CAPÍTULO I. INTRODUCCIÓN
1.2 MICROSOFT SILVERLIGHT
1.2.2 ARQUITECTURA
The MCS oers a broad set of indicators of child development at dierent stages of life (for more details see Hansen,2012 and Johnson, 2012).
I measure cognitive outcomes of children aged ve with several subtests of the British Ability Scales (BAS). These comprise the Naming Vocabulary (NV), the Picture Similar- ity (PS) and the Pattern Construction (PC) subtests. The rst subtest assesses children's expressive language skills and general language development. During the test a child is show pictures and they have to name the dierent objects in the pictures. The Picture Similarity test measures a child's ability to solve problems. The task is to assign a card to the most similar card out of four other cards. The third subtest measures a broad set of skills like spatial awareness, dexterity and coordination. In the test, the child has to assemble at squares or solid cubes according to a specied design.
Cognitive abilities in seven year olds are assessed by two BAS subtests, Pattern Construc- tion and Word Reading (WR), and the National Foundation for Educational Research (NFER) Number Skills test. The Word Reading subtests assesses the English reading ability of a child. The task is to read aloud a series of words in an increasing order of diculty. The NFER Number Skills test is based on the NFER Progress in Maths test and is designed to assess child's numeracy skills. In particular, it measures a child's basic ability to solve mathematical tasks.
Noncognitive abilities are measured by the Child Self Regulation Child Social Behaviour Questionnaire (CSR) and the Strength and Diculties Questionnaire (SDQ) for both age groups (for more details see Johnson, 2012). They provide assessments of dierent subdomains of noncognitive development. The subdomains covered by the CSR are (i) independence and self regulation, (ii) emotional dysregulation and (iii) cooperation. The latter domain is only partially covered for ve year old children, as only two of ve items have been collected. The subdomains covered by SDQ are (i) emotional symptoms, (ii) conduct problems, (iii) hyperactivity/ inattention, (iv) peer problems and (v) prosocial behaviour. In accordance with recent economic research, I treat hyperactivity as a mental disorder and therefore belonging to health capital rather than noncognitive abilities (see,
for instance, Currie and Stabile, 2006 and Ding et al.,2009).
Table 3.1: Development of children aged ve to seven
Age 5 Age 7
mean sd min max mean sd min max Lone parent Cognitive BAS NV 107.34 13.80 10 170 BAS PS 81.80 12.33 10 119 BAS PC 86.41 19.44 10 143 114.80 15.70 10 177 BAS WR 100.95 31.18 10 191 NFER Math 9.38 2.87 0 15 Noncognitive SDQ ES 1.50 1.57 0 7 1.73 1.88 0 9 SDQ CP 1.86 1.61 0 8 1.70 1.68 0 8 SDQ PR 1.39 1.51 0 9 1.45 1.63 0 10 SDQ PB 8.38 1.69 1 10 8.50 1.63 0 10 CSR IND 2.52 0.35 1.20 3 2.50 0.36 1.20 3 CSR ED 1.43 0.52 1 3 1.79 0.47 1 3 CSR COO 1.56 0.45 0 2 2.56 0.33 1 3 N 1422 1568 Two parents Cognitive BAS NV 111.06 14.36 10 170 BAS PS 83.11 11.06 10 119 BAS PC 89.92 18.16 10 149 118.38 15.99 10 173 BAS WR 110.00 29.44 10 222 NFER Math 9.98 2.67 0 15 Noncognitive SDQ ES 1.21 1.45 0 10 1.36 1.62 0 10 SDQ CP 1.33 1.38 0 10 1.19 1.40 0 10 SDQ PR 0.97 1.31 0 10 1.01 1.42 0 10 SDQ PB 8.47 1.58 0 10 8.70 1.54 1 10 CSR IND 2.54 0.33 1 3 2.53 0.36 1.20 3 CSR ED 1.30 0.47 1 3 1.66 0.46 1 3 CSR COO 1.60 0.42 0 2 2.63 0.32 1.40 3 N 7322 7047
Note: Own calculations based on MCS. BAS NV: BAS Naming Vocabulary Test; BAS PS: BAS Picture Similarity Test; BAS PC: BAS Pattern Construction Test; BAS WR: BAS Word Reading Test; NFER Math: NFER Progress in Math Test; SDQ ES: SDQ emotional symptoms subdomain score; SDQ CP: SDQ conduct problems subdomain score; SDQ PR: SDQ peer relationship problems subdomain score; SDQ PB: SDQ prosocial behaviour subdomain score; CSR IND: CSR independence/ self-regulation; CSR EMO: CSR emotional disregulation; CSR COO: CSR cooperation.
Table (3.1)1displays the summary statistics of the development indicators by age and type
of household. Children in lone parent households seem to perform worse in cognitive tests than children in two parent households. In addition to that, they exhibit worse values with respect to assessments of noncognitive abilities based on the SDQ, with the exception of prosocial behaviour. However, the observed dierences are likely to be insignicant given the standard deviations.
Table 3.2: Lagged health of children aged ve to seven
Age 5 Age 7
mean sd min max mean sd min max Lone parents # chronic 0.21 0.49 0 4 0.24 0.54 0 4 # hospital 0.34 0.93 0 10 0.17 0.61 0 10 SDQ HA 4.24 2.40 0 10 3.60 2.42 0 10 N 1422 1568 Two parents # chronic 0.17 0.43 0 4 0.23 0.53 0 5 # hospital 0.25 0.73 0 10 0.16 0.57 0 10 SDQ HA 3.65 2.27 0 10 3.04 2.29 0 10 N 7322 7047
Note: Own calculations based on MCS. # chronic: number of long- standing health conditions; # hospital: number of hospital admis- sions; SDQ HA: SDQ hyperactivity/ inattention subdomain score.
The latent variable child health is assessed by two manifest indicators, namely, the number of longstanding health conditions and the number of hospital admissions. I focus on these rather objective measures for two reasons: rstly, subjective measures have often been subject to criticism (for example Johnston et al., 2010) due to the possibility of reporting bias, and secondly, these child health indicators facilitate intertemporal comparisons of the impact of health on developmental outcomes as subjective parental assessments are not available for children under ve years. In addition to that, I also consider children's mental health. As noted above, I prefer the hyperactivity/ inattention subdomain of the SDQ rather than diagnosed cases, since diagnoses may be aected by other variables than the specic mental condition. It will be of special interest to compare the impact
1The descriptive tables within this section show the results of a balanced sample resulting after listwise
deletion which ignores missing values on hyperactivity risk factors. These are only relevant for the analysis of the inuence of mental health. Listwise deletion considering those variables would unnecessarily reduce sample size for the analysis of the eect physical health has on developmental outcomes.
of physical health to that of mental health, as recent studies nd that mental conditions exhibit a greater inuence than physical health (Currie and Stabile,2006 and Ding et al.,
2009). Table (3.2) provides basic descriptive statistics for the three health measures that are considered in the following analysis. Again, children in lone parent households seem to have worse outcomes than children in two parent families. This holds true for physical as well as mental health. However, dierences with respect to physical health diminish as children get older, while children in two parent households still seem to have lower hyperactivity or inattention scores on average. Comparing means and standard deviations to each other indicates overdispersion concerning the number of hospital admissions.