Researcher: People who participate in opinion surveys often give answers they believe the opinion surveyor expects to hear, and it is for this reason that some opinion surveys do not reflect the actual views of those being surveyed.
Conclusion (So)
If a survey is well constructed, survey respondents' desire to meet surveyors' expectations has no effect on the survey's results.
Evidence
Even though people often try to give answers that the surveyor expects to hear, in well-constructed surveys the questions are worded so as to provide respondents with no indication of which answers the surveyor might expect.
Evaluate
This is a pretty plausible argument so we might not have any organic reactions to it. Hence, we switch into Anti-Conclusion debate mode:
Given that ... the questions are worded so as to provide respondents with no indication of which answers the surveyor might expect,
How could it still be that ... respondents' desire to please the surveyor's expectations could still have some effect?
Maybe the respondents just come to their own conclusions about what answer they think would please the surveyor, and thus answer that way.
So even though the questions weren't designed to indicate a preferred answer, just like a Rorschach test, we can all see in it what we want to.
Goal
Look for an answer that addresses the possibility that even though the questions don't signal some preferred answer, the respondents desire to meet surveyors' expectations could still tilt the data somehow.