Linear Models in Statistics- Take-away Assignment Assessment on the Linear Models module is in two parts: a take-away assignment and an unseen (but open book) exam (in the University Assessment Period, after the Christmas Vacation). Each part is worth 50% of the total mark for the module. This handout is about the take-away assignment. It gives you details of what you are required to do to complete the assignment successfully. You will be sent, by email, a data set that provides global comparisons of a number of variables related to the health and wealth of different nations around the world. The variables in the SPSS file are: country: name of the country. region: one of six regions around the world, coded numerically 1-6 but given names for ease of understanding in the values column of variable view in SPSS: 1 = Eastern Europe, 2 = South America, 3 = Western Europe and USA, 4 = Middle East, 5 = Asia, 6 = Africa birthrat: live birth rate per 1,000 of population deathrat: death rate per 1,000 of population infdeath: infant death rate per 1,000 of population aged 12 months or less malelife: life expectancy at birth for males femalife: life expectancy at birth for females gnp: gross national product (standardised in dollars). In this variable alone there are some missing data cells. These have been coded with the dummy value 999999 Each data set is unique – yours will be different from those of other people who are taking this module. This has been done to prevent collusion (see further information below). Spend some time thinking about the data set, and the kinds of hypotheses you might be able to test using the data set. In your write up, you will need to explain briefly which hypotheses you decided to test, which statistical tools you have chosen to test these hypotheses, and why. You will then need to analyse the unique data set that you receive. It is up to you how you analyse the data, but the techniques used must be both appropriate and from among those you (will) have been taught this term. Use your judgement about how many tests you carry out and how much output you produce. Don't use every technique in every conceivable way. What you want to do is to use the data and your analyses of the data to tell a coherent story. There is no one correct set of analyses or one correct story. Different data sets may suggest different stories. For the assignment you need to write up what you find in these analyses as if you were writing up a results section in an academic paper, but should should give somewhat more attention than is usual in a journal paper to questions about why particular analyses were suitable, and in particular about whether the assumptions of the analysis methods were satisfied, how much it matters if they were not, whether there is anything that can be doneabout failure to satisfy the assumptions, and whether you need to temper your conclusions. The write up should include appropriate graphs and/or tables of results. Do NOT write a full paper. However, as indicated above, you should say, at the beginning, which hypotheses you are testing. In addition, you should write a very short conclusion in which you summarize what the results tell you. The write up should not exceed 1000 words. Figures and tables are not included in the word count, though you should not use too many of them. Please note that you should NOT copy tables directly from SPSS. Look at journal articles to see how figures and tables are usually presented there. Or refer to the APA Publication Manual for full details of APA format. Collusion: this assignment is an individual assignment and the write up that you hand in should be your own work. This does not mean that you cannot discuss the assignment with other people on the module (or with people not on the module, for that matter). However, you should generate your own set of hypotheses to test, decide for yourself which tests to perform, and write the account of what you have done by yourself. It would be surprising and suspicious, for example, if two students presented exactly the same sets of analyses, even if they described those analyses in their own words, and reported different (descriptive and inferential) statistics because their data sets were different. Collusion amounts to misconduct, and if you are found guilty of misconduct you may lose some or all of the marks for this piece of work. See http://www.sussex.ac.uk/adqe/standards/academicmisconduct for more details of the University's policy on plagiarism and collusion. Your assignment should be submitted to the Psychology Office (PEV1-1B2) at the end of the Autumn Teaching Period (Sussex Direct will give details of the exact deadline). There will be standard penalties for assignments handed in late. Alan Garnham ([email protected]) October 2015