Problem Set #3

docx

School

Boston University *

*We aren’t endorsed by this school

Course

717

Subject

Economics

Date

Feb 20, 2024

Type

docx

Pages

Uploaded by AgentSnail2582

Problem Set 3: Measures of Disease Frequency and Association [100 points] Please adhere to the “Rules for Collaborating”. For most questions, you can provide a full and thoughtful answer in 1-2 sentences. Include your R code and show calculations for full credit. Part I. Eviction Lab Court-ordered eviction is the forced removal of people from rental properties. Evictions are a contributor to housing insecurity and are associated with mental distress among renters. Dr. Matthew Desmond is the principal investigator of Eviction Lab at Princeton University. His team collects national data to better understand the burden, causes, and consequences of eviction. We will use this data to quantify the eviction burden, and to explore associations between various factors and eviction. Before we begin, let’s review the general eviction process (which can vary by state): · Eviction filing – a landlord asks a court to begin eviction proceedings against a renter household · Eviction judgment – a judge hears testimony, reviews evidence, and renders a decision · Eviction enforcement – a court’s decision to evict a renter household is executed 1. The state of Maryland historically has one of the highest burdens of eviction in the U.S. In 2018, Maryland had about 2,230,527 total households (e.g., owner, renter, other), among which an estimated 803,000 people belonged to renter households. That same year, there were roughly 559,000 new eviction filings. a) Calculate the prevalence of eviction filings in Maryland in 2018. (5 points) Prevalence = 559,000/2,230,527 = 25% in 2018 b) Interpret the prevalence of eviction filings you calculated in part a. (5 points) The prevalence of eviction filings was 25% in Maryland in 2018 c) Calculate the 1-year risk of eviction filings in Maryland in 2018. Note: This measure is commonly referred to as the “eviction filing rate”, even though it is not a true rate. (5 points) Prevalence = 559,000 / 803,000 = 0.696 = 69.6% in 2018

d) Interpret the 1-year risk of eviction filings you calculated in part c. (5 points) The renters of Maryland in 2018 had 69.6% risk of being evicted 2. We have downloaded the 2016 Massachusetts Cities dataset from Eviction Lab. It represents data on 244 Massachusetts cities. The data can be found in the Eviction2016MACities.csv dataset. The variables are described in the table below: Variable Name Description name Name of the Massachusetts city population City population size in 2016 poverty.rate Percent of the population with income in the past year below poverty level pct.renter.occupied Percent of occupied housing units that are renter occupied median.gross.rent Median gross rent in city median.household.income Median household income in city median.property.value Median property value in city rent.burden % of income that is spent on rent (≥50% is displayed as 50%) pct.white Percent of city that identifies as White pct.af.am Percent of city that identifies as African American pct.hispanic Percent of city that identifies as Hispanic pct.am.ind Percent of city that identifies as Native American pct.asian Percent of city that identifies as Asian pct.nh.pi Percent of city that identifies as Native Hawaiian or Other Pacific Islander

pct.multiple Percent of city that identifies as more than one race pct.other Percent of city that identifies as “Other” race renter.occupied.households Count of renter occupied households eviction.filings The number of eviction cases in the city evictions The number of eviction judgements in which renters were ordered to leave eviction.rate Eviction rate in city (#evictions/#renter occupied households) eviction.filing.rate Eviction filing rate (#eviction filings/#renter occupied households) The variables in this dataset are continuous. We are going to derive 3 dichotomous variables so that we can examine factors associated with high eviction rates. The R code is provided for you. Read the dataset into RStudio using the following code: EvictedMA<-read.csv(file.choose()) attach(EvictedMA) According to the U.S. Census Bureau, 12.7% of the population was living in poverty in 2016. (This is referred to as the poverty rate). Let’s create a new dichotomous variable called “high.poverty”. Cities with poverty rates above 12.7% will be coded as 1, cities with poverty rates below this cut point will be coded as 0. Run the following code to create this new variable: high.poverty=ifelse(poverty.rate>12.7, 1, 0) #Cities with high poverty rates (above 12.7%) are coded as 1, cities with low poverty rates are coded as 0 Next, we are going to create a new variable called “high.rent”. Cities with high rent burden (rent is more than 30% of income) will be coded as 1, cities with low rent burden will be coded as 0. Run the following code to create this new variable: (0 points) high.rent=ifelse(rent.burden>30, 1, 0)

