Re: Is permutations t-test in UCINET affected by the size of the categories
|
Well, it is not an artifact of group size. The t-test is about means, so it doesn¡¯t matter whether one group is larger than the other ¨C the smaller group could have the larger mean. BUT ¡ there could be sociological confounds. Suppose people are homophilous, so that positive people send ties to positive people, and negative to negative. Then of course the large positive group will be more central than the small negative group on most centrality measures. ? steve ?
toggle quoted message
Show quoted text
From: ucinet@...
Sent: Tuesday, April 30, 2019 10:28
To: ucinet@...
Subject: [UCINET] Is permutations t-test in UCINET affected by the size of the categories? ? I have a network where 75% of nodes have a certain attribute (let's say positive) and 25% of nodes are negative. If I run permutations t-tests in UCINET to test the difference in the means of various centrality measures (ex: closeness centrality). Could the results of the t-tests be an artifact because of the difference in size of the two categories. In other words, could positive nodes statistically have higher mean closeness centrality because they are more prevalent than the negative nodes? Postdoctoral Research Fellow
|
