181
age 0.743** 0.731** 1.350**
(0.167) (0.109) (0.189)
Attended KG school 0.860 0.970* 0.987
(0.586) (0.452) (0.663)
Ate meal at home 0.482 0.266 0.594
(0.677) (0.500) (0.725)
Language Hausa 0.867 0.528 0.452
(0.674) (0.669) (1.109)
Absent from school -0.948* -0.319 -0.238
(0.481) (0.414) (0.636)
Teacher absent -1.564** -0.570 -1.444**
(0.485) (0.385) (0.559)
Family wealth: q2 0.506 -0.119 0.220
(0.768) (0.571) (0.850)
Family wealth: q3 1.652* 0.251 -0.0156
(0.764) (0.657) (0.943)
Family wealth: q4 2.180* 0.535 0.727
(0.876) (0.639) (0.990)
Family wealth: q5 2.151* 1.151 1.839
(1.019) (0.839) (1.300)
School: TPR -0.00124 0.00113 0.00228
(0.00156) (0.00141) (0.00223)
School: female HT 1.160 1.297* 1.525
(0.996) (0.651) (1.086)
School: library 2.525** 1.111+ 1.689+
(0.887) (0.625) (0.949)
School: electricity 1.194 1.302+ 2.388*
(0.849) (0.689) (1.134)
School: drinking
water -1.970** -1.467** -2.514**
(0.577) (0.443) (0.654)
School : toilet 2.100** 0.983* 1.522*
(0.576) (0.404) (0.639)
Constant -2.594 1.597 5.214
(2.520) (1.815) (3.236)
Observations 2,831 2,832 2,827
R-squared 0.126 0.184 0.216
Notes: (1) Survey weights have been used to estimate the regression model. (2) Robust
standard errors are reported.
Given that the overall level of learning is low, perhaps it is more meaningful to study the
determinants of zero scores instead of total scores. Table 3.4.3 presents Probit estimates of the
determinants of zero scores in five Hausa sub-tasks. Only marginal effects are reported. Once
again, there is no systematic return to time spent in school – being enrolled in grade 3
significantly lowers the probability of a zero score only in case of CNEPM and ORF. Students of
government schools are significantly more likely to have zero scores while older students have
a significantly lower probability of zero scores. Differences in pre-school attendance do not
matter. The (language) teacher absence significantly increases zero score probability in case of
CLSPM and reading comprehension. The provision of toilet, electricity and library also lower the
probability of zero scores. However, family wealth doesn’t matter with the exception of CLSPM.