Seventy four boys with reading disabilities and a matched control group of
normal readers were followed during their compulsory schooling (to grade 9)
. Word decoding performance was assessed with the Wordchains test on three
occasions. For each student a linear growth function was estimated. The slo
pe parameter was used as the dependent variable in a multiple regression an
alysis with a large number of explanatory factors from grades 2 and 3 inclu
ding phonological skills, reading and spelling, non-verbal intelligence, mo
tor skills, eye movements, teacher ratings of school behaviour and parents'
education. The proportion of variance in growth explained for the reading
disabled boys amounted to 25%, the most powerful predictors being early rea
ding performance and non-verbal intelligence. More surprisingly, neither te
acher ratings nor family background gave useful predictive information. For
the control group none of the independent variables explained the variance
of growth in reading.