Living Well
Mark LaCour
2025-03-07
Importing, cleaning, coding, wrangling data
dat <- read.csv("~/Desktop/Junk/Google Drive/Research/--Research -- Advance/0Distrust/wave1/wave1.coded.csv")
dat$avg.pc=rowMeans(dat[,96:102])
is.male=ifelse(dat$gender=="Male",1,0)
is.other.gender=ifelse(dat$gender!="Male" & dat$gender!="Female",1,0)
is.white=ifelse(dat$race=="White",1,0)
is.black=ifelse(dat$race=="Black or African American",1,0)
is.asian=ifelse(dat$race=="Asian",1,0)
is.hispanic=ifelse(dat$hispanic=="Hispanic or Latino or Spanish Origin",1,0)
is.married=ifelse(dat$marital.status=="Married",1,0)
is.christian=ifelse(dat$religion=="Christian (Catholic, Methodist, Episcopalian, etc.)",1,0)
dat$is.christian=is.christian
is.none=ifelse(dat$religion=="Not religious (e.g., atheist, agnostic)",1,0)
dat$is.none=is.none
is.other=ifelse(dat$is.christian==0 & dat$is.none==0,1,0)
dat$number.children=ifelse(dat$number.children=="None",0,dat$number.children)
dat$number.children=ifelse(dat$number.children=="5+",5,dat$number.children)
dat$number.n.house=ifelse(dat$number.n.house=="5+",5,dat$number.n.house)
# nrow(dat) # 991
# sum(dat$age>100,na.rm=T) # 2 participants probably made a typo when putting their age in
dat=dat[dat$age<100,] # removing themSome demographic, descriptive statistics
# mean(dat$age,na.rm=T) avg. age = 45.53575
# sd(dat$age,na.rm=T) sd = 15.6828Average age was 45.54, SD = 15.68
#table(dat$gender)
# Female: 488, 0.4924319
# Male: 470, 0.4742684
# Other: 33, 0.0332997
# table(dat$race)
# Asian: 59, 0.05953582
# Black: 130, 0.1311806
# White: 673, 0.679112
# Other: 129, 0.1301715
# table(dat$hispanic)
# Hispanic: 131, 0.1321897
# Not hisp: 844, 0.851665
# Didn't say: 16, 0.01614531
# table(dat$income)/nrow(dat)
# $0-$29,999 183, 0.184662
# $30,000-$59,999 238, 0.2401615
# $60,000-$89,999 220, 0.221998
# $90,000-$119,999 138, 0.1392533
# $120,000-$149,999 98, 0.09889001
# $150,000+ 112, 0.1130172
#table(dat$education)
# Some high school, 12, 0.01210898
# High school diploma, or, 147, 0.14833502
# Some college 212, 0.21392533
# Associate degree 116, 0.11705348
# Bachelor’s Degree 344, 0.34712412
# Master’s Degree 123, 0.12411705
# Doctorate Degree or 37, 0.03733602Which variables to focus on?
#colnames(dat)
# age, education, income, pols, avg.pcI looked at the variables that were recorded in this data set. Unfortunately, we didn’t ask about whether respondents lived in a rural/suburban/urban context or whether they lived in a red/blue/swing state. That would’ve been helpful. But age, education, and income seem like promising starts. “pols” = political affiliation. Most Louisianians are conservative, and I assume this rural area will be more conservative than some other parts (e.g., New Orleans). “avg.pc” is a global score of “patient centered care”. Patient centered care is very big in healthcare. It measures the degree to which respondents feel their medical care paid attention to them, explained things clearly, didn’t rush them, etc. I feel like this is going to be higher among wealthier people / areas.
Different sources of hesitancy towards formal healthcare systems
1. “I promptly go to the doctor (or hospital) when I think I need to.”
summary(lm(scale(dat$avoid1.R)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid1.R) ~ dat$age + dat$education +
## dat$income + dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.8717 -0.7098 -0.2043 0.7677 2.4766
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.919216 0.183051 10.485 <2e-16 ***
## dat$age -0.001542 0.002002 -0.770 0.4413
## dat$education -0.027955 0.023149 -1.208 0.2275
## dat$income -0.037646 0.020417 -1.844 0.0655 .
## dat$pols 0.020049 0.025563 0.784 0.4331
## dat$avg.pc -0.457327 0.037844 -12.085 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9281 on 944 degrees of freedom
## (39 observations deleted due to missingness)
## Multiple R-squared: 0.1477, Adjusted R-squared: 0.1432
## F-statistic: 32.72 on 5 and 944 DF, p-value: < 2.2e-16The only significant predictor for endorsing this statement is patient-centered experiences. The more patient-centered expeirences they have, the less strongly they endorse this statement.
