Notes to Pages 179–189Chapter 13. Family Conflict
- Epigraph quoted in Layard and Dunn (2009).
- Office for National Statistics. http://www.ons.gov.uk/peoplepop
ulationandcommunity/birthsdeathsandmarriages/divorce/datasets/di
vorcesinenglandandwaleschildrenofdivorcedcouples.
Note that in the ALSPAC sample almost all the children lived ini-
tially with both biological parents. This is no longer the case, but in
our sample we are unable to study the impact of being born to a single
mother. - Duncan and Hoffman (1985); Weitzman (1985).
- See Amato and Keith (1991) and Amato, Loomis, and Booth
(1995) for the United States. Cherlin et al. (1991) also look at the UK. - See for example, Antecol and Bedard (2007); Bratberg, Elseth
Rieck, and Vaage (2014); Cooper et al. (2011); Ermisch, Francesconi,
and Pevalin (2004); Fronstin, Greenberg, and Robins (2001); Kiernan
(1997); and Prevoo and ter Weel (2015). - On this issue, see also Amato, Loomis, and Booth (1995); Han-
son (1999); and Jekielek (1998). We also reran Table 13.5 omitting as
controls both mother’s and father’s mental health— since these could
be affected by conflict and/or separation and thus act as mediating vari-
ables. When this was done the coefficients in the bottom line rose to
0.08 (.07) 0.11 (.08) 0.05 (04)
This interaction affect was thus larger but still not significant,
given the sample size.- NFPI (2000).
Chapter 14. Schooling
- Hanushek (1970).
- One could do a value- added calculation for each child charac-
teristic— by subtracting the score at 11 from the score at 16. But we
prefer to estimate a freely determined coefficient, rather than imposing
a coefficient of unity on the lagged dependent variable. - This estimate is an underestimate due to measurement error in
emotional health but an overestimate due to the omission of variables,
such as neighborhood effects. - (i) On class size and academic performance, using Tennessee’s
Project STAR, Hanushek (1999) found no effect of class size while