First
Grading Rubric for Assignment One
Project Component
Points
Introduction
Choose a main outcome variable that is continuous:
• BMI
• urinary arsenic
• blood lead
• blood cadmium
• blood mercury
• blood selenium
• blood manganese
• number of alcoholic drinks per day
• number of days per year with 4 – 5 alcoholic drinks
5
Provide background information on chosen outcome variable
• Define health concerns attached to this variable
• Include any national prevalence and incidence estimates related to this variable
• Include proper in-text citations in AMA style
15
Define a proper research question, with a main exposure variable and the chosen outcome variable
5
Provide a summary of other risk factors and covariates that may affect this exposure-outcome relationship
• For each covariate, briefly describe how it may be causally related to the outcome of interest
10
Describe any disparities in the outcome of interest based on identified covariates (e.g., do prevalence data only exist for 2 race groups, or only exist for 1 age group, or similar limitations?)
10
State a testable hypothesis that includes:
• Exposure of interest
• Outcome of interest
• Potential Confounders or Effect Modifiers*
10
Included 5 – 10 references supporting the background research into exposure and outcome variables and related covariates
• Must be formatted in AMA style, 10th edition
10
Methods
Present initial sample size and explain how final sample size will be smaller
• Typically due to application of inclusion and exclusion criteria, which should be defined here
5
Present study design and explain the appropriateness of this design for the research question
5
Explain data collection procedures for each of the following:
• Exposure of interest
• Outcome of interest
• All covariates of interest, especially those identified as potential confounders or effect modifiers
10
Explain analysis plan including
• Univariate analysis
• Bivariate analysis
• Stratified analysis
• Multivariable linear analysis
10
Include which software and version will be used to conduct all analyses
5
Total Points
100
*Since students have only covered these topics superficially, we are mainly looking for them to identify covariates that may influence the exposure-outcome relationship. It is less important that they correctly identify a variable as a “confounder” at this stage than they identify that a variable may compromise our ability to measure the exposure-outcome relationship.
SECOND
Grading Rubric for Assignment Two
Project Component
Points
Table and Graph
Table has a title with the following:
• Study name
• Study location
• Years of data collection
• Study sample size (n=)
5
Table has been formatted to GPH standards:
• Single spacing for all tables
• ½ inch margins on all pages
• Arial 11-point font
5
Table presentation is professional:
• All variable levels appear together on the same page (do not have the variable name at the end of one page and some of its levels on the next page)
• All footnotes appear on the same page as their variables
5
Variables are formatted correctly:
• Table contains all 33 variables in the dataset
• Variables are properly labeled (NOT dataset variable names)
10
All continuous variables have the following statistics:
• Mean
• Standard deviation
• Range
• Median
• Interquartile Range
• Assessment of normality of distribution in table footnotes
15
All categorical variables have the following statistics:
• Frequency (n=)
• Percent of total
• Labels and values for all of their categories
15
One graph (histogram, boxplot, or scatterplot) showing the distribution of data for the continuous outcome variable, including:
• correctly labeled x-axis
• correctly labeled y-axis
• accurate graph title with variable name
10
Narrative
Narrative has been formatted to GPH standards:
• 1.5 spacing for all text
• ½ inch margins on all pages
• Arial 11-point font
5
One paragraph for the graph of the continuous outcome variable, including:
• Description of the findings, including an explanation of the graph
• Identifying any outliers on the graph and describing methods of addressing these outliers in future analyses (for example, sample restriction)
10
Two paragraphs for the univariate analysis table, including:
• Description of the sample by person, place, and time
• Descriptions of the findings for your main exposure and main outcome variables
• Descriptions of the findings for any covariates that had unusual or noteworthy values (for example, very high or very low values)
• Descriptions of the findings for any covariates that you plan to include in future analyses (for example, demographic variables)
• List of any instances where variable levels were collapsed or variables were recoded, if those exist
20
Total Points
100
THIRD
Grading Rubric for Assignment Three
Project Component
Points
Exposure Variable Table and Outcome Variable Table
Both tables have titles with the following:
• Type of analysis (bivariate)
• Main variable of interest (either exposure or outcome)
• Study name and location
• Years of data collection
• Study sample size (n=)
5
Tables have been formatted to GPH standards:
• Single spacing for all tables
• ½ inch margins on all pages
• Arial 11-point font
5
Presentation of tables is professional:
• All variable levels appear together on the same page (do not have the variable name at the end of one page and some of its levels on the next page)
• All footnotes appear on the same page as their variables
• Report significant findings as ‘p < 0.001’ not as ‘p = 0.000’
5
Variables are formatted correctly:
• Each table contains the same chosen 10 – 15 variables
• Variables are properly labeled (NOT dataset variable names)
• Table is independently formatted, not cut-and-paste Stata output
