MEMORANDUM

The data set for the assignment is about breast cancer. Some of these data have been changed for the purposes of this assignment so the results from the analysis of this file may not reflect findings in the literature or current clinical knowledge. However, please analyse these data as if they had come from a real study.

Background

A medical researcher has collected data from 1207 consecutive women with breast cancer in one large hospital. The researcher is interested in the characteristics of the woman (age, tumour size, histological grade, HRT use) and wishes to describe them in a clear and concise way. The researcher wishes to see whether there are any associations between the characteristics (age, tumour size, histological grade, HRT use) and the vital status of the women at the end of follow-up.

Task
Your task is to analyse the data with respect to the following questions to describe the main features of the dataset "breast cancer assignment 2011.sav". You need to prepare a word processed document containing your results. The document must not be hand written. Imagine that this document will be read by people who do not have access to the data. You are expected to use full sentences for your answers, not just a few words. It is NOT NECESSARY OR REQUIRED to read about breast cancer in the medical literature, or to compare your results to results from articles in the medical literature.

NOTE: ASSIGNMENT DOCUMENTS THAT CONTAIN LARGE AMOUNTS OF UNNECESSARY SPSS OUTPUT WILL BE PENALISED. EXTRACT THE APPROPRIATE INFORMATION AND THEN INCORPORATE INTO YOUR DOCUMENT. ALTERNATIVELY EXPORT ALL OF YOUR SPSS OUTPUT (USING FILE>EXPORT) TO A WORD PROCESSOR SUCH AS MICROSOFT WORD AND COPY ONLY THE RELEVANT PARTS TO A SEPARATE DOCUMENT THAT YOU WILL SUBMIT. YOUR DOCUMENT SHOULD NOT BE THE SAME AS ANYONE ELSE'S DOCUMENT. YOU DO NOT NEED TO USE TURNITIN, BUT IT IS VERY CLEAR TO MARKERS IF WORK FROM DIFFERENT STUDENTS IS THE SAME.
Deadline
2 printed copies of the assignment should be handed to Melanie McCann, Room 0.064, Graduate School Office by 12:00 NOON on MONDAY 14 NOVEMBER 2011. ONLY YOUR STUDENT ID SHOULD BE APPEAR ON THE ASSIGNMENT NOT YOUR NAME.

1. SUMMARISE THE DATA for all subjects. Put the appropriate summary statistics, e.g. mean, median, frequencies, percentages, for the different variables into a table to make them concise. You can see examples of consise summaries presented in tables in academic journal articles. Use appropriate graphical summaries to decide which summary statistics are appropriate, but do not include these graphical summaries in your submitted document.
DO NOT PRESENT EVERY POSSIBLE SUMMARY STATISTIC. THINK ABOUT WHICH ONES ARE APPROPRIATE. DO NOT CUT AND PASTE LARGE AMOUNTS OF UNNESSARY INFORMATION FROM SPSS INTO THE DOCUMENT THAT YOU INTEND TO SUBMIT.

2. Use appropriate statistical tests to investigate whether there are any associations (relationships) between vital status (whether or not the patient died) and the patient's characteristics (2a. age, 2b. tumour size (pathsize) 2c. histological grade, 2d. HRT use). For each of the four comparisons in turn (2a to 2d) with vital status answer the questions below using one or more properly constructed sentences to answer each question. Present the important summary information for each comparison and say what it might mean for the wider population of women with breast cancer. Answer all of the following questions for 2a age. Then answer them all for 2b, etc.
What are the null and alternative hypotheses? (They should be appropriate for the data type and the test that you are going to use.)
What test did you use to investigate whether this null hypothesis was true?
Why was this the correct test to use and what assumption(s), if any, did you need to make in order to use this test?
If assumptions were made, how did you check whether the assumptions were valid?
What was your conclusion about whether or not there was a statistically significant relationship?
Using important numbers from your computer output (means, standard deviations, medians, interquartile ranges, counts, proportions, percentages, differences between means, means of differences, confidence intervals and/or p-values, etc., as appropriate) to summarise the size and direction of the relationship found in the sample.
Make an inference for women with breast cancer in the wider population about any association between the variable being considered and the chance of dying within a similar follow-up period?

3. Did the age of the patients differ between the three histological grades? Check that all assumptions of your chosen method are valid and explain how you checked this.