Import Dataset

library(readr)
library(expss)
Dissertation_Dataset <- read_csv("Dissertation_Dataset.csv", 
    col_types = cols(FED_STATE = col_factor(levels = c("Indian Affairs", "Federal", "Local", "State", "Private", "Regional")), HEALTH_CJ = col_factor(levels = c("Law Enforcement", "Health", "Generalist", "Other")), SENATE = col_factor(levels = c("0", "1", "2"))))
Dissertation_Dataset = apply_labels(Dissertation_Dataset,
                      HEALTH_CJ = "Health vs. Criminal Justice Groups",
                      FED_STATE = "Private, Public, & Level of Gov't")

Summary of Variables

The folllowing is a summary of descriptive statistics for the variables of interest. Please see the chart below it for a definition of each variable.

library(kableExtra)
library(psych)
options(knitr.table.format = "html")
Count_Data <- Dissertation_Dataset[7:49]
sum_table <- describe(Count_Data)

kable(cbind(sum_table), caption = "Summary Stats Count Variables", booktabs = T) %>%
    kable_styling(font_size=12,latex_options = c("striped","scale_down")) %>%
    row_spec(0, bold=T, color = "white", background = "black") %>%
    column_spec(1, bold=T)
Summary Stats Count Variables
vars n mean sd median trimmed mad min max range skew kurtosis se
HEALTH_CJ* 1 144 1.972222 0.8924269 2.0 1.844828 0.0000 1 4 3 0.9916517 0.4678727 0.0743689
FED_STATE* 2 144 3.395833 1.2858496 3.0 3.396552 1.4826 1 6 5 0.0042446 -1.4272499 0.1071541
ps_OPR 3 144 34.076389 110.6041312 0.0 6.939655 0.0000 0 705 705 4.5262394 21.9248207 9.2170109
ps_OPRS_H 4 144 57.243056 140.0133483 0.0 18.655172 0.0000 0 648 648 2.9813662 8.2867048 11.6677790
ps_prescrip 5 144 12.826389 44.3629862 0.0 1.215517 0.0000 0 276 276 4.4182595 20.7450086 3.6969155
ps_Prescribers 6 144 23.597222 63.6338213 0.0 7.000000 0.0000 0 389 389 3.5454209 13.6041732 5.3028184
ps_Prescrib_Ed 7 144 4.472222 21.1600443 0.0 0.000000 0.0000 0 169 169 5.6656580 34.4148461 1.7633370
ps_vital 8 144 17.694444 128.0398363 0.0 0.000000 0.0000 0 1443 1443 9.9286633 104.8480298 10.6699864
ps_Manuf 9 144 12.395833 78.6083697 0.0 0.000000 0.0000 0 878 878 9.5397241 99.7276241 6.5506975
ps_FDA 10 144 6.333333 57.4216401 0.0 0.000000 0.0000 0 667 667 10.7209997 119.0504687 4.7851367
ps_Left 11 144 7.826389 47.7619871 0.0 0.000000 0.0000 0 506 506 8.6277872 82.2819784 3.9801656
ps_BadApp 12 144 72.131944 174.1233887 0.0 27.120690 0.0000 0 1203 1203 3.6030090 15.2237202 14.5102824
ps_Foreign 13 144 96.937500 290.9585129 0.0 14.698276 0.0000 0 1951 1951 3.7315216 15.4344554 24.2465427
pd_Disease 14 144 38.340278 98.7730437 0.0 14.594828 0.0000 0 851 851 4.8186381 31.5997937 8.2310870
pd_Quality 15 144 10.687500 61.4633291 0.0 0.000000 0.0000 0 630 630 7.9383697 71.4752903 5.1219441
pd_PSE 16 144 19.791667 122.7181588 0.0 0.000000 0.0000 0 1384 1384 9.7738479 103.6113885 10.2265132
pd_CJ 17 144 41.680556 217.9217044 0.0 4.853448 0.0000 0 2444 2444 9.5353772 100.0991385 18.1601420
pd_MAT 18 144 22.118056 69.9126825 0.0 4.275862 0.0000 0 581 581 4.8128317 29.3526491 5.8260569
pd_Access 19 144 38.770833 141.5748940 0.0 9.051724 0.0000 0 1464 1464 7.5913989 69.7376950 11.7979078
ds_PSE 20 144 31.493056 89.2544094 0.0 7.241379 0.0000 0 518 518 3.5578856 13.1018563 7.4378675
ds_Prevent 21 144 154.145833 361.9478185 36.0 78.120690 53.3736 0 2718 2718 5.1494407 31.1674924 30.1623182
ds_Stigma 22 144 17.743056 65.6199853 0.0 1.172414 0.0000 0 459 459 4.6235712 22.7786603 5.4683321
ds_MAT 23 144 122.805556 263.4593765 0.0 54.008621 0.0000 0 1320 1320 2.7309331 7.5306121 21.9549480
ds_Access 24 144 134.881944 206.5048636 54.5 89.931034 80.8017 0 1236 1236 2.6928503 9.1255366 17.2087386
ds_Quality 25 144 35.326389 138.9626820 0.0 8.931035 0.0000 0 1479 1479 8.2980834 79.6697730 11.5802235
ds_ODR 26 144 98.326389 190.6399539 0.0 53.879310 0.0000 0 1222 1222 3.1220471 12.3221127 15.8866628
ds_Samari 27 144 5.916667 29.2910402 0.0 0.000000 0.0000 0 244 244 6.3417633 42.4056401 2.4409200
ds_Coordinate 28 144 31.159722 73.6317299 0.0 13.905172 0.0000 0 501 501 4.0834581 21.0974342 6.1359775
ds_Divert2PH 29 144 41.868056 124.1610868 0.0 8.672414 0.0000 0 926 926 4.2100316 21.0434041 10.3467572
ss_Take 30 144 24.784722 65.0401880 0.0 7.008621 0.0000 0 308 308 3.0990896 9.2591265 5.4200157
ss_PDMP 31 144 144.812500 263.5796603 1.5 85.637931 2.2239 0 1438 1438 2.6789505 8.1230675 21.9649717
ss_Reg 32 144 10.312500 51.4106556 0.0 0.000000 0.0000 0 488 488 6.7109009 52.6919868 4.2842213
ss_New 33 144 16.416667 89.3887166 0.0 0.000000 0.0000 0 611 611 5.5551223 30.1890068 7.4490597
ss_Guide 34 144 32.229167 81.2566269 0.0 8.741379 0.0000 0 454 454 2.9105477 8.3120832 6.7713856
