Table 4.

Fixed effects from the generalized mixed-effects logistic regression model predicting autocorrect errors

Word Type	Term	Beta	Std error	χ2	p value
All target words collapsed	Intercept	−2.477	0.158	245.292	<0.001
	Paragraph Type	0.587	0.17	11.902	<0.001
	Part of Speech	1.039	0.19	30.012	<0.001
	Tau/Aβ42 ratio	0.601	0.243	6.12	0.013
	Reading Times	−0.074	0.092	0.637	0.425
	Interval between Biomarker & Autocorrect Task	−0.068	0.085	0.642	0.423
	Age	0.129	0.088	2.128	0.145
	Sex = Male	−0.364	0.173	4.438	0.035
	Education	−0.168	0.095	3.12	0.077
	APOE-4 status = Absent	−0.195	0.184	1.127	0.288
	Edit Distance	−1.047	0.107	95.072	<0.001
	Word Length	0.205	0.086	5.731	0.017
	Paragraph Type x Part of Speech	−0.443	0.318	1.943	0.163
	Paragraph Type x Tau/Aβ42 ratio	−0.145	0.297	0.239	0.625
	Part of Speech x Tau/Aβ42 ratio	0.176	0.303	0.337	0.562
	Paragraph Type x Part of Speech x Tau/Aβ42 ratio	0.237	0.517	0.21	0.647
Dominant Function Words	Intercept	−1.845	0.294	39.257	< .001
	Tau/Aβ42 ratio	0.673	0.241	7.82	0.005
	Reading Times	−0.181	0.129	1.974	0.160
	Interval between Biomarker & Autocorrect Task	−0.12	0.094	1.621	0.203
	Age	0.189	0.099	3.666	0.056
	Sex = Male	−0.382	0.195	3.826	0.050
	Education	−0.275	0.103	7.169	0.007
	APOE-4 status = Absent	−0.147	0.205	0.509	0.476
	Edit Distance	−1.407	0.232	36.733	< .001
	Word Length	0.524	0.166	9.937	0.002 a
Dominant Content words	Intercept	−2.579	0.323	63.929	< .001
	Tau/Aβ42 ratio	0.724	0.442	2.684	0.101
	Reading Times	0.018	0.154	0.014	0.905
	Interval between Biomarker & Autocorrect Task	0.016	0.139	0.013	0.911
	Age	−0.002	0.148	0	0.987
	Sex = Male	−0.313	0.276	1.29	0.256
	Education	0.135	0.146	0.856	0.355
	APOE-4 status = Absent	−0.275	0.295	0.874	0.350
	Edit Distance	−0.814	0.231	12.411	< .001
	Word Length	0.007	0.171	0.002	0.968
Nondominant Function Words	Intercept	−2.214	0.406	29.749	< .001
	Tau/Aβ42 ratio	0.52	0.352	2.187	0.139
	Reading Times	0.369	0.167	4.891	0.027
	Interval between Biomarker & Autocorrect Task	−0.179	0.124	2.082	0.149
	Age	0.215	0.136	2.485	0.115
	Sex = Male	−0.201	0.262	0.59	0.443
	Education	0.096	0.14	0.474	0.491
	APOE-4 status = Absent	−0.514	0.276	3.47	0.062
	Edit Distance	−1.021	0.228	20.092	< .001
	Word Length	0.16	0.284	0.318	0.573
Nondominant Content Words	Intercept	−3.376	0.351	92.736	< .001
	Tau/Aβ42 ratio	0.803	0.505	2.527	0.112
	Reading Times	0.414	0.215	3.697	0.055
	Interval between Biomarker & Autocorrect Task	−0.298	0.194	2.364	0.124
	Age	0.173	0.202	0.726	0.394
	Sex = Male	−0.185	0.411	0.203	0.652
	Education	−0.031	0.21	0.021	0.884
	APOE-4 status = Absent	−0.564	0.427	1.742	0.187
	Edit Distance	−1.122	0.287	15.308	< .001
	Word Length	0.132	0.206	0.415	0.520

^aThis effect was in the opposite direction as the overall length effect (shown in Figure 1c). However, after removing two long targets (though/through; always/around) the effect was no longer significant. We did not experimentally manipulate or control target word length or edit distance, as such these effects must be interpreted with caution.