Originally, statistics was the branch of political science dealing with the collection, classification, and discussion of numerical facts relating to the condition of a State or community. Now it is the study of numerical data, their classification and analysis. It embraces every department of activity and knowledge to which numerical comparison can be applied, but properly applies to social facts, and its greatest use is in economics and public administration.
The usefulness of statistics is seriously reduced by the ease with which they may be slewed. For example: statistically air transportation is safer than travelling by motor car when comparing accidents over the distance travelled, that is per mile travelled. However, when one compares the statistics of fatalities over the number of journeys made, irrespective of distance, then travelling by aeroplane is ten times more likely to be fatal than travelling by motor car (according to the British Royal Society for the Prevention of Accidents).
A further example of how statistics can be slewed is in the popular use of averages, properly the mean average. The reliability of a mean average in establishing the general value of a group of values is strongly dependant upon the large number of elements being compared. For example, if a government wishes to show that the average wage being paid to workers is far higher than it really is, they need only to include a few extraordinary high wage values in the set of figures to offset the more prevalent low values. For example, take a set of 1000 elements each of which has a value between 18,000 and 20,000. Obviously the mean average will accurately describe a value between 18,000 and 20,000. Now add a single value of 500,000 to the set and the mean average will rise by over 450, and yet the most common value will remain the same, between 18,000 and 20,000.
Careful selection of values for inclusion in statistics can also be used to slew the results. A survey of members of the public sounds objective, and gives the impression of being representative of the populace. However, a survey of the public in which only men wearing business suits are selected will in all likelihood produce very different results to a survey in which equal proportions of men and women of varying ages, ethnic origins, and modes of dress are sampled.
Similarly, a survey on morality carried out among adults leaving church on a Sunday morning, should be expected to reveal a different result to a survey carried out among adults leaving a night club in the early hours of a Sunday morning, and yet both could be honestly described as a survey of adults.
The willingness with which the general public accept the findings of statistics, and the difficulty in establishing the objectivity of otherwise of such findings, has long been a powerful weapon in the arsenal of propaganda used by politicians and by advertising firms. Research Statistics
Abbreviations
Research Statistics
