Syllabus

Schedule

Course Weeks

• Week 1: 8/29 – 9/5
• Week 2: 9/5 – 9/12
• Week 3: 9/12 – 9/19
• Week 4: 9/19 – 9/26
• Week 5: 9/26 – 10/3
• Week 6: 10/3 – 10/10
• Week 7:  10/10 – 10/17
• Week 8: 10/17 – 10/21 (Sunday)

Course Schedule

Week 1: Fundamentals of Biostatistics in Public Health Research

Weekly Outcomes:

• Review datasets for use in a report and the development of research questions.
• Determine the types of variables in a dataset for use in the development of a report.

Readings:

• Sullivan Textbook: Chapter 1
• Review the follow websites about types of data: Laerd Statistics

Videos:

• Video: Course Introduction [Dr. Liam O’Brien]
• Weekly Lecture: Week 1 [Dr. Liam O’Brien]

Assignments:

• Discussion: Begin by introducing yourself to the class. What is your background in Biostatistics? What are you most excited about? In what ways will this course come in handy for your Public Health Career? [Ungraded]
• Discussion:Introduction to Final Project – Choosing Variables, Developing a Question) — For the final project (and associated discussions and assignments) you will work with pre-existing data from one of these two studies:
  1. The Framingham Heart Study, a long term ongoing cohort study which collects data to examine risk factors associated with cardiovascular disease
  2. The Bay State Medical Center Birth Weight Study, a case control study which collected data to examine risk factors associated with low birth weight

  Enter the Final Project Datasets and Supplemental Materials folder (also available via the sidebar) and read the two MS-Word documents there which describe the background of these two studies and provide a variable codebook for the data available. After reviewing these documents and the Excel files associated with them, select whichever one of these datasets aligns best with your research interests.

  Then, in your initial post:

  • Identify 10 variables in your selected data set that might be of interest
  • Note whether each variable is categorical (nominal, dichotomous, ordinal) or continuous (interval, ratio) as described in the ‘Laerd Statistics’ reading this week
  • Propose 5 research questions that could be asked using these variables (note that each question should reference two variables)
  • Explain why you find these questions of interest

  For example, using data from the Birth Weight Dataset, one might ask if there is an association between the presence of uterine irritability (categorical-dichotomous) and low birth weight (categorical-dichotomous), as uterine irritability is a risk factor for giving birth to low weight babies. 

Week 2: Studies and Experiments

Weekly Outcomes:

• Examine a research paper to determine if it is an observational study or an experiment.
• Describe how a representative sample is taken from a population through the description of a study provided in a research paper.
• Review a research paper to understand how the experiment of study was designed.
• Develop research questions based upon a chosen dataset for the development of a report. 

Readings:

• Sullivan Textbook: Chapters 2.1 through 2.4
• Abram KM, Teplin LA, McClelland GM, Dulcan MK. Comorbid Psychiatric Disorders in Youth in Juvenile Detention. Arch Gen Psychiatry. 2003;60(11):1097-1108. doi:10.1001/archpsyc.60.11.1097. http://archpsyc.jamanetwork.com/article.aspx?articleid=208029

Videos:

• Weekly Lecture: Week 2

Assignments:

• Discussion: Research Questions — Referring back to the feedback you received from your peers and your instructor on your potential research questions and their associated variables, finalize three research questions relating to your selected dataset. Make sure the variables used in the final questions you select are of the following format:
  1. Continuous response and continuous explanatory variable
  2. Categorical response and categorical explanatory variable
  3. Continuous response and categorical explanatory variable (note: the categorical variable must be dichotomous for this question)

  Each research question should have a clear explanatory and response variable. Review, as necessary, the ‘Laerd Statistics’ reading from last week which explains data types and the difference between response variables (also referred to as dependent or outcome variables) and explanatory variables (also referred to as independent or predictor variables).

  Then, in your initial post, list your three finalized research questions, indicating the variable types.

  For example, using data from the Framingham Heart Study dataset, one question might be: ‘Is there a significant difference in total serum cholesterol levels [TOTCHOL] (continuous-ratio) (response) between smokers and non-smokers [CURSMOKE] (categorical-dichotomous) (explanatory)?’

  In your response post to your peers, provide feedback on their finalized research questions, which will become the foundation of each student’s final project. Consider the following questions as you review your peers’ research questions: 

  • Is each research question clearly stated? 
  • Does each research question relate to the dataset?
  • Do the three research questions meet the data type requirements? 
  • Does each research question identify response and explanatory variables correctly?

