QAP Quiz on Network Analysis

In QAPs and CUGs, why do we need a large sample of permuted networks to compare our observed network to?

• to adequately represent the random null-distribution

• to get a lower p-value

• to get higher regression coefficients

• to compute the Standard Error of the regression coefficients

The intercept estimate of -1.860 represents

• the baseline probability of a tie value to be 1

• the baseline log-odds of a tie if all other independent variables are 0

• the baseline probability of a tie value to be 0

• the baseline log-odds of a tie if all other independent variables are 1

We find a "seniority sender" coefficient of -0.045 with p = .009. How can this result be interpreted?

• We find no significant evidence for seniority of the sender (assuming significance level = 5%)

• Lawyers higher in seniority are significantly less likely to ask others for advice than lawyers with lower seniority (assuming alpha = 5%)

• Among lawyers asking for advice, 4.5% of the lawyers have a high level of seniority

• Lawyers higher in seniority are significantly less likely to be asked for advice than lawyers with lower seniority (assuming alpha= 5%)

Do we find support for hypothesis ``senior lawyers are less likely to receive advice'' at significance level = 5%?

• No, the coefficient estimate is greater than 0

• Yes, because the odds estimate is greater than 1

• Yes, because the p-value is less than 0.05

• No, because the p-value is greater than 0.05

We find a "same school" coefficient of 0.29 with p = .060. How can this result be interpreted?

• We find no significant evidence (at alpha = 0.05) for school homophily

• Among lawyers asking for advice, 29% of the lawyers went to the same school

• Those who went to the same school are significantly more likely to ask each other for advice

• lawyers have a 6% probability of having an advice tie if they went to the same school

Do we find support for hypothesis ``lawyers are more likely to ask lawyers with whom they worked with for advice'' at significance level = 5%?

• No, because the coefficient estimate is less than 0

• Yes, because the odds estimate is greater than 1

• Yes, because the p-value is less than 0.05

• No, because the p-value is greater than 0.05

What are the log-odds of a lawyer with level 1 seniority sending a tie to a lawyer with level 2 seniority from the same office (but not from the same school or with a work tie)?

• -1.860 - 0.045 + 1*0.036 + 1.062 = - 0.807

• -1.860 - 0.045 + 1*0.036 + 1.062 + 0.294 + 2.478 = 1.965

• - 0.045 + 2*0.036 + 1.062 = 1.089

• -1.860 - 0.045 + 2*0.036 + 1.062 =  - 0.771