LEC 461 – Assignment 1 Kanchan Mukherjee Introduction Your assignment uses the ‘voles’ dataset and follows on from the Week 5 Laboratory. There is some further analysis to complete and then a short report to be written on the whole analysis (lab-work from Weeks 1,3,5, and the further analysis). Deadline and submission You need to upload the electronic copy in moodle by 4pm of Monday, Week 18 (6th March). Further analysis One aim of the project is to determine which factors affect the weight of a vole, and how they affect it. As well as analysing weight, a team of ecologists are interested in the following: Once all of the environmental covariates (sex, site, year, season) are controlled for, which of the remaining variables in the data can be used to further increase the accuracy of predicting vole weight? Specifically, the covariates of interest are: anaplasma and babesia. We will not be looking at bartonella, moleflea or flea; feel free to delete these from the data set. Starting from your existing best model (determined as Model M in the Week 5 Lab task), use adjusted R2 to decide if either, both or neither of these covariates anaplasma and babesia (as main effects) help to improve the model fit. Perform model checking of your final model. Report The report will be short, specifically: • Follow the IMRAD structure (Introduction, Methods, Results, Analysis, Discussion). Though you may wish to combine Results and Analysis together into one section. • 4 pages maximum with sensible margins • Single spaced, 11pt font The content of the report will draw upon: 1. Relevant material from the lab tasks in Week 1 and Week 3. 2. The lab material from Week 5. 3. The additional analysis requested above. 1 The introduction should briefly describe the data and the aims of the analysis. It should be no more than two paragraphs long. The methods section should briefly describe the statistical approach(es) adopted, and explain why they were sensible. Again, two or three short paragraphs should suffice. The analysis and results section should start with relevant descriptive results from Week 1, any simple ANOVA results from Week 3 that you consider appropriate. If you have an appropriate graph (or graphs) from Week 1 that fits with the results from Weeks 3 and 5, include it in your report. Include also any further plots or tables which you realise will help enlighten the reader about the relationship of some variable with weight. You should include an abridged version of your model fitting results. Show in a table: the equations of the models considered, and their adjusted R2 values. In particular, you should be sure to include: the best main-effects model from the lab; the best model with interactions; and the models fitted with the extra covariate(s). For the final model (which may include anaplasma, babesia, or both) show parameter estimates (including a measure of uncertainty such as standard error) and write down the model equation. Here you should include at least one graph with some form of model checking (i.e. a residual plot) and briefly discuss the residual plots which you produced but have not shown. Discussion should describe the adequacy of the model fit and then focus on the biological meaning of the model formula and parameter estimates. For example “Males are on average ??g heavier than females; with 95% confidence this value is between ??g and ??g.” It should also draw together the results from your exploratory analyses (Week 1 and Week 3) with the full model fit; Where do they agree? Do they ever disagree? Finally, what are the limitations of the study? Refer to research papers/books to interpret biological results. Pro-tips on writing a good report • Do not exceed 4 pages! Material past the fourth page will not be marked. The only exception to this rule is a list of references, which may extend on to a fifth page. • Do not paste in tables from SPSS! Create your own tables: (1) these look better, and (2) SPSS produces a great deal of irrelevant output – your own table should include just the useful information. Consider pasting SPSS tables into Excel and modifying from there. • Every number, table or graph which you include should be mentioned and discussed (however briefly) in the text. If it is not important enough to be considered by you then why should I? You will lose marks if you include irrelevant output. • Do not show everything you did! The reader does not wish to wander with you through the blind alleys of your model-fitting, for example. • Justify what you do/say. Imagine questions someone might ask about the study: make sure the answers are in the report. • The results section should be a mixture of graphs, tables and text which refers to these graphs and tables! Restrict graphics and tables to what is most useful (usually 4 graphs should be plenty). A concise, well structured, table can convey a lot of information succinctly. 2 Marking In each report: • the majority of the credit will be for a good motivation of, description of and statistical interpretation of your analysis (including exploratory parts), for carrying out the analysis correctly and for doing substantial research and refereeing to research papers to interpret biological results. • the following will certainly contribute to your overall mark: writing in prose, structuring your report, labelling graphs and tables, and only including numbers, graphs and tables which are relevant. Clear and concise introduction (including introducing the data set) and methods sections. Conclusions from the analysis: have you answered the questions that were posed – clearly and correctly? Did you discover anything else from your analysis? • Please also see the accompanying document on additional marking criteria for LEC students. 3
