Misleading Statistics (Term 1 2014 Tuesday Thursday Class)

          People come across a lot of statistics every day.  Statistics are used especially by governments or companies. Statistics are used in different ways, for example, to advertise, to report, or to get attention. Without awareness of misleading statistics, people can get the wrong ideas. Therefore, if statistics are misleading, people’s decisions might be incorrect. Statistics are often misleading in many ways. Statistics should be interpreted with caution as they can be misleading; they can both lie and tell the truth.  This essay will discuss about many different types of misleading statistics such as intentionally misleading statistics, sample problems, comparing different situations, misleading surveys, missing information, correlation confusion and average problems.

                                                                                                                                    804 words

Intentionally Misleading Statistic:

Misleading statistics are being reported to people every day. Truthfully, they are designed deliberately to confuse people.Those statistics are reviewed and changed by experts. Therefore it requires the experts to interpret the misleading statistics.

One example of misleading statistics was Barack Obama’s presidential election during 2012.The narrative starts from the unemployment rate of Americans which declined sharply from 8.1% in August to 7.8% in September. This particular statistic has put many Americans in awe.The changes in percentage seem very small; however, considering the whole population of America, this actually turned everything upside down. Fortunately. This statistical problem was solved because one of the employees named Julious Buckmonwas caught faking the story and admitted receiving the order from higher positions. “Go Ahead, and fabricate it” (NY post, 2013) was a conversation line taken from the telephone call between Buckmon and a powerful anonymous. The journalist was making the report for 60,000 jobless households only in Philadelphia, eventually each of them stood for 5,000 households in USA. Moreover, “Buckmon filled up the surveys by himself instead, especially those who had not answered their phones or opened their doors.” (NY post, 2013). By faking the survey results, it appears as if many citizens were then employed but in reality, they were not. Consequently, it was considered to be one of the important factors in helping Barack Obama to win his election. It is shown that even official terms sometimes could be misleading and this kind of misleading report has changed the situations completely and eventually changing the whole nation. 

Another example was in the 20^thcentury; intentionally misleading statistics were the reasons that the United States lost a battle in the Vietnam War.Earlier, body count was not necessary but territory was. Even if one side started out with more territories or land areas and they ran out of soldiers, the other side could fight back easily if they had more troops. In the Vietnam War, the United States believed in the body count statistics that commanders in the battle reported. Furthermore,“It was the first war in the United States history that the body count was only a number; it was not how victorious is earned”. (Phillips, 2009) The battlefield commanders surely got benefits from this misleading statistic. For instance, in a battle in guerrilla war, hundreds of Vietnamese soldiers were killed, but only two of the United States troops died. The commanders inflated body count to get the statistic to appear more successful than they actually were. (Phillips, 2009) They used this statistic to make citizens believe that they did not lose tothe North Vietnamese.However, the result was the opposite from what the Americans were informed.The United States lost the war and the reports during the war was responsible for deceiving the American public.

            Moving on, from politics to the world of advertisements, let’s look at another example. There are many successful and false advertisements on products, especially cigarettes. Regardless of how critically the public think, those who work in the advertisement industry will eventually come up with a brand new idea and repeatedly trick us to be consumers of their products.They employ every possible way to create logic in their advertisement to show how excellent their company’s cigarettes are, compared to other brands. As shown in the picture above, they mentioned that “more doctors choose Camel than other cigarettes” (Camel doctor’s choice, 2014). The word “more” refers to doctors which makes people assume that there are hundreds or thousands doctors who smoke their cigarettes.By not informing the readers clearly about the number of doctors they mentioned, one will conceive that the cigarette must be better and harm their health less.It was written: “smoke only Camel for 30 days and you will experience how it touches your taste” (Camel doctor’s choice, 2014). This means that, in order to find out how good that brand’s cigarette is,one needs to spend at least a month on smoking their company’s cigarette which only seems like a scant period of time for the smokers. However, a month period is actually sufficient to get the smokers addicted and therefore, the customers are likely to become regular consumers of the product. Furthermore, those experts under the advertisement might not be independent doctors;they could be amateur doctors who were paid to smoke Camel.Furthermore, the letter “T” which resembles human throat will make consumers inevitably generalize that their throats will be safe from the diseases.Deplorably, this does not matter because cigarettes will not only harm the throat, but also the entire respiratory system, blood vessels and possibly cause brain tumors which are worse than throat infections.

