On polling and biases

So I'm reading "Thinking, Fast and Slow" by Daniel Kahneman, and this section really got to me, especially when I considered it in relation to polling (btw, this is paraphrased):

A survey was given to a group of German students. It included the following two questions:

How happy are you these days?
How many dates did you have last month?

The correlation between the answers was about zero.

Another group of students was given the questions, but in reverse order:

How many dates did you have last month?
How happy are you these days?

The results this time were completely different. In this sequence, the correlation between the number of dates and reported happiness was about as high as correlations between psychological measures can get.


So what does this say about our polling? In my mind, it says that it's more of an influencer than an actual measure of beliefs. Our minds are so easily primed without us noticing.

I have never actually been polled, but I would imagine that a pro-GOP polling outfit could prime the pump by starting off with questions like "How concerned are you about terrorism?" before asking about approval/disapproval of the president, while left-leaning ones would start with "How concerned are you about global warming?".