Another area that is often overlooked in assessing data is the treatment of missing items. While good attention is paid to differences in response rates - missing cases - the treatment of missing items - how the 'don't knows' and refusal to individual questions have been treated is sometimes ignored. Sometimes differences in distribution for a particular variable can be because one country has calculated with don't knows in the base while another country has calculated with the 'don't knows' set to missing. In some instances the differences in items missing can be large with the proportion of respondent who refuse to answer the income question being much higher in one country than in another or where the mode of the survey may have impacted on the willingness to respond. Before including variables in your analysis make sure to check that it is reasonably populated and the rates are similar across countries.