Lexical and sublexical reading predictors

Poor readers

Poor readers’ performance

ID	wr_time_z	wr_error_z	nwr_time_z	nwr_error_z	pa.bis	forward	backward	lexita.acc	it.ok.acc
LAE23	-2.88	-0.13	-3.36	-0.53	0.68	-0.10	-1.06	0.95	0.09
LAE38	-2.75	-0.26	-2.73	-1.70	0.52	0.52	-0.49	0.59	-1.17
LC19	-4.97	-0.65	-4.58	0.06	-1.48	-0.49	-1.10	-3.12	-2.29
MEN02	-2.96	-1.04	-2.10	-2.41	0.16	-0.49	-0.49	-0.76	-0.35
MEN10	-3.40	-0.69	-2.68	-0.53	0.12	-1.02	0.08	-0.20	-0.32
MEN29	-2.24	-2.36	-0.52	-4.55	-0.26	-0.10	-0.49	0.37	-0.72
MEN35	-0.12	-2.36	0.94	-2.54	-0.04	-0.10	0.08	-0.20	1.24

Compensated DYS

Compensated DYSs’ performance

ID	wr_time_z	wr_error_z	nwr_time_z	nwr_error_z	pa.bis	forward	backward	lexita.acc	it.ok.acc
LAUVEN	-0.58	-1.23	-1.05	-1.45	-0.71	0.71	0.71	-0.71	-0.71
LAE04	-4.58	1.04	-0.99	-0.43	0.79	-0.39	-0.07	-0.61	-0.79
LAE17	-1.15	-1.04	-1.33	-0.65	0.44	-1.50	0.74	0.26	-0.50
LC03	-2.75	-1.82	-0.44	-1.70	-1.29	-0.49	0.13	-0.08	-0.20
LC15	-3.00	-0.72	-1.47	-1.23	-0.99	0.45	-0.07	0.41	0.20
MEN22	-4.16	-0.69	-1.16	-1.34	-0.65	-1.02	-0.49	1.24	-0.45
MEN33	-3.33	0.43	-1.21	0.67	0.31	0.82	0.66	0.09	-0.32

Exclude participants

df %>% 
  filter(group == "DYS" | (group == "TD" & is.na(other_diagnoses))) %>% 
  mutate(ID = as.character(ID)) %>% 
  filter(!startsWith(ID, "VER")) -> df 

df %>%
  mutate(
    exclude = case_when(
      ID == "VER02" ~ 1, # ADHD - university - DYS
      ID == "VER04" ~ 1, # ADHD - university - DYS
      ID == "MEN12" ~ 1, # discalculia - 3rd - TD
      ID == "MEN18" ~ 1, # disgrafia, discalculia - 3rd - TD
      ID == "MEN26" ~ 1, # disgrafia - 5th - TD, 
      ID == "LAE33" ~ 1, # disgrafia, disortografia, 3rd - TD, 
      
      ### new exclusion criteria: 
      ID == "LAE38" ~ 1,
      ID == "LC19" ~ 1, 
      ID == "MEN29" ~ 1, 
      ID == "MEN02" ~ 1, 
      
      
      ID == "MEN10" ~ 1, 
      ID == "LAE23" ~ 1, 
      # ID == "MEN35" ~ 1, 
      TRUE ~ 0
    )
  ) %>%
  filter(exclude == 0) %>%
  
  # stricter crtiteria 
  # filter(group.exclusion != "PR") %>% 
  
  # less strict criteria 
  # filter(group == "DYS" | (group == "TD" & reading.score > -2)) %>%
  
  
  filter(ID != "VER01" & ID != "VER03") %>% 
  dplyr::select(-exclude) -> df

Descriptive statistics

Reading Performance

df %>% 
  lm(formula = wr_time_z ~ group) -> m1.reading
df %>% 
  lm(formula = wr_error_z ~ group) -> m2.reading
df %>% 
  lm(formula = nwr_time_z ~ group) -> m3.reading
df %>% 
  lm(formula = nwr_error_z ~ group) -> m4.reading

