The New Zealand Census provides an invaluable source of information on the estimated resident population of the country and its demographic, social and economic characteristics. The data derived from the census finds wide usage in research and analyses which inform numerous government policy decisions, including resource and funding allocation. Consequently, missing data caused by non-response to the census, whether unit (not returning completed census forms) or item (not providing a valid response to questions in the census forms) can cause a range of practical, methodological and ethical issues. Users of census datasets need to understand the implications of including or excluding the non-responding sub-population from their analyses. This study looked at the residual non-response category ‘Not Stated’ or ‘Not Elsewhere Included’ (NS/NEI) for the 2006 and 2013 Censuses. NS/NEI is an aggregate of two different population groups. The majority are substitute records created by Statistics New Zealand as an estimate of the population not responding to the census at all, and the remaining are people who did not provide a valid response to the question on the census form, the item non-respondents. Detailed analyses were conducted on the variable that has one of the highest rates of item non-response: the question on personal income. The aim was to examine the demographic and socio-economic characteristics of the non-respondents and therefore identify any patterns or over-representation of certain population groups, likely to have an impact on any analysis carried out using these data. This study therefore enables analysts to have a more evidence based approach towards dealing with this residual category. The analysis of the demographic and available key socio-economic characteristics of the item non-respondents (NS/NEI group excluding substitutes) to the income question showed patterned behavior with certain population groups over-represented - females, youth, those aged 65+ years, Pacific Islanders and Māori, people born overseas, people receiving government benefits, those not in the labour force or unemployed and people working as labourers and machinery operators and drivers. Consequently, the exclusion or inclusion (by say, pro-rata apportioning or imputation) of the NS/NEI count from the analysis of the income variable will adversely affect these groups in terms of representativeness. The users of census data need to take into account that the NS/NEI is not a homogeneous grouping. The imputed age-sex and spatial distribution of the majority of the substitutes is modelled on the population enumerated on census night and consequently, the demographic profile of the substitutes is more representative of census respondents. The substitutes included in the NS/NEI category can therefore substantially alter the demographic profile of the actual item non-respondents. To achieve an accurate profile, it is important to first exclude the substitutes from analyses. Decisions around the inclusion or exclusion of the NS/NEI count from any analysis can then be taken by ascertaining whether any particular population groups are over-represented among the item non-respondents and how the exclusion is likely to impact on the conclusion(s) reached from the analysis of that particular variable. Alternatively, for inclusion in the analysis, this will provide a more informed base for modelling the probable responses for the item non-respondents.