Take a photo of a barcode or cover
Reviews
Everydata: The Misinformation Hidden in the Little Data You Consume Every Day by John H. Johnson
sweemeng's review against another edition
5.0
If you already know some stats, this books is not that useful. But it highlights a lot of things people take from granted when presented with data(information). Definitely recommend for people to read, for people that knows some stats, it is a relaxing refresher
leamaura's review against another edition
3.0
How to read the news 101. There were only a few things I didn't know already, but the explanations were easy to follow and the facts taken as examples were interesting.
atxspacecowboy's review against another edition
3.0
Everydata = the data that surrounds us in our everyday lives
This book is a relatively simplistic overview of many principles of statistics such as, outliers, margins of error, statistical significance, sampling, cherry picking, confirmation bias, averages, probability, forecasts, etc.
The goal of the book is to make the reader a better educated consumer of “everydata.”
Notes:
A sample of the population in question is used and inferences are made on the whole when polling the entire population in question is not possible.
A striking feature of research in American psychology is that it’s conclusions are based not on a broad a cross-section of humanity but hey small corner of the human population; mainly person living in the United States.
The United States is less than 5% of the worlds population but his home to 68% of samples in studies.
But wait, there’s more; research is done consistently on college students, specifically undergraduate students in psychology classes.
According to one journal, 2/3 of the published studies done in America used undergrad psychology students for their samples.
Bad
Regarding sample size, bigger isn’t always better. You could sample every undergrad psychology student in the US and it wouldn’t give you a better picture of the US or world population as a whole.
For those who don’t like sampling:
“The next time you have a blood test, ask them to take it all out.”
-Arthur Nielsen, Sr.
5 things you can to do make better decisions using aggregated date, averages, and outliers.
1. know what a summary statistic is and what it isn’t; sometimes summaries don’t tell the full story.
2. Understand what type of average is being presented; mean, median, or mode. Example: the average person in the world has two arms (some have zero or one arm but very few have 3)
3. Ask, “An average of what?” Is it representative of a sample? Is it an average of averages?
4. See if all the data is treated equally. Weighted averages?
5. Identify outliers, understand the importance they can have on the data, and exclude where appropriate. (Ex Bill Gates lives on a street with 9 other people who earn $50K annually, their mean income would be over a $Billion)
Omitted variables are the main reason correlation does not equal causation.
5 things you can do to see if the data you’re seeing actually matters:
1. Ask if the results could be due to random chance (sample size)
2. Understand that many findings are based on probability.
3. Know that the data you see in headlines is often part a range (+/- confidence interval)
4. Even if the effect is statistically significant, look at the size or magnitude of the effect.
5. Consider the impact that data has on your life. Just because something is statistically significant and has a large magnitude it may not have an impact on your every day life.
How to be a smart consumer of data that is misrepresented or could be misinterpreted:
1. For charts and graphs take a good look at the X axis and Y axis. Scale and height can lead to misleading results.
2. Pay attention to the language; what people don’t say it often as important as what people do say.
3. Verify your source.
4. Make sure it’s not a mistake. Almost 1 in 5 large businesses have suffered a financial loss due to a spreadsheet error.
5. Interpret the data correctly. Sometimes the data is correct but it is misinterpreted because of confusion with fractions, decimals, or other user error. (1/3lb burgers don’t sell nearly as well as 1/4lb burgers because most people think 1/3 is smaller than 1/4)
“4 out of 5 doctors recommend Gerber baby food.”
So, only one didn’t recommend Gerber, right? Wrong. Actually, only about 12% of doctors surveyed recommended Gerber. How did they get to four out of five? Cherry picking. Cherry picking means you select anecdotal data to make your point while ignoring other data points that contradict it. At the time of this survey, many doctors (more than 25%) did not recommend baby food at all because of added sugar and fillers.
Here’s the actual data:
562 pediatricians surveyed
408 recommend baby food
76 recommended a specific brand
67 recommended Gerber
Based on the data, the following statements are true and give a more accurate accurate representation of the findings:
“Of the doctors we polled who recommended baby food, less than 1 (0.8) out of 5 recommended Gerber”
“Of doctors who recommend baby food, and of those who recommend a specific brand of baby food, 4/5 recommend Gerber”
“1 out of every 10 doctors we polled recommended Gerber”
...but none of those sell as much baby food as the cherry picked line.
The FTC (Federal Trade Commission) ruled that Gerber was misleading with cherry picked data.
How to be a smart consumer of forecasts:
1. Know that predictions of the future depend upon the past. If there are issues with the data of the past, it will negatively affect the accuracy of your forecast.
2. Understand the difference between deterministic (it will rain tomorrow) and probabilistic (there’s a 40% chance if rain) forecasts.
3. Understand the terminology; “forecast” and “prediction” are often used interchangeably, but they are different.
4. Understand that the accuracy of a forecast may change over time. Forecasting the final score of the baseball game in the seventh-inning would surely be more accurate than at the beginning of the game. More data.
5. Accept that there will always be some level of uncertainty.
In summary, five takeaways to be a good consumer of everydata:
1. Recognize data when you see or hear it. It’s everywhere.
2. Get your facts right. Verify the data you’re seeing is in fact accurate.
3. Understand where the data is coming from and who is presenting it (biases).
4. Watch out for the obvious data traps, e.g., correlation vs. causation.
5. Understand that interpreting data correctly will help you make good decisions.
This book is a relatively simplistic overview of many principles of statistics such as, outliers, margins of error, statistical significance, sampling, cherry picking, confirmation bias, averages, probability, forecasts, etc.