The materials presented above illustrate how data in various advertisements in and governments got manipulate and how false information was applied to deceive citizens.

           

Bibliography

1.      Camel doctors choice [Web, Graphic]. Retrieved from http://171.67.24.121/tobacco_web/images/tobacco_ads/doctors_smoking/more_doctors_smoke_camels/large/camels_doctors_choice.jpg [accessed 11 February 2014]

2.      John, C. (2013, 11 18). Census ‘faked’ 2012 election jobs report. Retrieved from http://nypost.com/2013/11/18/census-faked-2012-election-jobs-report/[accessed 11 February 2014]

3.    Phillips, M. (2009, June 1). Army deploys old tactic in pr war. Retrieved from http://online.wsj.com/news/articles/SB124380078921270039  [Accessed 09 February 2014]

Level: EAP 5

Class day & Time: Tue/Thurs 17:30-20:00

Title: How to Lie with the Statistic

Sample Problem

Some statistics are created based on samples, since the whole population is impossible to measure. Therefore, most of the samples are biased even if researchers try their best not to create any biases. However, some misleading statistics are established because surveyors are picking samples that would provide accurate numbers. The size of samples can lead to misleading statistics since some groups of samples are excluded, which can influence the result of the survey. Hence, it is essential to collect data from random samples. As an example, coin has two heads and eight tails side, after throwing a coin ten times, we might suppose that the probability of getting two heads side is only twenty percent, but the truth is that the percentage of getting either sides is fifty percent each. The result would have been more precise if the coin is flipped more than one hundred times (Chudler, Internet).

The size of samples, which is too small, also cannot represent the whole population. For instance, if we selected one hundred people and ask them whether they accept or refuse the discrimination of homosexuals, and fifty-five of them accepted, normally people would presume that fifty-five percent of all population approved. Thus, it could possibly be that the majority of the total population would not accept the discrimination against homosexual. The result might be different because surveyors might ask only those people who accepted the problem or people in their samples were more religious than those in reality (Spagnoli, 2009, Internet).

Non-response bias in online surveys is another problem that can create misleading statistics. Only those people who are interested in the topic would respond or complete the surveys. As an illustration, Responsive Management and the South Carolina Department of Natural Resources (SCDNR) conducted a survey via Internet to rate the participations and comments on saltwater fishing and shell fishing in South Carolina, especially to clearly indicate the possibility and accurate data of online surveys. They could be done this research because they had a sample out of the entire population (people who have a South Carolina Saltwater Recreational fisheries license) that could be compared to the online survey result. The data, which came from the database, show that only 16,100 out of the total population(103,000 people) could be contacted by e-mail addresses, which means that the online survey had eliminated eighty five percent of the possible samples(Responsive Management, Internet).

Furthermore, in order to influence the result or statistics, people are willing to complete the survey or vote more than once because of unverified respondents. A complicated issue is when companies offer a bonus for completing online surveys such as a chance to win a prize, discount, or other benefits, it could encourage multiple responses from one person. If people are keen on winning an item, they will probably find a way to complete the survey several times, which can increase their chances to win the prize (Responsive Management, Internet). Another purpose for people to vote twice is that they want to influence the result of the survey. As an example, if a survey is limited by only one email account, people will create new accounts to get more chances to take part in the survey. According to a poll on the Internet, published by Newsweek in year 2000, nine percent of people said that Nader was the only politician worth voting for. Consequently, people estimated he would get at least nine percent of the vote in the election, but he only got approximately three percent of the vote. There was a bias in this selected sample. The reason is that this sample was not randomly selected from the whole population. A disproportionate amount of Nader supporters took part in the poll, for the purpose to make his image as a leader (Alden, 2007, Internet).

Another factor that can lead to the misleading statistics is the place of samples. Some surveys need to be done in an exact place to get reliable information. For example, if Apple Company wants to know the number of people who will buy new IPhone Six with the price of more than one thousand US dollars, they should get their survey in the zone of wealthy people because the poor cannot afford to buy this product in such a high price. So, if the survey was conducted in the countryside the result would be inaccurate. 