10
A complete outcome variable table:
• All 10 – 15 variables, including the exposure variable, all relevant covariates, and all demographic variables
• Frequency (n=), percent of total, mean, and standard deviation for ALL levels of ALL variables
• Use of proper statistics (likely t-test or ANOVA)
• Footnotes indicating test performed
• Exact p-values reported for each variable
20
A complete exposure variable table:
• All 10 – 15 variables, including the outcome variable, all relevant covariates, and all demographic variables
• Frequency (n=) and percent of total for ALL levels of ALL variables, and for both exposed and unexposed groups
• Use of proper statistics (likely Pearson’s Χ2, although please use Fisher’s exact test for small cell sizes)
• Footnotes indicating test performed
• Exact p-values reported for each variable
20
Narrative
Narrative has been formatted to GPH standards:
• 1.5 spacing for all text
• ½ inch margins on all pages
• Arial 11-point font
5
One to two paragraphs for the outcome variable table, including:
• Description of the findings for the exposure variable, including the direction of the association
• Description of the findings for the demographic variables
• Description of the findings for any variables that are unusual or noteworthy
• Identify any variables which might be possible confounders or effect modifiers
15
One to two paragraphs for the exposure variable table, including:
• Description of the findings for the outcome variable, including the direction of the association
• Description of the findings for the demographic variables
• Description of the findings for any variables that are unusual or noteworthy
• Identify any variables which might be possible confounders or effect modifiers
15
Total Points
100
FORUTH AND LAST
Individual Project – Spring 2022
OVERVIEW
Students will identify a research question and write a background and methods section including an analytic plan, and then conduct univariate, bivariate, stratified, and multivariable analyses to test their hypotheses. They will write up final results and conclusions after the final analysis. Students will receive extensive feedback on all assignments so that they are prepared to conduct each subsequent analysis and ultimately report their findings in the style of peer-reviewed publications. Students will employ the NHANES dataset described below to identify a research question and complete this project.
DATA
For this project, instructors will provide a subset of data from the 2015-2016 National Health and Nutrition Examination Survey (NHANES) administered by the CDC. In addition, students will be provided with a codebook that contains information on all relevant variables.
This is an individual-level project. Students are to do all of the work independently. Any questions on assignments should be directed to the instructor or course assistant.
ANALYTIC RESOURCES
All analyses are to be conducted in STATA. STATA can be accessed through the virtual computer lab. If you are unfamiliar with STATA, there are introductory and ‘data wrangling’ STATA class offered by the NYU Library. The schedule can be found here.
TOPIC
Each student will identify ONE environmental exposure of interest and develop a research question to examine the relationship between that chosen environmental exposure and a continuous outcome variable. For an example of a research paper from the Annual Review of Public Health using this dataset, please see:
Hiatt RA, Brody JG. Environmental determinants of breast cancer. Annual Review of Public Health. 2018 Apr 1;39:113-33.
Continuous variables in this dataset are:
• BMI
• urinary arsenic
• blood lead
• blood cadmium
• blood mercury
• blood selenium
• blood manganese
• number of alcoholic drinks per day
• number of days per year with 4 - 5 alcoholic drinks
FORMATTING
Note: for all written assignments, the following formatting guidelines must be used:
• 1.5 spacing for all text
• Single spacing for all tables
• ½ inch margins on all pages
• Arial 11-point font
TABLES
• Table templates have been provided and students are strongly encouraged to use these.
• STATA output will not be accepted; submitting STATA output as a table will result in a grade of ‘0’ for that portion of the assignment. You must transfer your analytic findings into neat, correctly labeled tables preferably using the templates provided.
• All tables should:
◦ be numbered (Table 1, Table 2, etc.) and have complete titles with a clear explanation of the analysis, study sample size, study name if applicable, sampling location, and study years;
◦ be single-spaced;
◦ have column labels that are clear and concise;
◦ have row labels for variable names that use the complete word, with any acronyms explained in a Table footnote;
◦ not use dataset variable names as labels;
◦ use leading zeros (e.g., 0.01 not .01); and,
◦ be consistent with the number of significant numerical places (2-digit places to the right of the decimal for proportions, means, and measures of association and 3-digit places to the right for p-values are sufficient).
Assignment One: Introduction and Methods (due 2/7 by 10pm ET)
Introduction:
Once you have identified a main outcome variable, you will need to review the literature (you may use references from the Ann Rev PH paper) and other relevant reviews/papers to develop the “Introduction” section of your paper. The goal here is to begin the process of thinking about how you can clearly and succinctly state the “problem,” provide essential background data on the risk factors and covariates to put your study into context, and finally develop concrete and testable research questions with related hypotheses.
