|
< ><FONT face="Times New Roman" size=3>The term statistics is used in either of two senses.In common parlance it is generally employed synonymously with the word data.Thus someone may say that he has seen”statistics of industrial accidents in the United States.” It would be conducive to greater precision of meaning if we were not to use statistics in this sense,but rather to say “data (or figures ) of industrial accidents in the United States.”</FONT></P>
< ><FONT face="Times New Roman" size=3>“Statistics” also refers to the statistical principles and methods which have been developed for handling numerical data and which form the subject matter of this text.Statistical methods,or statistics, range form the most elementary descriptive devices, which may be understood by anyone , to those extremely complicated mathematical procedures which are comprehended by only the most expert theoreticians.It is the purpose of this volume not to enter into the highly mathematical and theoretical aspects of the subject but rather to treat of its more elementary and more frequently used phases.</FONT></P>
< ><FONT face="Times New Roman" size=3>Statistics may be defined as the collection, presentation, analysis, and interpretation of numerical data.The facts which are dealt with must be capable of numerical expression.We can make little use statistically of the information that dwellings are built of brick, stone, wood, and other materials; however, if we are able to determine how many or what proportion of,dwellings are constructed of each type of material, we have numerical data suitable for statistical analysis.</FONT></P>
<P ><FONT face="Times New Roman" size=3>Statistics should not be thought of as a subject correlative with physics, chemistry, economics, and sociology. Statistics is not a science; it is a scientific method. The methods and procedures which we are about to examine constitute a useful and often indispensable tool for the research worker. Without an adequate understanding of statistics, the investigator in the social sciences may frequently be like the blind man groping in a dark closet for a black cat that isn’t there. The methods of statistics are useful in an ever---widening range of human activities, in any field of thought in which numerical data may be had.</FONT></P>
<P ><FONT face="Times New Roman" size=3>In defining statistics it was pointed out that the numerical data are collected, presented, analyzed, and interpreted. Let us briefly examine each of these four procedures.</FONT></P>
<P ><FONT face="Times New Roman" size=3>COLLECTION Statistical data may be obtained from existing published or unpublished sources, such as government agencies, trade associations, research bureaus, magazines, newspapers, individual research workers, and elsewhere. On the other hand, the investigator may collect his own information, going perhaps from house to house or from firm to firm to obtain his data. The first-hand collection of statistical data is one of the most difficult and important tasks which a statistician must face. The soundness of his procedure determines in an overwhelming degree the usefulness of the data which he obtains.</FONT></P>
<P ><FONT face="Times New Roman" size=3>It should be emphasized, however, that the investigator who has experience and good common sense is at a distinct advantage if original data must be collected. There is much which may be taught about this phase of statistics, but there is much more which can be learned only through experience. Although a person may never collect statistical data for his own use and may always use published sources, it is essential that he have a working knowledge of the processes of collection and that he be able to evaluate the reliability of the data he proposes to use. Untrustworthy data do not constitute a satisfactory base upon which to rest a conclusion.</FONT></P>
<P ><FONT face="Times New Roman" size=3>It is to be regretted that many people have a tendency to accept statistical data without question. To them, any statement which is presented in numerical terms is correct and its authenticity is automatically established.</FONT></P>
<P ><FONT face="Times New Roman" size=3>PRESENTATION Either for one’s own use or for the use of others, the data must be presented in some suitable form. Usually the figures are arranged in tables or presented by graphic devices.</FONT></P>
<P ><FONT face="Times New Roman" size=3>ANALYSIS In the process of analysis, data must be classified into useful and logical categories. The possible categories must be considered when plans are made for collecting the data, and the data must be classified as they are tabulated and before they can be shown graphically. Thus the process of analysis is partially concurrent with collection and presentation.</FONT></P>
<P ><FONT face="Times New Roman" size=3>There are four important bases of classification of statistical data: (1) qualitative, (2) quantitative, (3) chronological, and (4) geographical, each of which will be examined in turn.</FONT></P>
<P ><FONT face="Times New Roman" size=3>Qualitative When, for example, employees are classified as union or non—union, we have a qualitative differentiation. The distinction is one of kind rather than of amount. Individuals may be classified concerning marital status, as single, married, widowed, divorced, and separated. Farm operators may be classified as full owners, part owners, managers, and tenants. Natural rubber may be designated as plantation or wild according to its source.</FONT></P>
<P ><FONT face="Times New Roman" size=3>Quantitative When items vary in respect to some measurable characteristics, a quantitative classification is appropriate. Families may be classified according to the number of children. Manufacturing concerns may be classified according to the number of workers employed, and also according to the values of goods produced. Individuals may be classified according to the amount of income tax paid.</FONT></P>