ss_Cautious 35 144 28.770833 108.7894431 0.0 3.956897 0.0000 0 952 952 5.9681218 41.0786119 9.0657869
ss_Prescrib_Ed 36 144 85.965278 182.0856024 0.0 39.413793 0.0000 0 929 929 2.9086055 8.7521426 15.1738002
ss_Crim_Enforce 37 144 249.472222 613.4331219 0.0 100.431034 0.0000 0 4113 4113 4.1663114 20.6558597 51.1194268
ss_Penalt 38 144 4.236111 17.0006113 0.0 0.000000 0.0000 0 106 106 4.2486691 18.0373984 1.4167176
Problem_Supply 39 144 295.930556 461.8098730 80.5 194.198276 119.3493 0 2367 2367 1.9541657 3.5299150 38.4841561
Problem_Demand 40 144 159.715278 329.0143293 30.0 80.172414 44.4780 0 2444 2444 3.7703214 18.1663251 27.4178608
Solutions_Demand 41 144 557.534722 713.6461574 301.0 409.879310 365.4609 0 3885 3885 2.1642430 5.0281759 59.4705131
Solutions_Supply 42 144 548.722222 724.0372768 332.5 403.913793 433.6605 0 4113 4113 2.3823470 7.1560209 60.3364397
TOTAL WORDS 43 144 2444.604167 1379.3977059 2235.0 2285.965517 1238.7123 561 8342 7781 1.2483781 1.9935355 114.9498088 
text_tbl <- data.frame(
  Variables = c("HEALTH_CJ","FED_STATE","ps_OPR","ps_OPRS_H","ps_prescrip","ps_Prescribers",
      "ps_Prescrib_Ed","ps_vital","ps_Manuf","ps_FDA","ps_Left","ps_BadApp","ps_Foreign","pd_Disease",
      "pd_Quality","pd_PSE","pd_CJ","pd_MAT","pd_Access","ds_PSE","ds_Prevent","ds_Stigma","ds_MAT",
      "ds_Access","ds_Quality","ds_ODR","ds_Samari","ds_Coordinate","ds_Divert2PH","ss_Take","ss_PDMP",
      "ss_Reg","ss_New","ss_Guide","ss_Cautious","ss_Prescrib_Ed","ss_Crim_Enforce","ss_Penalt",
      "Problem_Supply","Problem_Demand","Solutions_Demand","Solutions_Supply","TOTAL WORDS"),
  Description = c(
      "Categorical Variable, which groups speaker orgs based on either health focus, criminal justice (CJ) focus, or other",
      "Categorical Variable, which groups speaker orgs based on either federal gov agency, state agency, local agency, regional group of agencies, or private entities",
      "Word count of problem definition (WCPD) blaming the characteristics of opioid prescriptions",
      "WCPD blaming opioid prescriptions for heroin use",
      "WCPD generally blaming overprescription of opioids",
      "WCPD blaming prescribers for overprescribing",
      "WCPD blaming lack of prescriber education",
      "WCPD blaming 5th vital sign of pain",
      "WCPD blaming drug manufacturers",
      "WCPD blaming FDA",
      "WCPD blaming diversion of left-over prescriptions",
      "WCPD blaming bad apples for increasing drug supply",
      "WCPD blaming foreign actors for increasing drug supply",
      "WCPD acknowleding addiction as a disease",
      "WCPD blaming poor quality of addiction treatment",
      "WCPD blaming psychological, sociological, environmental or economical (PSEE) factors",
      "WCPD blaming the mischaracterization of the problem as a CJ problem rather than a health problem",
      "WCPD blaming poor access to Medication Assisted Treatment (MAT)",
      "WCPD blaming poor access to Treatment Generally",
      "Word count of solutions (WCS) addressing PSEE factors",
      "WCS addressing demand side prevention (excluse preventioon of supply tactics)",
      "WCS addressing stigma",
      "WCS increasing access to MAT",
      "WCS increasing access to Treatment Generally",
      "WCS increasing quality of Treatment",
      "WCS increasing access to Overdose Reversal Medications",
      "WCS passage or strengthening of Good Samaritan Laws",
      "WCS coordinating between CJ and Health actors",
      "WCS that are alterntives to incarceration, like drug courts or treatment",
      "WCS involving drug take back programs",
      "WCS regarding Prescription Drug Monitoring Programs (PDMP)",
      "WCS involving rescheduling, adding black box labels, or regulation of opioids",
      "WCS funding new drugs to address pain, with the intent of decreasing the prescribing of opioids",
      "WCS promoting prescriber guidelines",
      "WCS calling for more cautious prescribing practices generally",
      "WCS of prescriber or distributor education",
      "WCS commitment to criminal enforcement",
      "WCS increasing or creating new criminal penalities",
      "Total WC of all subcategories that define the problem as an issue of drug supply",
      "Total WC of all subcategories that define the problem as the demand for drugs",
      "Total WC of all subcategories that propose soltions aimed at decreasing the demand",
      "Total WC of all subcategories that propose solutions aimed at decreasing the supply of drugs",
      "Total WC of all words in each case"
  )
)