2. “I have avoided or delayed going to the hospital (or doctor).”
summary(lm(scale(dat$avoid2)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid2) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.3992 -0.7642 0.2230 0.6617 1.9566
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 2.117445 0.182141 11.625 < 2e-16 ***
## dat$age -0.008471 0.001992 -4.253 2.32e-05 ***
## dat$education -0.052740 0.023034 -2.290 0.0223 *
## dat$income -0.029752 0.020315 -1.465 0.1434
## dat$pols -0.000561 0.025436 -0.022 0.9824
## dat$avg.pc -0.389388 0.037656 -10.341 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9235 on 944 degrees of freedom
## (39 observations deleted due to missingness)
## Multiple R-squared: 0.146, Adjusted R-squared: 0.1415
## F-statistic: 32.29 on 5 and 944 DF, p-value: < 2.2e-16Being older means people endorse this stagement less strongly. Having higher educational attainment is also related to less agreement with this statement. As is patient-centered experiences.
3. “Some people avoid visiting their doctor even when they suspect they should go. Would you say this is true for you, or not true for you?”
summary(lm(scale(dat$avoid3)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid3) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.28424 -0.77474 0.06807 0.76996 2.05270
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.951336 0.182964 10.665 < 2e-16 ***
## dat$age -0.008339 0.002001 -4.168 3.36e-05 ***
## dat$education -0.014765 0.023138 -0.638 0.52355
## dat$income -0.056173 0.020407 -2.753 0.00603 **
## dat$pols 0.049533 0.025551 1.939 0.05285 .
## dat$avg.pc -0.401039 0.037826 -10.602 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9276 on 944 degrees of freedom
## (39 observations deleted due to missingness)
## Multiple R-squared: 0.1469, Adjusted R-squared: 0.1424
## F-statistic: 32.51 on 5 and 944 DF, p-value: < 2.2e-16Older people are less likely to agree with this statement. Same for wealthier people and people with more patient-centered experiences.
4. “The doctors/hospitals available to me are underfunded.”
summary(lm(scale(dat$avoid4)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid4) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.02465 -0.66790 -0.02555 0.58571 2.48384
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.093242 0.194057 5.634 2.33e-08 ***
## dat$age -0.001111 0.002121 -0.524 0.6005
## dat$education -0.000909 0.024515 -0.037 0.9704
## dat$income -0.054577 0.021622 -2.524 0.0118 *
## dat$pols -0.063057 0.027073 -2.329 0.0201 *
## dat$avg.pc -0.195499 0.040100 -4.875 1.27e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9828 on 943 degrees of freedom
## (40 observations deleted due to missingness)
## Multiple R-squared: 0.041, Adjusted R-squared: 0.03592
## F-statistic: 8.064 on 5 and 943 DF, p-value: 1.846e-07People with higher income, who are more conservative, and who have higher patient-centered experiences are less likely to endorse this statement. It’d be really great to see if any of these effects interact with whether they live in a rural/suburban/urban area.
5. “I worry about burnout and overworked healthcare workers.”
summary(lm(scale(dat$avoid5)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid5) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.6382 -0.5937 0.2039 0.7407 1.8483
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.001355 0.191774 5.222 2.18e-07 ***
## dat$age -0.006950 0.002098 -3.313 0.000958 ***
## dat$education 0.013357 0.024247 0.551 0.581868
## dat$income 0.021500 0.021389 1.005 0.315061
## dat$pols -0.124858 0.026778 -4.663 3.57e-06 ***
## dat$avg.pc -0.132820 0.039637 -3.351 0.000837 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.972 on 943 degrees of freedom
## (40 observations deleted due to missingness)
## Multiple R-squared: 0.05622, Adjusted R-squared: 0.05121
## F-statistic: 11.23 on 5 and 943 DF, p-value: 1.543e-10Older people, more conservative people, and people with more patient-centered experiences are less likely to endorse this statement.
