Skip to main content
. Author manuscript; available in PMC: 2019 Nov 12.
Published in final edited form as: BMC Psychiatry. 2019 Nov 8;19(1):348. doi: 10.1186/s12888-019-2352-8

Table 1. Results of Latent Class Analysis of Internet Addiction Test and Problematic Use of the Internet Subtypes in the Stellenbosch Sample.

Classes Log Likelihood BIC Entropy LMR p
IAT Total
1 -32299.8821 65192.978
2 -29177.0041 59547.851 .906 6235.245 <.001
3 -28257.5981 58309.669 .901 1835.959 .746
PUI Subtypes
1 -7750.3982 15887.233
2 -7376.1332 15531.478 .830 746.628 <.001
3 -7236.3283 15644.641 .761 278.900 .759

Note. LMR = Lo-Mendell-Rubin adjusted Likelihood Ratio Test when comparing the k to k – 1 class model; p = probability value for the Lo-Mendell-Rubin (LMR) adjusted Likelihood Ratio Test (LRT). IAT Total N = 1661. PUI Subtypes N = 564.

1

Best loglikelihood values initially obtained using 160, 32 and then replicated using 320, 64 random starting value perturbations.

2

Best loglikelihood values initially obtained using 80, 16 and then replicated using 160, 32 random starting value perturbations.

3

Best loglikelihood value initially required 1280, 256 random starting value perturbations and then replication using 2560, 512 random starts.