kable(text_tbl, booktabs = T)%>% 
  kable_styling(font_size=10,latex_options = c("striped","scale_down")) %>%
  group_rows("Interest Group Type", 1, 2) %>%
  group_rows("Problem Definition - Supply", 3, 13) %>%
  group_rows("Problem Definition - Demand", 14, 19) %>%
  group_rows("Solutions - Demand", 20, 28) %>%
  group_rows("Problem Definition - Supply", 29, 37) %>%
  group_rows("Theme Definitions", 38, 42) %>%
  group_rows("Other", 43, 43) %>%
  column_spec(2, width = "40em") %>%
  row_spec(0, bold=T, color = "white", background = "black")
Variables Description
Interest Group Type
HEALTH_CJ Categorical Variable, which groups speaker orgs based on either health focus, criminal justice (CJ) focus, or other
FED_STATE Categorical Variable, which groups speaker orgs based on either federal gov agency, state agency, local agency, regional group of agencies, or private entities
Problem Definition - Supply
ps_OPR Word count of problem definition (WCPD) blaming the characteristics of opioid prescriptions
ps_OPRS_H WCPD blaming opioid prescriptions for heroin use
ps_prescrip WCPD generally blaming overprescription of opioids
ps_Prescribers WCPD blaming prescribers for overprescribing
ps_Prescrib_Ed WCPD blaming lack of prescriber education
ps_vital WCPD blaming 5th vital sign of pain
ps_Manuf WCPD blaming drug manufacturers
ps_FDA WCPD blaming FDA
ps_Left WCPD blaming diversion of left-over prescriptions
ps_BadApp WCPD blaming bad apples for increasing drug supply
ps_Foreign WCPD blaming foreign actors for increasing drug supply
Problem Definition - Demand
pd_Disease WCPD acknowleding addiction as a disease
pd_Quality WCPD blaming poor quality of addiction treatment
pd_PSE WCPD blaming psychological, sociological, environmental or economical (PSEE) factors
pd_CJ WCPD blaming the mischaracterization of the problem as a CJ problem rather than a health problem
pd_MAT WCPD blaming poor access to Medication Assisted Treatment (MAT)
pd_Access WCPD blaming poor access to Treatment Generally
Solutions - Demand
ds_PSE Word count of solutions (WCS) addressing PSEE factors
ds_Prevent WCS addressing demand side prevention (excluse preventioon of supply tactics)
ds_Stigma WCS addressing stigma
ds_MAT WCS increasing access to MAT
ds_Access WCS increasing access to Treatment Generally
ds_Quality WCS increasing quality of Treatment
ds_ODR WCS increasing access to Overdose Reversal Medications
ds_Samari WCS passage or strengthening of Good Samaritan Laws
ds_Coordinate WCS coordinating between CJ and Health actors
Problem Definition - Supply
ds_Divert2PH WCS that are alterntives to incarceration, like drug courts or treatment
ss_Take WCS involving drug take back programs
ss_PDMP WCS regarding Prescription Drug Monitoring Programs (PDMP)
ss_Reg WCS involving rescheduling, adding black box labels, or regulation of opioids
ss_New WCS funding new drugs to address pain, with the intent of decreasing the prescribing of opioids
ss_Guide WCS promoting prescriber guidelines
ss_Cautious WCS calling for more cautious prescribing practices generally
ss_Prescrib_Ed WCS of prescriber or distributor education
ss_Crim_Enforce WCS commitment to criminal enforcement
Theme Definitions
ss_Penalt WCS increasing or creating new criminal penalities
Problem_Supply Total WC of all subcategories that define the problem as an issue of drug supply
Problem_Demand Total WC of all subcategories that define the problem as the demand for drugs
Solutions_Demand Total WC of all subcategories that propose soltions aimed at decreasing the demand
Solutions_Supply Total WC of all subcategories that propose solutions aimed at decreasing the supply of drugs
Other
TOTAL WORDS Total WC of all words in each case

