Surveys

A small scale survey is often a useful addition to a Final Year Project. If your Final Year Project as a whole is an empirically based investigation such as in a science-based discipline, then your survey is likely to more major but these comments will still apply. In the Social Sciences and Business Studies, a survey can usefully be conducted which will: Shows some initiative on your part Gives you the chance to display the survey and statistical analysis skills you have learnt in your course Helps to illustrate the major themes of your Final Year Project with some 'real' data However, there are some dangers with surveys -a badly conducted survey may be worse then no survey at all! Here are some basic questions (and answers to them!) to ask yourself before you contemplate your survey.. Q. Do you want to collect a reasonably small amount of data in numerical form from a large (N=30 or more) number of respondents ? A. A traditional questionnaire is probably indicated here. Keep the questionnaire reasonably short (12-15 questions maximum) and then analyse each question within it. Q. Do you want to collect much more detailed data from a smaller number of respondents? A. A traditional questionnaire is probably indicated here. Keep the questionnaire reasonably short (12-15 questions maximum) and then analyse each question within it. Q. Do you want to investigate one company, or department in a company, in real depth ? A. This calls for a case study approach. Your case study may well suggest further lines of enquiry in the future – do not argue that your case study proves a hypothesis, however! Questionnaire Analysis Q. Is your data continuous e.g. such as heights and weights? A. Use descriptive statistics and XY plots Test using 't-tests' or ANOVA Q. Is your data nominal or ordinal (e.g. circling one of a series of responses to a question?) A. Use medians and barcharts/ piecharts. Test using non-parametric measures such as chi-square In your questionnaire analysis, make sure that you : Justify the methodology you have used, indicating sampling frame, sampling technique,sample size, non-response rates. Show that you have performed a pilot survey. Used an appropriate statistical package for your results. For example SPSS, MINITAB and TurboStats are designed to undertake statistical analysis, whereas EXCEL is not. Use combinations of packages if necessary. Use the correct statistical tests and show you know how to interpret the results correctly e.g. a chi-square applied to a cross-tabulation. Be aware of the fact that your results can indicate: Results that are statistically significant but not socially significant. (e.g.difference in height of male v. female students.) Results that are socially significant because they are not statistically significant. (e.g. no significant difference in proportion of spend on advertising in high v. low profitability companies). Analysis of Interviews Stage 1 You need to turn your interview into text form as soon as possible. If you take extensive notes rather than tape-record, this will make this task easier. Be on the look-out for particularly good quotes that illustrate your theme. Stage 2 Count and tabulate (simple bar-charts) the number of times that particular themes occur. Stage 3 Illustrate your points with well-chosen quotations e.g. 'I found doing the production was immensely hard work but very well worthwhile' (2nd year Drama student) Case study Analysis Case Study Analysis You will have chosen your case study because it illustrates particular themes in the literature particularly well – for example, it may be a case study of a sportswear company that has dramatically increased its turnover by applying e-marketing. Your case study should illustrate particular factors or combinations of factors that make the case you have chosen worthy of study. The case study should give pointers to factors that might prove significant were they to be studied more extensively e.g. in a larger scale survey. If you study two or more organisations, attempt to account for the differences observed between them e.g. are differences attributable to management style, market position, type of industry? Remember that case studies effectively illustrate the dynamics of processes at work but they do not necessarily 'prove' a case to be true (nor can they when the sample size is 1!) Important issues to remember when you undertake survey work The following comments apply to whatever survey method you have chosen. It is possible, of course, to combine methods e.g. questionnaires with an interview or a case study. This makes for powerful social research because you are combining the representativeness of the survey with the ecological validity (or richness of data) of the interview or case study. Always gave a methodological justification for the methods you have chosen (in a few paragraphs). Ensure that your survey illustrates the themes of the survey and does not appear to be just an 'add-on' for the sake of it. Be careful to keep the survey within justifiable bounds and that it does not 'take over' the project, squeezing out time that would be better devoted to analysis. Do not attempt to say that your survey proves a hypothesis| Rather, to be more correct we would say that our findings are consistent with or not consistent with similar surveys found on the subject. If your own survey indicates a result which differs from similar surveys, attempt an explanation for the differences. Read and show that you have read a good textbook which discusses methodological and survey design issues. Pay particular heed to the advice given by a tutor on survey analysis issues. Make sure that you have necessary permissions and authorisations well in advance. You may require ethical approval in certain instances. Creative Research Systems offers a free aid for survey researchers with useful resources including explanation and guidance to statistical procedures. You can consult the document abstracted from my Final Year Project guide Survey Analysis to amplify these points.