If you can read this, either the style sheet didn't load or you have an older browser that doesn't support style sheets. Try clearing your browser cache and refreshing the page.

(Wall Street Journal)   110% of statisticians would like to see better use of statistics in 2013   (blogs.wsj.com) divider line 53
    More: Unlikely, Bialik  
•       •       •

1768 clicks; posted to Main » on 31 Dec 2012 at 9:34 AM (1 year ago)   |  Favorite    |   share:  Share on Twitter share via Email Share on Facebook   more»



53 Comments   (+0 »)
   
View Voting Results: Smartest and Funniest

Archived thread

First | « | 1 | 2 | » | Last | Show all
 
2012-12-31 08:57:43 AM  
Petty vandalism such as graffiti is down eighty percent, while heavy sack beatings are up a shocking nine hundred percent.
 
ZAZ [TotalFark]
2012-12-31 09:02:46 AM  
My pet peeve is when ads or articles cite murder rates or death rates by saying 'That's one murder every 10 minutes,' or, 'Someone is dying from starvation every five seconds,

The rate I prefer is years per death, e.g. if NYC has 10 million people and 10,000 murders per year the figure is 1,000 years. The figure for death from all causes is about 80 years, so murder in our hypothetical NYC is a significant cause of death but not the probable cause of death.

It's amazing, even in our complex modern world, how many assertions fail simple 'back of the envelope' reasonable estimates with elementary computations

I see this all the time in transportation planning. One study purported to show that speed cameras had been beneficial, when by the authors' own figures the cost of time wasted by slowing down exceeded all the alleged benefits.
 
2012-12-31 09:12:16 AM  
Statistically speaking, 9 out of 10 people enjoy gang rape.
 
2012-12-31 09:31:26 AM  
profile.ak.fbcdn.net
 
2012-12-31 09:32:51 AM  
Statistically speaking, 99% of us will be paying more taxes next year.

/probably
 
2012-12-31 09:38:34 AM  
oh man this headline is giving me a significant p-value. time for an F test, baby.
 
2012-12-31 09:43:06 AM  
94.7% of statistics are made up.
 
2012-12-31 09:45:11 AM  
Lies, damned lies and statistics.
 
2012-12-31 09:46:07 AM  
I think the percentage of this working is 33.33 repeating of course.
 
2012-12-31 09:47:14 AM  
Michael Dean, a senior marketing analyst in Minneapolis, wanted more clarity in weather news and forecasts. "How accurate are the five-, seven-, and 10- day forecasts?" Dean asked. "Can't someone collect data on the predicted temperature various days in advance, and then see what temperature it ends up being? What is the range of error by the number of days out for the forecast? What times of the year or what regions of the country does this range vary the most? I am thinking I should ignore anything longer than a five-day forecast, but those may be off a lot, too." (Some of Dean's questions are answered by the website Forecast Advisor.)

uh... waste of taxpayer money. big computers. supercomputing contests. can't predict way out of paper bag. recommend time-series historical profile analysis. i nailed the 'october' real estate market crash within three days using it. pulled my money outta the market. unfortunately ex-wife and narcotics addicted family stole it all.... human behavior is not-so-predictable-, unless you use a kardashian-oxycotin-bieber-sex-spears-coke model.
 
2012-12-31 09:47:24 AM  
Soooo, are we saying that politicians can't quote statistics anymore?

/ That should make presidential debates fun. Set any politician who quotes statistics on fire or something. We'd run out of politicians quick.
 
2012-12-31 09:47:36 AM  
100% of bogus statistics are made up by statisticians
 
2012-12-31 09:51:23 AM  

johnnieconnie: Statistically speaking, 99% of us will be paying more taxes next year.

/probably


Well only 50% will since only about 50% pay
 
2012-12-31 09:51:39 AM  
*margin of error +-110%
 
2012-12-31 09:51:53 AM  
I think Todd Snider got it pretty much right. Not entirely SFW lyrics, but Todd IS the real deal and funny as Hells.
 
2012-12-31 09:56:17 AM  
Kelly Jackson, who teaches at Camden County College in Blackwood, N.J., would like to see better practices in charting, where the Y-axis should always start at 0 when possible. "One of the problems my students have is interpreting data and graphs that don't use '0′ as the starting point," Jackson said. "Imagine a graph that starts vertically at 500 and shows bars of height 550 and 600."

This is one of mine. The news always zooms in the Y-axis, thus showing wild volatility that is normal in nature, but that smooths out when you zoom out and show the whole chart. You might not always have to show zero, but at least show the entire range of the data as far back as needed to tell the whole story.
 
2012-12-31 10:08:48 AM  
110%? That's like three quarters.
 
