»Me, myself and BI«

Bissantz ponders


« previous | next »

Pseudo-precision

368,472 of 818,271 deaths in Germany back in 2004 were caused by cardiovascular disease. At least that is what the German Federal Statistics Office says. Due to the notorious inaccuracy of death statistics, most critics only believe the first digit.

We tend to trust statistics that contain multiple digits. We forget, however, that data ascertainments are not and cannot be precise due to systematic errors (i.e. the low percentage of autopsies) as well as random mistakes in entering, reading and transferring data*.

Eliminate pseudo-precision in your reporting.

Table_pseudo_precision

Take a moment to think what conclusions you want to draw from your numbers.

Table_without_pseudo_precision

* Read the German book “How to lie with statistics” for other sickening, pseudo-precise examples with numbers. (Krämer, W., So lügt man mit Statistik, 3. Aufl., München 2002, S. 16 ff.)

  • del.icio.us
  • Technorati
  • Furl
  • Netvouz
  • Ma.gnolia
  • Blue Dot
  • Digg
  • MisterWong
  • YahooMyWeb
  • Simpy

Leave a response

Friday, October 31st, 2008

Test: Are you a good manager?

Sunday, October 19th, 2008

Rough cut bars may hurt

Friday, October 3rd, 2008

Do managers have to ride rabid tigers? Part II

Friday, September 19th, 2008

Do managers have to ride rabid tigers?

Friday, September 5th, 2008

Do time series charts really compare time series?

Friday, August 22nd, 2008

Reference bars

Friday, August 8th, 2008

Stacking is out

Friday, July 25th, 2008

The demise of decograms

Friday, July 11th, 2008

Killing the sacred cow of information

Friday, June 27th, 2008

The Seventh Dan: Put it in Writing


»Me, myself and BI« Bissantz ponders
DE Deutsch