From the Economist, Google Scholar, Scopus, SSRN, Publish or Perish, personal website of professors, etc. find some articles on a topic which interests you. Then you decide the topic by yourself within a social science, e.g. economics/business/political science/sociology) , in other words a topic you are knowledgeable about and (most importantly) a topic that CAN be analysed using statistical methods and regressions (e.g. qualitative and discourse analyses are NOT useful for this purpose, unfortunately). What should submit? Should submit the following THREE files: 1 A report 2 An unique do file which will allow me to repeat your whole analysis. 3 DTA: Your transformed data, i.e. in this STATA file I’ll need to see how you have transformed the data, e.g. maybe cleaned some outliers, imputed some missing values, transformed units of measure converted variables, etc. This means the the file is containing the data your do file has created after running on the source data. If the do file DO NOT run smoothly on this dta, I’ll need to take out marks from the CW, even if the analysis is sound. I’ll also mark based on how well your do file is commented and organised as a whole. What should be putted in the report? Think about a simple research question, it does not need to be sophisticated/complicated/new, it has to be sound and clear. The so called literature review has to be minimal (three papers, maybe…), just to show that you have made some very simple background readings. Once you have fond your research question use AS MANY AS POSSIBLE statistical tools we have covered (do not use tools/knowledge NOT included in the QM module, for example DO NOT use time series econometrics NOR panel econometrics) to show how you try to address, even in a simple way, your specific research question(s). Another suggestion: the multiple regression analysis is the more sophisticated method you have studied and it is therefore the one you need to rely more upon. However you need to have a analysis in the same fashion you have done in the tutorial. Is there an ‘ideal structure’? The honest answer is that for a CW in QM not really. It more about your ability to show that you know well your question and you use your tools (statistical formulae, graphical and tabular analyses and regression, etc. ect. etc. ect. etc. ect.) accordingly.