Variable Selection
Daniel Chen
November 14, 2017
library(forcats)
library(useful)## Loading required package: ggplot2library(glmnetUtils)
library(glmnet)## Loading required package: Matrix## Loading required package: foreach## Loaded glmnet 2.0-13##
## Attaching package: 'glmnet'## The following objects are masked from 'package:glmnetUtils':
##
## cv.glmnet, glmnetrm(list = ls())
df <- readRDS('./data/subset_recode.RDS')source('./analysis/chendaniely/model_utils.R')1 Reference Values
# lapply(lapply(df, levels), '[[', 1)# response variable levels
levels(df$Q13)## [1] "Yes, every year" "Yes, some years" "No, never"2 Subset response variables
never_every <- df[df$Q13 %in% c('Yes, every year', 'No, never'), ]
never_every$Q13 <- droplevels(never_every$Q13)
never_every$Q13 <- fct_relevel(never_every$Q13, 'No, never', after = 0L)
table(never_every$Q13, useNA = 'always')##
## No, never Yes, every year <NA>
## 819 908 0levels(never_every$Q13)## [1] "No, never" "Yes, every year"testthat::expect_equal(levels(never_every$Q13)[1], "No, never")never_some <- df[df$Q13 %in% c('Yes, some years', 'No, never'), ]
never_some$Q13 <- droplevels(never_some$Q13)
never_some$Q13 <- fct_relevel(never_some$Q13, 'No, never', after = 0L)
table(never_some$Q13, useNA = 'always')##
## No, never Yes, some years <NA>
## 819 423 0levels(never_some$Q13)## [1] "No, never" "Yes, some years"testthat::expect_equal(levels(never_some$Q13)[1], "No, never")never_someevery <- df[df$Q13 %in% c('Yes, every year', 'Yes, some years', 'No, never'), ]
never_someevery$Q13_nv_se <- as.character(never_someevery$Q13)
never_someevery$Q13_nv_se <- sapply(X = never_someevery$Q13_nv_se, FUN = recode_never_someevery)
never_someevery$Q13_nv_se <- as.factor(never_someevery$Q13_nv_se)
never_someevery$Q13_nv_se <- fct_relevel(never_someevery$Q13_nv_se, 'No, never', after = 0L)
table(never_someevery$Q13_nv_se, never_someevery$Q13,useNA = 'always')##
## Yes, every year Yes, some years No, never <NA>
## No, never 0 0 819 0
## Yes, some or every year 908 423 0 0
## <NA> 0 0 0 0levels(never_someevery$Q13_nv_se)## [1] "No, never" "Yes, some or every year"testthat::expect_equal(levels(never_someevery$Q13_nv_se)[1], "No, never")
never_someevery$Q13 <- never_someevery$Q13_nv_se
testthat::expect_equal(levels(never_someevery$Q13)[1], "No, never")3 Save out data
save(never_every, never_some, never_someevery, file = 'data/model_dataframes.RData')4 LASSO
Use LASSO to fit a model to use as a variable selection method
4.1 Never Every
x <- build.x(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work - 1,
data = never_every, contrasts = FALSE)
y <- build.y(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work,
data = never_every)
never_every_cv <- cv.glmnet(x = x, y = y, family = "binomial", nfolds = 10, weights = never_every$weight)plot(never_every_cv)
# coef(never_every_cv, s = 'lambda.min')4.2 Never Some
x <- build.x(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work - 1,
data = never_some, contrasts = FALSE)
y <- build.y(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work,
data = never_some)
never_some_cv <- cv.glmnet(x = x, y = y, family = "binomial", nfolds = 10, weights = never_some$weight)plot(never_some_cv)
# coef(never_some_cv, s = 'lambda.min')4.3 Never SomeEvery
x <- build.x(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work - 1,
data = never_someevery, contrasts = FALSE)
y <- build.y(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work,
data = never_someevery)
never_someevery_cv <- cv.glmnet(x = x, y = y, family = "binomial", nfolds = 10, weights = never_someevery$weight)plot(never_someevery_cv)
# coef(never_someevery_cv, s = 'lambda.min')4.4 Coeff Tables
coef_cv <- cbind(coef(never_every_cv, s = 'lambda.min'),
coef(never_some_cv, s = 'lambda.min'),
coef(never_someevery_cv, s = 'lambda.min'),
coef(never_every_cv, s = 'lambda.1se'),
coef(never_some_cv, s = 'lambda.1se'),
coef(never_someevery_cv, s = 'lambda.1se')
)
coef_cv <- as.data.frame(as.matrix(coef_cv))
coef_cv <- round(coef_cv, digits = 4)
coef_cv[coef_cv == 0] <- '.'