Uniqueness neglect
For this item, we averaged endorsements of the following 4 statements together:
- “Healthcare workers do not spend enough time speaking with me when I visit.”
- “Healthcare workers do not respect my opinions or concerns.”
- “Healthcare workers are too quick to compare my situation with other patients who aren’t like me.”
- “Healthcare workers don’t give enough consideration to my unique situation.”
uniqueness=scale(rowMeans(cbind(dat$avoid6,dat$avoid7,dat$avoid8,dat$avoid9)))
summary(lm(uniqueness~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = uniqueness ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.78280 -0.52399 -0.05161 0.49478 3.00072
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.062e+00 1.472e-01 20.809 <2e-16 ***
## dat$age -7.285e-05 1.606e-03 -0.045 0.964
## dat$education -2.014e-02 1.859e-02 -1.083 0.279
## dat$income -4.077e-03 1.638e-02 -0.249 0.804
## dat$pols 3.314e-02 2.054e-02 1.614 0.107
## dat$avg.pc -8.348e-01 3.039e-02 -27.470 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.7439 on 940 degrees of freedom
## (43 observations deleted due to missingness)
## Multiple R-squared: 0.4522, Adjusted R-squared: 0.4493
## F-statistic: 155.2 on 5 and 940 DF, p-value: < 2.2e-16The only correlate of “uniqueness neglect” is patient-centered experiences.
10. “Some medical trends are very new and untested.”
summary(lm(scale(dat$avoid9)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid9) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.7744 -0.5696 -0.1027 0.5333 2.7705
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 2.737573 0.162376 16.859 <2e-16 ***
## dat$age -0.002199 0.001771 -1.242 0.2147
## dat$education -0.042875 0.020503 -2.091 0.0368 *
## dat$income 0.005833 0.018080 0.323 0.7471
## dat$pols 0.035788 0.022637 1.581 0.1142
## dat$avg.pc -0.704569 0.033530 -21.013 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.8212 on 942 degrees of freedom
## (41 observations deleted due to missingness)
## Multiple R-squared: 0.3328, Adjusted R-squared: 0.3293
## F-statistic: 93.99 on 5 and 942 DF, p-value: < 2.2e-16People with higher educational attainment and more patient-centered experiences tend to endorse this item more strongly.
11. “Too many things in the medical field keep changing.”
summary(lm(scale(dat$avoid11)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid11) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.08704 -0.75782 -0.01839 0.77085 2.39574
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.282753 0.191419 1.477 0.140
## dat$age 0.003110 0.002096 1.484 0.138
## dat$education -0.017474 0.024225 -0.721 0.471
## dat$income -0.011918 0.021351 -0.558 0.577
## dat$pols 0.155817 0.026741 5.827 7.74e-09 ***
## dat$avg.pc -0.193852 0.039578 -4.898 1.14e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9705 on 942 degrees of freedom
## (41 observations deleted due to missingness)
## Multiple R-squared: 0.0633, Adjusted R-squared: 0.05833
## F-statistic: 12.73 on 5 and 942 DF, p-value: 5.415e-12People who are more conservative tend to endorse this statement more strongly. People with more patient-centered experiences tend to endorse it less strongly.
12. “Mental healthcare professionals (e.g., psychiatrists, counselors) promote awareness of mental health disorders so they can make more money on clients.”
summary(lm(scale(dat$avoid12)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid12) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.0653 -0.7183 -0.1279 0.6607 2.6742
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.627865 0.182075 3.448 0.000589 ***
## dat$age 0.002654 0.001991 1.333 0.182878
## dat$education -0.055092 0.023026 -2.393 0.016923 *
## dat$income -0.030832 0.020308 -1.518 0.129290
## dat$pols 0.243370 0.025427 9.571 < 2e-16 ***
## dat$avg.pc -0.286216 0.037642 -7.604 6.94e-14 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9231 on 944 degrees of freedom
## (39 observations deleted due to missingness)
## Multiple R-squared: 0.1542, Adjusted R-squared: 0.1498
## F-statistic: 34.43 on 5 and 944 DF, p-value: < 2.2e-16People with higher educational attainment and more patient-centered experiences are tend to endorse this statement less strongly. People who are more conservative tend to endorse it more strongly.
