53
54 In our analysis we subtract the score of dependence from 6 and we rename the variable
‘independence’ so that a higher score indicates the opposite, namely that the pupil is not dependent on the teacher.
Conflict:
- This kid and I seem to be in constant battle
- This child feels that I am treating him/her unfairly
- If this kid is in a bad mood, I know it's going to be a long and difficult day
- This child's feelings towards me can be completely unpredictable or change abruptly - Dealing with this child requires a lot of energy from me
Here too we subtract the score of conflict from 6 and we rename the variable ‘no-conflict so that a higher score indicates the opposite, namely that the pupil and the teacher are getting along very well.
Closeness:
- This child talks honestly with me about his/her feelings and experiences - If this child is sad, it will seek comfort in me
- I have a cordial, warm relationship with this child - This child seems to feel safe with me
- Dealing with this child gives me a sense of effectiveness
A higher score indicates that the teacher experiences affection, warmth and open communication in the relationship with a student.
For the variables motivation, self-efficacy, relation with teacher, relation with classmates, the pupils themselves were asked questions to which they may respond with one of five options: (1) not at all true, (2) not true, (3) sometimes true/sometimes not, (4) true, and (5) exactly true. Here too the scores are set from 1 to 5.
Self-efficacy:
- I am certain that I can complete even the most challenging assignments in school.
- I can complete even the most challenging school assignments if I do my best.
- I can also learn difficult things at school
- I'm sure I'll be able to do everything at school this year - I can do almost anything at school if I just keep trying - I can do all my school work if I have enough time Motivation:
- I like it when I learn something new at school
- I like it when I have learned something in school that is important to me - I am satisfied when I have learned something in school that I understand
- I'd rather do difficult assignments that teach me something new than easy assignments - If I don't understand something right away at school, I do my best to understand
55 Relation with teacher:
- If I feel unhappy, I can talk about it with the teacher - I can talk to the teacher about my problems
- I have a good contact with the teacher - The teacher understands me
- I feel at ease with the teacher
- The teacher usually knows how I feel - I would not rather have a different teacher Relation with classmates:
- I do not feel lonely my class
- I get along well with my classmates
- I'd rather stay in this class and not be in a different class - We have a nice class
- I have a lot of contact with my classmates - I like interacting with the kids in my class
56 A-2 Average cognitive and non-cognitive skills by background variables
For each variable we report the differences in the means between groups of pupils. We use t-test to determine the significance of the difference between the means.
- By parental education: mean pupil of high-educated parents – mean pupil of low-educated parents
- By parental income: mean pupil of high-income parents – mean pupil of low-income parents - By migrant background: Mean pupil without a migrant background – mean pupil with a migrant
background
- By urbanicity: mean pupil in low urban areas – mean pupil in urban areas - By gender: mean boys – mean girls
Variables By parental
education
By parental income
By migrant background
By urbanicity
By gender
Cognitive skills
Nscct- figure composition 0,816*** 0,679*** 1,111*** 0,573*** 0,224***
Nscct-exclusion 0,696*** 0,674*** 0,900*** 0,453*** -0,088
Nscct-number series 0,738*** 0,646*** 0,692*** 0,288*** 0,088
Nscct-categories 1,006*** 0,752*** 1,400*** 0,719*** -0,430***
Nscct-analogies 1,513*** 1,060*** 1,482*** 0,815*** -0,973***
Nscct-total 4,769*** 3,811*** 5,585*** 2,848*** -1,179***
Reading comprehension 6,423*** 5,085*** 6,838*** 3,244*** -3,313***
Technical reading 1,914*** 0,569 -2,277*** -1,737*** -1,021
Math 6,320*** 5,350*** 7,413*** 4,133*** 5,544***
vocabulary 6,819*** 5,707*** 10,283*** 5,164*** 1,383***
Non-cognitive skills
Performance 0,095*** 0,099*** 0,198*** 0,172*** -0,255***
Behavior 0,144*** 0,110*** 0,160*** 0,114*** -0,366***
Working attitude 0,198*** 0,111*** 0,078*** 0,069*** -0,500***
Popularity 0,116*** 0,157*** 0,091*** 0,040 -0,122***
Self-efficacy -0,020 -0,064*** -0,205*** -0,141*** 0,073***
motivation -0,064*** -0,074*** -0,220*** -0,118*** -0,004
Relation w/ teacher -0,056*** -0,021 -0,035 -0,025 -0,127***
Relation w/ classmates -0,022 0,050*** 0,051 0,070*** 0,043***
independence 0,176*** 0,149*** 0,143*** 0,103*** -0,016
No-conflict 0,153*** 0,152*** 0,174*** 0,110*** -0,316***
closeness 0,053*** 0,082*** 0,121*** 0,082*** -0,274***
*** p<0.01, ** p<0.05, * p<0.1
57 A-3 Lubotsky-Wittenburg method
Consider un unobserved variable 𝑥∗ for which we have two observed proxies 𝑥1 and 𝑥2. You can think of 𝑥∗ as cognitive skills and 𝑥1 and 𝑥2 as two observed measures of cognitive skills.
