3.4.1. Comparability of overall design

You need to satisfy yourself that the surveys you are using are of sufficiently comparable design to support the analysis you intend. For surveys that are explicitly designed to support cross-national comparisons you need only consider whether the departures from the design specification are warranted and whether the variables of interest are subject to any "difference" between the countries. For example, if the survey used different modes in different countries (telephone, face to face, self-completion with interviewer, mail, web etc) is the mode likely to have an effect on how the question was answered or whether there are any issues around the cultural equivalence of the item?

Where you are using datasets that are not explicitly designed for this purpose you need to exercise your own judgement on comparability. It is increasingly common with international surveys for countries to be allowed to use whichever mode of survey collection is "usual" in that country. This means that in some countries the survey will be by face-to-face interview while in others the survey may have been by telephone or by post. Internet surveys are still not widely used as a method although some countries are experimenting with them as an additional mode. For example, the Australian Bureau of Statistics included the option to complete the 2006 Census on line.