The literature is equivocal on the existence of a systematic difference in item completion rates between the mail (postal) mode and the web mode, two self-administered data collection modes. The study addresses that issue in the context of a concurrent mixed-mode contact strategy with self-selection of the data collection mode. However, the study should be viewed as exploratory rather than confirmatory. In order to balance the differences in demographic properties of the respondents, who were free to choose the mode for their participation, a propensity score weighting was used; and in order to take into account the differing properties of the items comprising the questionnaire, a logistic regression was applied. The esults show a small but significant difference in base item completion rate in favour of the web mode, not attributable to any of the investigated item properties, demographic properties or data collection procedures. Additionally, several item level factors were found to differentially influence item completion rates in the two modes, indicating an interaction between item and mode—some of them with direct applicability to data collection improvement while others awaiting further work in order to understand them better.