampling Method; ESM) measures, for which they were paid separately. Participants completed three waves of trait measures and two waves of state measures. For the first two waves, trait surveys were collected immediately before beginning the ESM protocol.
1.3.2.1 Main Sample
For the main sample, participants were recruited from the psychology subject pool at Washington University in St. Louis. Participants were told that the study posted on the recruitment website was the first wave of a longer longitudinal study they would be offered the opportunity to take part in.
Participants were brought into the lab between October 2018 and December 2019, where a research assistant or the first author explained the study procedure to them and walked them through the consent procedure. If they consented, participants were led to a room where they could fill out a form to opt into the ESM portion of the study. They then completed baseline trait measures using the Qualtrics Survey Platform. After, the participants were debriefed, paid $10 in cash and, if they opted into the ESM portion of the study, the ESM survey procedure was explained to them.
Participants then received ESM surveys four times per day for two weeks (target n = 56). The survey platform was built by the first author using the jsPsych library (De Leeuw, 2015). Additional JavaScript controllers were written for the purpose of this study and are available on the first author’s GitHub. Start times were based on times that participants indicated they would like to receive their first survey based on their personal wake times. Surveys were sent every 4 hours, meaning that the surveys spanned a 12-hour period from the start time participants indicated. Participants received their first survey at their chosen time on the Monday following their in-lab session. They were compensated $.50 for each survey completed for a maximum of $28. To incentivize responding, participants who completed at least 50 surveys received a “bonus” for a total compensation of $30, which was distributed as an Amazon Gift Card.
1.3.3 Analytic Plan
The present study tested three methods of machine learning classification models, some of which have been used for idiographic prediction in other studies (Fisher & Soyster, 2019; Kaiser & Butter, 2020): (1) Elastic Net Regression (Friedman, Hastie, & Tibshirani, 2010), (2) The Best Items Scale that is Cross-validated, Correlation-weighted, Informative and Transparent (BISCWIT; Elleman, McDougald, Condon, & Revelle, 2020), and (3) Random Forest Models (Kim et al., 2019).
Because we have a large number of indicators to test, each of the methods used have variable selection features and, in some instances, other methods for reducing overfitting, as detailed below. To both reduce the number of indicators used in each test and to test which group of indicators are the most predictive of procrastination and loneliness, we will also test these in several sets: (1) Personality indicators (15), (2) Affective indicators (10), (3) Binary situation indicators (16), (4) DIAMONDS situation indicators (8), (5) Psychological indicators (personality + affect) (25), (6), Situation indicators (binary + DIAMONDS) (24), and (7) Full set (personality + affect + binary situations + DIAMONDS) (49). We will additionally test each of these with and without the 18 timing indicators, for a total set of 14 combinations of the 67 features.
In each of these methods, we used cumulative rolling origin forecast validation, which was comprised of the first 75% of the time series, and held out the remaining 25% of the data set for the test set. In the rolling origin forecast validation, we used the first one-third of the time series as the initial set, five observations as the validation set, and set skip to one, which roughly resulted in 10-15 rolling origin “folds.”
Out of sample prediction was tested based on classification error and area under the ROC (receive operating characteristic) curve (AUC). Classification error is a simple estimate of the percentage of the test sample that was correctly classified by the model. In addition, the AUC will capture the trade-off between sensitivity and specificity across a threshold. In the present study, we used an AUC threshold of .5, which indicates binary classification at chance levels. ROC visualizations plot 1 - specificity (i.e. false positive rate: false positives / (false positives + true negatives)) against sensitivity (i.e. true positive rate: true positives / (true positives + false positives)).