On the whole, there is always a built-in bias in most of the surveys or statistics, which can create misleading statistics (Chudler, Internet). It usually takes a lot of hard work and preparation to get a good stratified sample. However, most researchers do not give any attention to this. Therefore, most surveys are undependable. Moreover, in order to get an accurate information, a sample should be selected randomly and make sure that sample is a stratified sample (Math, Internet).

Word Count: 807words

Bibliography:

- Alden, L., (2007). Statistics can be misleading [online]. California. Available from: http://econoclass.com/misleadingstats.html [Accessed 01 January 2014].

-Author Unknown, Specializing in survey research on Natural Resource and Outdoor Recreation Issues [online]. Harrisonburg, Responsive Management. Available from: http://www.responsivemanagement.com/news_from/2010-05-04.htm.

[Accessed 01 January 2014].

- Chudler, E., How to Lie and cheat with Statistics [online]. Washington,. Available from: http://faculty.washington.edu/chudler/stat3.html. [Accessed 01 January 2014].

- Math, G. How to Lie with Statistics [online]. Mexican, Calvin College. Available from: http://www.calvin.edu/academic/economics/faculty/bios/HaneyDocs/page-58952330.html. [Accessed 01 January 2014].

- Spagnoli, F., (2009). Lies, Damned Lies, and Statistics (9): Too Small Sample Sizes in Surveys [online]. Belgium. Available from: http://filipspagnoli.wordpress.com/2009/09/10/lies-damned-lies-and-statistics-9-too-small-sample-sizes-in-surveys/ [Accessed 01 January 2014].

 

Writing Essay

“Statistics should be interpreted with caution as they can be misleading; they can both lie and tell the truth.”

            Statistics can be misleading in many different situations but only three of them have been chosen to talk about in this essay. They are differences in numbers, definitions and situations.

Numbers are one of the main reasons that figures can be confusing. As people are too attracted by numbers, they are not fully aware of other details given. Sometimes appearances can be deceiving such as mistaking figures and biased numbers that are measured technically in different situations. Some examples are introduced in the following cases. Statistics in the rate of pregnancy was misleading for the reason of comparing unequal numbers. The percentage of teenagers having kids in conservative religious areas was higher than the national average, which demonstrated that the rate of unwed mothers had increased. So, the statistics was confused as the result of the growing percentage of married teens. (Responsible Thinking, n.d.). Moreover, another example is that there was a claim about the unemployment percentage after a new president was elected. However, the president’s defenders talked back as the employment percentage also rose. Due to the process of population growth, therefore the total rate of both employment and unemployment had increased. So this means that both of the parties were true. (Responsible Thinking, n.d.).

Another cause of misleading statistics in comparing different situations is the situation itself, which can end up with two different consequences. For example, two students have a bet about receiving the better grades from the same courses that they have taken last quarter. As you can see, the figure 1 below illustrates the grades of 2 different students. By just looking at it, one of the students declared that it was a tie since they both received 2 As, 1 B, and 2 Cs. However, the other student claimed that he won the bet by having higher GPA for the quarter. To find the GPA, we must multiply the grades’ points with the subjects’ credits individually. Then add all the products and divide it by the total credits. So, this example gives us one story with different endings, which is causing the statistics to be confusing. We can say they are tied since they both received the same grades. On the other hand, if they are looking at the GPA, there would be one winner for sure (Misleading Graphs and Statistics, n.d.). In addition, statistics can be misleading in many other ways like in various situations. One example of this is that percentages can also make us confused if we compare the diverse groups of people. This can often come out with headlines such as “Mississippi has the maternal death rate in century” (note: MI is chosen for illustrative purpose only). The implication is that Mississippi did not have a good health care system. Recently it also turned out that the population of black people have a worse maternal death rate. For example, in Mississippi 1 in 1000 black women die and 1 in 2000 white women in childbirth.  In Illinois, 1 in 9000 black women and 1 in 18000 white women die when giving birth. Mississippi has a lower rate for death groups but it has a population which is 50% black and 50% white, so about 1 in 15000 women die in childbirth. In contrast, if Illinois in 10% black and 90% white, then its death rate will only be 9 in 15000 in childbirth, which is slightly lower rate than in Mississippi (Stats n.d.).

Table 1: the grades of 2 different students.

(Allow 4 points for A, 3 points for B, 2 points for C, 1 point for D, and 0 point for F.)

