We compare two alternative mixed mode survey designs with face-to-face data collection in terms of differences in estimates. Both mixed mode designs involve face-to-face, telephone and web interviewing. One design uses modes sequentially; the other offers respondents an explicit choice of mode. All three samples are probability samples of the Dutch population, selected from the same frame in the same way and administered the same survey instrument, namely the questionnaire of round 4 of the European Social Survey. We find differences and consequently urge caution in comparing estimates based on data collected using different (mixes of) modes.
measurement error mode effects nonresponse bias social desirability bias telephone interviewing web surveys