florida_twins

[ben-domingue]

decompose to multiple tables?

Study from Florida about twins 9 years and older. Items related to twins' home and school environment, behavioral tendencies, and attitudes about schooling/home/social aspects

[saviranadela]

adjusted based on codebook wave 1,2,3

some minor items in the dataset don't exist in the codebook (such as ci, cj, missing social), so i don't include them

[saviranadela]

data:
florida_twins.zip

code:

library(tidyverse)
library(readr)

# ------ WAVE 1 DATASET
df1 <- read_csv('Wave 1 Child LDBase.csv')

names(df1) <- tolower(names(df1))

auths <- df1 |>
  select(student_id0,
         contains('auth'))

authsb <- auths |>
  select(student_id0,
         contains('b')) |>
  # recode invalid responses as NA, change remaining responses to 0/1 binary
  mutate(across(starts_with('auth'), ~if_else(. == 0, NA, .)),
         across(starts_with('auth'), ~if_else(. == 1, 0, .)),
         across(starts_with('auth'), ~if_else(. == 2, 1, .)))

# save names of vars in this df to drop and later merge back onto the full df
auth_names <- names(authsb[2:ncol(authsb)])

df1 <- df1 |>
  # drop unneeded variables
  select(-all_of(auth_names),
         -reading_grades0,
         -reading_grades1,
         -starts_with('npar'),
         -starts_with('nnpar'),
         -starts_with('par'),
         -starts_with('artcorrect'),
         -starts_with('artfalse'),
         -starts_with('arttpe'),
         -starts_with('artppk'),
         -starts_with('artspk'),
         -starts_with('classsp'),
         -starts_with('classcm'),
         -starts_with('classca'),
         -starts_with('schattach'),
         -starts_with('schbond'),
         -starts_with('schactiv'),
         -starts_with('schneg'),
         -starts_with('reading_social'),
         -starts_with('reading_grades'),
         -starts_with('reading_curiosity'),
         -starts_with('reading_competition'),
         -starts_with('reading_involvement'),
         -starts_with('reading_work'),
         -starts_with('reading_efficacy'),
         -starts_with('reading_recognition'),
         -starts_with('panas_pa'),
         -starts_with('panas_na'),
         -starts_with('cadsyv_pos'),
         -starts_with('cadsyv_dar'),
         -starts_with('cadsyv_pro'),
         -starts_with('cadsyv_neg'),
         -starts_with('cadsyv_soc'),
         -starts_with('cadsyv_resp'),
         -starts_with('cadsyv_dis'),
         -starts_with('friends_bad'),
         -starts_with('friends_school'),
         -starts_with('friends_good'),
         -contains('info_sharing'),
         -bin,
         -ain,
         -id1,
         -contains('total'),
         -starts_with('childchaos'),
         -starts_with('bg_id'),
         -starts_with('identifier'),
         -starts_with('tid'),
         -`...1`,
         -twinid,
         -gender_master,
         -pair_gender,
         -multiple,
         -zygparsum,
         -zyg_par,
         -fid) |>
  left_join(authsb, by='student_id0') |>
  select(-starts_with('student_id')) |>
  # add participant ID
  rename(family_id = famid,
         age = qage) |>
  pivot_longer(cols = -c(id0, family_id, age),
               names_to = 'item',
               values_to = 'resp',
               values_drop_na = T) |>
  mutate(wave = 1)


# ----- WAVE 2 DATASET

df2 <- read_csv('wave2multitwinq718 LDBase_0.csv')
names(df2) <- tolower(names(df2))