1.4 Demographics
1.4.0.1 Trait
participants <- googlesheets4::sheets_read("https://docs.google.com/spreadsheets/d/1r808gQ-LWfG98J9rvt_CRMHtmCFgtdcfThl0XA0HHbM/edit?usp=sharing", sheet = "ESM") %>%
select(SID, Name, Email) %>%
mutate(new = seq(1, n(), 1),
new = ifelse(new < 10, paste("0", new, sep = ""), new))
1
old_names <- trait_codebook$`New #`
# wave 1 trait
baseline <- sprintf("%s/04-data/01-raw-data/baseline_05.07.20.csv", res_path) %>%
read_csv() %>%
filter(!row_number() %in% c(1,2) & !is.na(SID) & SID %in% participants$SID) %>%
select(SID, StartDate, gender, YOB, race, ethnicity) %>%
mutate(SID = mapvalues(SID, participants$SID, participants$new)) %>%
mutate(wave = 1,
gender = factor(gender, c(1,2), c("Male", "Female")),
YOB = substr(YOB, nchar(YOB)-4+1, nchar(YOB)),
race = mapvalues(race, 1:7, c(0,1,3,2,3,3,3)),
ethnicity = ifelse(!is.na(ethnicity), 3, NA),
race = ifelse(is.na(ethnicity), race, ifelse(ethnicity == 3, ethnicity))) %>%
select(-ethnicity)
save(baseline,
file = sprintf("%s/04-data/01-raw-data/cleaned_combined_2020-05-06.RData", res_path))load(url(sprintf("%s/04-data/01-raw-data/cleaned_combined_2020-05-06.RData", res_path)))
dem <- baseline %>%
select(SID:race) %>%
mutate(age = year(ymd_hms(StartDate)) - as.numeric(YOB),
StartDate = as.Date(ymd_hms(StartDate)),
race = factor(race, 0:3, c("White", "Black", "Asian", "Other"))) %>%
select(-YOB)
dem %>%
summarize(n = length(unique(SID)),
gender = sprintf("%i (%.2f%%)",sum(gender == "Female"), sum(gender == "Female")/n()*100),
age = sprintf("%.2f (%.2f)", mean(age, na.rm = T), sd(age, na.rm = T)),
white = sprintf("%i (%.2f%%)"
, sum(race == "White", na.rm = T)
, sum(race == "White", na.rm = T)/n()*100),
black = sprintf("%i (%.2f%%)"
, sum(race == "Black", na.rm = T)
, sum(race == "Black", na.rm = T)/n()*100),
asian = sprintf("%i (%.2f%%)"
, sum(race == "Asian", na.rm = T)
, sum(race == "Asian", na.rm = T)/n()*100),
other = sprintf("%i (%.2f%%)"
, sum(race == "Other", na.rm = T)
, sum(race == "Other", na.rm = T)/n()*100),
StartDate = sprintf("%s (%s - %s)", median(StartDate),
min(StartDate), max(StartDate)))## # A tibble: 1 × 8
## n gender age white black asian other StartDate
## <int> <chr> <chr> <chr> <chr> <chr> <chr> <chr>
## 1 208 154 (71.96%) 19.51 (1.27) 69 (32.24%) 34 (15.89%) 67 (31.31%) 30 (14.02%) 2019-03-29 (2018-10-17 - 2019-12-05)dem %>%
kable(., "html"
, col.names = c("ID", "Start Date", "Gender", "Race/Ethnicity", "Age")
, align = rep("c", 5)
, caption = "<strong>Table S1</strong><br><em>Descriptive Statistics of Participants at Baseline<em>") %>%
kable_styling(full_width = F) %>%
scroll_box(height = "900px")| ID | Start Date | Gender | Race/Ethnicity | Age |
|---|---|---|---|---|
| 02 | 2018-10-17 | Female | White | 18 |
| 01 | 2018-10-17 | Female | Black | 19 |
| 03 | 2018-10-17 | Female | Asian | 19 |
| 04 | 2018-10-18 | Male | Other | 19 |
| 05 | 2018-10-18 | Male | White | 19 |
