Question interpretation in web surveys may not only depend on the textual content
but also on visual design aspects. Research has shown that respondents seem to make
use of interpretative heuristics when answering questions potentially influencing
their answer behavior. In this study, we investigate the implications of the ‘middle
means typical’ (MMT) heuristic, which suggests that respondents perceive the middle
option of a scale as the most typical one. For this purpose, we use data from a survey
experiment embedded in the probability-based German Internet Panel (N = 4,679) varying
the inclusion of a non-substantive “Don’t know” option (with or without separation
from the substantive options) and scale polarity (unipolar or bipolar). The four questions
under investigation were adopted from the Big5 inventory dealing with agreeableness
and openness. The results suggest that the MMT heuristic has a minor impact on answer
behavior, as the separation of non-substantive options did not affect answer distributions
and response times (as a measure of response effort). However, scale polarity influenced
answer behavior and response times. Similar to what has been observed in previous
studies, unipolar scales elicited more middle answers and bipolar scales elicited
more positive answers. Bipolar scales also resulted in longer response times. Although
design violations against the MMT heuristic do not seem to impact answer behavior,
we still recommend exercising caution when designing scales with non-substantive options.
We also highlight the necessity of testing scales differing with respect to polarity.