The chart below shows the breakdown of the number of cases (testimonies) for categories of health vs. criminal justice organizations.

library(ggplot2)
## 
## Attaching package: 'ggplot2'
## The following objects are masked from 'package:psych':
## 
##     %+%, alpha
## The following object is masked from 'package:expss':
## 
##     vars
library(tidyverse)
## ── Attaching packages ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── tidyverse 1.2.1 ──
## ✔ tibble  1.4.2     ✔ dplyr   0.7.8
## ✔ tidyr   0.8.2     ✔ stringr 1.3.1
## ✔ purrr   0.2.5     ✔ forcats 0.3.0
## ── Conflicts ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ ggplot2::%+%()     masks psych::%+%()
## ✖ ggplot2::alpha()   masks psych::alpha()
## ✖ dplyr::between()   masks expss::between()
## ✖ dplyr::compute()   masks expss::compute()
## ✖ dplyr::filter()    masks stats::filter()
## ✖ dplyr::first()     masks expss::first()
## ✖ stringr::fixed()   masks expss::fixed()
## ✖ purrr::keep()      masks expss::keep()
## ✖ dplyr::lag()       masks stats::lag()
## ✖ dplyr::last()      masks expss::last()
## ✖ purrr::modify()    masks expss::modify()
## ✖ purrr::modify_if() masks expss::modify_if()
## ✖ dplyr::na_if()     masks expss::na_if()
## ✖ tidyr::nest()      masks expss::nest()
## ✖ dplyr::recode()    masks expss::recode()
## ✖ stringr::regex()   masks expss::regex()
## ✖ purrr::transpose() masks expss::transpose()
## ✖ dplyr::vars()      masks ggplot2::vars(), expss::vars()
library(forcats)
q <- ggplot(Dissertation_Dataset, aes(HEALTH_CJ)) +
  geom_bar(width = 0.8, fill = "green") +
  geom_text(stat='count', aes(label=..count..), vjust=-0.75)
q + scale_x_discrete(limits=c("Generalist", "Other", "Law Enforcement", "Health"))

The amount of testimony offerred by health actors was almost double that of law enforcement actors (78 cases vs. 43). Comparatively, both law enforcement and health groups far outweighed the amount of testimony offerred by generalists (n=7) or other advocacy groups (n=16).