	Estimate	SE	t	p
Word reading fluency
(Intercept)	-4.15	0.30	-13.63	< .001
groupTD	3.74	0.38	9.79	< .001
Word reading accuracy
(Intercept)1	-2.75	0.31	-8.89	< .001
groupTD1	2.84	0.39	7.34	< .001
Pseudoword reading fluency
(Intercept)2	-3.03	0.25	-12.28	< .001
groupTD2	2.96	0.31	9.58	< .001
Pseudoword reading accuracy
(Intercept)3	-3.04	0.38	-7.90	< .001
groupTD3	2.90	0.48	6.02	< .001

group	M	sd	range
Pseudoword reading accuracy
DYS	-3.04	3.05	-11.78 - 0.75
TD	-0.14	1.21	-4.15 - 1.29
Pseudoword reading fluency
DYS	-3.03	1.93	-8.35 - -0.44
TD	-0.07	0.81	-1.81 - 1.8
Word reading accuracy
DYS	-2.75	2.58	-9.07 - 1.04
TD	0.09	0.78	-2.36 - 1.18
Word reading fluency
DYS	-4.15	2.49	-14.12 - -0.58
TD	-0.42	0.85	-2.58 - 1.65

Summary.

group	measure	M	SD	range
Spoonerism accuracy
DYS	Accuracy	30.41	7.28	10-40
TD	Accuracy	36.98	3.09	26-40
DYS	BIS	-1.44	2.12	-6.47-1.68
TD	BIS	0.99	0.91	-2.12-2.13
DYS	RT	12.17	5.60	3.92-24.27
TD	RT	5.63	2.46	1.72-11.61
Italian orthographic knowledge
DYS	Total score	29.55	10.18	12-50
TD	Total score	46.35	9.36	25-59
DYS	False alarms	3.66	3.31	0-11
TD	False alarms	1.71	2.18	0-12
DYS	Correct responses	33.21	9.73	14-51
TD	Correct responses	48.06	9.47	25-60
Verbal short-term memory
DYS		5.97	0.91	4-8
TD		6.63	1.11	4-9
Visual attention span
DYS		0.66	0.11	0.38-0.9
TD		0.78	0.11	0.62-1
Vocabulary
DYS		52.69	8.05	17-60
TD		57.06	2.01	48-60
Verbal working memory
DYS		4.24	1.90	0-8
TD		5.39	1.40	3-8

Correlations

	WRT	WRE	NWRT	NWRE	PA	STM	WM	VOC	IOK	VAS
PA	0.67	0.74	0.59	0.69	1.00	0.40	0.44	0.33	0.48	0.55
STM	0.30	0.37	0.31	0.34	0.40	1.00	0.29	0.02	0.24	0.33
WM	0.26	0.26	0.29	0.31	0.44	0.29	1.00	0.14	0.26	0.41
VOC	0.32	0.50	0.37	0.40	0.33	0.02	0.14	1.00	0.32	0.26
IOK	0.70	0.52	0.68	0.40	0.48	0.24	0.26	0.32	1.00	0.49
VAS	0.45	0.53	0.48	0.41	0.55	0.33	0.41	0.26	0.49	1.00

Inferential analysis - tasks

Spoonerism

RT

	Sum Sq	Df	F value	Pr(>F)
(Intercept)	2351.246	1	10782.690	0.000
age	0.172	1	0.790	0.377
group	10.288	1	47.179	0.000
Residuals	16.790	77	NA	NA

term	estimate	std.error	statistic	p.value
(Intercept)	9.315	0.090	103.840	0.000
age	-0.049	0.055	-0.889	0.377
groupTD	-0.786	0.114	-6.869	0.000

Accuracy

	LR Chisq	Df	Pr(>Chisq)
age	8.483	1	0.004
group	30.674	1	0.000
age:group	6.987	1	0.008

term	estimate	std.error	statistic	p.value
(Intercept)	3.360	0.039	85.162	0.000
age	0.121	0.042	2.907	0.004
groupTD	0.249	0.046	5.421	0.000
age:groupTD	-0.125	0.047	-2.639	0.008

group	age.trend	SE	df	asymp.LCL	asymp.UCL	z.ratio	p.value
DYS	0.121	0.042	Inf	0.039	0.202	2.907	0.004
TD	-0.004	0.023	Inf	-0.049	0.040	-0.190	0.849

BIS

	Sum Sq	Df	F value	Pr(>F)
(Intercept)	79.761	1	41.129	0.000
age	19.832	1	10.226	0.002
group	126.177	1	65.062	0.000
age:group	15.181	1	7.828	0.007
Residuals	147.388	76	NA	NA

term	estimate	std.error	statistic	p.value
(Intercept)	-1.861	0.290	-6.413	0.000
age	1.015	0.317	3.198	0.002
groupTD	2.844	0.353	8.066	0.000
age:groupTD	-1.040	0.372	-2.798	0.007

group	age.trend	SE	df	lower.CL	upper.CL	t.ratio	p.value
DYS	1.015	0.317	76	0.383	1.647	3.198	0.002
TD	-0.025	0.193	76	-0.410	0.360	-0.128	0.899