For the introduction, please write 3 – 4 paragraphs that include:
• information on your chosen outcome variable, including any national prevalence and incidence estimates
• a brief summary of the risk factors and covariates proposed in the research question and a clear explanation of how they may be casually related to the outcome of interest
• an explanation of the disparities in the outcome of interest based on identified covariates and as related to your research question
• clearly stated and testable research question with appropriate hypothesis/hypotheses that include exposure of interest, outcome of interest, potential confounders, and potential effect measure modifiers
• references – minimum of 5, maximum of 10 – that provide support for the above information and are formatted in AMA Manual of Style, 10th Edition
Methods:
The methods section of an epidemiological research paper is considered the ‘road map’ of the study in so far as it should provide sufficient detail for another investigator to be able to replicate the study.
For the methods section, please write 3 – 4 paragraphs that include:
• explanation of initial and final sample size and any inclusion/exclusion criteria applied to arrive at final sample size
• explanation of the study design and justification of the appropriateness of the selected design
• explanation of data collection procedures for each of the following: main exposure of interest, main outcome of interest, and all covariates of interest – and identify which ones are potential confounders or effect measure modifiers
• explanation of analysis plan – including clear explanation of univariable, bivariable, and stratified analysis (including testing for confounding and effect measure modification), and multivariable linear regression
• include software and version on which analyses will be conducted
Assignment Two: Univariate Analysis (due 2/28 by 10pm ET)
Conduct univariate analyses on all variables identified in your research question and analytic plan, including your main exposure, outcome, all demographic variables, and all relevant covariates. Please use the table templates provided.
When working with continuous variables:
a. for tables: present means (SE), medians (IQR), minimum/maximum ranges for all variables of interest;
b. prepare graphs showing the distribution of the data for each continuous variable
c. identify any outliers and explain how you will handle these data points.
When working with categorical variables:
a. for tables: present breakdown of n(%) for each category within the variable
b. if any additional collapsing/recoding of categories is to be conducted, explain these decisions.
The following components should be submitted in one document:
1. one table with the appropriate univariate statistics for exposure, outcome, and each covariate you are using
2. graphs demonstrating the distribution of each continuous variable you will be using
3. about 1 – 2 paragraphs for the table describing your findings, and one paragraph for the graphs describing it.
Assignment Three: Bivariate Analysis (due 4/4 by 10pm ET)
Conduct bivariable analyses for the exposure and outcome variables using all variables identified in the research question and analysis plan.
When analyzing continuous covariates, students should plan to use the t-test or ANOVA, whichever is more appropriate. When analyzing categorical covariates, students should plan to use Pearson’s Χ2 (or Fisher’s exact test for small cell sizes) unless otherwise indicated. The UCLA IDRE site can offer guidance on which statistical test to choose.
Table templates for bivariate analyses have been provided, and students should utilize these.
The following components should be submitted in one document:
1. two tables with the appropriate bivariable data and statistical tests displaying the information as per your analysis plan:
a. one table comparing main exposure and covariates of interest and their mean values (with SD) of the outcome variable using the appropriate statistics, and
b. one table comparing main outcome and covariates of interest either
i. for categorical variables: among those with and without the exposure using the appropriate statistics, or
ii. for continuous variables: with their mean values (with SD) of the outcome variable using the appropriate statistics.
2. about 1 – 2 paragraphs for each table describing the information in that table. In addition, please explain whether your findings support the presence of potential confounders and/or effect modifiers.
Assignment Five: Multivariable Linear Regression, Results, and Conclusions
(due 5/10 by 10pm ET)
Conduct the appropriate multivariable linear analyses to test the research question identified. These analyses should be based on the analytic plan to evaluate the association between the outcome and exposure, controlling for other factors (as previously determined).
The analysis portion of this assignment has one component:
1. one table containing the beta coefficients, SEs, t-statistics, p-values, and 95% confidence intervals for all variables in your multivariable analysis. Underneath the table you should report the model summary statistics of the F-value and its p-value, r-squared, and adjusted r-squared.
For assistance in understanding linear regression STATA output, see this IDRE page.
The written portion of this assignment has two components:
1. Results section: about 2 – 3 paragraphs discussing the findings from the multivariable analysis, including:
a. Findings on the main exposure-outcome association
b. Interpretations of the intercept and all beta coefficients, including p-values and 95% confidence intervals
c. Interpretations of model summary statistics
d. Reports of missing variable values, particularly for exposure and outcome variables
2. Discussion section: about 3 – 4 paragraphs discussing the main findings, including:
a. Relating the original research question to your findings
b. Discussing the significance of findings to public health priorities and/or interventions
c. Identify any selection or information biases, and describe the impact of these on the main association
d. Suggest directions for future research based on findings
DO NOT PLAGARIZE.
For more information on what constitutes plagiarism, please read the following: https://wts.indiana.edu/writing-guides/plagiarism.html and GPH’s Statement of Academic Integrity.
In order to ensure that papers meet the above requirements, all papers will first be submitted to Turnitin via NYU Brightspace. Only doc or docx files are accepted by Turnitin.
