Data Lies

Data Lies

I've received a really important e-mail from LinkedIn today, it said I have 252 contacts I might know and I could add them to grow my network.

First I get a top-notch Math lessons, my small network of 474 (let's call this number n) contacts would grow into a network of 726 (and this one N) contacts!

That's amazing: 474 + 252 = 726

And then, I am presented me with an breath-taking graphical representation of this growth:

Reading the numbers and having a look at the picture, something is a bit... weird... can't you see it?

Well, let's call A to the area of the dark (bigger) circle, measuring its radius R at approximately 72 pixels, 

A = PI * R^2 ~= 16286 pixels.

The radius r = 28.5 of the lighter (smaller) circle, giving an area a of:

a = PI * r^2 ~= 2551 pixels.

The relation between the improved network I'd have and my current one is:

N/n ~= 1.5

whilst the relation between the areas of the two circles is:

A/a = 16286 / 2551 ~= 6.4

Let's review the "Lie Factor" (http://www.infovis-wiki.net/index.php/Lie_Factor)

 

 

 

 Size of effect in Graphic (g) = (16286 - 2551) / 2551 ~= 5.4

Size of effect in data (d) = (726 - 474) / 474 ~= 0.5

Lie Factor = g/d ~= 10.8

To ensure the Integrity of a graphic, its Lie Factor should have a value between 0.95 and 1.05. If the value is less or greater, it indicates a substantial (and often intended) distortion, far beyond minor inaccuracies (e.g. caused by plotting).

4 circles of size 474 fits easily in the 726 circle.

I thought: maybe they've used radii to apply the growth factor...

R/r = 71 / 28.5 ~= 2.5

No. Lie Factor > 2.5.

Now, I don't like being lied to, not even in unimportant facts as a poor infovis. What was the point of this? Imagine other visualisations / info-graphics you might have read...

A' = a*1.5 ~= 3826 pixels 

R' = sqrt(A' / PI) ~= 35 pixels

Seriously, why?

16 mins ago, this whole thing happened again.

Like
Reply

Linkedin es todo mentiroso y bombero omen! ajajaj

I liked this article! If they show bars instead of circles the effect could have more impact. It's basically a "misunderstood" of how things scales up depending on how many dimensions we are using. To increase an area the double you have to multiply by Sqrt(2) and not by 2. Let's assume it's just a misunderstood. I liked it! Good analysis.

To view or add a comment, sign in

More articles by Víctor E. Esteve

Others also viewed

Explore content categories