colnames(coef_cv) <- c('ne_min', 'ns_min', 'nse_min',
'ne_1se', 'ns_1se', 'nse_1se')
knitr::kable(coef_cv, digits = 2)| ne_min | ns_min | nse_min | ne_1se | ns_1se | nse_1se | |
|---|---|---|---|---|---|---|
| (Intercept) | -0.2854 | -0.604 | 0.2132 | -0.277 | -0.6188 | 0.3039 |
| PPGENDERFemale | 4e-04 | 0.0574 | 0.0629 | . | . | . |
| PPGENDERMale | . | . | -1e-04 | . | . | . |
| ppagecat18-24 | -0.1632 | 0.2377 | . | -0.0068 | . | . |
| ppagecat25-34 | -0.4156 | . | -0.1966 | -0.2977 | . | -0.096 |
| ppagecat35-44 | -0.186 | . | -0.1701 | -0.0681 | . | -0.0626 |
| ppagecat45-54 | . | . | . | . | . | . |
| ppagecat55-64 | 0.3827 | -0.2168 | 0.1284 | 0.2861 | . | . |
| ppagecat65-74 | 0.7812 | . | 0.5041 | 0.6732 | . | 0.313 |
| ppagecat75+ | 1.278 | -0.1683 | 0.8933 | 1.0712 | . | 0.5469 |
| PPEDUCATLess than high school | -0.0339 | -0.2665 | -0.1533 | . | . | . |
| PPEDUCATHigh school | . | -0.6302 | -0.1821 | . | -0.1718 | -0.011 |
| PPEDUCATSome college | . | . | . | . | . | . |
| PPEDUCATBachelor_s degree or higher | 0.4469 | 0.1004 | 0.3128 | 0.3703 | . | 0.3092 |
| PPETHMWhite, Non-Hispanic | . | . | . | . | . | . |
| PPETHMBlack, Non-Hispanic | -0.1732 | -0.0497 | -0.15 | -0.0468 | . | -0.0058 |
| PPETHMHispanic | . | . | . | . | . | . |
| PPETHMOther, Non-Hispanic | . | 0.0815 | 0.0024 | . | . | . |
| PPETHM2+ Races, Non-Hispanic | -0.1913 | -0.3549 | -0.4103 | . | . | . |
| incomeunder $10k | -0.3039 | -0.0835 | -0.3229 | -0.0962 | . | -0.1399 |
| income$10k to $25k | -0.3861 | -0.2455 | -0.373 | -0.2088 | . | -0.1975 |
| income$25k to $50k | -0.0933 | . | -0.1005 | . | . | . |
| income$50k to $75k | . | . | . | . | . | . |
| income$75k to $100k | . | . | . | . | . | . |
| income$100k to $150k | 0.2184 | 0.0667 | 0.1687 | 0.1369 | . | 0.0808 |
| incomeover $150k | 0.4586 | . | 0.3203 | 0.3367 | . | 0.1684 |
| maritalsingle | -0.1839 | -0.2617 | -0.2506 | -0.191 | . | -0.1912 |
| maritalpartnered | . | . | . | . | . | 4e-04 |
| PPMSACATMetro | . | 0.2311 | 0.1203 | . | . | . |
| PPMSACATNon-Metro | . | -0.0033 | . | . | . | . |
| PPREG4Midwest | -0.0221 | . | . | . | . | . |
| PPREG4Northeast | . | . | . | . | . | . |
| PPREG4South | . | -0.1469 | -0.0164 | . | . | . |
| PPREG4West | 0.036 | 0.3971 | 0.2439 | . | 0.0536 | 0.113 |
| workunemployed | 0.225 | . | 0.2038 | 0.1243 | . | 0.0738 |
| workemployed | -0.0283 | . | . | -0.0189 | . | . |
Using the lambda1se values we select the following demographic variables:
ppagecat PPEDUCAT PPETHM income marital work
5 Bootstrap LASSO
total <- 100
pb <- txtProgressBar(min = 0, max = total, style = 3)##
|
| | 0%all_kept_variables <- list()
for (i in 1:total) {
boot_never_every <- sample(x = 1:nrow(never_every), size = nrow(never_every), replace = TRUE)
boot_never_every <- never_every[boot_never_every, ]
boot_never_some <- sample(x = 1:nrow(never_some), size = nrow(never_some), replace = TRUE)
boot_never_some <- never_some[boot_never_some, ]
boot_never_someevery <- sample(x = 1:nrow(never_someevery), size = nrow(never_someevery), replace = TRUE)
boot_never_someevery <- never_someevery[boot_never_someevery, ]
x <- build.x(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work - 1,
data = boot_never_every, contrasts = FALSE)
y <- build.y(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work,
data = boot_never_every)
