Table 1 Results of the generalized linear mixed model (GLMM).

Fixed effect	Estimate (β)	SE	Z	p-value	Signif.
ChatGPT-4o (baseline)	− 0.265	0.117	− 2.267	0.023	*
DeepSeek-R1 vs ChatGPT-4o	− 1.829	0.155	− 11.769	< 0.001	***
Year 2018	0.037	0.137	0.268	0.788
Year 2019	0.350	0.139	− 3.173	0.002	**
Year 2020	0.333	0.138	− 2.419	0.016	*
Year 2021	− 0.337	0.138	− 2.420	0.016	*
Unit 2	0.177	0.105	1.676	0.094
Unit 3	0.344	0.105	3.279	0.001	**
Unit 4	− 0.017	0.107	− 0.164	0.870
DeepSeek-R1 × Year 2018	0.330	0.205	1.612	0.107
DeepSeek-R1 × Year 2019	− 0.050	0.223	− 0.224	0.822
DeepSeek-R1 × Year 2020	− 0.625	0.240	− 2.602	0.009	**
DeepSeek-R1 × Year 2021	− 0.361	0.230	− 1.571	0.116

* p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed). In the GLMM model, Unit 1 was set as the reference category for subject unit and is not shown in the table.

Quick links

Search