A SYSTEM OF PROGRAMMES FOR ANALYSING SOCIAL DATA

As social research increasingly encompasses a wider range of disciplines, so the amount of analysis, considered pertinent to the investigation of particular problems also increases. Only ten years ago this quantity rarely exceeded what could be handled by sorting, tabulation or mechanical calculation. More recently, the scope of research analysis would be severely restricted if such techniques were all that was available. Fortunately this is not the case. Concomitantly with this growth in data has been the development of computerised statistical procedures. And in turn, the greater power of computer calculation has encouraged researchers to consider different approaches to data analysis, rather than being forced to follow the traditional techniques. Factor analysis presents an obvious example here, but discriminant function and cluster analysis are equally useful. Some researchers point to dangers of this change and aruge that the automation of data analysis has led to neglect of thorough research designs and replaced them with complex post hoc calculations. To utterly deny these accusations would be foolish. But it is perfectly reasonable to support the automation of data processing whilst at the same time insisting upon the need for rigorous experimental design.