How to Mislead with Statistics
​
Mehmet Demir Soytorun
In debate settings and marketing, one of the most common ways of supporting one’s arguments and/or products is using statistics as evidence. Hard data is usually considered to be an objective and irrefutable way to support claims and persuade audiences. However, correct data can very well be used to mislead people, whether intentionally or unintentionally. In this article, I will explore three ways in which factually correct data can lead to malinformed conclusions.
Fractional vs. absolute increases
In 1995, a health crisis in the UK arose after the Committee on the Safety of Medicines declared that a new generation of birth control pills had an increased risk of blood clots by a seemingly ludicrous 100%. In the following years, the usage of oral contraception among young women fell from 40% to 27%. Because of this, thousands of unwanted pregnancies occurred and the now infamous 1995 pill scare is credited for being the cause of 13.000 abortions in the UK. At first glance, the decline in oral contraception usage may seem reasonable because of the 100% increase in blood clot risks. However, the huge percentage increase was over a tiny risk. The new blood clot risk of birth control pills were 0.028% compared to 0.014%. Most people would consider an increase of 0.014% to be hardly newsworthy, and yet it caused nationwide panic and thousands of abortions because of a simple misinterpretation.
Truncated graphs
Oftentimes, data is intentionally manipulated to deceive people, especially by those who seek to benefit from it, such as advertisers or politicians. Companies often use truncated graphs (where y-values start at a higher value than zero) in their advertisements in order to manipulate visual data to better suit their narrative. Take a look at the following Chevrolet ad from 1992:
​
​
​
​
​
​
​
​
​
​
​
​
​
​
The advertisement claims that over 98% of Chevy trucks remain on the road for over ten years, and the graph makes it look like Chevy trucks are incredibly reliable compared to other manufacturers. An unassuming customer might infer from the graph that Chevrolet trucks are nearly ten times as reliable as Nissan and twice as reliable as Toyota trucks. In reality, the difference in reliability is almost negligible. Correcting for the truncation, the graph would look like this:
​
​
​
​
​
​
​
​
​
​
​
Starting the graph at 95% zooms into a small portion of the graph. Every company’s trucks are more than 95% reliable over 10 years, but the truncation greatly exaggerates a 3% difference to look like 1000%.
Cumulative graphs
Another method used by companies to mislead customers are cumulative graphs. A cumulative graph is a graph that visualizes a running total, where each data point represents the sum of all previous values up to that point. While it can be used honestly to showcase growth or progress, it can also be used to mask declines or inconsistencies within the data set. One of the primary ways they are used to mislead people is to make a cumulative sales graph. Take a look at the graph below, used by Apple CEO Tim Cook during a presentation:
​
​
​
​
​​
(The pink bar graph and line were added afterwards by astute viewers)
Even though the cumulative sales graph above gives the impression of constant growth in sales, the actual quarterly sales tell a different story. There are periods of stagnation and even decline in the year, and most importantly, there is not a super clear positive trend of increasing quarterly sales. This disparity showcases how cumulative data is misleading when used to interpret individual, quarter-on-quarter performance.
In conclusion, having a correct interpretation of the data at hand is just as important as having accurate data. Data that is absolutely correct can be cherry picked, distorted and put into misleading graphs to suit a narrative. If you torture the data for long enough, it will confess to anything.​
Works Cited
“Question: An Advertisement for Chevrolet Trucks Uses the Following Bar Graph. 100% - Percentage of Trucks Sold over the Last 10 Years That Are Still on the Road: Chevy Ford Toyota 95% Nissan Datsun More than 98% of All Chevy Trucks Sold in the Last 10 Years Are Still on the Road. over the Years, No Other Truck Is That Dependable, Foreign or Domestic. What Do The.” Solved An Advertisement for Chevrolet Trucks Uses the | Chegg.Com, 3 May 2020, www.chegg.com/homework-help/questions-and-answers/advertisement-chevrolet-trucks-uses-following-bar-graph-100-percentage-trucks-sold-last-10-q49255032. Accessed 25 Nov. 2024.
Stowe, Lisa. “ How to Create a Cumulative Frequency Graph.” Study.Com, study.com/skill/learn/how-to-create-a-cumulative-frequency-graph-explanation.html. Accessed 25 Nov. 2024.
Woollaston, Victoria. “How Apple Exaggerated Sales of Its iPad: Chart Shown at Launch Is ‘misleading’ Because It Fails to Show Recent Dip in Sales.” Daily Mail Online, Associated Newspapers, 24 Oct. 2013, www.dailymail.co.uk/sciencetech/article-2473131/How-Apple-exaggerated-sales-iPad-Chart-shown-launch-misleading-fails-recent-dip-sales.html. Accessed 25 Nov. 2024.
“Health | U-Turn over Pill Scare.” BBC News, BBC, 7 Apr. 1999, news.bbc.co.uk/2/hi/health/313848.stm. Accessed 25 Nov. 2024.
A;, Furedi. “The Public Health Implications of the 1995 ‘Pill Scare.’” Human Reproduction Update, U.S. National Library of Medicine, 1999, pubmed.ncbi.nlm.nih.gov/10652971/. Accessed 25 Nov. 2024.
“Health | Research Spells out Pill Risks.” BBC News, BBC, 10 Nov. 2000, news.bbc.co.uk/1/hi/health/1015822.stm. Accessed 25 Nov. 2024.