df2 <- df2 |>
  select(-`...1`,
         -starts_with('bg_id'),
         -starts_with('bfcrr_twin'),
         -contains('hair'),
         -contains('eyes'),
         -starts_with('bfcrr_peas'),
         -starts_with('bfcrr_sim'),
         -starts_with('bfcrr_pretend'),
         -starts_with('btid'),
         -starts_with('fid'),
         -twinid,
         -gender_master,
         -pair_gender,
         -multiple,
         -contains('info_sharing'),
         -contains('total'),
         -id1,
         -ain,
         -bin,
         -zygparsum,
         -zyg_par,
         -starts_with('bpals_parentperformance'),
         -starts_with('bpals_parentdissonance'),
         -starts_with('artcorrect'),
         -starts_with('artfalse'),
         -starts_with('arttpe'),
         -starts_with('bpals_classroomapproach'),
         -starts_with('bpals_classroomavoid'),
         -starts_with('bpals_classroommaster'),
         -starts_with('bpals_teacheravoid'),
         -starts_with('bpals_teacherapproach'),
         -starts_with('bpals_teachermastery'),
         -starts_with('bpanas_na'),
         -starts_with('bpanas_pa'),
         -starts_with('bcadsyv_pos'),
         -starts_with('bcadsyv_dar'),
         -starts_with('bcadsyv_pro'),
         -starts_with('bcadsyv_neg'),
         -starts_with('bcadsyv_soc'),
         -starts_with('bcadsyv_resp'),
         -starts_with('bcadsyv_dis'),
         -starts_with('bfriends_bad'),
         -starts_with('bfriends_school'),
         -starts_with('bfriends_good'),
         -contains('sub'),
         -contains('bdweckincremental'),
         -contains('bdweckentity'),
         -contains('qoccup'),
         -starts_with('bgritconsistency'),
         -starts_with('bgritperseverance'),
         -starts_with('bart')) |>
  rename(age = bq2age) |>
  pivot_longer(cols = -c(id0, age, famid),
               names_to = 'item',
               values_to = 'resp',
               values_drop_na = T) |>
  mutate(wave = 2) |>
  rename(family_id = famid)


# ------- WAVE 3 DATASET

df3 <- read_csv('w3multitwinq818 LDBase.csv')

names(df3) <- tolower(names(df3))

df3 <- df3 |>
  select(-`...1`,
         -starts_with('bg_id'),
         -twinid,
         -contains('gender'),
         -multiple,
         -id1,
         -fid,
         -contains('hair'),
         -contains('eyes'),
         -starts_with('cfcrr_peas'),
         -starts_with('cfcrr_sim'),
         -starts_with('cfcrr_pretend'),
         -contains('info_sharing'),
         -contains('total'),
         -contains('hemt'),
         -contains('occup'),
         -contains('_text'),
         -zygparsum,
         -zyg_par,
         -cpar140,
         -contains('info_sharing'),
         -starts_with('artcorrect'),
         -starts_with('artfalse'),
         -starts_with('cpanas_na'),
         -starts_with('cpanas_pa'),
         -contains('friends_bad'),
         -contains('friends_school'),
         -contains('friends_good'),
         -contains('sub'),
         -contains('cdweckincremental'),
         -contains('cdweckentity'),
         -starts_with('cdbi_sum'),
         -starts_with('cleq_schoolstress'),
         -starts_with('cleq_interpersonalstress'),
         -starts_with('cleq_schoolevents'),
         -starts_with('cleq_interpersonalevents'),
         -starts_with('cleq_stress'),
         -starts_with('cleq_events'),
         -contains('cart'),
         -starts_with('cweeklytv')) |>
  rename(age = cq3age) |>
  mutate(across(contains('hwk'), ~ . + 1),
         across(contains('grades'), ~if_else(. == 7, NA, .))) |>
  pivot_longer(cols = -c(id0, age, famid),
               names_to = 'item',
               values_to = 'resp',
               values_drop_na = T) |>
  mutate(wave = 3) |>
  rename(family_id = famid)


# ---- merge three waves datasets

df <- rbind(df1, df2, df3)

length(unique(df$item))

df_chaos <- df %>%
  filter(grepl("chaos", item))

df_grades <- df %>%
  filter(grepl("grades", item))

df_nes <- df %>%
  filter(grepl("nes", item))

df_par <- df %>%
  filter(grepl("par", item))

df_class <- df %>%
  filter(grepl("class", item))

df_sch <- df %>%
  filter(grepl("sch", item))

df_read <- df %>%
  filter(grepl("read", item))

df_panas <- df %>%
  filter(grepl("panas", item))

df_cads <- df %>%
  filter(grepl("cads", item))