Course	Student 1	Student 2
Maths (4 credits)	A	C
Chemistry (4 credits)	A	C
English (3 credits)	B	B
Psychology (3 credits)	C	A
Tennis (1 credits)	C	A

Source: Misleading Graphs and Statistics.(n.d.). Retrieved from:http://faculty.atu.edu/mfinan/2043/section31.pdf

A final cause of misleading statistics in situations is when definitions are mixed up and being used in different meanings. Confusing figures can be used in different definitions and in different situations. One example of this is that a report from US News and World Report illustrated that Alaskans did not treat their children properly. Many people would look at this figure and think that it is true but is it really true? If people examined the comparisons carefully between the states chosen to compare, there will be something wrong. The difference in the abuse rate is probably made up from different criteria in those areas. Like in Alaska, which is probably considered as the worst state in child abuse, indicated that until a child’s health or welfare is harmed or hurt then that would be written down as abuse. However for Pennsylvania, which is perhaps known as the best state in treating children, defined abuse as a failure to act as parents. Also in the survey, North Dakota had not been reported. So, North Dakota might have been the best or the worst state. No one knows. The survey was conducted by the states using different things to compare. It would have been better if they only compared states with the same meanings. (Alden,2005-7)

            To sum up, there are a variety of ways that statistics can mislead people. It is just that people should be more aware of what they see because it might leads them to think differently. Statistics can both be honest or deceiving.

.
Bibliography

Alden, L. (2005-7). Statistics can be misleading. Retrieved from: http://www.econoclass.com/misleadingstats.html

Misleading Graphs and Statistics. (n.d.). Retrieved from: http://faculty.atu.edu/mfinan/2043/section31.pdf

Responsible Thinking. (n.d.). Principles. Retrieved from http://www.truthpizza.org/logic/stats.html

Stats. (n.d.). Percentages. Retrieved from http://www.stats.org/faq_percentages.htm

Surveys can be a Virus for Statistics.

People use a lot of surveys in their daily lives to support their argument or help themselves to make a decision.  According to the Oxford dictionary, a survey is the investigation of opinion, behavior, et cetera of a particular group of people, which is usually done by asking them questions (Oxford Advanced Learner’s Dictionary, (2010). Schools, businesspeople, restaurants or shop owners often use surveys. Seeing many different surveys each day, people tend to let their guard down when reading one and people can easily misunderstand the real content. Surveys can be misleading regarding the design of questions, survey methodology, samples, and intepretation.

One way in which surveys can be biased is how questions in a survey are designed. There are two things which are worth considering about how to ask questions. First, the questions in the surveys are clear so that people have problems in understanding them. Technical words, slang or jargon are sometimes used and the questions are sometimes confusing or wordy. For instance, a surveyor asked, “What do you think about parking?” This question is not clear about what  the surveyor want to know. Participants will try to make a guess if it is asked about the general parking or a person’s ability to park (Driscoll & Brizee, 2010). The questions can also angle for a particular answer by encouraging participant to respond in a certain way (Driscoll & Brizee, 2010). For example, the question “Should killing babies be legalized?” can give the wrong impression to the participants and can be rephrased as “Should abortion be legalized?”. If the first question was asked, the response would be negative. However, the responses to the second question would be more postive than that to the first one. It is clearly that the wording of a query is very influential to its answer.

In addition, different methods of doing surveys can gather different results. Methods of survey refer to the various ways that the surveys are done such as voting, answering multiple choice questions, giving ideas on the website or by writing on the provided sheet. A common problem is opinion polls. Statistics based on the polls cannot be trusted if the polls are constructed to encourage a particular answer . Polls can easily be rigged to get a desired amswer by the way the questions are phrased since it only attracts  responses from those who are interested in the discussion topic. Also, in the case of doing oral surveys, the recruiter sometimes tells them what to answer and they will say anything just to get their $75 or $100, (Craig, 2007). However, only a small percentage of participants will not participate. On the other hand, those who are not interested in the survey or feel irritated to be asked by the surveyors would give a nonchalantly answer. For instance, they will just choose the shortest answer or something they feel like saying. Clearly, both opinion polls and oral surveys have disadvantages that surveyors must consider.

