How to Lie With Statistics by Darrell Huff - Book Report/Review Example

? [Your School] How to Lie With Statistics by Darrell Huff Introduction The book “How to lie with statistics”, by Darrell huff was written in 1954 and describes the fallacies created using statistical inferences and materials. The book is written in a simple and concise style with illustrations in cartons, making it an entertaining read. Huff uses numerous examples in the book to bring out the many ways that statistics can be used in lying. He discusses how statistical aspects such as dubious graphs and biased figures are used by organizations and people to mislead others either unintentionally or intentionally. Huff aims at arming people with information on how to analyze all the research data provided and looking out for sampling errors and hidden biases. Sampling Biases Sampling is the initial process in any research and aims at collecting data for analysis after formulation of the research topic or question. A complete enumeration of the target population is recommended as this increases the accuracy and precision. However, due to the cost implications and time, most researchers use a sample of the target population (Huff 10). Although a sample is supposed to represent all the characters of the population without any bias, this is rarely the case especially during organization marketing surveys and political opinion polls. For example, a sitting mayor seeking re-election in a certain city may pick a sample population to determine his popularity from his hometown county, where he got most of the votes during the last election. This sample population from his home town will give false results about his popularity and he will use these ratings to boost his campaign for re-election. Using this sample locks out people with a different opinion of the mayor from participating in the survey. Researches and surveys mostly use questionnaires or direct interviews in collecting data from the sample population. Interviews and questionnaires can be structured in a manner that misleads the respondent. For example, if Shell-BP is carrying out a survey on whether customers uses its lubricant, they can simply ask a respondent in a questionnaire or interview; “Have you ever used shell BP lubricant?” This question is open and will give misleading results, since it doesn’t ask how often the customer uses the lubricant and during which period they used it, as well as under what circumstances they used the lubricant. Sometimes, despite using the right wording during an interview or in a questionnaire, respondents might give inadequate information or lie. For example, in a social survey to find out the infidelity rate of married men in a certain population, an interviewer can ask a man how many times he has cheated during his marriage. Despite the assurance of confidentiality, some respondents might give false answers to such a question as they are scared of their partners finding out. Well Chosen Average According to Huff, one way that statistics are used in misleading the mass is by stating that a certain number is the average. In layman’s terms the meaning of average is taken as the mean which is calculated by adding up all data points and dividing the total by the number of the data points that were used. However, other measures of central tendency including the mode that refers to the most frequent data point, or the medium referring to the number where half of the data points are lower and the other half are higher, can also be used as average (Huff 45). Researchers do not reveal to the population whether they are using the mode, mean or medium as average in their surveys, although most people will assume that average automatically refers to the mean. Depending on what a person attempts to verify, they can use to use the term “average” in reporting their results whereas they have used either the mode, median or mean in proving their points. Little Figures That Are Not Here Darrell also illustrates the fallacies that can result from statistics if a small sample is chosen to represent the larger population (58). A small sample leads to biased results with a lot of errors, affecting the credibility of the survey. For example, if a media house is carrying out a survey to determine their popularity among listeners, they may use a sample size of 1,000. If 600 listeners interviewed attest that they listen to the station, the direct popularity percentage calculated is at 60 percent. However, if the station is to use a sample size of 1,200 respondents, the popularity percentage changes to 50 percent. On the other hand if the station chooses a large sample of 2400 the popularity percentage stands at 25. This is a clear indication that researchers can manipulate the sample size to achieve the desired effects, although the sample size should neither be too small or too big for improved accuracy and precision. Another issue brought out in the book is that most information and data used in surveys and researches is collected from individuals yet the results are interpreted as a representative of the entire population. The sample selected is assumed to represent all the characteristics of the population which is not always the truth. For example if a health survey is carried out to determine the prevalence of Sexually Transmitted Disease (STDs) among men and women , the individual results from the health facilities might show that 62 percent of men have sought treatment for STDs as opposed to 38 percent of women. These results will be interpreted to mean that men have a high prevalence rate of getting STDs compared to women, while this might not be reflecting the real situation of the entire population because maybe men feel more comfortable seeking treatment for STDs while women shy away from going to hospital although they suffer from STDs. Eye-Catching Graphs After analyzing the data, statisticians use eye-catching graphs to present the outcome of the survey. Huff illustrates that modifying the presentation of the same data can give two different impressions and neither of the two graphs are lying (70). The book also shows how you can interchange the figures on the X-axis and those on the Y-axis to get two different graphs with different meanings. A researcher can plot values on either the X-axis or the Y-axis depending on the desirable effects they want to achieve. Semi Attached Figure Statistical surveys and researches are not carried out on the entire population but a selected group (Huff 87). They can be carried out on a target population which refers to the group from which inferences will be made from, or a study population on which an experiment or treatment is conducted on. An example of a target population is if you wish to carry out a survey on the number of smokers affected by throat cancer, the smokers are the target population. A study population can be a group of obese women who are put on a strict diet and intensive exercises for a period of 8 weeks and their weight loss is monitored and recorded. Whether researchers use a target or study population in making statistical inferences, all the other potential risk factors are held constant apart from the factors being studied. For example, if a smoker gets cancer during the study it is assumed that it is a result of smoking although it might be a hereditary disease that has affected the smoker’s family generation over a period of time. An obese woman might loose weight due to stress or sickness during the study, but at the end of the survey it is assumed that it happened because of dieting and exercises. Fallacies in such studies can be removed by taking into consideration other risk factors that can affect the factor under consideration instead of assuming they do not exist. Trends Huff shows how trends are used in creating lies by observing how things change over a short period of time, predicting the future using statistics and coming up with an eye-catching headline (95). An example is if there is an increase in the rate of a disease like bird flu during winter over a period of five years, a pharmaceutical company marketing bird flu vaccine can use the frequency of the disease, to make bird flu look like epidemic, just by predicting that this trend is likely to continue during all future winters. This will grab media headlines and cause panic in population and many people will go for vaccination to prevent the disease. Such a pharmaceutical company aims at selling a vaccine simply by forecasting the future trend using statistical lies. Conclusion Huff aims at sensitizing people to look out for phony statistics as well as recognize usable and sound data. He advises that one should always question who stated a certain statistical fact, why they stated it and if they have any knowledge of what they are saying. Huff states that people should always check if the conclusions drawn can be justified by the raw data and most of all, make sure that the statistics given make sense. By looking keenly and consciously into all the statistics, people will tell the difference between true statistics and vague and false statistics that have been manipulated with an aim of misleading the mass. Work Cited Huff Darrel. How to lie with statistics. New York: Penguin Publishers, 1954. Read More