Combining survey data with administrative data strengthens researchers' analyses, but we usually need the consent of each respondent to do so. Several studies have shown that the highest consent rates are achieved by asking the linkage consent question at the beginning of the questionnaire. However, asking this question at the beginning may impact respondents' answers throughout the interview. We hypothesize that asking for consent could affect responses in one of two contradictory ways. First, respondents might fear that their answers will be checked and thus expend more effort to give correct data (better respondents hypothesis). Second, respondents might think that their responses are no longer important and thus expend less effort(worse respondents hypothesis). We test these two hypotheses using a survey that asked half of the respondents for linkage consent at the beginning of the survey, and the other half at the end. Since the consenters in the two groups are not equivalent, we use the entropy balance technique to simulate a random distribution over both groups and calculate the effect of asking the consent question at the beginning on data quality, operationalized as measurement error (responses compared to administrative data) and item nonresponse. Preliminary results show that the linkage consent question does not affect response behavior, in the web or in CATI, which is good news for researchers who want to ask the linkage consent question in the beginning of the questionnaire, and also for methodologists who use linked data to investigate measurement error.