Another way that surveys can confuse people is related to samples of participants. This means that samples are just selected for any purposes which can lead to a big confusion. For instance, during the US presidential campaign of Franklin D. Roosevelt in the 1990s, the polls were conducted only among people who had telephones. After the survey, pollsters predicted that FDR would not win the election, but actually he did. The reason for  this inaccurate prediction is that pollsters only asked rich people who had telephones and that was just a part of the population. This was not accurate which could not represent the overall opinion of people (Huff, 1954).

 Small samples can mislead the result of the survey. A business company also uses this idea to gain popularity among public. In advertisements, only few people were interviewed regarding their opinion of using the product. Then, they imply that the majority of people are using their product. Therefore, this method is very unreliable since the sample size was quite small and not randomly selected.

Besides, survey results can be misinterpreted in many ways. One way that survey results can be misinterpreted is assuming a correlation between two variables as causation. During the survey, surveyors manipulate a variable to measure its effects on another variable. For example, a surveyor went to ask people both males and females whether they use the Internet and found out that 28% of males and 65% of females use the Internet. If the surveyor estimates that sex causes the difference in Internet use, it will be completely wrong. The surveyor can only say that sex and Internet use have correlation. Also, the interpretation of the word “most” can lead to misleading statistics. Sometimes when the surveyor does the survey, he or she found that one response is the most frequent response. Then the surveyor interprets the survey that most people do this or like this. It is incorrect unless the response is from the majority of participants (over 50%). For example, after a person does a survey about the popularity of politicians, he or she obtains the result, in which 48% of participants likes Prime Minister Hebrew, 36% likes Senator Myung, and 16% likes Princess Scheherazade. The surveyor cannot estimate that most people like Prime Minister Hebrew even though his percentage is higher than others. To sum up, survey results can be misinterpreted ("Common errors in," 2011).

In conclusion, there are four main problems in doing a survey: biased questions, methodology problems, sample problems, and wrong interpretations. Surveyors must consider carefully about these points in order to make an accurate survey.

Bibliography

Charney, (2007), The top 10 ways to get misleading poll results, New York. Revived from http://www.charneyresearch.com/2007July9_CampaignElections_Top10ways.htm

Common errors in the interpretation of survey data . (2011). Retrieved from http://www.oesr.qld.gov.au/about-statistics/analytical-methods/common-errors-interpretation-survey.pdf

Driscoll, D. L., & Allen, B. (2014, April 17). Creating good interview and survey questions. Retrieved from https://owl.english.purdue.edu/owl/resource/559/06/

Huff, D. (1954). How to lie and cheat with statistics. Retrieved from http://faculty.washington.edu/chudler/stat3.html

Oxford Advanced Learner’s Dictionay. (2010) (8^th ed.)

Word counts: 1058, Class: EAP 5, Time: 5:30-8:00

Missing information, a form of correlation confusion,is one of the most common mistakes that makes statistics become misleading. These kind of statistics are confusing to the audiences because they lack necessary information on several variables or a whole unit, and sometimes no information is provided at all. There are five distinct characteristics of the missing information, and the first is called missing completely without reason which means that the data is not acquired from the original source because of a research method. The second type, missing without reason, occurs due to the survey questions. The third kind is missing information that depends on inexperienced predictions. In this category, only selected group of people are questioned and this makes the statistic very inaccurate if the researchers generalized the results. The last characteristic, censoring, happens when the source of information refuses to give an answer to the survey (Andrew, 2013). Furthermore, correlation errors and causation can be found when lurking variables appear. This means that hidden variables can also affect the true variables. An example for this is that the pollution in China increased because there were lots of Christians in China. A lurking variable of time exist in there. Although, Christianity has spread in China, that country improved its industry (Beckey Roselius).

The main reason why many people make missing information statistics is because they can use statistics for their advantage and manipulate them to satisfy their desires. Amidst the many groups of people, the most common groups that twist statistics are advertisers and politicians (Derek R, 1982). The main goal of advertisers is usually to sell as many goods as possible to consumers. As for politicians, the objectives of their manipulations are to make themselves look good and win elections (Derek R, 1982). For example, there are instances when researchers do surveys on income rate and target people from every class. When filling in the surveys, people who have incredibly high or low income might incline to not answering questions concerning their gross pay due to various reasons. This kind of behavior will lead to the creation of a statistic that is missing some core information, and the situation of this dismissal is called censoring (David C., 2012).

