28 March 2008

data anal

someone famously said "there are three kinds of lies: lies damn lies and the statistics". the third is being talked about now.

In an engineering chemistry lab exam i was supposed to draw a graph based on the experiemntal results. i do not remember what the thing was about. i got three points on the graph; there were supposed to be co-linear. ofcourse there were not! then primary school teachers teach us to drop a point and draw a stright line. the examiner said you have rigged the results! "where have i?" i asked. you can see the points but the graph is approximated. in summary i lost some marks.

darmashastra says that any one who screws a chemistry lab sample salt should be failed. luckily i had obtained a second sample after pouring water to the salt in the test tube. so i was not unhappy to lose some marks in the second experiemnt.

now, this is not context contrast so we have to talk about data analysis, data anal for short. a magazine here said Saskatoon is the most dangerous place to live in North America! no one can be more happy! this means that after living here i can live anywhere in North America! i looked at their methedology! they have % in 200 , 300 even 400 range! if any % is more than 100 i get skeptical: this means they should instead say "n times". 200% is same as 2 times. % has a specific purpose: expressing a value relative to hundred.

these magazines ask 50 people and call it a survey. it is like asking 1000 software engineers in Bengaluru on who will win the next election; none of them is going to vote! How does their answer matter.

this data anal is a murky thing. i get scared when i read data anal in other papers and then of my own !

No comments:

Post a Comment

Please leave a note about what you think about this write up. Thanks.