Standard spreadsheet packages such as Excel can perform multiple regression analysis and are sufficient for most routine tasks. A regression equation can be calculated via menus and dialogue boxes and no knowledge of the formulae is required. However, when problems such as autocorrelation (see below) are present, specialised packages such as TSP, Microfit or Stata are much easier to use and provide more comprehensive results.
We also introduce a new example in this section, estimating a demand equation for imports into the UK over the period 1973-2003. There are a number of reasons for this switch, for we could have continued with the birth rate example (you are asked to do this in the exercises). First, it allows us to work through a small 'research project' from beginning to end, including the gathering of data, data transformations, interpretation of results, etc. Second, the example uses time-series data and this allows us to bring out some of the particular issues that arise in such cases. Time-series data do not generally constitute a random sample of observations such as we have dealt with in the rest of this book. This is because the observations are constrained to follow one another in time rather than being randomly chosen. The proper analysis of time-series data goes far beyond the scope of this book; however, students often want or need to analyse such data using elementary techniques. This chapter therefore also emphasises the checking of the adequacy of the regression equation for such data. For a fuller treatment of the issues, the reader should consult a more advanced text such as Maddala (1992) or Thomas (1993).
Was this article helpful?