13. “I sometimes hesitate to see a doctor because I don’t have childcare (one of the options should be “I don’t have children”)”
child.only=subset(dat,dat$dat.number.children !="0",1,0) # only people with a non-zero amount of children included
summary(lm(scale(dat$avoid13)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
,child.only)
)##
## Call:
## lm(formula = scale(dat$avoid13) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc, data = child.only)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.4071 -0.6091 -0.3021 0.2722 3.2229
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.371553 0.231030 5.937 4.69e-09 ***
## dat$age -0.018293 0.002523 -7.251 1.15e-12 ***
## dat$education 0.072192 0.028047 2.574 0.0103 *
## dat$income -0.022353 0.024849 -0.900 0.3687
## dat$pols 0.009471 0.030350 0.312 0.7551
## dat$avg.pc -0.208371 0.046476 -4.483 8.65e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9419 on 664 degrees of freedom
## (319 observations deleted due to missingness)
## Multiple R-squared: 0.1132, Adjusted R-squared: 0.1065
## F-statistic: 16.94 on 5 and 664 DF, p-value: 8.751e-16Older people and people with more patient-centered experience tend to endorse this statement less strongly. People with higher educational attainment tend to endorse it more strongly.
14. I sometimes hesitate to see a doctor or go to the hospital because I worry about the financial costs.
summary(lm(scale(dat$avoid14)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid14) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.4423 -0.8309 0.2592 0.7530 1.7689
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 2.030513 0.183483 11.066 < 2e-16 ***
## dat$age -0.013086 0.002007 -6.519 1.15e-10 ***
## dat$education -0.026017 0.023197 -1.122 0.262336
## dat$income -0.071706 0.020475 -3.502 0.000483 ***
## dat$pols -0.026993 0.025640 -1.053 0.292699
## dat$avg.pc -0.284812 0.037945 -7.506 1.41e-13 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9299 on 942 degrees of freedom
## (41 observations deleted due to missingness)
## Multiple R-squared: 0.1361, Adjusted R-squared: 0.1315
## F-statistic: 29.69 on 5 and 942 DF, p-value: < 2.2e-16Older people, people with higher income, and people with fewer patient-centered expriences tend to endorse this item less strongly.
15. I sometimes hesitate to see a doctor or go to the hospital because I worry about missing work.
summary(lm(scale(dat$avoid15)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid15) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.8511 -0.7685 -0.1704 0.8052 2.0125
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.625158 0.185134 8.778 < 2e-16 ***
## dat$age -0.015701 0.002026 -7.752 2.34e-14 ***
## dat$education 0.029846 0.023406 1.275 0.203
## dat$income -0.014208 0.020660 -0.688 0.492
## dat$pols 0.003540 0.025870 0.137 0.891
## dat$avg.pc -0.274318 0.038286 -7.165 1.57e-12 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9383 on 942 degrees of freedom
## (41 observations deleted due to missingness)
## Multiple R-squared: 0.1224, Adjusted R-squared: 0.1177
## F-statistic: 26.27 on 5 and 942 DF, p-value: < 2.2e-16Older people and those with higher patient-centered care tend to endorse this statement less strongly.
16. “I sometimes hesitate to see a doctor or go to the hospital because I worry about my insurance.”
summary(lm(scale(dat$avoid16)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid16) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.33384 -0.84220 0.08532 0.79186 1.94851
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.897568 0.185904 10.207 < 2e-16 ***
## dat$age -0.009195 0.002035 -4.519 7.01e-06 ***
## dat$education 0.002359 0.023559 0.100 0.92027
## dat$income -0.063055 0.020745 -3.040 0.00243 **
## dat$pols -0.049341 0.025943 -1.902 0.05749 .
## dat$avg.pc -0.321027 0.038450 -8.349 2.43e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9416 on 942 degrees of freedom
## (41 observations deleted due to missingness)
## Multiple R-squared: 0.1169, Adjusted R-squared: 0.1122
## F-statistic: 24.95 on 5 and 942 DF, p-value: < 2.2e-16Older people, those with higher income, and those with more patient-centered experiences tend to ednorse this statement less strongly.
