Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Perfect Example of Data Manipulation
1 point by efojs on July 5, 2022 | hide | past | favorite | 3 comments
There is this article with a title:

"Tattoos and piercings are more common among those who experienced childhood abuse and neglect"

(not going to make a link for them)

What message do you get out of this title? — that most tattooed people were abused in childhood, don't you?

How about their conclusion:

> Results showed that around 40% of participants had at least one tattoo or piercing and approximately 25% of participants reported significant child abuse or neglect. Among the participants reporting child abuse, 48% had a tattoo or piercing, while only 35% of people not reporting child abuse had a tattoo or piercing.

It says that 48% were abused — almost half. Impressive, ha?

However, if you plot what they say, or recalculate, or rebuild the sentence, you get that:

only 12% are tattooed and were abused, and ~35% are just tattooed.

Just saying.



Oh, I'm sorry, even now I lied here, but not intentionally:

> and ~35% are just tattooed.

~35% (37% to be exact) not of all respondents, but of those who have tattoos (from those 40%).

From all respondents it would be 28% vs. 12%.


> Among the participants reporting child abuse, 48% had a tattoo or piercing

Based on the quoted text I'd read that (union of abuse and tattoo/piercing) as 48%, not 12%.


Yes, it is the power of this manipulation — even seeing data does not help, but:

"Among the participants reporting child abuse (25% of all respondents), 48% had a tattoo or piercing" ― out of 100 people, 25 were abused and 48% of them (i.e. 12 people) have tattoos → 12 of 100 respondents have tattoos and were abused, while 28 just have tattoos.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: