This assignment (PROJECT – Statistical analysis in R software, computer exercise) is based on statistical analysis of public health datasets (the datasets are currently available in the FLO site). The content of the analysis covers descriptive analysis and presentation of relevant data (module 2), hypothesis testing (modules 4-5), and correlation and regression (module 6). The questions are entirely based on the work you have been doing in your modules 2, 4-6 will cover problems associated with data preparation, visualization and summary statistics, hypothesis testing, correlation and regression. I am happy for you to discuss the problems with your colleagues but your responses to the questions must be completely your own. The statistical analysis is based on computer exercise using R/R Studio.
MARKING CRITERIA
Each question will be marked on its merits. There is often more than one way to
answer the questions, so I am happy to mark if the answer makes sense! For each
question, marks have been sliced by sub-section. Penalties to be applied if deadline is not met.
NOTE
Please provide your chunk of code and R output for each of the questions. If you
have any difficulties to generate a report from R Markdown (using Knit), then copy your chunk of code and R output in the word document.
2
QUESTION 1
MODULE 2 DATA PREPARATION & VISULAIZATION (20 MARKS)
Data were extracted from a cross sectional study of adults aged 50+ years with
depressive symptoms from the World Health Organization (WHO) Study on Global
AGEing and Adult Health (SAGE). The SAGE study design and procedures are
described elsewhere
(http://apps.who.int/healthinfo/systems/surveydata/index.php). The article from
SAGE data was published in Sleep Health 2020 Feb;6(1):92-99. doi:
10.1016/j.sleh.2019.08.009
The suicide attempt item asked, during this period, did you ever try to end your life? This variable is the main outcome variable and response options were 1=yes or 0=no for this item. Participants were asked to rate their sleep quality for the previous two nights with response options very good, good, moderate, poor, and very poor. We recategorized participants reporting sleep quality into two categories 1=poor/very poor or 0=good/very good/moderate sleep quality. For sleep length, participants were asked how many hours they slept the previous two nights. They provided responses in hours and minutes and calculated the mean duration across the two nights. SAGE interviewers reported participants’ sex (male or female) and age in single years. The SAGE dataset is currently located in the FLO site (https://flo.flinders.edu.au/mod/resource/view.php?id=3522598 ).
a. The SAGE.xlsx data file is currently located in the FLO site. Import the data file into R by using appropriate R function. Rename the variable centre by country.
b. The value levels of each variable are described as numbers. Using mutate function, label the value levels of the following variables
◦ gender (1=Male, 2=Female)
◦ attempt (0=No, 1=Yes)
◦ health (1=Bad/very bad, 2=Moderate, 3=Good/very good)
◦ sleepqual (0=Other, 1=Poor/very poor sleep)
c. Filter the “China” data and visualise the following variables using an appropriate geom function. Discuss the findings from each of the diagrams.
◦ Histogram for age
◦ Simple bar diagram for gender
◦ To explore the sleep length between male and female participants, draw a box plot of sleep length by gender.
◦ To explore the suicidal attempt between male and female participants, draw a clustered bar chart of suicidal attempt by gender.
3
QUESTION 2
MODULES 4-5 HYPOTHESIS TESTS (20 MARKS)
The retrospective cohort study identified pregnant women with a BMI >30kg/m2 recorded at their first antenatal visit, from within the ObstetriX and eMaternity databases from January 2013 to December 2017, within the Northern Sydney Local Health District in Sydney, NSW, Australia. The neonatal outcomes by obesity classes are provided in the following table. The data “neonata.xlsx” is currently available in the FLO site. The details of the following table can be found in the article “Obesity class impacts adverse maternal and neonatal outcomes independent of diabetes” through the following link https://www.frontiersin.org/articles/10.3389/fendo.2022.832678/full. The article is also available in the FLO.
a. Perform an analysis in R programming and complete the table below (Hints: Use an appropriate hypothesis test for each of the neonatal outcomes to calculate the p values)
b. Write a paragraph from the findings and make a conclusion whether neonatal outcomes are significantly associated with BMI obesity group.
4
QUESTION 3
MODULE 6 CORRELATION & REGRESSION (20 MARKS)
Data were extracted from a cross sectional study of adults aged 50+ years with
depressive symptoms from the World Health Organization (WHO) Study on Global
AGEing and Adult Health (SAGE). The SAGE study design and procedures are
described elsewhere
(http://apps.who.int/healthinfo/systems/surveydata/index.php). The article from
SAGE data was published in Sleep Health 2020 Feb;6(1):92-99. doi:
10.1016/j.sleh.2019.08.009
The age of participants was recorded as singles years of age. For sleep length,
participants were asked how many hours they slept the previous two nights. The
SAGE dataset is currently located in the FLO site
(https://flo.flinders.edu.au/mod/resource/view.php?id=3522598).
Interrogate the dataset provided to answer the following question:
Is age associated with sleep length?
a. Write a statistical analysis plan (SAP) to test the association between age and sleep length. (Hint. The statistical analysis plan should cover (i) the visual inspection; (ii) correlation coefficient; (iii) regression modelling; (iv) assumptions; (v) model coefficients and p values; (vi) coefficient of determination)
b. Perform an analysis according to your SAP in R programming.
c. Discuss the findings and draw a conclusion from the discussion.
5