df_friends <- df %>%
  filter(grepl("friends", item))

df_auth <- df %>%
  filter(grepl("auth", item))

df_pals <- df %>%
  filter(grepl("pals", item))

df_dweck <- df %>%
  filter(grepl("dweck", item))

df_grit <- df %>%
  filter(grepl("grit", item))

df_cqbar <- df %>%
  filter(grepl("cqbar", item))

df_hwk <- df %>%
  filter(grepl("hwk", item))

df_tech <- df %>%
  filter(grepl("tech", item))

df_media <- df %>%
  filter(grepl("media", item))

df_dbi <- df %>%
  filter(grepl("dbi", item))

df_leq <- df %>%
  filter(grepl("leq", item))

df_game <- df %>%
  filter(grepl("game", item))

write.csv(df_chaos, "florida_twins_chaos.csv", row.names=FALSE)
write.csv(df_grades, "florida_twins_grades.csv", row.names=FALSE)
write.csv(df_nes, "florida_twins_nes.csv", row.names=FALSE)
write.csv(df_par, "florida_twins_par.csv", row.names=FALSE)
write.csv(df_class, "florida_twins_class.csv", row.names=FALSE)
write.csv(df_sch, "florida_twins_sch.csv", row.names=FALSE)
write.csv(df_read, "florida_twins_read.csv", row.names=FALSE)
write.csv(df_panas, "florida_twins_panas.csv", row.names=FALSE)
write.csv(df_cads, "florida_twins_cads.csv", row.names=FALSE)
write.csv(df_friends, "florida_twins_friends.csv", row.names=FALSE)
write.csv(df_auth, "florida_twins_auth.csv", row.names=FALSE)
write.csv(df_pals, "florida_twins_pals.csv", row.names=FALSE)
write.csv(df_dweck, "florida_twins_dweck.csv", row.names=FALSE)
write.csv(df_grit, "florida_twins_grit.csv", row.names=FALSE)
write.csv(df_cqbar, "florida_twins_cqbar.csv", row.names=FALSE)
write.csv(df_hwk, "florida_twins_hwk.csv", row.names=FALSE)
write.csv(df_tech, "florida_twins_tech.csv", row.names=FALSE)
write.csv(df_media, "florida_twins_media.csv", row.names=FALSE)
write.csv(df_dbi, "florida_twins_dbi.csv", row.names=FALSE)
write.csv(df_leq, "florida_twins_leq.csv", row.names=FALSE)
write.csv(df_game, "florida_twins_game.csv", row.names=FALSE)

[ben-domingue]

a few notes:

• id0 -> id
• age -> cov_age
• do we need to tie this to issue florida_twins_behavior #734 ? i am not sure if these two are redundant. [apologies!]
• some of the scales (auth=400, cads=228, cqbar=100, friends=226, panas=120, read=108) have so many items that i'm concerned something is going wrong. as one example, the cads scale doesn't seem like it should be anywhere near 228 items long. i think we need to dig more into those.

thank you @saviranadela !!!

[ben-domingue]

@saviranadela did we finalize this and i just forgot to close it?

[saviranadela]

hi Ben! oh no you're good, i'm still working on florida twins, will update it asap! @ben-domingue

[saviranadela]

there's one thing I'm currently stuck on:
as an example, according to the codebook, CADS should have 57 items. however, i noticed that one item can have multiple derivatives (for example, CADS1 is written as CADS10 and CADS11 in the dataset, and CADS11 has CADS110 and CADS111). i'm not entirely sure what each of these represents, as the codebook doesn’t provide much information. i’m still digging into it, but I wanted to check if you happen to recall any details about these! @ben-domingue

[ben-domingue]

hm. i don't know but that is a question we can also bring to the PIs; if it isn't well-documented we can maybe just go to the source rather than guess. if you think we should do that, maybe draft some language and i can put it into an email to the PIs and cc you?