Plot

Partial correlations

STM and WM

STM

	LR Chisq	Df	Pr(>Chisq)
age	0.000	1	0.987
group	1.144	1	0.285

term	estimate	std.error	statistic	p.value
(Intercept)	1.786	0.078	22.768	0.000
age	-0.001	0.047	-0.017	0.987
groupTD	0.105	0.098	1.065	0.287

WM

	LR Chisq	Df	Pr(>Chisq)
age	1.537	1	0.215
group	6.333	1	0.012

term	estimate	std.error	statistic	p.value
(Intercept)	1.417	0.093	15.188	0.000
age	0.065	0.052	1.244	0.214
groupTD	0.281	0.113	2.485	0.013

Plot

Partial correlations

STM

WM

IOK

Total Score

	LR Chisq	Df	Pr(>Chisq)
age	17.834	1	0
group	153.473	1	0

term	estimate	std.error	statistic	p.value
(Intercept)	3.352	0.035	95.270	0
age	0.077	0.018	4.239	0
groupTD	0.499	0.041	12.044	0

Correct Answers

	LR Chisq	Df	Pr(>Chisq)
age	0.249	1	0.618
group	89.263	1	0.000
age:group	5.212	1	0.022

term	estimate	std.error	statistic	p.value
(Intercept)	3.511	0.036	97.704	0.000
age	-0.020	0.040	-0.499	0.618
groupTD	0.377	0.041	9.131	0.000
age:groupTD	0.101	0.044	2.285	0.022

group	age.trend	SE	df	asymp.LCL	asymp.UCL	z.ratio	p.value
DYS	-0.020	0.04	Inf	-0.097	0.058	-0.499	0.618
TD	0.081	0.02	Inf	0.043	0.120	4.121	0.000

False Alarms

	LR Chisq	Df	Pr(>Chisq)
age	19.861	1	0.000
group	36.845	1	0.000
age:group	10.615	1	0.001

term	estimate	std.error	statistic	p.value
(Intercept)	1.423	0.097	14.608	0.000
age	-0.533	0.121	-4.414	0.000
groupTD	-0.891	0.147	-6.052	0.000
age:groupTD	0.527	0.161	3.272	0.001

group	age.trend	SE	df	asymp.LCL	asymp.UCL	z.ratio	p.value
DYS	-0.533	0.121	Inf	-0.770	-0.296	-4.414	0.000
TD	-0.006	0.106	Inf	-0.215	0.202	-0.061	0.952

Plot

Partial correlations

Voc.

Accuracy

	LR Chisq	Df	Pr(>Chisq)
age	5.653	1	0.017
group	10.489	1	0.001
age:group	4.180	1	0.041

term	estimate	std.error	statistic	p.value
(Intercept)	3.932	0.029	133.394	0.000
age	0.075	0.031	2.375	0.018
groupTD	0.113	0.035	3.211	0.001
age:groupTD	-0.074	0.036	-2.043	0.041

group	age.trend	SE	df	asymp.LCL	asymp.UCL	z.ratio	p.value
DYS	0.075	0.031	Inf	0.013	0.137	2.375	0.018
TD	0.000	0.018	Inf	-0.036	0.036	0.018	0.986

RT

	Sum Sq	Df	F value	Pr(>F)
(Intercept)	1584.187	1	21534.856	0.000
age	0.020	1	0.271	0.604
group	3.054	1	41.509	0.000
Residuals	5.664	77	NA	NA

term	estimate	std.error	statistic	p.value
(Intercept)	7.646	0.052	146.748	0.000
age	-0.017	0.032	-0.520	0.604
groupTD	-0.428	0.066	-6.443	0.000

Plot

VAS

	Sum Sq	Df	F value	Pr(>F)
(Intercept)	11.589	1	971.198	0.000
age	0.012	1	0.982	0.325
group	0.288	1	24.095	0.000
Residuals	0.919	77	NA	NA

term	estimate	std.error	statistic	p.value
(Intercept)	0.654	0.021	31.164	0.000
age	0.013	0.013	0.991	0.325
groupTD	0.131	0.027	4.909	0.000

