In line with this view, the empirical evidence shows that people do not talk with the same discussants all the time, but that, in general, they select discussion partners from those who are available (Bello & Rolfe, 2014; Huckfeldt & Sprague, 1995).
Selection of political discussants does not necessarily mean ending pre-existing relationships or befriending all Liberal Democrats that one meets; it can be as simple as choosing to sit at the opposite end of the table from politically conservative Aunt Edna at family gatherings.
H1: People are more likely to retain political discussants with whom they agree than with whom they disagree.
We can expect two possible outcomes: Disagreement could be noxious even for intimate relationships and could thus lead to a higher likelihood of discarding discussants (this outcome could be compatible with the Italian case study, which is characterized by strong pressure from the family environment).
Finally, respondents were asked to provide the initials of the first and last names of their main discussants to clearly identify them later.
The most straightforward method in the literature to operationalize network disagreement has been to ask respondents about the political preferences of a number of discussants (usually up to five) with whom they have engaged.
The discussant's initials are used to determine whether the discussants are the same in both waves.
Although it is clear that men and women often respond differently to toxicants, the discussants
agreed that predicting just how the sexes will respond, and when they will respond differently, has not proved simple because, Silverstone said, of the limited amount of data at this time.
By treating the question wording as a randomly applied experimental stimulus, we are able to address whether the "political" discussants are different from the "important matters" discussants and whether these differences have consequences for the effectiveness of political communication.
Interviewers then asked a battery of questions about each discussant in the sequential order in which the discussants were named.
This terminology focuses attention on the message sent by discussants and received by main respondents, which is the only context we are equipped to address with our research.
When the interviewer asked a respondent how she expected one of her discussants to vote in November, she was being asked to connect that person with a particular political preference.
Our measure is based on the mean time that is required for respondents to report their judgments regarding the political preferences of discussants in their residual networks.
This is a potential problem for our research because the order in which discussants are named may be substantively meaningful (Burt 1986; Huckfeldt et al.
Finally, the dummy variable does not have a statistically discernible effect, that is, response times do not differ between first-named discussants and others once we control for question sequence and mean network response time.