Cosmos


COSMOS - The SAO Encyclopedia of Astronomy › E

Errors

  • In astronomy almost every quantity we observe has an associated error often
    expressed as a $ \pm $ quantity after the value, ie:

    $ M = 4.1 \pm 1.0 M_\odot $

    suggests that our best estimate of the mass $ M $ is 4.1 M$ _\odot $ but that
    it could easily be 1.0 M$ _\odot $ more or less than that.

    Sometimes astronomers choose to drop the $ \pm $ and just put the error in the last digit in parentheses (brackets). So in the example above this would be somewhat confusingly written as:

    $ M = 4.1(10) M_\odot $

    Novices sometimes think this means $ 4.1\pm10.0 $ - it does not!

    More awkwardly, if the best estimate was $ M=4.0 M_\odot $ then we could have written:

    $ M = 4(1) M_\odot $

    which is equivalent to

    $ M = 4.0(10) M_\odot $!

    Normal or Gaussian Distributions

    gaussian
    In nature when we make a series of measurements they
    often follow a Gaussian or Normal distribution like that shown above.

    What does this really mean? Well, astronomers usually set the error to mean there is a 67% chance that the true value falls within one listed error of the value. If the measurements are distributed "normally", ie in a Gaussian fashion, then there is about a 96% chance the value is within twice the listed error and a >99% chance it is within three times the error. The error is often referred to as "sigma". So a 1-sigma result is quite poor, whereas a 3-sigma result reasonably secure. The sigma comes from the standard deviation of a Gaussian distribution.


Syndicate content