17. “I sometimes hesitate to see a doctor or go to the hospital because I worry that I will be embarrassed.”
summary(lm(scale(dat$avoid17)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid17) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.7892 -0.7251 -0.2803 0.6527 2.5008
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.629744 0.185983 8.763 < 2e-16 ***
## dat$age -0.012379 0.002037 -6.076 1.79e-09 ***
## dat$education 0.043300 0.023529 1.840 0.0660 .
## dat$income -0.045869 0.020744 -2.211 0.0273 *
## dat$pols -0.010417 0.026004 -0.401 0.6888
## dat$avg.pc -0.296238 0.038450 -7.704 3.32e-14 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9429 on 943 degrees of freedom
## (40 observations deleted due to missingness)
## Multiple R-squared: 0.1113, Adjusted R-squared: 0.1066
## F-statistic: 23.63 on 5 and 943 DF, p-value: < 2.2e-16Older people, those with higher income and those with more patient-centered experiences endorse them item less strongly.
18. “I sometimes hesitate to see a doctor or go to the hospital because I don’t have easy access to transportation.”
summary(lm(scale(dat$avoid18)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid18) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.6093 -0.6277 -0.2422 0.3144 3.1082
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 2.071048 0.180694 11.462 < 2e-16 ***
## dat$age -0.007553 0.001976 -3.822 0.000141 ***
## dat$education -0.015777 0.022851 -0.690 0.490085
## dat$income -0.166056 0.020154 -8.239 5.74e-16 ***
## dat$pols -0.084019 0.025234 -3.330 0.000903 ***
## dat$avg.pc -0.258328 0.037356 -6.915 8.60e-12 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9161 on 944 degrees of freedom
## (39 observations deleted due to missingness)
## Multiple R-squared: 0.1607, Adjusted R-squared: 0.1562
## F-statistic: 36.15 on 5 and 944 DF, p-value: < 2.2e-16Older people, those with higher income, those who are more conservative, and people with more patient-centered experiences tend to endorse this item less strongly.
Fearing bad news
Endorsements of the following items were averaged together:
- I sometimes hesitate to see a doctor or go to the hospital because they might give me “bad news”
- I avoid seeing my doctor because I fear I may have a serious illness
- I avoid seeing my doctor because it makes me think about dying.
bad.news=scale(rowMeans(cbind(dat$avoid19,dat$avoid22,dat$avoid23)))
summary(lm(scale(bad.news)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(bad.news) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.6762 -0.8033 -0.1731 0.7643 2.4740
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.413505 0.190108 7.435 2.35e-13 ***
## dat$age -0.003736 0.002079 -1.797 0.0727 .
## dat$education -0.020942 0.024024 -0.872 0.3836
## dat$income -0.012441 0.021195 -0.587 0.5574
## dat$pols -0.010686 0.026564 -0.402 0.6876
## dat$avg.pc -0.300221 0.039270 -7.645 5.15e-14 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9626 on 941 degrees of freedom
## (42 observations deleted due to missingness)
## Multiple R-squared: 0.07019, Adjusted R-squared: 0.06525
## F-statistic: 14.21 on 5 and 941 DF, p-value: 2.015e-13People with more patient-centered experiences tended to endorse these statements less strongly.
20. I sometimes hesitate to see a doctor or go to the hospital because I might feel embarrassed when they weigh me.
summary(lm(scale(dat$avoid20)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid20) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.2303 -0.7143 -0.4622 0.4776 2.6908
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.6793517 0.1937938 3.506 0.000477 ***
## dat$age -0.0030202 0.0021184 -1.426 0.154281
## dat$education 0.0457155 0.0244827 1.867 0.062175 .
## dat$income 0.0185951 0.0216074 0.861 0.389683
## dat$pols 0.0008783 0.0270399 0.032 0.974095
## dat$avg.pc -0.2178051 0.0400315 -5.441 6.76e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9815 on 943 degrees of freedom
## (40 observations deleted due to missingness)
## Multiple R-squared: 0.03942, Adjusted R-squared: 0.03432
## F-statistic: 7.739 on 5 and 943 DF, p-value: 3.805e-07People with more patient-centered experiences tend to endorse this item less strongly.
