Reference no: EM132389133
Assignment -
The dataset Dementia_Deaths_1997_2014.xlsx contains data on deaths from dementia in Australia between 1997 and 2014, and data on the population estimates.
In this dataset:
sex: 1=Male, 2=Female
age: 60=60-64, 65=65-69..........., 80=80-84, 85=85+
In Questions 1 to 4 we will assess how dementia deaths vary by age group, sex, calendar year and birth cohort.
1(a) Create a Poisson model to assess the association between the rate of dementia mortality and age.
1(b) Use the model coefficients to create a graph suitable for a scientific report which shows how the rate of dementia mortality changes with age.
1(c) Interpret the results from your model and graph.
2(a) Add sex and calendar year to your model from 1(a). Assess whether these terms improve your model and choose your preferred version of the model.
2(b) Interpret the results from your preferred model from 2(a).
3(a) Create an estimated year of birth (birth cohort) variable. Fit a Poisson model to assess the outcome dementia mortality, which includes age, sex, and birth cohort as exposures. Interpret the results.
3(b) Use the AIC and BIC measures to compare this model to your 'preferred' model from Q2b. Indicate your overall preferred model from the models fitted in Q1-Q3. Justify your choice.
4(a) Comment on the level of over-dispersion in your chosen Poisson model.
4(b) Refit this model using the Negative Binominal Distribution and interpret your results.
State and justify whether you would prefer to present this model or the Poisson model from 4(a).
4(c) Explain why we would be unlikely to be able to sensibly interpret model coefficients if we included age, calendar year and birth year (cohort) in the same model.
For Question 5 we will use the dataset Shopping_v2.csv.
Table 1: Description of variables in the Shopping_v2 dataset.
|
Variable
|
Description
|
|
studyno
|
Unique subject identifier
|
|
Gender
|
1=Male
2=Female
|
|
Age
|
Age in years
|
|
Education
|
Highest Level of Education Achieved (UK)
1=Degree or Higher
2=A-levels
3=GCSEs
4=Other
5=No Qualifications
|
|
Ease
|
Ease of Travel to Main Shop
1=Very Easy
2=Fairly Easy
3=Difficult
|
Q5. Fit ordinal logistic regression models to assess the association between the exposures age, sex, and education and the outcome 'Ease of travel to the main shop'.
Carefully interpret the results from your chosen model and assess whether the model assumptions hold.
Note - Please provide a full record of your software code and software output in an appendix.