• Written Assignment: In this assignment, review the “Comorbid Psychiatric Disorders in Youth in Juvenile Detention” paper.  Read through the article paying careful attention to the section describing how the subjects were selected and the study or experiment was designed, then answer the following questions.

  • What is the population that the researchers are studying?

  • Describe how the data was collected:

    • How was the sample chosen?

    • Do you think the sample is representative of the population?  Explain.

    • Is this an experiment or a study? Make sure to clearly explain how you came to this conclusion and support your claim.

    • Describe how this experiment or study was designed.

  • Do you think the results of this experiment/study could be applied to youth detention centers, in your own state or region?  Explain.

  Your written response should be no more than 2 double-spaced pages. This assignment will be graded using the Week 2 Written Assignment Rubric.

  Note: As you complete this assignment, keep in mind that the formatting of the journal article you are reviewing is similar to what you will be submitting for your final project.

Week 3: Summarizing Data

Weekly Outcomes:

• Utilize appropriate graphical techniques to display data.
• Utilize appropriate numerical summaries to summarize data.
• Utilize appropriate verbal summaries to interpret results

Readings:

• Sullivan Textbook: Chapter 4 and 12

Videos: 

• Weekly Lecture: Week 3
• Stata Tutorial: Introduction
• Stata Tutorial: How to Copy and Paste Data From Excel into Stata
• Stata Tutorial: Bar Graph
• Stata Tutorial: Boxplot
• Stata Tutorial: Histogram 
• Stata Tutorial: Scatterplot
• Stata Tutorial: Mosaic Plot [Dr. Liam O’Brien]

Assignments: 

• Written Assignment:

  An important first step in data analysis is to examine the variables that will be used to answer research questions by creating numerical and graphical summaries.  To do this, you will utilize Stata, a popular statistical analysis package created and maintained by StataCorp LLC.  

  Steps to purchase:

  • Choose your country
  • Click on Student
  • Click on New Purchase
  • You will need at least a 6 month license of Stata/IC

  Stata versions of both the Framingham Heart Study dataset and the Birth Weight Study dataset are available in the “Final Project Instructions, Datasets and Supplemental Materials” link in the course sidebar.  Though you may use the Excel versions of the data to do the statistical analysis, the Stata versions are highly recommended as have variable and value labels, which will make your resulting outputs more easily readable.

  Be sure you review the lecture and the Stata tutorial videos before completing this assignment.  Also note the ‘Help with Stata’ link on the left course launch which provides additional resources should you need assistance with Stata.

  Restate your research questions. For each of the variables that are used in your research questions, create single numerical and graphical summaries using Stata.  Note that the type of numerical and graphical summary that is appropriate for a given variable is related to its data type as described below:

  	Numerical summary	Graphical summary
  Categorical	Table of frequencies and percentages	Bar chart
  Continuous	Means, standard deviation, and sample size	Histogram or boxplot

  Then, for each of the two-variable associations described in your research questions, create appropriate numerical and graphical summaries as shown below:

  	Numerical summary	Graphical summary
  Continuous-continuous	Correlation	Scatterplot
  Categorical-categorical	Two-way contingency table with frequencies and percentages	Mosaic plot
  Continuous-categorical	Means, standard deviation, and sample size of the continuous variable for each category	Side by side box plots of the continuous variable for each category

  Examine the numerical and graphical summaries you created.  Provide a one sentence verbal summary for each indicating what it tells you about the data.

  Please submit the research questions, numerical summaries, graphical summaries, and verbal summaries all in a single Word document.  This assignment will be graded using the Week 3 Written Assignment Rubric.

Week 4: Confidence Intervals

Weekly Outcomes: 

• Construct confidence intervals using Stata.
• Interpret confidence intervals.