[saviranadela]

(saw our IRW data dictionary and just realized that Lucy handled this, not you. i'm really sorry for the confusion in my questions!)

sorry that i took a very long time with this, the data is quite a mess and they have lots of assessment tools.
i've cleaned the code and the table (like removing reverted scales, renaming the same measurement, etc)

data:
florida_twins.zip

code:

library(tidyverse)
library(readr)

# ------ WAVE 1 DATASET
df1 <- read_csv('Wave 1 Child LDBase.csv')

names(df1) <- tolower(names(df1))

# colnames(df1)

auths <- df1 |>
  select(student_id0,
         contains('auth'))

authsb <- auths |>
  select(student_id0,
         contains('b')) |>
  # recode invalid responses as NA, change remaining responses to 0/1 binary
  mutate(across(starts_with('auth'), ~if_else(. == 0, NA, .)),
         across(starts_with('auth'), ~if_else(. == 1, 0, .)),
         across(starts_with('auth'), ~if_else(. == 2, 1, .)))

# save names of vars in this df to drop and later merge back onto the full df
auth_names <- names(authsb[2:ncol(authsb)])

df1 <- df1 |>
  # drop unneeded variables
  select(-all_of(auth_names),
         -reading_grades0,
         -reading_grades1,
         # -starts_with('npar'),
         # -starts_with('nnpar'),
         # -starts_with('par'),
         -starts_with('artcorrect'),
         -starts_with('artfalse'),
         -starts_with('arttpe'),
         -starts_with('artppk'),
         -starts_with('artspk'),
         -starts_with('classsp'),
         -starts_with('classcm'),
         -starts_with('classca'),
         -starts_with('schattach'),
         -starts_with('schbond'),
         -starts_with('schactiv'),
         -starts_with('schneg'),
         -starts_with('reading_social'),
         -starts_with('reading_grades'),
         -starts_with('reading_curiosity'),
         -starts_with('reading_competition'),
         -starts_with('reading_involvement'),
         -starts_with('reading_work'),
         -starts_with('reading_efficacy'),
         -starts_with('reading_recognition'),
         -starts_with('panas_pa'),
         -starts_with('panas_na'),
         -starts_with('cadsyv_pos'),
         -starts_with('cadsyv_dar'),
         -starts_with('cadsyv_pro'),
         -starts_with('cadsyv_neg'),
         -starts_with('cadsyv_soc'),
         -starts_with('cadsyv_resp'),
         -starts_with('cadsyv_dis'),
         -starts_with('friends_bad'),
         -starts_with('friends_school'),
         -starts_with('friends_good'),
         -contains('info_sharing'),
         -bin,
         -ain,
         -id1,
         -contains('total'),
         -starts_with('childchaos'),
         -starts_with('bg_id'),
         -starts_with('identifier'),
         -starts_with('tid'),
         -`...1`,
         -twinid,
         -gender_master,
         -pair_gender,
         -multiple,
         -zygparsum,
         -zyg_par,
         -fid) |>
  left_join(authsb, by='student_id0') |>
  select(-starts_with('student_id')) |>
  # add participant ID
  rename(family_id = famid,
         cov_age = qage,
         id = id0) |>
  pivot_longer(cols = -c(id, family_id, cov_age),
               names_to = 'item',
               values_to = 'resp',
               values_drop_na = T) |>
  mutate(wave = 1)

df1 <- df1[!(grepl("^n", df1$item) & !grepl("^nes", df1$item)),] # remove reverted scales


# ----- WAVE 2 DATASET

df2 <- read_csv('wave2multitwinq718 LDBase_0.csv')
names(df2) <- tolower(names(df2))