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

#Cities with high rent burden (above 30%) are coded as 1, cities with low rent burden are coded as 0 Finally, we are going to create a new variable called “confirmed.eviction”. Cities with eviction numbers above 1 will be coded as 1, cities with eviction numbers under 1 will be coded as 0. Run the following code to create this new variable: (0 points) confirmed.eviction=ifelse(evictions>1, 1, 0) #Cities with confirmed evictions (above 1) are coded as 1, cities with fewer evictions (meaning there are possible data entry errors) are coded as 0 a) What type of study design was used for the 2016 Massachusetts Cities study? Briefly justify your response. (5 points) The study design used for the 2016 Massachusetts Cities study is an ecologic study design as it uses population-level data to assess the overall frequency of the outcome in a series of populations (cities in this case). b) Fill in the descriptive statistics table below with the appropriate statistics based on the variable type. Copy and paste your code below. We are looking for 13 values. (7 points) table( high.poverty ) prop.table(table( high.poverty )) table( high.rent ) prop.table(table( high.rent )) table( confirmed.eviction ) prop.table(table( confirmed.eviction )) Variable Name Descriptive Statistics (n=243) High poverty (high.poverty) Yes No 31 (12.8%) 212 (87.2%) High rent burden (high.rent) Yes No 113 (48.5%) 120 (51.5%)

Confirmed eviction (confirmed.eviction) Yes No 199 (81.9%) 44 (18.1%) c) We are interested in assessing the association between high poverty and confirmed eviction in Massachusetts in 2016. Fill in the 2x2 table below using the table command in R. We are looking for 6 values. (6 points) table(confirmed.eviction, high.poverty) High poverty Low poverty Confirmed eviction 30 169 No eviction 1 43 Total 31 212 d) Calculate the prevalence difference for the association between high poverty, compared to low poverty, and confirmed eviction in Massachusetts in 2016. Report your answer per 100 cities. (5 points) Prevalence difference = (30/31) – (169/212) = 0.171 = 17.1 per 100 e) Interpret the prevalence difference you calculated in part d. (5 points) Among individuals living in high poverty, there were 17.1 excess cases of eviction per 100 compared to those living in low poverty in 2016. f) If high poverty had been prevented in 100 cities living in high poverty, how many cases of confirmed eviction might have been avoided in Massachusetts in 2016? (3 points) If high poverty had been prevented in 100 cities living in high poverty, 17.1 cases of confirmed eviction might have been avoided g) Let’s next focus on the association between high rent burden and confirmed eviction in Massachusetts in 2016. Create a 2x2 table using the table command in R. Include row and column labels. We are looking for 6

values, and 4 labels. Refer to part c to help you with formatting your table. (8 points) table(confirmed.eviction, high.rent) High rent Low rent Confirmed eviction 100 99 No eviction 13 21 Total 113 120 h) Calculate the prevalence ratio for the association between high rent burden, compared to low rent burden, and confirmed eviction in Massachusetts in 2016. (5 points) Prevalence ratio = (100/113) / (99/120) = 1.07 i) Interpret the prevalence ratio you calculated in part h. (5 points) The prevalence of confirmed eviction was 1.07 times greater in individuals with high rent than in individuals with low rent Part II. COVID-19 and Evictions 3. Read the following abstract: Importance Although evictions have been associated with adverse mental health outcomes, it remains unclear which stages of the eviction process are associated with mental distress among renters. Variation in COVID-19 pandemic eviction protections across US states enables identification of intervention targets within the eviction process to improve renters’ mental health. Objective To measure the association between the strength of eviction protections (ie, stages blocked by eviction moratoriums) and mental distress among renters during the COVID-19 pandemic. Design, Setting, and Participants This cohort study used individual-level, nationally representative data from the Understanding Coronavirus in America Survey to measure associations between state eviction moratorium protections and mental distress. The sample of 2317 respondents included renters with annual household incomes less than