#### Partial correlations

Summary Inf 1

	Sum Sq	F value	Pr(>F)
Spoonerism (RT)
(Intercept)	2351.246	10782.690	0.000
age	0.172	0.790	0.377
group	10.288	47.179	0.000
Spoonerism (BIS)
(Intercept)	79.761	41.129	0.000
age	19.832	10.226	0.002
group	126.177	65.062	0.000
age:group	15.181	7.828	0.007
VAS
(Intercept)	11.589	971.198	0.000
age	0.012	0.982	0.325
group	0.288	24.095	0.000

rowname	Chisq	Pr(>Chisq)
Spoonerism ACC
age	8.483	0.004
group	30.674	0.000
age:group	6.987	0.008
Verbal short-term memory
age	0.000	0.987
group	1.144	0.285
Verbal working memory
age	1.537	0.215
group	6.333	0.012
Italian orthographic knolwedge (total score)
age	17.834	0.000
group	153.473	0.000
Italian orthographic knolwedge (correct responses)
age	0.249	0.618
group	89.263	0.000
age:group	5.212	0.022
Italian orthographic knolwedge (false alarms)
age	19.861	0.000
group	36.845	0.000
age:group	10.615	0.001
Vocabulary
age	5.653	0.017
group	10.489	0.001
age:group	4.180	0.041

Analysis 1 (original approach)

Lexical reading

df %>% 
  dplyr::select(ID,group, 
                wr_time_z:nwr_error_z, 
                pa.bis, 
                forward, 
                backward, 
                it.ok.acc, 
                va.span.acc) %>% 
  
  pivot_longer(names_to = "measure", 
               values_to = "score",
               3:6) %>% 
  mutate(dimension = if_else(startsWith(measure, "wr"), "lexical", "sublex.")) %>% 
  mutate(measure = case_when(
    measure == "wr_time_z" | measure == "nwr_time_z" ~ "time",
    measure == "wr_error_z" | measure == "nwr_error_z" ~ "error"
  )) %>% 
  
  rowwise() %>%
  mutate(
    verbal.wm = mean(c_across(forward:backward))
  ) %>%
  ungroup() %>% 
  
  
  mutate(across(pa.bis:va.span.acc, ~scale(.x))) %>% 
  mutate(across(pa.bis:va.span.acc, ~as.numeric(.x))) %>% 
  mutate(measure = as.factor(measure),
         group = as.factor(group),
         verbal.wm = scale(verbal.wm), verbal.wm = as.numeric(verbal.wm)) %>% 
  
  filter(dimension == "lexical")  %>% 
  
  lmer(formula = score ~ 
       
       # group * it.ok.acc * measure + 
       measure * it.ok.acc +
       group * it.ok.acc +
       # it.ok.acc + 
       # group * va.span.acc * measure + 
       # group * va.span.acc +
       # measure * va.span.acc +
       # va.span.acc +  
       # group * pa.bis * measure +
       # group * pa.bis +
       # pa.bis * measure +
        pa.bis +
       # group * verbal.wm * measure +
       # group * verbal.wm +
       # verbal.wm * measure +
       # verbal.wm + 
         
       # group * measure + 
     
     (1|ID)
     
     ) -> lex.mod

	Chisq	Df	Pr(>Chisq)
(Intercept)	18.347	1	0.000
measure	22.901	1	0.000
it.ok.acc	10.253	1	0.001
group	9.438	1	0.002
pa.bis	22.699	1	0.000
measure:it.ok.acc	11.500	1	0.001
it.ok.acc:group	9.491	1	0.002

measure	it.ok.acc.trend	SE	df	lower.CL	upper.CL	t.ratio	p.value
error	0.371	0.188	113.491	-0.001	0.744	1.973	0.051
time	0.963	0.188	113.491	0.590	1.336	5.117	0.000

group	it.ok.acc.trend	SE	df	lower.CL	upper.CL	t.ratio	p.value
DYS	1.193	0.266	75	0.663	1.723	4.481	0.000
TD	0.142	0.207	75	-0.271	0.555	0.685	0.496

contrast	it.ok.acc	estimate	SE	df	t.ratio	p.value
DYS - TD	-1	-2.224	0.472	75	-4.708	0.000
DYS - TD	0	-1.173	0.382	75	-3.072	0.003
DYS - TD	1	-0.122	0.549	75	-0.223	0.824

