Comparing indicators based on the CIV Survey 2007
Comparing indicators based on CIV Survey 2007 results
In the same way that survey estimates (or results) are subject to sampling variability, the difference between two survey estimates is also subject to sampling variability. It is the difference between the estimates may vary from the figures that would have been obtained had the entire population been surveyed.
A difference between two estimates is statistically significant if there is statistical evidence that there is a difference (i.e. it is unlikely to have occurred by chance). The statistical meaning of ‘significance’ differs from the usual sense of the word, in that the size or importance of the difference is not necessarily relevant.
The size of the difference required for it to be unlikely to produce statistically significant difference depends on the respective standard errors of the estimates being compared. Estimates at the total Victorian level have lower standard errors than those at LGA level. Therefore, a smaller difference is needed for comparisons between an LGA with the Victorian average than when comparing the differences between two LGAs.
In comparing LGA estimates with the Victorian average, the minimum difference required for that difference to be significant generally ranges from about four percentage points (when the estimates are very low or very high) to about six percentage points (when the estimates are closer to 50%). In comparing estimates for different LGAs, the minimum difference required generally ranges between 5-8 percentage points.