Alternative Logit Specifications Stat HW 5

Data

library(haven)
setwd("/Volumes/Jyoti/Stat 2 /PROJECT")
ddi <- read_ipums_ddi("/Volumes/Jyoti/Stat 2 /PROJECT/nhis_00004.xml")
data <- read_ipums_micro(ddi)

## Use of data from IPUMS NHIS is subject to conditions including that users
## should cite the data appropriately. Use command `ipums_conditions()` for more
## details.

data<- haven::zap_labels(data)

names(data) <- tolower(gsub(pattern = "_",replacement =  "",x =  names(data)))

Recode the variables

#depression level
data$depfeelevl <- as.factor(data$depfeelevl)

data$depfeelevl<- car::Recode(data$depfeelevl,
                       recodes="1='alot'; 2='a little'; 3='between little and alot'; 7:9=NA; else=NA",
                       as.factor=T)
data$depfeelevl<-relevel(data$depfeelevl, ref='alot')


# medication for depression
data$deprx <- as.factor(data$deprx)

data$deprx<- car::Recode(data$deprx,
                       recodes="1='no'; 2='yes';else=NA",
                       as.factor=T)
data$deprx<-relevel(data$deprx, ref='yes')


#currently Pregnant 
data$pregnantnow<-as.factor(data$pregnantnow)
data$curpreg<-car::Recode(data$pregnantnow,
                          recodes="0='yes';else=NA",
                          as.factor=T)
        
#education level
data$educ<-as.factor(data$educ)
data$educ<-Recode(data$educ,
                        recodes="102 ='noschool'; 201='hsdiploma'; 301='somecol'; 
                       400= 'undergrad'; 501= 'masters';else=NA", as.factor = T)
data$educ<-fct_relevel(data$educ,ref = 'noschool' ) 

#employment status

data$empstat<- car::Recode(data$empstat,
                       recodes="100='Employed'; 200='unemployed';else=NA",
                       as.factor=T)
data$empstat<-relevel(data$empstat, ref='Employed')

Survey design

#First we tell R our survey design
options(survey.lonely.psu = "adjust")

library(dplyr)
sub<-data%>%
  select(depfeelevl, curpreg, educ, deprx, empstat,sampweight,strata) %>%
  filter( complete.cases(.))


#First we tell R our survey design

options(survey.lonely.psu = "adjust")
des<-svydesign(ids=~1,
               strata=~strata,
               weights=~sampweight,
               data =sub)

1. Define an ordinal or multinomial outcome variable of your choosing and define how you will recode the original variable.

Level of depression with 1 being a lot, 2 being a little and 3 being somewhere between little and a lot 1=‘alot’; 2=‘a little’; 3=‘between little and alot’;else=NA",

2. State a research question about what factors you believe will affect your outcome variable.

What is the association between education and level of depression in pregnant women?

Other predictor variables: 1:Currently taking medication for depression 2: Employment

3. Fit the ordinal or the multinomial logistic regression models to your outcome.

#Multinomial Model

mfit<-svy_vglm(depfeelevl~educ+empstat+deprx,
           family=multinomial(refLevel = 1),
           design = des)
mfit%>%
  tbl_regression()

## ! `broom::tidy()` failed to tidy the model.

## x No tidy method for objects of class svy_vglm

## ✓ `tidy_parameters()` used instead.

## ℹ Add `tidy_fun = broom.helpers::tidy_parameters` to quiet these messages.

## x Unable to identify the list of variables.
## 
## This is usually due to an error calling `stats::model.frame(x)`or `stats::model.matrix(x)`.
## It could be the case if that type of model does not implement these methods.
## Rarely, this error may occur if the model object was created within
## a functional programming framework (e.g. using `lappy()`, `purrr::map()`, etc.).

Characteristic	Beta	95% CI1	p-value
(Intercept):1	0.50	-0.60, 1.6	0.4
(Intercept):2	0.60	-0.41, 1.6	0.2
educhsdiploma:1	-0.84	-1.9, 0.26	0.13
educhsdiploma:2	-0.28	-1.3, 0.73	0.6
educmasters:1	-1.5	-2.6, -0.38	0.008
educmasters:2	-0.53	-1.6, 0.49	0.3
educsomecol:1	-0.70	-1.8, 0.40	0.2
educsomecol:2	-0.22	-1.2, 0.79	0.7
educundergrad:1	-1.4	-2.5, -0.32	0.012
educundergrad:2	-0.42	-1.4, 0.60	0.4
empstatunemployed:1	0.26	0.05, 0.47	0.015
empstatunemployed:2	-0.09	-0.22, 0.04	0.2
deprxno:1	-1.5	-1.7, -1.3	<0.001
deprxno:2	-0.84	-1.0, -0.69	<0.001
1 CI = Confidence Interval

3.1) Describe the results of your model

Results of the analysis show that compared to currently pregnant women who never went to school, women who had masters degree were less likely (250%) to respond “a little” or (47%) less likely to respond “between a little and a lot”. Similarly, compared to currently pregnant women who never went to school, women who had undergraduate degree are 140% less likely to respond “a little” or (42%) less likely to respond “between a little and a lot”. In cases where the pregnant women responded “a little” the results were statistically significant.

In terms of employment and depression level among pregnant women it was found that currently pregnant and unemployed women are 74% more likely to respond “a little” compared to employed women.

Finally, in terms of employment and depression level among pregnant women it was found that compared to pregnant women who are currently taking depression medication, pregnant women who are not taking medication are 250% less likely to respond “a little” or 16% less likely to respond “between a little and a lot”.