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

$75 000 who reported no mental distress at baseline and were all followed between March 10 and September 3, 2020, prior to the federal eviction moratorium order by the Centers for Disease Control and Prevention. Exposures Strength of state moratorium protections as a categorical variable: none, weak (blocking court hearings, judgments, or enforcement without blocking notice or filing), or strong (blocking all stages of the eviction process beginning with notice and filing). Main Outcomes and Measures Moderate to severe mental distress was measured using the 4-item Patient Health Questionnaire. Linear regression models were adjusted for state COVID-19 incidence and mortality, public health restrictions, and unemployment rates. Results The sample consisted of 2317 individuals (20 853 total observations) composed largely (1788 [78%] weighted) of middle-aged adults (25-64 years of age) and women (1538 [60%]); 640 respondents (23%) self-reported as Hispanic or Latinx, 314 respondents (20%) as non-Hispanic Black, and 1071 respondents (48%) as non- Hispanic White race and ethnicity. Relative to no state-level eviction moratorium protections, strong protections were associated with a 12.6% relative reduction in the risk of mental distress, whereas weak protections were not associated with a statistically significant reduction (risk ratio, 0.96; 95% CI, 0.86-1.06). Conclusions and Relevance This analysis of the Understanding Coronavirus in America Survey data found that strong eviction moratoriums were associated with protection against mental distress, suggesting that distress begins early in the eviction process with notice and filing. This finding is consistent with the idea that to reduce mental distress among renters, policy makers should focus on primary prevention of evictions. a) The investigators reported risk ratios, not prevalence ratios. Why is risk the appropriate measure for mental distress occurrence in this study? Briefly justify your response. (5 points) Risk is the appropriate measure for mental distress as it used to measure participants that are followed over a period of time (in this case March to September), whereas prevalence is used to obtain a snapshot of a population in time. If we were to use prevalence, we would miss a lot of data. b) The investigators reported that relative to no state-level eviction moratorium protections, strong protections were associated with a 12.6% relative reduction in the risk of mental distress. This is the excess relative risk. What is the corresponding risk ratio for this association? (5 points) Excess Relative Risk = (RR – 1) x 100% -12.6 = (RR-1) x 100 -0.126 = RR -1

RR= 0.87 In other words, those with hypertension had 2.7 times the risk, which is the same as a 170% increase in risk compared to those without hypertension during the study period c) Suppose that there was substantial loss to follow-up among the 2317 participants in this study. Which measure of disease frequency ( prevalence, risk, or incidence rate ) would be most appropriate to report? Briefly justify your response. (5 points) Given there is substantial loss to follow up among the participants, Incidence rate would be the most appropriate to report because it considers losses to follow up, appropriate for dynamic populations, and populations with long follow-up times. Using Cumulative incidence would not be plausible as it is used for fixed populations with short follow up times, relatively few losses to follow up, and does not consider losses to follow up. 4. The variation in state eviction moratorium protections has also allowed researchers to study the impact of eviction on COVID-19 incidence and mortality. A study examined the association between having an eviction moratorium (yes or no), and the incidence rate of COVID-19 mortality in the U.S. In a sample of people living in states with no eviction moratorium during a 6-month period in 2020, there were 728 COVID-19 deaths over a total of 35,567 person-months. Among people living in states with an eviction moratorium during this same period, there were 4,042 COVID-19 deaths over a total of 1,034,699 person-months. a) Use the data provided to fill in the table below. Include row and column labels. We are looking for 4 values, and 3 labels. Refer to Question 2, part c to help you with formatting your table. (6 points) Eviction moratorium No eviction moratorium Covid-19 Deaths 4,042 728 Total person-months 1,034,699 35,567 b) Calculate the incidence rate ratio for the association between having no eviction moratorium, compared to having an eviction moratorium, and COVID- 19 deaths in the U.S. during this 6-month period in 2020. (5 points) Incidence Rate Ratio =(728/35,567) / (4,042/1,034,699) = 5.24

c) Interpret the incidence rate ratio you calculated in part b. (5 points) States without eviction moratorium had 5.24 times the rate of Covid-19 deaths compared to states with eviction moratorium during the 6-month study period in 2020

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

Problem Set #3

Related Documents