SPSS Survival Analysis

Reproductive decision-making is partly determined by trade offs in energetic allocation between different physiological needs. Mothers who are in poor health are hypothesised to have an increased need to trade off immune function and tissue repair with reproduction than healthy women, resulting in an expectation of longer time intervals between the birth of a child and the birth of the next child for women in poor health.

The data in the file 2018_assessment.sav are variables from the 1970 British Cohort dataset using the data collected at age 34.

1. Using the general hypothesis given as an introduction to this assessment, briefly and clearly state how that hypothesis will be operationalised through a specific research question. Make sure that you state which are the control variables, which are the predictor variables of theoretical interest and what your outcome variable will be.

2. Descriptive statistics, produce the following tables and figures with brief and clear legends:

a. Produce a table of descriptive statistics for all variables.

b. Produce a life table and survival graph, the interval you choose should be informative.

c. Produce Kaplan-Meier graphs for your categorical predictor variables.

d. Produce Cox regressions for your continuous predictor variables.

3. Describe what you found, particularly:

a. In the table of descriptives for all variables

b. In the life table and survival graph, with a focus on the proportion surviving an interval and the rate of censoring across intervals.

c. In the Kaplan-Meier graphs discuss whether the assumption of proportional hazard rate is met in each binary variable.

d. In the Cox regression for continuous variables, discuss whether the assumption of proportional hazard across time was met.

4. Perform a Cox regression analysis to test the study question that you outlined in question 1 including all predictor and control variables in the model. Include 95% confidence intervals. Include the correlation of estimates. Present the relevant SPSS output of the results with legends. Include the proportional hazards graph.

5. Interpret and comment on the results of the Cox regression analysis.

a. How many individuals were included in your analysis? How many had missing data?

b. Discuss the model as a whole and the way the individual predictor and control variables contributed to the model.

c. Discuss the correlation matrix noting anything unusual. If there is nothing unusual state that.

d. Did your findings support the hypothesis?

e. Interpret whether the proportional hazards assumption has been met.

Marking Criteria: