. Author manuscript; available in PMC: 2015 Dec 17.

Published in final edited form as: J Alcohol Drug Depend. 2015 Oct 19;3(5):224. doi: 10.4172/2329-6488.1000224

Table 3.

Descriptive network statistics.

	Sunshine			Jefferson
Wave=	1	2	3	1	2	3
Baseline
Density	0.008	0.005	0.005	0.002	0.001	0.001
Average degree	5.82	4.11	3.47	3.59	2.13	1.75
Number of ties	3399	3059	2162	3551	3100	1927
Missing fraction	0.24	0.03	0.18	0.35	0.05	0.28
Moran’s I¹	0.28	0.22	0.15	0.19	0.23	0.11
Moran’s I¹@ distance=2	0.22	0.19	0.19	0.19	0.21	0.20
Number of off list nominations	0.79	1.69	1.60	1.62	2.44	1.84
Dyad counts
Mutual	635	603	433	593	505	287
Asymmetric	1498	1778	1084	1451	1991	1067
Null	168103	274015	192859	486522	1058200	606399
Jaccard distance²
Wave 1 ==> 2		0.27			0.21
Wave 2 ==> 3			0.26			0.22
Tie changes between observations	0 => 1	1 => 0	1 => 1	0 => 1	1 => 0	1 => 1
Wave 1 ==> 2	1196	2075	1234	1181	2443	982
Wave 2 ==> 3	1149	1621	949	1083	1588	756
Alcohol use changes	Increase	Decrease	No change	Increase	Decrease	No change
Wave 1 ==> 2	174	124	305	249	242	669
Wave 2 ==> 3	157	172	301	251	245	662
Drunkenness freq. changes
Wave 1 ==> 2	112	142	347	153	184	812
Wave 2 ==> 3	210	163	304	235	213	882

Moran’s I is a measure of network-attribute autocorrelation (−1 to 1; Moran 1950).

The fraction of stable nominations among new, lost, and stable ones during the period [52].