difference between low quality high duration and high quality high duration. For language (1.01-
0.529), this gives an estimate of 0.481. For pre-reading the difference is somewhat smaller at
0.254 (0.622-0.368).

The original EPPE analyses modelled the quality measure as a continuous scale in the value
added analyses of children’s cognitive progress over the pre-school period (but the ‘home’ group
were not included in these models). The results also confirmed a separate significant effect for
quality (as measured by the ECERS-E instrument) over and above a larger effect for duration.

COSTS IN RELATION TO PRE-SCHOOL EFFECTS

Limitations of the cost data

There are a number of important caveats to the use of cost data in these analyses. The DfES
Sure Start unit has provided estimates of the cost of different types of provision. In addition to
the clearly stated problems in obtaining these costs it must be noted that only average costs of
the different sectors have been made available to us (previous research projects have pointed to
the difficulties in obtaining relevant figures). Hence it is impossible to separate the impact of cost
versus other measures of interest (such as quality or qualifications of staff) at centre level. We
have robust data concerning quality and staff qualification levels for individual centres and have
also calculated value added residuals estimates of the effectiveness of individual centres for a
range of cognitive and social/behavioural outcomes. However, type and cost are both
confounded in the data available. Therefore, estimates of the impact of centre type are used to
provide effect sizes of each form of provision.

In relation to cost we must interpret these effect sizes with considerable caution because types of
provision will differ both between type and within type in terms of quality, qualifications of staff
and overall costs. Two sets of analyses are presented to explore effect sizes associated with
each type of provider.

a) Value added - where the effect on relative progress over the pre-school period is
calculated for each type of provision just for the pre-school sample. Here the local
authority day nurseries are used as the reference group (LA day nurseries have the
smallest effect sizes in the VA models).

b) Contextualised - where the effect on attainment (at entry to primary school) is
calculated for each type of provision versus none (i.e. the reference group is the
home sample, for which no pre-school centre costs has been incurred).

Table B11 presents the effect sizes associated with each type of provider based on value added
analyses (in these analyses the ‘home’ group are not included). The results are also expressed
in terms of months of development (in parenthesis).

Table B11 Value added model : Effect sizes by type of provider

79

<<   <   91   92   93   94   95   96   97   >   >>