boot_never_every_cv <- cv.glmnet(x = x, y = y, family = "binomial", nfolds = 10, weights = boot_never_every$weight)
x <- build.x(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work - 1,
data = boot_never_some, contrasts = FALSE)
y <- build.y(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work,
data = boot_never_some)
boot_never_some_cv <- cv.glmnet(x = x, y = y, family = "binomial", nfolds = 10, weights = boot_never_some$weight)
x <- build.x(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work - 1,
data = boot_never_someevery, contrasts = FALSE)
y <- build.y(Q13 ~ PPGENDER + ppagecat + PPEDUCAT + PPETHM + income + marital +
PPMSACAT + PPREG4 + work,
data = boot_never_someevery)
boot_never_someevery_cv <- cv.glmnet(x = x, y = y, family = "binomial", nfolds = 10, weights = boot_never_someevery$weight)
coef_cv <- cbind(coef(boot_never_every_cv, s = 'lambda.min'),
coef(boot_never_some_cv, s = 'lambda.min'),
coef(boot_never_someevery_cv, s = 'lambda.min'),
coef(boot_never_every_cv, s = 'lambda.1se'),
coef(boot_never_some_cv, s = 'lambda.1se'),
coef(boot_never_someevery_cv, s = 'lambda.1se')
)
coef_cv <- as.data.frame(as.matrix(coef_cv))
coef_cv <- round(coef_cv, digits = 4)
coef_cv[coef_cv == 0] <- '.'
colnames(coef_cv) <- c('ne_min', 'ns_min', 'nse_min',
'ne_1se', 'ns_1se', 'nse_1se')
kept_variable <- coef_cv > 0
all_kept_variables[[i]] <- kept_variable
setTxtProgressBar(pb, i)
}##
|
|= | 1%
|
|= | 2%
|
|== | 3%
|
|=== | 4%
|
|=== | 5%
|
|==== | 6%
|
|===== | 7%
|
|===== | 8%
|
|====== | 9%
|
|====== | 10%
|
|======= | 11%
|
|======== | 12%
|
|======== | 13%
|
|========= | 14%
|
|========== | 15%
|
|========== | 16%
|
|=========== | 17%
|
|============ | 18%
|
|============ | 19%
|
|============= | 20%
|
|============== | 21%
|
|============== | 22%
|
|=============== | 23%
|
|================ | 24%
|
|================ | 25%
|
|================= | 26%
|
|================== | 27%
|
|================== | 28%
|
|=================== | 29%
|
|==================== | 30%
|
|==================== | 31%
|
|===================== | 32%
|
|===================== | 33%
|
|====================== | 34%
|
|======================= | 35%
|
|======================= | 36%
|
|======================== | 37%
|
|========================= | 38%
|
|========================= | 39%
|
|========================== | 40%
|
|=========================== | 41%
|
|=========================== | 42%
|
|============================ | 43%
|
|============================= | 44%
|
|============================= | 45%
|
|============================== | 46%
|
|=============================== | 47%
|
|=============================== | 48%
|
|================================ | 49%
|
|================================ | 50%
|
|================================= | 51%
|
|================================== | 52%
|
|================================== | 53%
|
|=================================== | 54%
|
|==================================== | 55%
|
|==================================== | 56%
|
|===================================== | 57%
|
|====================================== | 58%
|
|====================================== | 59%
|
|======================================= | 60%
|
|======================================== | 61%
|
|======================================== | 62%
|
|========================================= | 63%
|
|========================================== | 64%