Correlation Confusion

Statistics can be misleading because correlation confusion can distract people by forcing the connection between causes and effects, which are not obviously related to each other. This happens because the conclusions are drawn without careful attention to cases’ analysis (The Nizkor Project, 2012). The chance of this kind of correlation seems to increase when small sample sizes of statistics are chosen. As a result, some companies like to use this way to advertise their products (Haney, 2014).

In addition, correlation errors in misleading statistics appear because there is confusion between causes and effects. This occurs because the person, who made the statistics, committed the fallacy. Consequently, with lack of evidence, the conclusion confuses the cause and the effect. Then, some people will find the statistics misleading because they have different ideologies. For example, one said that violence on television should be banned because it will make people like violence. Another one said that violence on television exists because people like violence. That is why, it is hard to decide which one is the cause and which is the effect because people have different perspectives (The Nizkor Project, 2012).

People sometimes believe that statistics are misleading because if they consider about variables critically, there will be one cause and two effects. For example, the more exercises people do, the healthier and fitter they are. This also means that too much exercise is not good for people because there would be no time to create new muscle tissue or to recover their body (Haney, 2014). In addition, according to Courtney Taylor, there is a study that illustrates a correlation between jewellery sales and the amount of daylight. Higher sales seem to appear when the daylight lasts longer. However, the amount of sales still increases on Valentine’s Day and Christmas during the winter that has less daylight (Taylor, 2014).

Average Problem

Statistics with accurate data and correct creation techniques can be misleading when involving the word average. The word average can have three different definitions: mean, mode, and median.  For this reason people can manipulate the data and give the wrong impression that might fool the audiences. The word mean is the sum of all values that is divided by the numbers that represent those values. Median is the value that is at the center of the whole set of data. Mode is the values that occur most within the set of values (Mcmillian, Preston, Wolfe & Yu, 2006).

These three meaning of averages can be used to deceive people in different ways. Mean can be uses to show high average when there is a very huge gap separating data. This happen because the maximum number in the data is a very large number which would affect the total sum creating the illusion of a very high average number (BeckeyRoselius).   Median sometime could fool people the same way as the mean did the outcome would be that the mean has a slightly larger value than the median. However the difference is when the whole set of data has an even number of values.  People can choose random value that is located in the middle. For example 1, 2, 3, 100, 1000 and 2000 the one that is presenting the word average can deceive other people by maybe choosing 100 or 3. It might seem easy to identify the fraud with this small list of number here but with a larger list it might have been more difficult and one could be easily fooled. Mode is not used as much to scam people except for maybe changing the meaning of the average and makes the average number look small.

Bibliography

Mcmillian, A.Preston, D. Wolfe, J., Su, Y., (2006, 11 21). Basic statistics: mean, median, average, standard deviation, z-scores, and p-value. Retrieved from https://controls.engin.umich.edu/wiki/index.php/Basic_statistics:_mean,_median,_average,_standard_deviation,_z-scores,_and_p-value

Taylor, C. (2014). Common statistics mistakes. Retrieved from http://statistics.about.com/od/HelpandTutorials/a/Common-Statistics-Mistakes.htm[Viewed Date: February 4th, 2014]

The Nizkor Project. (2012). Fallacy: Confusing cause and effect. Retrieved from http://www.nizkor.org/features/fallacies/confusing-cause-and-effect.html [Viewed Date: February 15th, 2014]

David C., H. (2012). Treatment of missing data--part 1. Retrieved from http://www.uvm.edu/~dhowell/StatPages/More_Stuff/Missing_Data/Missing.html

Derek Rowntree, 1982, Statistics without tears: A primer for non-mathematicians. Harmondsworth: Penguin.

 [Acessed at: 01/Feb/2014]

BeckeyRoselius, H. (n.d.). Calvin. Retrieved from http://www.calvin.edu/academic/economics/faculty/bios/HaneyDocs/

Andrew, G. (2013). Missing-data imputation. Retrieved from http://www.stat.columbia.edu/~gelman/arm/missing.pdf

            To sum up, this essay has described some examples of misleading statistics. The information in these statistics are not clear enough to decide whether it is reliable or not. Thus, one needs to analyze the statistic carefully before making any decisions.