Skip to content

Observation – Statistics Can be Deceiving

October 18, 2011

For example…

From the U.S. Census, Table 1105:

National Summary, Fatal Motor Vehicle Accidents: 1990 – 2009

So the conclusion based on the data is:

Sober people are twice as likely to die in car crashes than drunks. So a further conclusion might be that drunks are twice as lucky than sober people.

Don’t drink and drive, regardless of what the data says.

  1. James permalink
    October 18, 2011 8:31 pm

    The percentages presented in the table simply means the proportion of sober/drunk people involved in fatal accident, which are not the same as likelihood or probability of getting involved in accidents. Thus even though sober people has twice the percentage than the drunk people here, it doesn’t mean sober people are twice likely to be killed in car accident. To find the true likelihood we need to investigate the number of sober/drunk people on road and then take ratios seperately. Assuming proportion of sober to drunk driver is probably 1000:1, a fatal accident ratio of 2:1 actually indicates the drunk are much more likely (500 times more likely in this case) to be involved, and thus suffer fatal injury in car accidents.

  2. October 18, 2011 9:23 pm

    You are absolutely correct.

    Surely you understood by my tone that I was taking a rather satirical view of statistics. I was merely pointing out something using a tactic our politicians use on a regular basis. Taking a set of numbers and drawing an absurd conclusion based on the raw data without factoring in the data that’s not in the table.

    It’s like the statement: Thirty one percent of all traffic fatalities are caused by drunk drivers. That means sixty-nine% is caused by sober drivers?

    But I digress.

Comments are closed.