<P ><FONT face="Times New Roman" size=3>Chronological Chronological data or time series show figures concerning a particular phenomenon at various specified times. For example, the closing price of a certain stock may be shown for each day over a period of months of years; the birth rate in the United States may be listed for each of a number of years; production of coal may be shown monthly for a span of years. The analysis of time series, involving a consideration of trend, cyclical period (seasonal ), and irregular movements, will be discussed.</FONT></P>
<P ><FONT face="Times New Roman" size=3>In a certain sense, time series are somewhat akin to quantitative distributions in that each succeeding year or month of a series is one year or one month further removed from some earlier point of reference. However, periods of time—or, rather, the events occurring within these periods—differ qualitatively from each other also. The essential arrangement of the figures in a time sequence is inherent in the nature of the data under consideration.</FONT></P>
<P ><FONT face="Times New Roman" size=3>Geographical The geographical distribution is essentially a type of qualitative distribution, but is generally considered as a distinct classification. When the population is shown for each of the states in the United States, we have data which are classified geographically. Although there is a qualitative difference between any two states, the distinction that is being made is not so much of kind as of location.</FONT></P>
<P ><FONT face="Times New Roman" size=3>The presentation of classified data in tabular and graphic form is but one elementary step in the analysis of statistical data. Many other processes are described in the following passages of this book. Statistical investigation frequently endeavors to ascertain what is typical in a given situation. Hence all type of occurrences must be considered, both the usual and the unusual.</FONT></P>
<P ><FONT face="Times New Roman" size=3>In forming an opinion, most individuals are apt to be unduly influenced by unusual occurrences and to disregard the ordinary happenings. In any sort or investigation, statistical or otherwise, the unusual cases must not exert undue influence. Many people are of the opinion that to break a mirror brings bad luck. Having broken a mirror, a person is apt to be on the lookout for the unexpected”bad luck “ and to attribute any untoward event to the breaking of the mirror. If nothing happens after the mirror has been broken, there is nothing to remember and this result (perhaps the usual result )is disregarded. If bad luck occurs, it is so unusual that it is remembered, and consequently the belief is reinforced. The scienticfic procedure would include all happenings following the breaking of the mirror, and would compare the “resulting” bad luck to the amount of bad luck occurring when a mirror has not been broken.</FONT></P>
<P ><FONT face="Times New Roman" size=3>Statistics, then, must include in its analysis all sorts of happenings. If we are studying the duration of cases of pncumonia, we may study what is typical by determining the average length and possibly also the divergence below and above the average. When considering a time series showing steel—mill activity, we may give attention to the typical seasonal pattern of the series, to the growth factor( trend) present, and to the cyclical behaviour. Sometimes it is found that two sets of statistical data tend to be associated.</FONT></P>
<P ><FONT face="Times New Roman" size=3>Occasionlly a statistical investigation may be exhaustive and include all possible occurrences. More frequently, however, it is necessary to study a small group or sample. If we desire to study the expenditures of lawyers for life insurance, it would hardly be possible to include all lawyers in the United States. Resort must be had to a sample;and it is essential that the sample be as nearly representative as possible of the entire group, so that we may be able to make a reasonable inference as to the results to be expected for an entire population. The problem of selecting a sample is discussed in the following chapter.</FONT></P>
<P ><FONT face="Times New Roman" size=3>Sometimes the statistician is faced with the task of forecasting. He may be required to prognosticate the sales of automobile tires a year hence, or to forecast the population some years in advance. Several years ago a student appeared in summer session class of one of the writers. In a private talk he announced that he had come to the course for a single purpose: to get a formula which would enable him to forecast the price of cotton. It was important to him and his employers to have some advance information on cotton prices, since the concern purchased enormous quantities of cotton. Regrettably, the young man had to be disillusioned. To our knowledge, there are no magic formulae for forecasting. This does not mean that forecasting is impossible; rather it means that forecasting is a complicated process of which a formula is but a small part. And forecasting is uncertain and dangerous. To attempt to say what will happen in the future requires a thorough grasp of the subject to be forecast, up-to-the-minute knowledge of developments in allied fields, and recognition of the limitations of any mechanica forecasting device.</FONT></P>
<P ><FONT face="Times New Roman" size=3>INTERPRETATION The final step in an investigation consists of interpreting the data which have been obtained. What are the conclusions growing out of the analysis? What do the figures tell us that is new or that reinforces or casts doubt upon previous hypotheses? The results must be interpreted in the light of the limitations of the original material. Too exact conclusions must not be drawn from data which themselves are but approximations. It is essential, however, that the investigator discover and clarify all the useful and applicable meaning which is present in his data.</FONT></P>
<P ><FONT size=3><FONT face="Times New Roman"> <p></p></FONT></FONT></P>
<P ><FONT size=3><FONT face="Times New Roman"> <p></p></FONT></FONT></P> |
|