A writer’s style has certain measurable characteristics-specifically, the repetitive use of certain words and the length of his sentences.  Quite recently (in 1962) two American professors of mathematics announced that they had used a computer to determine the authorship of a dozen 18th century essays that had been the center of a major literary controversy for well over 100 years.  The essays were undoubtedly written by either Alexander Hamilton or James Madison, but no one could be certain which.  With the computer, a count was made of certain key words in the problematical essays and in the known works of Hamilton and -adison When these counts were compared, the statistical result added up to apparent proof that 11 of the essays were Madison’s.

Ten years planning, a quarter of a billion dollars holdem strategies, and two years’ market research went into producing the Edsel motorcar, launched by Ford Motors in 1957.  Research predictions indicated a high probability of success for this new car.  But research cannot entirely eliminate risk: The Edsel didn’t meet the expected response.  It was withdrawn from production after two years.

Some statistical measurements can be made with accuracy-like counting the population of a country or measuring the amount of land washed away from instance, in commerce, to assess the probable welcome awaiting a new product.  In the advertising industry (which would be called upon to make such an assessment for a manufacturer) the method is called “market research.”  But this term usually refers to investigations of the actual response of the public to various products. What I am concerned with here is the public reaction to a new product – a testing-of-the-odds process that is called “sampling.”  A manufacturing company that is concerned with launching a new product-say, a cake mix-will hire an advertising agency to handle the project. The agency’s job will be the laborious business of collecting statistics.  Obviously, they will need figures relating to the sales of existing cake mixes; to population densities in different areas; to the areas in which most home cooking is done; to the potential receptiveness of the public to the idea of bought cake mixes; and so on.

A thorough sampling will also take statistical account of the attitudes of husbands and children, as well as wives, to different kinds of cake; of the income groups in which cake is an acceptable food; of nutritional values and of the composition and times of meals in general.  And the agency may well explore some unusual notions.  For instance, common sense may indicated that the  cake making market is exclusively formed of family units, but statistical research that thousands of young unmarried people are aching for just such an activity as cake making.
Some of these statistics- for instance, sales and population-will be readily available; others will have to be sought.  They will be found  by asking people questions.  And the form the questionnaire takes will itself be the subject of intensive research, since statistics from some other sampling project may prove that direct and oblique questioning have different advantages about poker .
The consensus among advertising agencies is that the best method of questioning is by knocking on doors and asking the people who open them the questions.  The casual approach in the street and the postal inquiry are less effective.  And it is at the stage of deciding which doors to knock on that the laws of probability first come into the sampling operation.

The ideal would be to get an unaffected answer to every question in the questionnaire from every interrogable person in the country.  But by applying the laws of probability to certain known factors (like the number of interrogable people and their distribution among areas, income groups, and age groups) the agency may find that an effectively ideal result can be gained by asking, say, one person in 5000.  The choice of the particular 5000 and the particular one person will also be made by referring to the laws of probability, which may provide some useful indications.  They may reveal, for instance, that if three doors are to be knocked on in three different streets in the same town, they had better not all have the same street number, if three Number Ones are chosen, the answers may be affected by the fact that Number Ones are corner houses with a slight social edge on the rest of the street.
Summarily, then, the statistics of actual measurement and those involving the laws of probability will both be used to prepare the way for the sampling.  And when the sampling is completed, both will again be used to analyze it and give the manufactures the information they need in order to launch their gamble- for putting a new product on the market is always a gamble, as the Ford people learned when they tried to see Edsels to the American public.
The statistics brought to light by the sampling will provide the agency with further statistics showing the kind of advertising campaign that will probably be most effective in promoting the sales of the cake mix.  If, for instance, the sampling shows that young unmarried people play poker game are unexpectedly interested in cake mixes, it becomes important to spend a proportion of the advertising allotment on telling them about the new one.  Telling them (and everybody else) involves, among other things, the statistics of magazine circulation.  And magazine circulation is in itself a problem in which the laws of probability can be used again.
Suppose, for example, the makers of the cake mix decided to advertise in the two biggest circulation home magazines in the country (to get at the big market) and also in a magazine bought mainly by young unmarried people (to get at the unexpected smaller market).  The problem of duplicated circulation then arises, for it is likely that the young unmarried people will see the home magazines as well as their own.  The laws of probability can tell the advertising agency the chances of any reader’s seeing one, two, or all three of the advertisements.  (Specifically, assuming that one reader in five sees the advertisement in any one magazine, the chances that the same reader will see it in two magazines are 13 in 125, and in three 1 in 125.)
As you can see even from this quick sweep of the statistical horizon  beyond advertising and commerce (a sweep that has ignored the algebraic complexities), there are a great many ways in which the laws of probability can help manufactures to reduce the risk (the gamble) when bringing out a new product.
Another high-powered modern industry insurance depends to a great extent on statistics and probability.  The business of the modern life insurance actuary (the word derives from actuaries, the shorthand writer who recorded the decisions of the Roman magistrates) depends on the ability to compute premiums, which are the insured person’s contributions to cover a specified risk.  The insurance company is, in effect, gambling on the life expectancy of its customers.  The amounts of premiums are determined in much the same way that bookmakers and other professional poker gamblers determine the odds or percentages that will give them a good chance of winning against a long run (that is, against the practically unlimited time and capital of all their customers).  Both the bookie and the actuary use he laws of probability; but where the bookie uses his knowledge of horses racing , jockeys, etc., the actuary uses populations statistics or mortality tables.

These tables show recorded birth and death rates over a number of years, and from them the actuary can work out your chances of survival during a given future period.  The amount of the premium is, of course, based on these chances; but added to it will be a further sum to cover the expenses of the insurance company, its profits and contingencies arising from short runs of luck that work against the company.  (I have purposely avoided any reference to “bad” luck; adjectives were invented by man, luck was around earlier.)