df2 <- df2 |>
  select(-`...1`,
         -starts_with('bg_id'),
         -starts_with('bfcrr_twin'),
         -contains('hair'),
         -contains('eyes'),
         -starts_with('bfcrr_peas'),
         -starts_with('bfcrr_sim'),
         -starts_with('bfcrr_pretend'),
         -starts_with('btid'),
         -starts_with('fid'),
         -twinid,
         -gender_master,
         -pair_gender,
         -multiple,
         -contains('info_sharing'),
         -contains('total'),
         -id1,
         -ain,
         -bin,
         -zygparsum,
         -zyg_par,
         -starts_with('bpals_parentperformance'),
         -starts_with('bpals_parentdissonance'),
         -starts_with('artcorrect'),
         -starts_with('artfalse'),
         -starts_with('arttpe'),
         -starts_with('bpals_classroomapproach'),
         -starts_with('bpals_classroomavoid'),
         -starts_with('bpals_classroommaster'),
         -starts_with('bpals_teacheravoid'),
         -starts_with('bpals_teacherapproach'),
         -starts_with('bpals_teachermastery'),
         -starts_with('bpanas_na'),
         -starts_with('bpanas_pa'),
         -starts_with('bcadsyv_pos'),
         -starts_with('bcadsyv_dar'),
         -starts_with('bcadsyv_pro'),
         -starts_with('bcadsyv_neg'),
         -starts_with('bcadsyv_soc'),
         -starts_with('bcadsyv_resp'),
         -starts_with('bcadsyv_dis'),
         -starts_with('bfriends_bad'),
         -starts_with('bfriends_school'),
         -starts_with('bfriends_good'),
         -contains('sub'),
         -contains('bdweckincremental'),
         -contains('bdweckentity'),
         -contains('qoccup'),
         -starts_with('bgritconsistency'),
         -starts_with('bgritperseverance'),
         -starts_with('bart')) |>
  rename(cov_age = bq2age, id = id0) |>
  pivot_longer(cols = -c(id, cov_age, famid),
               names_to = 'item',
               values_to = 'resp',
               values_drop_na = T) |>
  mutate(wave = 2) |>
  rename(family_id = famid)

df2$item <- substring(df2$item, 2) # remove prefix

df2 <- df2[!(grepl("^n", df2$item) & !grepl("^nes", df2$item)),] 


# ------- WAVE 3 DATASET

df3 <- read_csv('w3multitwinq818 LDBase.csv')

names(df3) <- tolower(names(df3))

df3 <- df3 |>
  select(-`...1`,
         -starts_with('bg_id'),
         -twinid,
         -contains('gender'),
         -multiple,
         -id1,
         -fid,
         -contains('hair'),
         -contains('eyes'),
         -starts_with('cfcrr_peas'),
         -starts_with('cfcrr_sim'),
         -starts_with('cfcrr_pretend'),
         -contains('info_sharing'),
         -contains('total'),
         -contains('hemt'),
         -contains('occup'),
         -contains('_text'),
         -zygparsum,
         -zyg_par,
         -cpar140,
         -contains('info_sharing'),
         -starts_with('artcorrect'),
         -starts_with('artfalse'),
         -starts_with('cpanas_na'),
         -starts_with('cpanas_pa'),
         -contains('friends_bad'),
         -contains('friends_school'),
         -contains('friends_good'),
         -contains('sub'),
         -contains('cdweckincremental'),
         -contains('cdweckentity'),
         -starts_with('cdbi_sum'),
         -starts_with('cleq_schoolstress'),
         -starts_with('cleq_interpersonalstress'),
         -starts_with('cleq_schoolevents'),
         -starts_with('cleq_interpersonalevents'),
         -starts_with('cleq_stress'),
         -starts_with('cleq_events'),
         -contains('cart'),
         -starts_with('cweeklytv')) |>
  rename(cov_age = cq3age, id = id0) |>
  mutate(across(contains('hwk'), ~ . + 1),
         across(contains('grades'), ~if_else(. == 7, NA, .))) |>
  pivot_longer(cols = -c(id, cov_age, famid),
               names_to = 'item',
               values_to = 'resp',
               values_drop_na = T) |>
  mutate(wave = 3) |>
  rename(family_id = famid)

df3$item <- substring(df3$item, 2)

df3 <- df3[!(grepl("^n", df3$item) & !grepl("^nes", df3$item)),]

