Home » 72 – Can We Trust Statistics?

72 – Can We Trust Statistics?

by Jill

  • Playing With The Pool – To ensure that your data is solid and a legitimate study, the pool of surveyed or studied people has to represent an actual population. Having a small pool of people or people from a particular community can taint the statistics and show a fake result.
  • Bad Questions and Bad Options – Asking questions that taint the answers or only providing limited results to answer the questions can ensure the results are entirely useless. Sometimes these surveys are run by polling agencies trying to provide a pre-determined outcome. People running the polls might use the poll to sway opinion with biased questions.
  • Show Only Great Results – People running surveys or studies will sometimes only show their results if it matches their intentions. However, all results should be shown because that is as valuable as an expected result.
  • Definitions Change – Definitions change over time. That means we often can’t compare years of data or different groups of statistics. Even physical conditions can change and will taint the results. For example, in sports, the fields and equipment change over time. For population studies, the population continually grows, so more people doing something this year might not be interesting if it does not increase compared to population increases.
  • What is a Study? – A study is a team of testing multiple variables against a population of people to see if something we change (a drug, a situation)  changes our outcomes beyond what we would expect in a natural result. For example, if I flip 100 coins and get 70 vs. 30, I expect it is still within chance. Or does it prove an abnormal situation caused the effect?
  • How Many People Do We Need? – The more people you have in a study, the more it will represent the general population, and the more valuable the data will be. The more questions you ask or the more groups you compare, the more people you will need to prove something.
  • Odds and Natural Chance – Coin flips and baby genders have a predictable result of outcomes. We want to know when we have a situation that caused an unexpected effect. If the choices are not removed when they are picked (in coins, we always have two sides), the odds stay the same. If we remove options like taking out bingo balls, the odds of each number being selected increase as balls are drawn.
  • Tyranny of the Mean – At times, people will compare a statistical average to compare two groups that cannot be reached. Even if team sports, you might hear the average scores of each team. But that really does not speak about the quality of the teams. Each team had a different schedule, accounting for the difference in scoring abilities.
  • Why Trend Towards the Mean – While every coin flip or baby has the same odds of one choice over another. The more you flip a coin or, the more babies are born, the more the count will meet the expected average. If you have a deviation from the predicted average, the trend will head towards the mean without intervention just because averages tend to work out. If a team is losing every game, they will start to win without changes because they tend to win their average game numbers.
  • Unrealistic Studies – There are times when studies are done in groups that do not represent our expected population. A drug is tested in mice, and it is not valuable enough to tell if it will happen in people. A cleaner might say it fights 99% of germs, but that might mean in a perfect situation on an ideal countertop.


  • Try to notice in this next week how many times people give you statistics that I want you to take a look at it and just think logically. Are there ways that the statistics were given about a specific thing? Could they be wrong? Or could they be right? See if you can see a little bit behind what’s going on that’s being given to you.


You may also like

Leave a Comment