¿ªÔÆÌåÓýThere is no single answer to that. In my own work, I create two versions of the data, one with everybody, and one with only respondents. I use them both, but for different things. For example, suppose I have measured ¡°who do you seek advice from?¡±. If I want to know how often people are sought after for advice, I use the big version of the data, because even if Dr. Big did not answer the survey, people did respond about him, so I want to know ¨C among those that responded ¨C how many people sought advice from Dr. Big and everyone else on the roster. ? For symmetric relations such as ¡°socialize with after work¡±, you could assume that if Dr. Big had answered the survey, he would have named exactly the people who named him. The missing values imputation procedure in UCINET would do that for you. It implements the methods of ?nidar?i? et al. ? ? steve ? From: ucinet@... <ucinet@...>
Sent: Tuesday, May 21, 2019 3:42 PM To: ucinet@... Subject: [UCINET] Actor that do not anwer the survey ? ? Hi there,? ? I applied a survey at an ICU with 201 actors (doctors, nurse techs, allied heath). ? In three months I got 133 answers out these 201.? ? To start the analyse, I must take away the name that did not answer the survey ou they should stay? ? I am follow the direction from this book - (Borgatti et al, 2013) - Analyzing social networks.? ? But I got this doubt, if I should left or not those names that did not answered the survey.? ? ? Thank you in advance! ? Vanessa?
|