Aside from categorizing groups as law enforcement or health groups, the groups could be categorized based on public or private groups.

d <- ggplot(data = Dissertation_Dataset, aes(x = FED_STATE)) + 
  geom_bar(fill="steelblue") +
  geom_text(stat='count', aes(label=..count..), vjust=-0.75)
d + scale_x_discrete(limits=c("Regional", "Indian Affairs", "Local", "State", "Private", "Federal"))

The amount of testimony provided by public actors greatly exceed that offerred by private actors (105 v. 39), with state, local and regional group testimony exceeding federal government testimony, but not by a great deal (54 v. 47).

If we combine the two ways of categorizing the groups testifying, Health v. Crimina Justice and Public vs. Private, we get the following: Federal Law Enforcement testimony = 29 Private Health group testimony = 27 Federal Health agency testimony = 17 State Law Enforcement testimony = 14 + Local Law Enforcement testimony = 13 + Regional Law Enforcement testimony=1 (Total = 28) State Health Agency testimony = 11 + Local Health Agency testimony = 7 (Total= 18)

z <- ggplot(Dissertation_Dataset, aes(FED_STATE, fill = HEALTH_CJ)) +
  geom_bar()
z + scale_x_discrete(limits=c("Regional", "Indian Affairs", "Local", "State", "Private", "Federal"))

Another way to display the information would be to look at the composition of groups as a proportion of each. See below.

ggplot(data = Dissertation_Dataset,aes(x = FED_STATE, fill = HEALTH_CJ)) + geom_bar(position = "fill")

Which house received the most testimony?

The Senate received almost double the amount of testimony as the house (n=89 vs. n=49). There was also 6 testimonies that were given to both the House and the Senate. This would be good to mention when describing the dataset. 0=House 1=Senate 2=Both

senate_tab <-table(Dissertation_Dataset$SENATE)
senate_tab
## 
##  0  1  2 
## 49 89  6
kable(cbind(senate_tab), caption = "Count of Senate Testimony", booktabs = T) %>%
    kable_styling(font_size=12)
Count of Senate Testimony
senate_tab
0 49
1 89
2 6

It seems from the graph below that the proprtion of law enforcement to health testimony varied by the House and the Senate. Since each body hears a separate set of testimony, it may be useful in the future to disaggregate and analyze separately the House vs. Senate testimony. Although, in this case, both the House and Senate voted near unanimously to CARA, so the different composition of testimony may not have been influential. Then again, CARA was introduced in the Senate and the House had produced a flurry of opioid bills at the same time. I believe they were later combined in reconciliation. Therefore, future analysis could compare the House vs. Senate testimony and the proposals that were supported by each house.

ggplot(Dissertation_Dataset, aes(SENATE, fill = HEALTH_CJ)) +
  geom_bar(position = "dodge")

#Proportion of FED_STATE that were HEALTH_CJ

ggplot(data = Dissertation_Dataset,aes(x = FED_STATE, fill = HEALTH_CJ)) + geom_bar(position = "fill")

#Findings of Overall Saliency of Themes: Supply vs. Demand; Problem vs. Solutions

ggplot(data=Dissertation_Dataset, aes(x = HEALTH_CJ, y = Problem_Demand, fill = FED_STATE)) +
  geom_bar(stat = "identity", position = "dodge") +
  labs(
    title = "Number of Words Describing the Problem as Demand",
    subtitle = "Comparing Types of Actors Problem Definition",
    caption = "Measured in word counts",
    x = "Health vs. Criminal Justice",
    y = "Word Count",
    fill = "Level"
    )

p <- ggplot(data=Dissertation_Dataset, aes(x = FED_STATE, y = Problem_Demand)) +
  geom_bar(stat = "identity", fill="Blue")
p + scale_x_discrete(limits=c("Regional", "Indian Affairs", "Federal", "State", "Local"))
## Warning: Removed 39 rows containing missing values (position_stack).

ggplot(Dissertation_Dataset, aes(x=Problem_Supply, y=Solutions_Supply)) +
    geom_point(shape=1)     # Use hollow circles

    #geom_smooth()            # Add a loess smoothed fit curve with confidence region
    #geom_smooth(method=lm, se=FALSE) #Add linear regression line,Don't add shaded confidence region
ggplot(Dissertation_Dataset, aes(x=Problem_Demand, y=Solutions_Demand)) +
    geom_point(shape=1)     # Use hollow circles

    #geom_smooth()            # Add a loess smoothed fit curve with confidence region
    #geom_smooth(method=lm, se=FALSE) #Add linear regression line,Don't add shaded confidence region

```