Summary
Random error is unmeasurable but can (under some circumstance) be quantified. Random sample error can be reduced through using larger sample sizes and appropriate sampling techniques. Demographers tend to ignore random measurement error in their data collection and analysis. The implications of random sampling error are revealed through the use of appropriate statistical analysis.
Systematic error (or bias) can arise from an inadequate sample design, problems in its implementation, non-response, or the collection of biased measures.
Analysis and assessment of the data, and actively searching for possible indicators that non-sampling errors are present in the data, allows users of the data to identify many types of systematic error. These errors should then be controlled for in the analysis of the data, or explicitly identified, so that the limitations of the data are better understood and the risk of producing misleading or spurious results is reduced.
Before embarking on detailed analysis of the data, or using the data for policy or planning purposes, users of the data should gauge the extent of systematic errors in the data and satisfy themselves that they do not make it impossible to draw meaningful conclusions.