Sublexical reading

df %>% 
  dplyr::select(ID,group, 
                wr_time_z:nwr_error_z, 
                pa.bis, 
                forward, 
                backward, 
                it.ok.acc, 
                va.span.acc) %>% 
  
  pivot_longer(names_to = "measure", 
               values_to = "score",
               3:6) %>% 
  mutate(dimension = if_else(startsWith(measure, "wr"), "lexical", "sublex.")) %>% 
  mutate(measure = case_when(
    measure == "wr_time_z" | measure == "nwr_time_z" ~ "time",
    measure == "wr_error_z" | measure == "nwr_error_z" ~ "error"
  )) %>% 
  
  rowwise() %>%
  mutate(
    verbal.wm = mean(c_across(forward:backward))
  ) %>%
  ungroup() %>%
  
  
  mutate(across(pa.bis:va.span.acc, ~scale(.x))) %>% 
  mutate(across(pa.bis:va.span.acc, ~as.numeric(.x))) %>% 
  mutate(measure = as.factor(measure),
         group = as.factor(group),
         verbal.wm = scale(verbal.wm), verbal.wm = as.numeric(verbal.wm)) %>% 
  
  filter(dimension == "sublex.")  %>% 
  
  lmer(formula = score ~ 
       
       # group * it.ok.acc * measure + 
       measure * it.ok.acc +
       # group * it.ok.acc +
       # it.ok.acc + 
       # group * va.span.acc * measure + 
       # group * va.span.acc +
       # measure * va.span.acc +
       # va.span.acc +  
       group * pa.bis * measure +
       # group * pa.bis +
       # pa.bis * measure +
       # pa.bis +
       # group * verbal.wm * measure +
       # group * verbal.wm +
       # verbal.wm * measure +
       # verbal.wm +
         
       # group * measure + 
     
     (1|ID)
     
     ) -> sublex.mod
# drop1(sublex.mod, test = "Chisq")

	Chisq	Df	Pr(>Chisq)
(Intercept)	21.851	1	0.000
measure	1.575	1	0.209
it.ok.acc	0.260	1	0.610
group	6.507	1	0.011
pa.bis	39.591	1	0.000
measure:it.ok.acc	12.064	1	0.001
group:pa.bis	3.341	1	0.068
measure:group	0.193	1	0.660
measure:pa.bis	29.643	1	0.000
measure:group:pa.bis	5.826	1	0.016

measure	it.ok.acc.trend	SE	df	lower.CL	upper.CL	t.ratio	p.value
error	-0.117	0.23	116.967	-0.573	0.338	-0.510	0.611
time	0.656	0.23	116.967	0.201	1.111	2.853	0.005

measure	group	pa.bis.trend	SE	df	lower.CL	upper.CL	t.ratio	p.value
error	DYS	1.627	0.259	116.967	1.115	2.139	6.292	0.000
time	DYS	0.264	0.259	116.967	-0.248	0.776	1.021	0.309
error	TD	0.682	0.441	116.967	-0.192	1.556	1.546	0.125
time	TD	0.527	0.441	116.967	-0.346	1.401	1.195	0.234

contrast	measure	pa.bis	estimate	SE	df	t.ratio	p.value
DYS - TD	error	-1	-2.327	0.836	116.967	-2.785	0.006
DYS - TD	time	-1	-1.349	0.836	116.967	-1.614	0.109
DYS - TD	error	0	-1.382	0.542	116.967	-2.551	0.012
DYS - TD	time	0	-1.612	0.542	116.967	-2.976	0.004
DYS - TD	error	1	-0.436	0.651	116.967	-0.671	0.504
DYS - TD	time	1	-1.875	0.651	116.967	-2.882	0.005

measure	it.ok.acc.trend	SE	df	lower.CL	upper.CL	t.ratio	p.value
error	-0.117	0.23	116.967	-0.573	0.338	-0.510	0.611
time	0.656	0.23	116.967	0.201	1.111	2.853	0.005

Summary Inf 2

	Chisq	Df	Pr(>Chisq)
Lexical reading predictors
(Intercept)	18.35	1	0.000
measure	22.90	1	0.000
it.ok.acc	10.25	1	0.001
group	9.44	1	0.002
pa.bis	22.70	1	0.000
measure:it.ok.acc	11.50	1	0.001
it.ok.acc:group	9.49	1	0.002
Sublexical reading predictors
(Intercept)1	21.85	1	0.000
measure1	1.58	1	0.209
it.ok.acc1	0.26	1	0.610
group1	6.51	1	0.011
pa.bis1	39.59	1	0.000
measure:it.ok.acc1	12.06	1	0.001
group:pa.bis	3.34	1	0.068
measure:group	0.19	1	0.660
measure:pa.bis	29.64	1	0.000
measure:group:pa.bis	5.83	1	0.016