Your final statistical report is due on 28 April (2pm) and it is worth 50% of your final mark. The length of your report should be approximately 2,000 words (tables, figures and the code excluded).
For your final report you need to conduct an independent analysis of the Understanding Society data to answer one of the questions below. Unlike statistical assignments, I do not suggest a detailed plan for your analysis. You will need to decide yourself which features of the data you want to present in your reports and what is the best of doing this.
Please choose one of the following questions.
-
Political scientists sometimes argue that people's political interest is usually stable over the life course. Test this statement empirically with the Understanding Society data. Check how average political interest changes at different points of electoral cycle.
-
Do people become happier or unhappier after childbirth? Explore how satisfaction with life in general and different aspects of it changes after childbirth, for men and women.
-
Explore how the household composition varies across ethnic groups in Britain.
If you want to explore a different question please feel free to talk to me about this. This is possible, but I need to approve your question first.
Below are some rules for completing the reports.
-
Two of the questions above involve an analysis of longitudinal data, and the third question requires an analysis of both individual and household-level data. This assignment tests your ability to work with multiple data sets (join them, reshape, organise as tidy data, etc.).
-
Your reports should have the maximum of five tables or graphs (you can decide yourself how many tables and graphs you want as long as the total number is no more than five). I will mark you down if you have more than five graphs and tables.
-
This assignment is not about complex statistical modelling. It tests your ability to work with data and present descriptive statistics in a clear and innovative way, both as tables and graphs.
-
Your reports should have a clear main message. Narrowing the topic down to one or two smaller questions and explore them carefully and in detail is better than presenting a lot of poorly interpreted numbers on different aspects of your main question.
-
I do not expect you to present a detailed literature review. That said, you may want to read two or three papers (possibly reviews) published in good journals in the last 10 to 15 years, and summarise the state of art in about two paragraphs of your report.
-
Your reports must be written in R Markdown and show all the code for the analysis (not included in word count). The code must start with reading the Understanding Society data and show all steps of the analysis and the output. You should conduct your analysis as a separate project on Github. Remember that you must never commit the Understanding Society data. Please knit your reports both as an .md file (to be shown on Github) and a pdf file (to be submitted on Bart). Your Bart submission should include a pdf version of your report, with a link to the Github repository.