The goal of the book is to make the reader a better educated consumer of “everydata.”
Notes:
A sample of the population in question is used and inferences are made on the whole when polling the entire population in question is not possible.
A striking feature of research in American psychology is that it’s conclusions are based not on a broad a cross-section of humanity but hey small corner of the human population; mainly person living in the United States.
The United States is less than 5% of the worlds population but his home to 68% of samples in studies.
But wait, there’s more; research is done consistently on college students, specifically undergraduate students in psychology classes.
According to one journal, 2/3 of the published studies done in America used undergrad psychology students for their samples.
Bad
Regarding sample size, bigger isn’t always better. You could sample every undergrad psychology student in the US and it wouldn’t give you a better picture of the US or world population as a whole.
For those who don’t like sampling:
“The next time you have a blood test, ask them to take it all out.”
-Arthur Nielsen, Sr.
5 things you can to do make better decisions using aggregated date, averages, and outliers.
1. know what a summary statistic is and what it isn’t; sometimes summaries don’t tell the full story.
2. Understand what type of average is being presented; mean, median, or mode. Example: the average person in the world has two arms (some have zero or one arm but very few have 3)
3. Ask, “An average of what?” Is it representative of a sample? Is it an average of averages?
4. See if all the data is treated equally. Weighted averages?
5. Identify outliers, understand the importance they can have on the data, and exclude where appropriate. (Ex Bill Gates lives on a street with 9 other people who earn $50K annually, their mean income would be over a $Billion)
Omitted variables are the main reason correlation does not equal causation.
5 things you can do to see if the data you’re seeing actually matters:
1. Ask if the results could be due to random chance (sample size)
2. Understand that many findings are based on probability.
3. Know that the data you see in headlines is often part a range (+/- confidence interval)
4. Even if the effect is statistically significant, look at the size or magnitude of the effect.
5. Consider the impact that data has on your life. Just because something is statistically significant and has a large magnitude it may not have an impact on your every day life.
How to be a smart consumer of data that is misrepresented or could be misinterpreted:
1. For charts and graphs take a good look at the X axis and Y axis. Scale and height can lead to misleading results.
2. Pay attention to the language; what people don’t say it often as important as what people do say.
3. Verify your source.
4. Make sure it’s not a mistake. Almost 1 in 5 large businesses have suffered a financial loss due to a spreadsheet error.
5. Interpret the data correctly. Sometimes the data is correct but it is misinterpreted because of confusion with fractions, decimals, or other user error. (1/3lb burgers don’t sell nearly as well as 1/4lb burgers because most people think 1/3 is smaller than 1/4)
“4 out of 5 doctors recommend Gerber baby food.”
So, only one didn’t recommend Gerber, right? Wrong. Actually, only about 12% of doctors surveyed recommended Gerber. How did they get to four out of five? Cherry picking. Cherry picking means you select anecdotal data to make your point while ignoring other data points that contradict it. At the time of this survey, many doctors (more than 25%) did not recommend baby food at all because of added sugar and fillers.
Here’s the actual data:
562 pediatricians surveyed
408 recommend baby food
76 recommended a specific brand
67 recommended Gerber
Based on the data, the following statements are true and give a more accurate accurate representation of the findings:
“Of the doctors we polled who recommended baby food, less than 1 (0.8) out of 5 recommended Gerber”
“Of doctors who recommend baby food, and of those who recommend a specific brand of baby food, 4/5 recommend Gerber”
“1 out of every 10 doctors we polled recommended Gerber”
...but none of those sell as much baby food as the cherry picked line.
The FTC (Federal Trade Commission) ruled that Gerber was misleading with cherry picked data.
How to be a smart consumer of forecasts:
1. Know that predictions of the future depend upon the past. If there are issues with the data of the past, it will negatively affect the accuracy of your forecast.
2. Understand the difference between deterministic (it will rain tomorrow) and probabilistic (there’s a 40% chance if rain) forecasts.
3. Understand the terminology; “forecast” and “prediction” are often used interchangeably, but they are different.
4. Understand that the accuracy of a forecast may change over time. Forecasting the final score of the baseball game in the seventh-inning would surely be more accurate than at the beginning of the game. More data.
5. Accept that there will always be some level of uncertainty.
In summary, five takeaways to be a good consumer of everydata:
1. Recognize data when you see or hear it. It’s everywhere.
2. Get your facts right. Verify the data you’re seeing is in fact accurate.
3. Understand where the data is coming from and who is presenting it (biases).
4. Watch out for the obvious data traps, e.g., correlation vs. causation.
5. Understand that interpreting data correctly will help you make good decisions.
psz's review against another edition
1.0
Simplistic and elementary. Very short but has even less content than you might think. Silver lining: very quick to get through.
Someone with zero understanding of statistics, surveys, polls, and samples might get much more out of this book than I did.
Someone with zero understanding of statistics, surveys, polls, and samples might get much more out of this book than I did.
silvia_linn's review against another edition
3.0
They’re trying to be freakanomics but without any astounding unexpected insights. It’s really a collection of evidences that you can’t trust anything. It was fine, but pretty shallow.
bretthardin's review against another edition
2.0
A good book about statistics for a layman. However, it is written in a strange friendly tone. Not an authoritative one. I was removed from the prose many times due to inline witticism.
donaldcramer's review against another edition
2.0
There are better pop math books out there but this may be a good one to share with my statistics class.