|
|========================================== | 65%
|
|=========================================== | 66%
|
|============================================ | 67%
|
|============================================ | 68%
|
|============================================= | 69%
|
|============================================== | 70%
|
|============================================== | 71%
|
|=============================================== | 72%
|
|=============================================== | 73%
|
|================================================ | 74%
|
|================================================= | 75%
|
|================================================= | 76%
|
|================================================== | 77%
|
|=================================================== | 78%
|
|=================================================== | 79%
|
|==================================================== | 80%
|
|===================================================== | 81%
|
|===================================================== | 82%
|
|====================================================== | 83%
|
|======================================================= | 84%
|
|======================================================= | 85%
|
|======================================================== | 86%
|
|========================================================= | 87%
|
|========================================================= | 88%
|
|========================================================== | 89%
|
|========================================================== | 90%
|
|=========================================================== | 91%
|
|============================================================ | 92%
|
|============================================================ | 93%
|
|============================================================= | 94%
|
|============================================================== | 95%
|
|============================================================== | 96%
|
|=============================================================== | 97%
|
|================================================================ | 98%
|
|================================================================ | 99%
|
|=================================================================| 100%5.1 Looking at results
5.1.1 Raw Counts
var_counts <- Reduce("+", all_kept_variables)
var_counts## ne_min ns_min nse_min ne_1se ns_1se
## (Intercept) 2 0 85 3 0
## PPGENDERFemale 47 69 73 14 22
## PPGENDERMale 0 0 0 0 0
## ppagecat18-24 0 79 12 0 35
## ppagecat25-34 0 50 0 0 5
## ppagecat35-44 0 11 1 0 0
## ppagecat45-54 0 19 6 0 3
## ppagecat55-64 100 0 61 91 0
## ppagecat65-74 100 24 99 100 1
## ppagecat75+ 100 0 100 100 0
## PPEDUCATLess than high school 6 1 1 0 0
## PPEDUCATHigh school 0 0 0 0 0
## PPEDUCATSome college 0 16 1 0 4
## PPEDUCATBachelor_s degree or higher 99 78 98 99 70
## PPETHMWhite, Non-Hispanic 49 11 34 35 0
## PPETHMBlack, Non-Hispanic 1 6 4 0 0
## PPETHMHispanic 22 42 30 1 9
## PPETHMOther, Non-Hispanic 35 60 50 8 29
## PPETHM2+ Races, Non-Hispanic 0 0 0 0 0
## incomeunder $10k 2 6 1 0 0
## income$10k to $25k 0 0 0 0 0
## income$25k to $50k 1 6 0 0 0
## income$50k to $75k 26 23 25 1 6
## income$75k to $100k 29 54 44 10 12
## income$100k to $150k 90 62 91 69 30
## incomeover $150k 99 40 95 90 11
## maritalsingle 0 0 0 0 0
## maritalpartnered 