21. “I avoid seeing my doctor because I feel uncomfortable when my body is being examined”
summary(lm(scale(dat$avoid21)~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = scale(dat$avoid21) ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.8543 -0.8145 -0.2259 0.9085 2.3187
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.543645 0.189867 8.130 1.34e-15 ***
## dat$age -0.006745 0.002076 -3.249 0.0012 **
## dat$education 0.002911 0.024011 0.121 0.9035
## dat$income -0.035811 0.021177 -1.691 0.0912 .
## dat$pols -0.031245 0.026515 -1.178 0.2389
## dat$avg.pc -0.288800 0.039253 -7.357 4.07e-13 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9626 on 944 degrees of freedom
## (39 observations deleted due to missingness)
## Multiple R-squared: 0.07907, Adjusted R-squared: 0.07419
## F-statistic: 16.21 on 5 and 944 DF, p-value: 2.325e-15Older people and those with more patient-centered experiences tend to endorse this statement less strongly.
Alternative medicine
The following items were averaged together:
- I prefer self-care or alternative care (e.g., the body will heal itself, alternative therapies)
- I believe in complementary and alternative medicine (e.g., herbal treatments, chiropractic).
- Natural remedies are better than drugs produced by the pharmaceutical industry.
alt.med=scale(rowMeans(cbind(dat$avoid24,dat$avoid25,dat$avoid26)))
summary(lm(alt.med~
dat$age+dat$education+dat$income+dat$pols+dat$avg.pc
))##
## Call:
## lm(formula = alt.med ~ dat$age + dat$education + dat$income +
## dat$pols + dat$avg.pc)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.16303 -0.72548 0.01783 0.64988 2.53472
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.283717 0.186290 1.523 0.128098
## dat$age 0.007018 0.002039 3.443 0.000601 ***
## dat$education 0.004242 0.023523 0.180 0.856947
## dat$income -0.043816 0.020742 -2.112 0.034910 *
## dat$pols 0.194592 0.026006 7.483 1.67e-13 ***
## dat$avg.pc -0.270697 0.038483 -7.034 3.85e-12 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9428 on 942 degrees of freedom
## (41 observations deleted due to missingness)
## Multiple R-squared: 0.1176, Adjusted R-squared: 0.1129
## F-statistic: 25.11 on 5 and 942 DF, p-value: < 2.2e-16People with higher income and more patient-centered experiences tend to like alternative medicine a bit less. Older people, and those who are more conservative tend to like alternative medicine a bit more.
Ordinal models
Relative trust
# Standardizing and coding some variables, not age though, as this is more interpretable in its normal units
dat$income=scale(dat$income)
dat$education=scale(dat$education)
dat$avg.pc=scale(dat$avg.pc)
dat$pols=scale(dat$pols)To what degree do people trust healthcare providers (HCPs) over social media users?
library(MASS)
# HCP VS. SOCIAL MEDIA USERS
pref=ifelse(dat$users=="Mostly trust the healthcare provider",1,dat$users)
pref=ifelse(dat$users=="Slightly favor the healthcare provider",2,pref)
pref=ifelse(dat$users=="Weight the provider and the users' advice equally",3,pref)
pref=ifelse(dat$users=="Slightly favor the online users",4,pref)
pref=ifelse(dat$users=="Mostly trust the online users",5,pref)
pref=factor(as.numeric(pref),levels=1:5)## Warning in factor(as.numeric(pref), levels = 1:5): NAs introduced by coerciondat$pref=pref
barplot(table(pref),ylim=c(0,700)) # # 900 x 640
Most people endorse 1 = “Mostly trust the healthcare provider”. Far fewer people endorse 2 = “Slightly favor the healthcare provider”.
