Differences in response patterns in a mixed mode: online/paper & pencil business survey

This paper investigates if significant differences exist between online and paper & pencil participants in a quarterly business survey in the German business?related services sector when respondents may freely choose to respond either online or by more conventional methods. It also analyzes the determinants of online participation and studies if item?nonresponse is reduced when respondents participate online. Online respondents, indeed, significantly differ from paper & pencil participants with respect to their ordinal sales and price assessments, with online participants being less optimistic on both counts than their paper & pencil counterparts. These differences are not attributable to observable firm characteristics as far as sales judgements are concerned and they disappear once it is controlled for observable firm heterogeneity in the case of price assessments. Significant differences in the judgement of demand, profit and employment as well as in the expectations concerning sales, prices, demand, profit and employment cannot be found. Binary probit model estimation results indicate that online participation is not significantly affected by sector affiliation, affiliation to East or West Germany nor information technology endowment. Firms with between 1 and 19 employees are, however, significantly less likely than larger firms to participate online. A significant correlation between online participation and unit?nonresponse does not exist, as a bivariate probit model with sample selection estimation results show. In general, item?nonresponse tends to be higher for online?respondents.