The
calculator at the right is designed to compute error ranges around survey
percentages automatically. If you would prefer to download a table that
will help you do it without the calculator, click here
for a printable PDF. Here's some background either way.
Standard
Error quantifies the uncertainty that comes from measuring only a sample
of a population rather than measuring the whole population. It is
determined by two variables:
|
|
-
The
sample size (the larger the sample the smaller the
Standard Error.)
-
The
percentage whose standard error is being calculated (percentages
closer to 0 or 100 have smaller Standard Errors.)
Standard
Error is used to calculate the range around an observed survey
percentage that includes the "real" number that would be
obtained if the entire population had been surveyed. This range is
usually expressed at a given level of certainty, called the Confidence
Level. The Confidence Level states the probability that a given error
range includes the "real" population number. In survey
research, Confidence Levels of 95%, 90% or 80% are most commonly used. A
level of 95% would mean that the "real" population percentage
would be included in an error range in at least 95% of the surveys if
they were repeated a large number of times. In other words, the odds
would be 19 to 1 that the estimate derived from the survey would be
correct within the calculated error range. The error range is calculated
by multiplying the Standard Error by a constant that is associated with
each Confidence Level.
The calculator above
does all this for you. Simply enter the desired Confidence Level, the sample size used
in your survey and the percentage whose error range you wish to
calculate. The resulting error range should be expressed as
plus/minus the observed percentage. For example, for a Confidence Level of 90%,
a sample size of 500 and a percentage of 60%, the error range would be +/- 3.6% points.
That
is, if the survey were repeated an infinite number of times, the observed percentage would fall between 56.4% and 63.6% at
least 95% of the time. The smaller the error range, the more
certain you can be that the survey percentage is correct.
|