Notes to Pages 189–198Krueger (2003) found that the internal rate of return from reducing
class size from 22 to 15 students is around 6%. Using other data Hoxby
(2000) found no effect, while Angrist and Levy (1999) showed that re-
ducing class size induces a significant and substantial increase in test
scores for fourth and fifth graders (although not for third graders)
using the effects of Maimonides’ Rule.
(ii) On class size and noncognitive development, Fredriksson, Ockert,
and Oosterbeek (2013) and Dee and West (2011) discover some bene-
ficial effects of smaller classes in primary schools (often rural), which
persist over time, while Jakobsson, Persson, and Svensson (2013) found
none using data on mental health problems and well- being among ad-
olescents in Swedish schools.
- Rockoff (2004), Aaronson, Barrow, and Sander (2007), Rivkin,
Hanushek, and Kain (2005), Kane and Staiger (2008), Chetty, Friedman,
and Rockoff (2014). - Flèche (2017).
- See also Jackson (2012) and Araujo et al. (2016) for teacher ef-
fects on non- test- score outcomes. - In online Table A14.1 we replace the teacher’s name by the
measured characteristics and teaching practices of the teacher. These
have almost no explanatory power, except in the case of academic per-
formance, which is shown to respond to teachers’ emotional health,
self- esteem, and confidence in teaching their subject. There is however
no effect of teacher’s experience on academic performance (in line
with most other research, e.g., Hanushek [1971], but also see Rockoff
[2004]). - Chetty, Friedman, and Rockoff (2014); Rothstein (2010); Jacob,
Lefgren, and Sims (2010). - See Flèche (2017).
Chapter 15. Measuring Cost- Effectiveness
in Terms of Happiness- This assumes cardinality and comparability across people. For
evidence on comparability across people, see Layard (2010). On cardi-
nality, if a variable is cardinal, this means that the difference between
a score of x and (x + 1) is the same as the difference between a score of
y and (y + 1), whatever the values of x and y. The evidence on whether
happiness measures are truly cardinal is limited, but Krueger and
Schkade (2008) found that test- retest differences were independent of