The distribution of any particular variable from a survey nearly never matches that from the best macro source. There are various reasons why there might be a discrepancy between your survey and the more official or higher quality macro data.

First, there is random sampling error. Since even the best surveys are random samples from the population, any particular sample is likely to differ from the population average just by chance. The nature of sampling error is well known and you can easily test whether the distribution of a variable in your survey is statistically significantly different from that specified by your high quality macro data.

Second, there may be differences in the sampling frame for the macro and micro data. Survey organizations are frequently unable to sample randomly from the entire population, and the choice of sampling frame can make a difference. For electoral turnout, surveys frequently sample all adults in the country, but official turnout figures often refer to the registered electorate or the voting eligible population.

Third, the nature of the measurement may not be quite the same for the different sources, e.g. the question wording may differ. Fourthly, there maybe effects of the way people are interviewed (e.g. by phone or face-to-face) and whom they are interviewed by.

Finally, there may be particular unit or item non-response bias in either the survey or the macro source.

The University of Manchester; Mimas; ESRC; RDI

Countries and Citizens: Unit 4 Combining macro and micro data by Steve Fisher, University of Oxford is licensed under a Creative Commons Attribution-Non-Commercial-Share Alike 2.0 UK: England & Wales Licence.