| 06 | 2018-10-18 | Female | Asian | 20 |
| 07 | 2018-10-18 | Female | Black | 20 |
| 08 | 2018-10-18 | Female | Black | 18 |
| 09 | 2018-10-18 | Female | 18 | |
| 10 | 2018-10-19 | Female | 19 | |
| 11 | 2018-10-19 | Female | Asian | 18 |
| 12 | 2018-10-19 | Female | White | 20 |
| 13 | 2018-10-19 | Female | White | 18 |
| 14 | 2018-10-19 | Female | Black | 19 |
| 16 | 2018-10-19 | Female | White | 18 |
| 15 | 2018-10-19 | Female | 20 | |
| 17 | 2018-10-19 | Male | Asian | 18 |
| 18 | 2018-10-22 | Female | White | 19 |
| 19 | 2018-10-22 | Female | Black | 20 |
| 20 | 2018-10-22 | Female | Asian | 18 |
| 21 | 2018-10-22 | Female | Asian | 19 |
| 22 | 2018-10-22 | Female | Black | 19 |
| 23 | 2018-10-22 | Male | White | 21 |
| 24 | 2018-10-22 | Male | Black | 20 |
| 25 | 2018-10-22 | Female | White | 18 |
| 27 | 2018-10-23 | Female | 18 | |
| 26 | 2018-10-23 | Female | 18 | |
| 28 | 2018-10-23 | Female | Other | 21 |
| 29 | 2018-10-23 | Female | Asian | 19 |
| 30 | 2018-10-23 | Female | Asian | 20 |
| 31 | 2018-10-24 | Female | White | 18 |
| 32 | 2018-10-24 | Female | Black | 20 |
| 33 | 2018-10-24 | Female | Asian | 18 |
| 34 | 2018-10-24 | Female | Black | 19 |
| 35 | 2018-10-26 | Female | Black | 18 |
| 36 | 2018-10-29 | Female | Asian | 21 |
| 37 | 2018-10-29 | Male | Other | 18 |
| 38 | 2018-10-29 | Male | Asian | 19 |
| 36 | 2018-10-29 | Female | Other | 20 |
| 37 | 2018-10-29 | Female | Asian | 18 |
| 41 | 2018-10-29 | Male | White | 19 |
| 38 | 2018-10-29 | Female | Black | 19 |
| 43 | 2018-10-29 | Female | White | 18 |
| 44 | 2018-10-30 | Female | Asian | 18 |
| 45 | 2018-11-01 | Female | 18 | |
| 46 | 2018-11-01 | Female | Asian | 22 |
| 48 | 2018-11-01 | Female | Asian | 21 |
| 47 | 2018-11-01 | Male | Asian | 23 |
| 49 | 2018-11-02 | Female | Asian | 20 |
| 51 | 2018-11-02 | Female | White | 20 |
| 50 | 2018-11-02 | Female | Other | 19 |
| 52 | 2018-11-05 | Male | Other | 21 |
| 53 | 2018-11-05 | Female | Asian | 19 |
| 52 | 2018-11-05 | Male | Asian | 21 |
| 53 | 2018-11-05 | Female | White | 19 |
| 56 | 2018-11-05 | Male | Asian | 18 |
| 58 | 2018-11-05 | Female | Asian | 21 |
| 57 | 2018-11-05 | Female | Asian | |
| 59 | 2018-11-06 | Female | Asian | 21 |
| 60 | 2018-11-06 | Male | White | 20 |
| 61 | 2018-11-06 | Male | White | 18 |
| 62 | 2018-11-06 | Male | Black | 18 |
| 63 | 2018-11-06 | Female | White | 20 |
| 64 | 2018-11-07 | Female | Other | 21 |
| 65 | 2018-11-07 | Male | White | 20 |
| 67 | 2018-11-07 | Female | Black | 19 |
| 66 | 2018-11-07 | Male | White | 20 |
| 68 | 2018-11-07 | Female | Asian | 18 |
| 69 | 2018-11-07 | Female | Asian | 18 |
| 70 | 2018-11-08 | Female | 19 | |
| 72 | 2018-11-08 | Female | Other | 18 |
| 71 | 2018-11-08 | Female | Other | 19 |
| 74 | 2018-11-08 | Female | Other | 22 |
| 73 | 2018-11-08 | Female | White | 18 |
| 75 | 2018-11-08 | Female | Asian | 19 |