Readings: 

• Sullivan Textbook: Chapters 6.1 through 6.6.1 (not 6.6.2 or 6.6.3)

Videos: 

• Weekly Lecture: 
  • Week 4 Part 1
  • Week 4 Part 2
• Stata Tutorial: Creating Confidence Intervals [Dr. Liam O’Brien]

Assignments:

• Discussion: Confidence Intervals — The focus this week is on confidence intervals. Utilize the following information and answer the questions below. A 95% confidence interval obtained from a sample of 100 outpatients for the true population mean normal mean systolic blood pressure is given by (114 mmHG, 120 mmHG). 
  • Provide a correct interpretation of this interval. Can you think of other interpretations that would also be correct?
  • This confidence interval came from a single sample, would we get the same interval if we obtained a different set of 100 patients? What does this imply about your interpretation of the given interval? 
  • If we wanted a 99% confidence interval instead, can you tell whether it would be narrower or wider? Can you tell by how much? 
• Quiz: Weeks 1-4 — You must use Stata to perform any statistical  calculations or tests required  (the answers have been calibrated for its use and using a different tool could result in slightly different answers). Note that in addition to statistical test procedures described in the lectures, Stata also offers a calculator (see Data->Other Utilities) for the calculation of simple proportions.

Week 5: Hypothesis Testing

Weekly Outcomes: 

• Utilize the basic steps of hypothesis testing to evaluate research questions involving a single sample .
• Relate information provided by a hypothesis test to information provided by a confidence interval.

Readings: 

• Sullivan Textbook: Chapters 7.1 through 7.3 and 7.6 (skip 7.4 and 7.5)

Videos:

• Weekly Lecture:
  • Week 5 Part 1
  • Week 5 Part 2 
• Stata Tutorial: Conducting One-Sample Hypothesis Tests [Dr. Liam O’Brien]
• Stata Tutorial: Conducting a Paired T-Test [Dr. Liam O’Brien]

Assignments: 

• Discussion: P-values — Everyone has seen or heard of a p-value. Consider your exposure and define what the p-value represents. Why can we compare it to alpha (the significance level of the test)? Last week we discussed confidence intervals. Consider what a p-value tells you that a confidence interval does not? What does a confidence interval tell you that a p-value does not? Which one is more useful? Make sure your response is in your own words and that your ideas are supported by reputable sources. 

  In your response to your peer’s initial post compare your answers. How do your p-value definitions differ? Consider your peers’ thoughts on the information gained from a p-value and a confidence interval. Do you agree or disagree? Why?

• Written Assignment — Now that you have used descriptive statistics and examined the data numerically and also graphically, you are ready to begin answering your research questions using inferential statistics.  Descriptive statistics describe the data in the sample, while inferential statistics allows one to take those results and hopefully make inferences about the population from which the sample was drawn.  To do this, one must perform hypothesis testing.

  In this assignment, you will define the steps needed to evaluate the continuous response and continuous explanatory variableresearch question and perform the appropriate hypothesis test with Stata.  Make sure you explicitly show all five steps and consider any necessary assumptions that were discussed in the lecture.

  These steps are:

  1. Define the parameter of interest
  2. State the hypotheses
  3. Determine the test statistic and p-value considering any necessary assumptions
  4. Decide whether to reject or not reject the null hypothesis
  5. Clearly state a conclusion in the context of the problem

  For example, suppose we have collected data from 50 subjects on the average number of hours slept per night and the average number of days per week of 20+ minutes of moderate exercise.  We are interested in seeing if there is any relationship between hours slept and number of exercise days.

  The 5-steps would be as follows:

  Parameter of interest: Population correlation between the average number of hours slept per night and the number of days the subjects participated in 20+ minutes of moderate exercise.  (Note: population correlation is the appropriate parameter of interest as our two variables are continuous)

  Hypothesis: H0: ρ = 0; HA: ρ not = 0

  Test statistic: I may not get a test statistic with Stata, but it can be calculated to be 2.02 for this data. The p-value is 0.048 and I have 48 degrees of freedom.  I should check my scatter plot to test if it looks roughly linear as that is an assumption, and if it does not note that here.

  Decision: Since the p-value is less than my alpha level of .05, I reject the null hypothesis of no correlation.

  Conclusion: I conclude that the correlation is significantly different (and larger) than 0.  In the context of my research question, there is a moderately weak but statistically significant correlation between average hours slept and days per week of moderate exercise.

  Your submission needs to clearly discuss each step to properly evaluate your primary research question. You should divide up your document with subheadings for each step. Be sure to restate your research questions at the beginning of your assignment.

Week 6: Hypothesis Testing Part 2

Weekly Objectives: 

• Discuss how the size of  a sample affects conclusions through the examination of a study’s confidence intervals.
• Evaluate research questions through the completion of appropriate hypothesis testing involving two samples.