dat$income=scale(dat$income)
dat$education=scale(dat$education)
dat$avg.pc=scale(dat$avg.pc)
dat$pols=scale(dat$pols)
dat$health.insurance=ifelse(dat$health.insurance=="Yes",1,0)
ordinal.mod=polr(pref~
age+income+education+avg.pc+pols,
dat
,Hess=TRUE)
# Getting some p-values
ctable <- coef(summary(ordinal.mod))
p <- pnorm(abs(ctable[, "t value"]), lower.tail = FALSE) * 2
ctable <- cbind(ctable, "p value" = p)
round(ctable,3)## Value Std. Error t value p value
## age -0.024 0.006 -4.173 0.000
## income -0.110 0.093 -1.185 0.236
## education -0.016 0.093 -0.171 0.864
## avg.pc -0.642 0.089 -7.225 0.000
## pols 0.380 0.086 4.422 0.000
## 1|2 0.154 0.262 0.590 0.556
## 2|3 2.233 0.295 7.566 0.000
## 3|4 2.233 0.295 7.567 0.000
## 4|5 3.717 0.414 8.984 0.000Age is a significant predictor of endorsing “lower” categories (more on the side of HCPs). Income and education are not signficantly related. Average patient centered experiences “avg.pc” is a significant predicter of endorsing “lower” categories (more on the side of HCPs). People who are more conservative (higher in “pols”) are more likely to endorse “higher” categories (more on the side of social media users.)
require(effects)## Loading required package: effects## Loading required package: carData## lattice theme set by effectsTheme()
## See ?effectsTheme for details.plot(Effect(focal.predictors = c("age"),
xlevels=list(age=seq(18,100,length.out=20)),
mod = ordinal.mod),
rug = FALSE,
style="stacked",
ylab="p(response)",xlab="Age")## Warning in Analyze.model(focal.predictors, mod, xlevels, default.levels, : the
## predictors income, education, avg.pc, pols are one-column matrices that were
## converted to vectors
plot(Effect(focal.predictors = c("avg.pc"),
xlevels=list(avg.pc=seq(-2.5,2.5,length.out=20)),
mod = ordinal.mod),
rug = FALSE,
style="stacked",
ylab="p(response)",xlab="avg.pc")## Warning in Analyze.model(focal.predictors, mod, xlevels, default.levels, : the
## predictors income, education, avg.pc, pols are one-column matrices that were
## converted to vectors
plot(Effect(focal.predictors = c("pols"),
xlevels=list(pols=seq(-2.5,2.5,length.out=20)),
mod = ordinal.mod),
rug = FALSE,
style="stacked",
ylab="p(response)",xlab="Liberal --> Conservative")## Warning in Analyze.model(focal.predictors, mod, xlevels, default.levels, : the
## predictors income, education, avg.pc, pols are one-column matrices that were
## converted to vectors
To what degree do people trust HCPs over friends and family?
pref=ifelse(dat$friends.family=="Mostly trust the healthcare provider",1,dat$friends.family)
pref=ifelse(dat$friends.family=="Slightly favor the healthcare provider",2,pref)
pref=ifelse(dat$friends.family=="Weight the provider and my friends and family equally",3,pref)
pref=ifelse(dat$friends.family=="Slightly favor my friends and family",4,pref)
pref=ifelse(dat$friends.family=="Mostly trust my friends and family",5,pref)
pref=factor(as.numeric(pref),levels=1:5)
dat$pref=pref
barplot(table(pref),ylim=c(0,700))
There are a lot more people who either balance HPCs with friends and
family (3) or weight the advice of friends and family over HPCs to some
degree (4+)
ordinal.mod=polr(pref~
age+income+education+avg.pc+pols,
dat
,Hess=TRUE)
# Getting some p-values
ctable <- coef(summary(ordinal.mod))
p <- pnorm(abs(ctable[, "t value"]), lower.tail = FALSE) * 2
ctable <- cbind(ctable, "p value" = p)
round(ctable,3)## Value Std. Error t value p value