| 76 | 2018-11-08 | Female | Black | 20 |
| 77 | 2018-11-08 | Male | Black | 21 |
| 79 | 2018-11-08 | Male | Asian | 18 |
| 80 | 2018-11-08 | Female | White | 21 |
| 82 | 2018-11-09 | Female | Black | 20 |
| 81 | 2018-11-09 | Female | Asian | 22 |
| 83 | 2018-11-09 | Female | White | 18 |
| 84 | 2018-11-09 | Female | White | 20 |
| 85 | 2018-11-14 | Female | White | 18 |
| 86 | 2018-11-14 | Female | Asian | 21 |
| 87 | 2018-11-14 | Female | Asian | 21 |
| 89 | 2018-11-14 | Female | White | 19 |
| 88 | 2018-11-14 | Female | Asian | 18 |
| 90 | 2018-11-20 | Female | Other | 21 |
| 91 | 2018-11-20 | Female | Black | 21 |
| 93 | 2018-11-28 | Female | White | 20 |
| 92 | 2018-11-28 | Male | Black | 20 |
| 94 | 2018-11-28 | Female | Asian | 21 |
| 95 | 2018-11-28 | Female | Black | 21 |
| 96 | 2018-11-28 | Female | White | 21 |
| 97 | 2018-11-28 | Male | Asian | 19 |
| 98 | 2018-11-29 | Female | Asian | 18 |
| 99 | 2018-11-29 | Female | Other | 21 |
| 100 | 2018-11-29 | Female | Black | 19 |
| 101 | 2018-11-29 | Female | White | 18 |
| 102 | 2018-11-29 | Female | White | 19 |
| 103 | 2019-03-15 | Female | White | 20 |
| 104 | 2019-03-15 | Female | 22 | |
| 106 | 2019-03-22 | Female | Asian | 21 |
| 105 | 2019-03-22 | Female | White | 20 |
| 107 | 2019-03-22 | Female | White | 19 |
| 109 | 2019-03-29 | Female | White | 19 |
| 108 | 2019-03-29 | Female | White | 20 |
| 111 | 2019-04-05 | Female | White | 19 |
| 110 | 2019-04-05 | Male | Other | 22 |
| 112 | 2019-04-05 | Female | Black | |
| 113 | 2019-04-05 | Female | Asian | 23 |
| 114 | 2019-04-05 | Male | White | 20 |
| 116 | 2019-04-12 | Female | White | 21 |
| 115 | 2019-04-12 | Female | White | 19 |
| 118 | 2019-04-12 | Female | Asian | 21 |
| 117 | 2019-04-12 | Female | White | 20 |
| 119 | 2019-04-12 | Male | Asian | 19 |
| 121 | 2019-04-12 | Female | White | 20 |
| 122 | 2019-04-12 | Male | White | 20 |
| 123 | 2019-04-12 | Female | Asian | 20 |
| 124 | 2019-04-12 | Female | Black | 23 |
| 126 | 2019-04-19 | Male | Other | 22 |
| 125 | 2019-04-19 | Female | Asian | 21 |
| 128 | 2019-04-19 | Male | Black | 23 |
| 127 | 2019-04-19 | Female | White | 21 |
| 129 | 2019-04-19 | Female | Other | 20 |
| 130 | 2019-04-19 | Female | White | 19 |
| 131 | 2019-04-19 | Male | White | 20 |
| 133 | 2019-04-26 | Male | Asian | 20 |
| 132 | 2019-04-26 | Female | White | 21 |
| 134 | 2019-04-26 | Female | Asian | |
| 136 | 2019-09-11 | Male | Asian | 20 |
| 135 | 2019-09-11 | Male | White | 19 |
| 138 | 2019-09-13 | Female | White | 19 |
| 137 | 2019-09-13 | Female | White | 19 |
| 139 | 2019-09-17 | Female | Black | 19 |
| 141 | 2019-09-18 | Female | White | 21 |
| 142 | 2019-09-13 | Male | Asian | 18 |
| 143 | 2019-09-19 | Female | 18 | |
| 146 | 2019-09-19 | Female | Asian | 20 |
| 148 | 2019-09-20 | Female | Black | 20 |
| 147 | 2019-09-20 | Female | White | 20 |
| 149 | 2019-09-20 | Female | White | 22 |
| 150 | 2019-09-20 | Female | Asian | 18 |
| 151 | 2019-09-20 | Female | Other | 18 |
| 152 | 2019-09-20 | Male | White | 18 |
| 153 | 2019-09-27 | Female | Asian | 21 |
| 154 | 2019-09-27 | Male | White | 18 |
| 155 | 2019-09-27 | Male | White | 19 |
| 156 | 2019-09-27 | Male | White | 20 |
| 157 | 2019-09-27 | Female | White | 21 |
| 158 | 2019-09-27 | Female | Asian | 21 |
| 159 | 2019-09-27 | Female | Other | 18 |
| 160 | 2019-09-27 | Female | Asian | 18 |
| 161 | 2019-09-27 | Female | White | 18 |
| 162 | 2019-09-27 | Female | Black | 19 |
| 164 | 2019-10-04 | Female | 18 | |
| 163 | 2019-10-04 | Male | Black | 19 |
| 165 | 2019-10-04 | Female | White | 22 |
| 166 | 2019-10-04 | Male | Other | 19 |
| 168 | 2019-10-04 | Female | Asian | 19 |
| 167 | 2019-10-04 | Male | Asian | 19 |
| 169 | 2019-10-18 | Male | Asian | 21 |
| 170 | 2019-10-18 | Female | Asian | 19 |
| 171 | 2019-10-18 | Male | Other | 18 |
| 172 | 2019-10-18 | Female | Other | 19 |
| 174 | 2019-10-18 | Male | White | 19 |
| 173 | 2019-10-18 | Female | Asian | 18 |
| 175 | 2019-10-28 | Male | Black | 19 |
| 176 | 2019-10-28 | Male | Other | 19 |
| 177 | 2019-10-30 | Male | Asian | 21 |
| 179 | 2019-11-02 | Male | Other | 21 |
| 181 | 2019-11-02 | Male | Black | 21 |
| 180 | 2019-11-02 | Female | White | 20 |
| 182 | 2019-11-03 | Female | Black | 19 |
| 184 | 2019-11-03 | Female | Asian | 20 |
| 183 | 2019-11-03 | Female | Black | 21 |
| 185 | 2019-11-07 | Female | Other | 19 |
| 186 | 2019-11-07 | Female | Asian | 21 |
| 188 | 2019-11-08 | Female | Other | 18 |
| 187 | 2019-11-08 | Male | White | 20 |
| 190 | 2019-11-08 | Male | White | 21 |
| 189 | 2019-11-08 | Female | Other | 20 |
| 192 | 2019-11-08 | Female | Black | 18 |
| 191 | 2019-11-08 | Male | Other | 18 |
| 199 | 2019-11-11 | Female | Asian | 19 |
| 200 | 2019-11-11 | Female | Asian | 18 |
| 201 | 2019-11-11 | Female | Asian | 19 |
| 202 | 2019-11-11 | Male | Asian | 18 |
| 203 | 2019-11-14 | Male | White | 18 |
| 204 | 2019-11-14 | Female | Other | 20 |
| 205 | 2019-11-14 | Female | Asian | 19 |
| 206 | 2019-11-15 | Female | White | 19 |
| 197 | 2019-11-15 | Female | 18 | |
| 207 | 2019-11-15 | Female | Other | 18 |
| 193 | 2019-11-15 | Male | White | 18 |
| 196 | 2019-11-15 | Female | White | 21 |
| 195 | 2019-11-15 | Male | Black | 21 |
| 197 | 2019-11-15 | Male | White | 18 |
| 209 | 2019-11-20 | Female | Asian | 19 |
| 210 | 2019-11-20 | Female | 18 | |
| 211 | 2019-11-20 | Female | 19 | |
| 212 | 2019-11-20 | Male | White | 20 |
| 213 | 2019-11-22 | Male | Asian | 19 |
| 214 | 2019-11-22 | Female | Other | 19 |
| 216 | 2019-11-22 | Female | Asian | 19 |
| 215 | 2019-11-22 | Female | White | 19 |
| 217 | 2019-12-02 | Male | 19 | |
| 218 | 2019-12-03 | Male | Asian | 19 |
| 219 | 2019-12-05 | Male | White | 19 |
| 220 | 2019-12-05 | Female | Other | 20 |
| 221 | 2019-12-05 | Female | Asian | 21 |
| 222 | 2019-12-05 | Female | Black | 20 |