2012-12-31 10:10:21 AM  
www.global-air.com

My college political science professor said, ''If you don't get anything else out of this class, remember this: All statistics are made up''. (new window)
 
2012-12-31 10:12:02 AM  
As long as pirate attacks keep going down, I don't care what global warming lies they try to tell us.
 
2012-12-31 10:13:49 AM  

ZAZ: The rate I prefer is years per death, e.g. if NYC has 10 million people and 10,000 murders per year the figure is 1,000 years. The figure for death from all causes is about 80 years, so murder in our hypothetical NYC is a significant cause of death but not the probable cause of death.


414 of them in 2012 as of a few days ago, just over 4% of your guess. Try Baltimore or New Orleans.
 
2012-12-31 10:15:58 AM  
47% of statistics are made up.
 
2012-12-31 10:15:58 AM  
johnnieconnie: Well only 50% will since only about 50% pay

I thought it was 47%?
 
2012-12-31 10:18:59 AM  

thorthor: johnnieconnie: Well only 50% will since only about 50% pay

I thought it was 47%?


With a loose enough margin of error you're always right.
 
2012-12-31 10:20:23 AM  

ThrobblefootSpectre: As long as pirate attacks keep going down, I don't care what global warming lies they try to tell us.


What they don't tell you is that a reduction in pirate attacks usually means an increase in zombie and/or ninja attacks.
 
2012-12-31 10:20:41 AM  
Personal media statistic annoyance #1: Reporting rate change for very small numbers, very common in medical research journalism. .

Article: Drug X caused a 200% increase in cancer!!!!
Reality: Increased from 0.000001% to 0.000003%

Yes, it's a 200% increase, because 900 people will get sick instead of 300, and it should be reported, but it implies a high risk of cancer, not a 0.000002% increase in your chance of getting cancer.

Personal media statistic annoyance #2: Rate change on fluctuating numbers. Less common, but even more annoying. "Value X dropped 20% over the previous term, but increased 25% in the last few years!". Implying 25% is greater than 20%, so things are worse/better than they were... except the 25% increase brings it back to baseline.
 
2012-12-31 10:23:14 AM  

plcow: This is one of mine. The news always zooms in the Y-axis, thus showing wild volatility that is normal in nature, but that smooths out when you zoom out and show the whole chart. You might not always have to show zero, but at least show the entire range of the data as far back as needed to tell the whole story.


Showing zero is something that should always be done. I have seen far to many graphs which, at a glance, look like they show very interesting data. But then you look at the axis and you'll see that the y-axis goes from 500 to 550 with point one being 505 and point 2 being 545. Since people suck at remembering where they got information form there will be people who remember that the graph showed a very large difference, the second bar was 9x the size, but the other people who will keep correcting them will note that the difference was less than 10%.

Example: you want people to stop eating crayons because it is bad for your health. Which of these graphs would you show them?

www.piccer.nl

Lying with statistics doesn't require mad skills or faking data, just present it in a way that people are likely to agree with you.

/A 6 foot statistician walks crosses a river which is on average 3 foot deep
//He drowned
 
2012-12-31 10:25:18 AM  

PhDemented: Personal media statistic annoyance #1: Reporting rate change for very small numbers, very common in medical research journalism. .

Article: Drug X caused a 200% increase in cancer!!!!
Reality: Increased from 0.000001% to 0.000003%

Yes, it's a 200% increase, because 900 people will get sick instead of 300, and it should be reported, but it implies a high risk of cancer, not a 0.000002% increase in your chance of getting cancer.


And this. Forgot about it for a bit. When people mention changes in percentages you need to provide a baseline.
 
2012-12-31 10:29:19 AM  
Five out of three Americans don't understand fractions.
 
2012-12-31 10:34:26 AM  

utah dude: uh... waste of taxpayer money. big computers. supercomputing contests. can't predict way out of paper bag. recommend time-series historical profile analysis. i nailed the 'october' real estate market crash within three days using it. pulled my money outta the market. unfortunately ex-wife and narcotics addicted family stole it all.... human behavior is not-so-predictable-, unless you use a kardashian-oxycotin-bieber-sex-spears-coke model.


In aggregate, it's reasonably predictable. You can't tell who is going to do what, but in bulk, you can you guess fairly correctly how many will do what.
 
2012-12-31 10:34:35 AM  

vudukungfu: 110%? That's like three quarters.


half the time that's spot-on. like Dad always said: cheap at half the price (/) -or- six of one, half-dozen of the other.

/ it all adds up
 
2012-12-31 10:40:25 AM  
A parsimonious model is better, don't overfit.
 
2012-12-31 10:42:17 AM  

DerAppie: plcow: This is one of mine. The news always zooms in the Y-axis, thus showing wild volatility that is normal in nature, but that smooths out when you zoom out and show the whole chart. You might not always have to show zero, but at least show the entire range of the data as far back as needed to tell the whole story.