# ---- merge three waves datasets

df <- rbind(df1, df2, df3)

print(unique(df$item), max = 2000)


df_chaos <- df %>%
  filter(grepl("chaos", item))

df_grades <- df %>%
  filter(grepl("grades", item))

df_nes <- df %>%
  filter(grepl("nes", item))

df_par <- df %>%
  filter(grepl("par", item))

df_class <- df %>%
  filter(grepl("class", item))

df_sch <- df %>%
  filter(grepl("sch", item))

df_read <- df %>%
  filter(grepl("read", item))

df_panas <- df %>%
  filter(grepl("panas", item))

df_cads <- df %>%
  filter(grepl("cads", item))

df_friends <- df %>%
  filter(grepl("friends", item))

df_auth <- df %>%
  filter(grepl("auth", item))

df_pals <- df %>%
  filter(grepl("pals", item))

df_dweck <- df %>%
  filter(grepl("dweck", item))

df_grit <- df %>%
  filter(grepl("grit", item))

df_cqbar <- df %>%
  filter(grepl("cqbar", item))

df_hwk <- df %>%
  filter(grepl("hwk", item))

df_tech <- df %>%
  filter(grepl("tech", item))

df_media <- df %>%
  filter(grepl("media", item))

df_dbi <- df %>%
  filter(grepl("dbi", item))

df_leq <- df %>%
  filter(grepl("leq", item))

df_game <- df %>%
  filter(grepl("game", item))

write.csv(df_chaos, "florida_twins_chaos.csv", row.names=FALSE)
write.csv(df_grades, "florida_twins_grades.csv", row.names=FALSE)
write.csv(df_nes, "florida_twins_nes.csv", row.names=FALSE)
write.csv(df_par, "florida_twins_par.csv", row.names=FALSE)
write.csv(df_class, "florida_twins_class.csv", row.names=FALSE)
write.csv(df_sch, "florida_twins_sch.csv", row.names=FALSE)
write.csv(df_read, "florida_twins_read.csv", row.names=FALSE)
write.csv(df_panas, "florida_twins_panas.csv", row.names=FALSE)
write.csv(df_cads, "florida_twins_cads.csv", row.names=FALSE)
write.csv(df_friends, "florida_twins_friends.csv", row.names=FALSE)
write.csv(df_auth, "florida_twins_auth.csv", row.names=FALSE)
write.csv(df_pals, "florida_twins_pals.csv", row.names=FALSE)
write.csv(df_dweck, "florida_twins_dweck.csv", row.names=FALSE)
write.csv(df_grit, "florida_twins_grit.csv", row.names=FALSE)
write.csv(df_cqbar, "florida_twins_cqbar.csv", row.names=FALSE)
write.csv(df_hwk, "florida_twins_hwk.csv", row.names=FALSE)
write.csv(df_tech, "florida_twins_tech.csv", row.names=FALSE)
write.csv(df_media, "florida_twins_media.csv", row.names=FALSE)
write.csv(df_dbi, "florida_twins_dbi.csv", row.names=FALSE)
write.csv(df_leq, "florida_twins_leq.csv", row.names=FALSE)
write.csv(df_game, "florida_twins_game.csv", row.names=FALSE)

this should be the right one.
(auth=200, cads=114, friends=42, panas=40, read=108)
(and i don't know what cqbar means as it’s not explained anywhere in the codebook, but I included it for the sake of completeness) @ben-domingue

[saviranadela]

some of the scales are doubled because of the 0 and 1 suffix i mentioned previously. i haven’t figured out the exact reason yet, but i came to think this isn't really a big deal as i believe it likely represents two different facets or sub-items related to the same general question. however, if you think it’s necessary to confirm with the PIs, we can email them like this:

Dear Dr. Sara A. Hart,

I hope this email finds you well.

[Introduction here]. We are currently working with the Florida Twin Project dataset for our Item Response Warehouse and came across an observation that we’d like to seek your guidance on. Specifically, we noticed that some items include suffixes like "0" and "1." For instance, the first item of CADS appears to be broken down into "cadsyv10" and "cadsyv11."

We were unable to find an explanation for these suffix distinctions in the provided codebook and we were wondering if you could kindly clarify the difference between these two components. Are they measuring different subdimensions of the same construct, or is there another rationale for their distinction?

Your insight would be greatly appreciated as it would help us better understand the dataset and ensure accurate processing. 

Thank you for your time and assistance. Looking forward to your response.

I’m happy to send the email myself and cc you, if that works better. Let me know what you think is best!

otherwise, if you think it doesn't really matter, then the above file should be ok to be QAed!