Another approach is to consider the residual effects attributable to individual pre-school centres
after controlling for all measured child, family, home and contextual factors. These residuals
provide measures of effectiveness and can be regarded as the outcome of pre-school quality
differences. These quality differences are in turn the result of a range of differences between
pre-school centres. The advantage of this approach is that the residual centre effects can be
regarded as a proxy for the cumulative effect of all quality differences.
For pre-reading the range of centre residuals effects is 10.43 points on the pre-reading scale.
This is a difference between the best and the worst centre in a distribution that is approximately
normal. Let’s take the centre that is one standard deviation below the mean as an averagely bad
centre in terms of quality, and the centre that is one standard deviation above the mean as an
averagely good centre in terms of quality. The difference between these is 2 SD units, which is
4.2. This could be regarded as a measure of the effect size of quality (average bad compared
with average good). It is a relatively conservative estimate as it is considerably smaller than
comparing the very worst with the very best, and the levels of quality compared are ones that are
frequently present in the population of pre-school centres. This effect size for quality of 4.2 pre-
reading units is equivalent to 4.15 months of development.
Similar computations for language give an effect size for quality of 2.48 months of development,
and for early number concepts an effect size for quality of 3.36 months of development.
Finally, the relative effects of increasing family income versus increasing the time the child
attends pre-school are explored. This question is best considered where the duration of pre-
school is considered as months of attendance rather than sessions per week as the EPPE
results clearly show an effect for number of months while 5 sessions are not significantly different
in their impact from 10 sessions a week. The various calculations on income and duration
effects show that for children in low income families, one extra year of pre-school (i.e. 3 rather
than 2 years) is associated with a similar effect upon pre-reading as increasing family income by
£10,000 p.a. The costs table overleaf show that any form of pre-school provision costs a lot less
than £10,000, particularly if only 5 sessions a week are provided.
Conclusion
The calculation of effect sizes for specific sub groups of children allows comparison with the
effects attributable to other child, family or home environment characteristics. Of particular policy
relevance for this paper are the comparisons with the size of family income effects and those of
different durations of pre-school. In addition the analyses reported here have extended the study
of the impact of different levels of quality and duration of pre-school.
82