Showing zero is something that should always be done. I have seen far to many graphs which, at a glance, look like they show very interesting data. But then you look at the axis and you'll see that the y-axis goes from 500 to 550 with point one being 505 and point 2 being 545. Since people suck at remembering where they got information form there will be people who remember that the graph showed a very large difference, the second bar was 9x the size, but the other people who will keep correcting them will note that the difference was less than 10%.

Example: you want people to stop eating crayons because it is bad for your health. Which of these graphs would you show them?


My pet peeve is not labeling axes. What is y? "Crayon-related fatalities per million children"? If so, your graph shows the benefit of crayon-eating, and therefore we should make it mandatory in all schools for children to eat their crayons.
 
2012-12-31 10:42:33 AM  
There's a big difference between statistics and "marketing analytics", subby. Like the difference between a thinking cranium and a bag of hammers.
 
2012-12-31 11:00:58 AM  
Mark Train was a damn optimist . (" there are liars, damn liars and there are Statisticians )
 
2012-12-31 11:02:43 AM  
It could be 110% if they asked 11 people and one of them died before the study was published.
 
2012-12-31 11:05:13 AM  

Jenna Tellya: DerAppie: plcow: This is one of mine. The news always zooms in the Y-axis, thus showing wild volatility that is normal in nature, but that smooths out when you zoom out and show the whole chart. You might not always have to show zero, but at least show the entire range of the data as far back as needed to tell the whole story.

Showing zero is something that should always be done. I have seen far to many graphs which, at a glance, look like they show very interesting data. But then you look at the axis and you'll see that the y-axis goes from 500 to 550 with point one being 505 and point 2 being 545. Since people suck at remembering where they got information form there will be people who remember that the graph showed a very large difference, the second bar was 9x the size, but the other people who will keep correcting them will note that the difference was less than 10%.

Example: you want people to stop eating crayons because it is bad for your health. Which of these graphs would you show them?

My pet peeve is not labeling axes. What is y? "Crayon-related fatalities per million children"? If so, your graph shows the benefit of crayon-eating, and therefore we should make it mandatory in all schools for children to eat their crayons.


Y is unimportant as these are fictional datapoints. If I had actual data I'd have labeled the axis. Just make up a label and then think about which one you'd choose. And then think about which one you'd choose if you wanted to discredit the findings.
 
2012-12-31 11:05:29 AM  
9 out of 5 statisticians surveyed confirmed that we need more statistics.
 
2012-12-31 11:16:56 AM  
You think you'd always want to start at zero, but then I give my students data spanning from around 40-60, ask them to freehand a boxplot, and of course the majority start at zero and attempt to cram the boxplot into whatever tiny space is left on the page.

It's more important to label the axes and make sure you an appropriate range for your type of data and your graph. Are you displaying counts or percentages? You should (probably) start at zero. Are you making time plot or a scatterplot? For god's sake, just have the axes cover the range of the data.
 
2012-12-31 11:36:04 AM  

rcf1105: You think you'd always want to start at zero, but then I give my students data spanning from around 40-60, ask them to freehand a boxplot, and of course the majority start at zero and attempt to cram the boxplot into whatever tiny space is left on the page.

It's more important to label the axes and make sure you an appropriate range for your type of data and your graph. Are you displaying counts or percentages? You should (probably) start at zero. Are you making time plot or a scatterplot? For god's sake, just have the axes cover the range of the data.


THIS. I get annoyed when laypeople look at plots of global average temperature in degrees Kelvin and screech about how they don't include zero on the Y-axis.
 
2012-12-31 11:45:25 AM  
My pet peeve is when journalists (or farkers) use nonsensical wording, like "200% decrease".
 
2012-12-31 11:53:19 AM  

LouDobbsAwaaaay: I get annoyed when laypeople look at plots of global average temperature in degrees Kelvin and screech about how they don't include zero on the Y-axis.


abattybroad.files.wordpress.com
What "laypeople" may look like.
/97% of them cannot math.
 
2012-12-31 11:57:43 AM  
8 out of 10 cats love this thread
 
2012-12-31 12:11:01 PM  

utah dude: oh man this headline is giving me a significant p-value. time for an F test, baby.


I gotta go Type 1 right now...stomach sensing I might have to go Type 2 as well.

Power.
 
2012-12-31 12:25:52 PM  

Summer Glau's Love Slave: LouDobbsAwaaaay: I get annoyed when laypeople look at plots of global average temperature in degrees Kelvin and screech about how they don't include zero on the Y-axis.

[abattybroad.files.wordpress.com image 752x500]
What "laypeople" may look like.
/97% of them cannot math.