11 10 9 10 7
## PPMSACATMetro 37 94 72 14 51
## PPMSACATNon-Metro 2 0 0 0 0
## PPREG4Midwest 9 26 5 0 1
## PPREG4Northeast 28 13 21 5 0
## PPREG4South 9 0 0 0 0
## PPREG4West 57 96 98 21 81
## workunemployed 97 42 98 81 1
## workemployed 0 0 0 0 0
## nse_1se
## (Intercept) 100
## PPGENDERFemale 23
## PPGENDERMale 0
## ppagecat18-24 3
## ppagecat25-34 0
## ppagecat35-44 0
## ppagecat45-54 0
## ppagecat55-64 25
## ppagecat65-74 99
## ppagecat75+ 100
## PPEDUCATLess than high school 0
## PPEDUCATHigh school 0
## PPEDUCATSome college 0
## PPEDUCATBachelor_s degree or higher 99
## PPETHMWhite, Non-Hispanic 17
## PPETHMBlack, Non-Hispanic 0
## PPETHMHispanic 4
## PPETHMOther, Non-Hispanic 18
## PPETHM2+ Races, Non-Hispanic 0
## incomeunder $10k 0
## income$10k to $25k 0
## income$25k to $50k 0
## income$50k to $75k 2
## income$75k to $100k 13
## income$100k to $150k 59
## incomeover $150k 77
## maritalsingle 0
## maritalpartnered 9
## PPMSACATMetro 38
## PPMSACATNon-Metro 0
## PPREG4Midwest 0
## PPREG4Northeast 1
## PPREG4South 0
## PPREG4West 73
## workunemployed 75
## workemployed 05.1.2 Counts > 50
var_counts > 50## ne_min ns_min nse_min ne_1se ns_1se
## (Intercept) FALSE FALSE TRUE FALSE FALSE
## PPGENDERFemale FALSE TRUE TRUE FALSE FALSE
## PPGENDERMale FALSE FALSE FALSE FALSE FALSE
## ppagecat18-24 FALSE TRUE FALSE FALSE FALSE
## ppagecat25-34 FALSE FALSE FALSE FALSE FALSE
## ppagecat35-44 FALSE FALSE FALSE FALSE FALSE
## ppagecat45-54 FALSE FALSE FALSE FALSE FALSE
## ppagecat55-64 TRUE FALSE TRUE TRUE FALSE
## ppagecat65-74 TRUE FALSE TRUE TRUE FALSE
## ppagecat75+ TRUE FALSE TRUE TRUE FALSE
## PPEDUCATLess than high school FALSE FALSE FALSE FALSE FALSE
## PPEDUCATHigh school FALSE FALSE FALSE FALSE FALSE
## PPEDUCATSome college FALSE FALSE FALSE FALSE FALSE
## PPEDUCATBachelor_s degree or higher TRUE TRUE TRUE TRUE TRUE
## PPETHMWhite, Non-Hispanic FALSE FALSE FALSE FALSE FALSE
## PPETHMBlack, Non-Hispanic FALSE FALSE FALSE FALSE FALSE
## PPETHMHispanic FALSE FALSE FALSE FALSE FALSE
## PPETHMOther, Non-Hispanic FALSE TRUE FALSE FALSE FALSE
## PPETHM2+ Races, Non-Hispanic FALSE FALSE FALSE FALSE FALSE
## incomeunder $10k FALSE FALSE FALSE FALSE FALSE
## income$10k to $25k FALSE FALSE FALSE FALSE FALSE
## income$25k to $50k FALSE FALSE FALSE FALSE FALSE
## income$50k to $75k FALSE FALSE FALSE FALSE FALSE
## income$75k to $100k FALSE TRUE FALSE FALSE FALSE
## income$100k to $150k TRUE TRUE TRUE TRUE FALSE
## incomeover $150k TRUE FALSE TRUE TRUE FALSE
## maritalsingle FALSE FALSE FALSE FALSE FALSE
## maritalpartnered FALSE FALSE FALSE FALSE FALSE
## PPMSACATMetro FALSE TRUE TRUE FALSE TRUE
## PPMSACATNon-Metro FALSE FALSE FALSE FALSE FALSE
## PPREG4Midwest FALSE FALSE FALSE FALSE FALSE
## PPREG4Northeast FALSE FALSE FALSE FALSE FALSE
## PPREG4South FALSE FALSE FALSE FALSE FALSE
## PPREG4West TRUE TRUE TRUE FALSE TRUE
## workunemployed TRUE FALSE TRUE TRUE FALSE
## workemployed FALSE FALSE FALSE FALSE FALSE
## nse_1se
## (Intercept) TRUE
## PPGENDERFemale FALSE
## PPGENDERMale FALSE
## ppagecat18-24 FALSE
## ppagecat25-34 FALSE
## ppagecat35-44 FALSE
## ppagecat45-54 FALSE
## ppagecat55-64 FALSE
## ppagecat65-74 TRUE
## ppagecat75+ TRUE
## PPEDUCATLess than high school FALSE
## PPEDUCATHigh school FALSE
## PPEDUCATSome college FALSE
## PPEDUCATBachelor_s degree or higher TRUE
## PPETHMWhite, Non-Hispanic FALSE
## PPETHMBlack, Non-Hispanic FALSE
## PPETHMHispanic FALSE
## PPETHMOther, Non-Hispanic FALSE
## PPETHM2+ Races, Non-Hispanic FALSE
## incomeunder $10k FALSE
## income$10k to $25k FALSE
## income$25k to $50k FALSE
## income$50k to $75k FALSE
## income$75k to $100k FALSE
## income$100k to $150k TRUE
## incomeover $150k TRUE
## maritalsingle FALSE
## maritalpartnered FALSE
## PPMSACATMetro FALSE
## PPMSACATNon-Metro FALSE
## PPREG4Midwest FALSE
## PPREG4Northeast FALSE
## PPREG4South FALSE
## PPREG4West TRUE
## workunemployed TRUE
## workemployed FALSE5.1.3 Proportion
var_prop <- var_counts / total
var_prop## ne_min ns_min nse_min ne_1se ns_1se
## (Intercept) 0.02 0.00 0.85 0.03 0.00
## PPGENDERFemale 0.47 0.69 0.73 0.14 0.22
## PPGENDERMale 0.00 0.00 0.00 0.00 0.00
## ppagecat18-24 0.00 0.79 0.12 0.00 0.35
## ppagecat25-34 0.00 0.50 0.00 0.00 0.05
## ppagecat35-44 0.00 0.11 0.01 0.00 0.00
## ppagecat45-54 0.00 0.19 0.06 0.00 0.03
## ppagecat55-64 1.00 0.00 0.61 0.91 0.00
## ppagecat65-74 1.00 0.24 0.99 1.00 0.01
## ppagecat75+ 1.00 0.00 1.00 1.00 0.00
## PPEDUCATLess than high school 0.06 0.01 0.01 0.00 0.00
## PPEDUCATHigh school 0.00 0.00 0.00 0.00 0.00
## PPEDUCATSome college 0.00 0.16 0.01 0.00 0.04
## PPEDUCATBachelor_s degree or higher 0.99 0.78 0.98 0.99 0.70
## PPETHMWhite, Non-Hispanic 0.49 0.11 0.34 0.35 0.00
## PPETHMBlack, Non-Hispanic 0.01 0.06 0.04 0.00 0.00
## PPETHMHispanic 0.22 0.42 0.30 0.01 0.09
## PPETHMOther, Non-Hispanic 0.35 0.60 0.50 0.08 0.29
## PPETHM2+ Races, Non-Hispanic 0.00 0.00 0.00 0.00 0.00
## incomeunder $10k 0.02 0.06 0.01 0.00 0.00
## income$10k to $25k 0.00 0.00 0.00 0.00 0.00
## income$25k to $50k 0.01 0.06 0.00 0.00 0.00
## income$50k to $75k 0.26 0.23 0.25 0.01 0.06
## income$75k to $100k 0.29 0.54 0.44 0.10 0.12
## income$100k to $150k 0.90 0.62 0.91 0.69 0.30
## incomeover $150k 0.99 0.40 0.95 0.90 0.11
## maritalsingle 0.00 0.00 0.00 0.00 0.00
## maritalpartnered 0.11 0.10 0.09 0.10 0.07
## PPMSACATMetro 0.37 0.94 0.72 0.14 0.51
## PPMSACATNon-Metro 0.02 0.00 0.00 0.00 0.00
## PPREG4Midwest 0.09 0.26 0.05 0.00 0.01
## PPREG4Northeast 0.28 0.13 0.21 0.05 0.00
## PPREG4South 0.09 0.00 0.00 0.00 0.00
## PPREG4West 0.57 0.96 0.98 0.21 0.81
## workunemployed 0.97 0.42 0.98 0.81 0.01
## workemployed 0.00 0.00 0.00 0.00 0.00
## nse_1se
## (Intercept) 1.00
## PPGENDERFemale 0.23
## PPGENDERMale 0.00
## ppagecat18-24 0.03
## ppagecat25-34 0.00
## ppagecat35-44 0.00
## ppagecat45-54 0.00
## ppagecat55-64 0.25
## ppagecat65-74 0.99
## ppagecat75+ 1.00
## PPEDUCATLess than high school 0.00
## PPEDUCATHigh school 0.00
## PPEDUCATSome college 0.00
## PPEDUCATBachelor_s degree or higher 0.99
## PPETHMWhite, Non-Hispanic 0.17
## PPETHMBlack, Non-Hispanic 0.00
## PPETHMHispanic 0.04
## PPETHMOther, Non-Hispanic 0.18
## PPETHM2+ Races, Non-Hispanic 0.00
## incomeunder $10k 0.00
## income$10k to $25k 0.00
## income$25k to $50k 0.00
## income$50k to $75k 0.02
## income$75k to $100k 0.13
## income$100k to $150k 0.59
## incomeover $150k 0.77
## maritalsingle 0.00
## maritalpartnered 0.09
## PPMSACATMetro 0.38
## PPMSACATNon-Metro 0.00
## PPREG4Midwest 0.00
## PPREG4Northeast 0.01
## PPREG4South 0.00
## PPREG4West 0.73
## workunemployed 0.75
## workemployed 0.005.1.4 Conclusion
drop:
- PPGENDER
- PPETHM
- PPMSACAT
- marital
keep:
- ppagecat
- PPEDUCAT
- income
- PPREG4
- work