## age -0.016 0.004 -3.759 0.000
## income -0.137 0.071 -1.938 0.053
## education -0.008 0.072 -0.117 0.907
## avg.pc -0.521 0.067 -7.720 0.000
## pols 0.397 0.067 5.883 0.000
## 1|2 -0.435 0.206 -2.111 0.035
## 2|3 0.781 0.208 3.757 0.000
## 3|4 1.902 0.225 8.452 0.000
## 4|5 3.700 0.330 11.218 0.000Higher age is still associated with more trust towards HPCs. Neither income nor education are significantly related to HPCs vs. friends and family. More patient-centered care expeirences is associated with more trust in HPCs. Being more conservative leads to more trust in frieds and family compared to HPCs
plot(Effect(focal.predictors = c("age"),
xlevels=list(age=seq(20,100,length.out=20)),
mod = ordinal.mod),
rug = FALSE,
style="stacked",
ylab="p(response)",xlab="Age")## Warning in Analyze.model(focal.predictors, mod, xlevels, default.levels, : the
## predictors income, education, avg.pc, pols are one-column matrices that were
## converted to vectors
plot(Effect(focal.predictors = c("avg.pc"),
xlevels=list(avg.pc=seq(-2.5,2.5,length.out=20)),
mod = ordinal.mod),
rug = FALSE,
style="stacked",
ylab="p(response)",xlab="avg.pc")## Warning in Analyze.model(focal.predictors, mod, xlevels, default.levels, : the
## predictors income, education, avg.pc, pols are one-column matrices that were
## converted to vectors
plot(Effect(focal.predictors = c("pols"),
xlevels=list(pols=seq(-2.5,2.5,length.out=20)),
mod = ordinal.mod),
rug = FALSE,
style="stacked",
ylab="p(response)",xlab="Liberal --> Conservative")## Warning in Analyze.model(focal.predictors, mod, xlevels, default.levels, : the
## predictors income, education, avg.pc, pols are one-column matrices that were
## converted to vectors
To what degree do people trust HCPs over doing their own research??
pref=ifelse(dat$own.research=="Mostly trust the healthcare provider",1,dat$own.research)
pref=ifelse(dat$own.research=="Slightly favor the healthcare provider",2,pref)
pref=ifelse(dat$own.research=="Weight the provider and my own research equally",3,pref)
pref=ifelse(dat$own.research=="Slightly favor my own research",4,pref)
pref=ifelse(dat$own.research=="Mostly trust my own research",5,pref)
pref=factor(as.numeric(pref),levels=1:5)
dat$pref=pref
barplot(table(pref),ylim=c(0,700))
This data so far shows the most variability, but the majority still
place the most wait in HPCs over doing their own research.
ordinal.mod=polr(pref~
age+income+education+avg.pc+pols,
dat
,Hess=TRUE)
# Getting some p-values
ctable <- coef(summary(ordinal.mod))
p <- pnorm(abs(ctable[, "t value"]), lower.tail = FALSE) * 2
ctable <- cbind(ctable, "p value" = p)
round(ctable,3)## Value Std. Error t value p value
## age -0.006 0.004 -1.571 0.116
## income -0.126 0.066 -1.918 0.055
## education -0.051 0.069 -0.743 0.458
## avg.pc -0.717 0.066 -10.911 0.000
## pols 0.332 0.064 5.227 0.000
## 1|2 -0.665 0.200 -3.327 0.001
## 2|3 0.778 0.200 3.882 0.000
## 3|4 2.103 0.215 9.787 0.000
## 4|5 3.463 0.266 13.005 0.000Age is not a predictor of whether people place more trust in HPCs or their own research. People with higher patient-centered experiences are more likely to trust HPCs relative to their own resarch. People who are more conservative are more likely to trust their own research relative to HPCs
plot(Effect(focal.predictors = c("avg.pc"),
xlevels=list(avg.pc=seq(-2.5,2.5,length.out=20)),
mod = ordinal.mod),
rug = FALSE,
style="stacked",
ylab="p(response)",xlab="avg.pc")## Warning in Analyze.model(focal.predictors, mod, xlevels, default.levels, : the
## predictors income, education, avg.pc, pols are one-column matrices that were
## converted to vectors
plot(Effect(focal.predictors = c("pols"),
xlevels=list(pols=seq(-2.5,2.5,length.out=20)),
mod = ordinal.mod),
rug = FALSE,
style="stacked",
ylab="p(response)",xlab="Liberal --> Conservative")## Warning in Analyze.model(focal.predictors, mod, xlevels, default.levels, : the
## predictors income, education, avg.pc, pols are one-column matrices that were
## converted to vectors