Unfortunately, those laypeople also complain about the dimensions on my Y-axis.
 
2012-12-31 12:39:03 PM  

LouDobbsAwaaaay: Summer Glau's Love Slave: LouDobbsAwaaaay: I get annoyed when laypeople look at plots of global average temperature in degrees Kelvin and screech about how they don't include zero on the Y-axis.

[abattybroad.files.wordpress.com image 752x500]
What "laypeople" may look like.
/97% of them cannot math.

Unfortunately, those laypeople also complain about the dimensions on my Y-axis.


Hi Yo!

/Actually, that was pretty good.
 
2012-12-31 02:04:26 PM  

Summer Glau's Love Slave: LouDobbsAwaaaay: Summer Glau's Love Slave: LouDobbsAwaaaay: I get annoyed when laypeople look at plots of global average temperature in degrees Kelvin and screech about how they don't include zero on the Y-axis.

[abattybroad.files.wordpress.com image 752x500]
What "laypeople" may look like.
/97% of them cannot math.

Unfortunately, those laypeople also complain about the dimensions on my Y-axis.

Hi Yo!

/Actually, that was pretty good.


And here I was fixated instead on the "degrees" mistake, goddammitsomuch
 
2012-12-31 02:06:48 PM  

plcow: Kelly Jackson, who teaches at Camden County College in Blackwood, N.J., would like to see better practices in charting, where the Y-axis should always start at 0 when possible. "One of the problems my students have is interpreting data and graphs that don't use '0′ as the starting point," Jackson said. "Imagine a graph that starts vertically at 500 and shows bars of height 550 and 600."

This is one of mine. The news always zooms in the Y-axis, thus showing wild volatility that is normal in nature, but that smooths out when you zoom out and show the whole chart. You might not always have to show zero, but at least show the entire range of the data as far back as needed to tell the whole story.


On the contrary, for some things the (relatively small looking) difference IS what's important.  In those cases, showing a graph that starts from 0 would be worse than useless.

There ARE times when it would be better to focus on the entire range starting from zero, but it's oversimplifying at best (and downright dumb at worst) to say every graph should start at 0.

To be fair, that isn't quite what you were saying, I realize.  I'm responding more to what Kelly Jackson said than what was posted.
 
2012-12-31 02:16:46 PM  

ciberido: plcow: Kelly Jackson, who teaches at Camden County College in Blackwood, N.J., would like to see better practices in charting, where the Y-axis should always start at 0 when possible. "One of the problems my students have is interpreting data and graphs that don't use '0′ as the starting point," Jackson said. "Imagine a graph that starts vertically at 500 and shows bars of height 550 and 600."

This is one of mine. The news always zooms in the Y-axis, thus showing wild volatility that is normal in nature, but that smooths out when you zoom out and show the whole chart. You might not always have to show zero, but at least show the entire range of the data as far back as needed to tell the whole story.

On the contrary, for some things the (relatively small looking) difference IS what's important.  In those cases, showing a graph that starts from 0 would be worse than useless.

There ARE times when it would be better to focus on the entire range starting from zero, but it's oversimplifying at best (and downright dumb at worst) to say every graph should start at 0.

To be fair, that isn't quite what you were saying, I realize.  I'm responding more to what Kelly Jackson said than what was posted.


I completely agree. You use the right tool for the job.

If you're trying to show recent changes in trends in the deficit, you don't need to include the zero on either the deficit or the year axis.
 
2012-12-31 02:39:25 PM  

LouDobbsAwaaaay: THIS. I get annoyed when laypeople look at plots of global average temperature in degrees Kelvin and screech about how they don't include zero on the Y-axis.


On the temperature deal... 80 degrees F is not twice as hot as 40 degrees F.

I get sad when it takes me a few tries to explain that to students.
 
2012-12-31 02:57:08 PM  

LouDobbsAwaaaay: rcf1105: You think you'd always want to start at zero, but then I give my students data spanning from around 40-60, ask them to freehand a boxplot, and of course the majority start at zero and attempt to cram the boxplot into whatever tiny space is left on the page.

It's more important to label the axes and make sure you an appropriate range for your type of data and your graph. Are you displaying counts or percentages? You should (probably) start at zero. Are you making time plot or a scatterplot? For god's sake, just have the axes cover the range of the data.

THIS. I get annoyed when laypeople look at plots of global average temperature in degrees Kelvin and screech about how they don't include zero on the Y-axis.


Laypeople look at plots of global average temperature in degrees Kelvin?
 
Displayed 50 of 53 comments

First | « | 1 | 2 | » | Last | Show all

View Voting Results: Smartest and Funniest


This thread is archived, and closed to new comments.

Continue Farking
Submit a Link »






Report