Readings: 

• Sullivan Textbook: Chapters 7.5, 7.7, and 7.9
• Paolo Boffetta, Paolo et al. Fruit and Vegetable Intake and Overall Cancer Risk in the European Prospective Investigation Into Cancer and Nutrition (EPIC)JNCI J Natl Cancer Inst (2010) 102 (8): 529-537 doi:10.1093/jnci/djq072 first published online April 6, 2010 http://jnci.oxfordjournals.org/content/102/8/529.full

Videos: 

• Weekly Lecture:
  • Week 6 Part 1
  • Week 6 Part 2
• Stata Tutorial: Two-Sample Hypothesis Tests [Dr. Liam O’Brien]
• Stata Tutorial: Performing Hypothesis Test for Contingency Tables [Dr. Liam O’Brien]
• Stata Tutorial: Correlation Test [Dr. Liam O’Brien]

Assignments: 

• Discussion: Confidence Intervals Part 2 — Read the Faber (2014) article on the influence of sample size and then the Boffetta et al (2010) article of the relationship between fruit and vegetable consumption and reduced cancer risk. Examine the confidence intervals reported in the abstract (note that a hazard ratio (HR) of 1 means no change in risk) and comment on their sizes. Are they statistically significant? How does the size of the samples affect your opinion of the conclusions reached?
• Written Assignment — Continue with hypothesis testing using Stata for your remaining two research questions.  One question uses a categorical response and categorical explanatory variable, and the last question uses a continuous response and categorical-dichotomous explanatory variable.

  The five steps are:

  • Define the parameter of interest

  • State the hypotheses

  • Determine the test statistic and p-value considering any necessary assumptions

  • Decide whether to reject or not reject the null hypothesis

  • Clearly state a conclusion in the context of the problem

   Be sure to watch the lectures before beginning this assignment as the parameter, the nature of the hypotheses, the statistical test needed, etc. do change when the data type changes, so you will not be performing a correlation test for these two questions.

  Your submission needs to clearly discuss each step to properly evaluate your two remaining research questions. You should divide up your document with subheadings for each step. Be sure to restate your research questions at the beginning of the assignment.

• Reminder: Next week your Final Project is due. 

Week 7: Linear Regression

Weekly Outcomes: 

• Perform a linear regression to interpret the slope and intercept of the regression line
• Utilize a dataset to complete a professional report. 

Readings:

• Sullivan Textbook: Chapter 9.3

Videos: 

• Weekly Lecture: Week 7
• Stata Tutorial: Simple Linear Regression
• Stata Tutorial: Linear Regression [Dr. Liam O’Brien]

Assignments:

• Discussion: Linear Regression — Up until this point your have assessed two continuous variables using a correlation. This week you will run a simple linear regression to address your research question relating your two continuous variables together. Take the regression equation you obtain from Stata and interpret the slope of the line. What information does this give you that you weren’t able to obtain by running a simple correlation? Note that this regression does not need to be included in your final project. 

  In your response to your peer’s initial post comment on other variables that might potentially confound the relationship between the two displayed variables. 

• Final Project Submission — As a Public Health professional presentation of research findings is an important skill. Now that you have answered your three research questions it is time to consolidate your results and appropriate statistical inferences you drew from them into a final report, which should be similar to a research article one would submit for publication.

  Please see the final project document for details on the required components of this report and guidelines for submission.

  GPH-716: Final Project Document

Week 8: Analysis of Variance

Weekly Outcomes: 

• Perform a one-way ANOVA using STATA to compare means across more than two groups.

Readings: 

• Sullivan Textbook: Chapter 7.8

Videos: 

• Weekly Lecture: Week 8
• Stata Tutorial: Oneway ANOVA [Dr. Liam O’Brien]

Assignments:

• Written Assignment — Using the provided dataset below, determine whether the mean birthweight for infants differs significantly according to mothers’ smoking status by using a one way ANOVA in Stata. Make sure to include all 5 hypothesis testing steps in your writeup.  If you found a significant difference, how might you determine which specific groups differ?  Are there any potential problems with the method you have proposed?
  • Define the parameter of interest.
  • State the hypotheses
  • Determine the test statistic and p-value considering any necessary assumptions
  • Decide whether to reject or not reject the null hypothesis
  • Clearly state a conclusion in the context of the problem
  • Specific Comparisons: How would you figure out which specific groups differ?  
  • Potential Problems: What are the potential problems with the method you have proposed to figure out which groups differ?

Datasets:

smoking.xls  (Excel file)  

smoking.dta  (Stata file)

Your paper should use subheadings that divide the submission into the sections above. 

Student Resources

Policies