𝑦 = 𝛽𝑥∗+ 𝜀 𝑥1 = 𝜌1𝑥∗+ 𝑢1 𝑥2 = 𝜌2𝑥∗+ 𝑢2
We assume that 𝑥∗ is uncorrelated with 𝜀 and that 𝑢1 and 𝑢2 are uncorrelated with 𝑥∗ and 𝜀. We allow 𝑢1 and 𝑢2 to be correlated and denote their covariance by 𝜎12. Suppose in a first stage that 𝜌1 = 𝜌2 = 1, so that the two variables are equally correlated with 𝑥∗. We then have 𝑐𝑜𝑣(𝑦, 𝑥1) = 𝑐𝑜𝑣(𝑦, 𝑥1) = 𝛽𝑣𝑎𝑟(𝑥∗) = 𝛽𝜎𝑥2. Regressing 𝑦 on either 𝑥1 or 𝑥2 yields measurement error and therefore attenuation bias. Regardless, one could ask which of two proxies yield the least bias and thus is better to use. Leamer (1983) suggests to pick the variable with the highest 𝑅-squared and the lowest variance. Another way is to average both variables which allows to use all the information available but this will not necessarily yield a smaller variance. The contribution of Lubotsky &
Wittenberg (2006) is that they thought of a linear combination of 𝑥1 and 𝑥2 that minimizes the error variance. In particular, let 𝑢0 be the linear combination of the error terms from equation (3) and (4) as follows : 𝑢0 = 𝛿1𝑢1+ 𝛿2𝑢2 , with 𝛿1+ 𝛿2 = 1. 𝛿1 that minimizes20 the variance of 𝑢0 is:
𝛿1 = 𝜎22 − 𝜎12 𝜎12− 2𝜎12+ 𝜎22
where 𝜎12 and 𝜎22 are the variances of the errors 𝑢1 and 𝑢2 respectively. We can then construct the variance of 𝑢0 as follows:
𝑣𝑎𝑟(𝑢0) = 𝜎22𝜎12− 𝜎122 𝜎12 − 2𝜎12+ 𝜎22 Thus the estimate of 𝛽 is given asymptotically by:
𝑏∗ = 𝛽 𝜎𝑥2
𝜎𝑥2+ 𝜎22𝜎12− 𝜎122 𝜎12− 2𝜎12+ 𝜎22 Proof:
20 𝑣𝑎𝑟(𝑢0) = 𝑣𝑎𝑟(𝛿1𝑢1+ 𝛿2𝑢2) = 𝛿12𝜎12+ 2𝛿1(1 − 𝛿1)𝜎12+ (1 − 𝛿1)2𝜎22 , since 𝛿1+ 𝛿2= 1
𝑑𝑣𝑎𝑟(𝑢0)
𝑑𝛿1 = 0 ⇒ 2𝛿1𝜎12+ 2𝜎12− 4𝛿1𝜎12− 2𝜎22+ 2𝛿1𝜎22= 0 ⇒ 𝛿1= 𝜎22− 𝜎12
𝜎12− 2𝜎12+ 𝜎22 3
4
58 Let 𝑥′ be the optimal linear combination of 𝑥1 and 𝑥2 such as 𝑥′ = 𝑥∗+ 𝑢0
and we know 𝑦 = 𝛽𝑥∗+ 𝜀 𝑏∗= 𝑐𝑜𝑣(𝑥′, 𝑦)
𝑣𝑎𝑟(𝑥∗+ 𝑢0)=𝑐𝑜𝑣(𝑥∗+ 𝑢0, 𝛽𝑥∗+ 𝜀)
𝑣𝑎𝑟(𝑥∗) + 𝑣𝑎𝑟(𝑢0) =𝛽𝑣𝑎𝑟(𝑥∗) + 𝑐𝑜𝑣(𝑥∗, 𝜀) + 𝛽𝑐𝑜𝑣(𝑢0, 𝑥∗) + 𝑐𝑜𝑣(𝑢0, 𝜀) 𝑣𝑎𝑟(𝑥∗) + 𝑣𝑎𝑟(𝑢0)
Since we assumed that 𝑢1 and 𝑢2 are uncorrelated with 𝑥∗ and 𝜀 and we know that 𝑢0 is a linear combination of 𝑢1 and 𝑢2, we have: 𝑐𝑜𝑣(𝑥∗, 𝜀) = 0 , 𝑐𝑜𝑣(𝑢0, 𝑥∗) = 0 , 𝑐𝑜𝑣(𝑢0, 𝜀) = 0
𝑏∗= 𝛽𝑣𝑎𝑟(𝑥∗) 𝑣𝑎𝑟(𝑥∗) + 𝑣𝑎𝑟(𝑢0) 𝑏∗= 𝛽 𝜎𝑥2
𝜎𝑥2+ 𝜎22𝜎12− 𝜎122 𝜎12− 2𝜎12+ 𝜎22
We will however need the variances 𝜎12 and 𝜎22 which we do not know. It turns out that when running the regression of 𝑦 on both 𝑥1 and 𝑥2, the sum of their coefficients exactly equals the coefficient 𝑏∗ obtained on the optimally weighted combination of the proxies. This is because 𝜌2 = 1, so that the two variables 𝑥1 and 𝑥2 are equally correlated with 𝑥∗. When this is not the case, a weighted sum of the coefficients of 𝑥1 and 𝑥2 is needed. The weights are related to the individual regression coefficients of the proxies.
In order to account for the sampling error, we use bootstrap with 6000 repetitions.
59 A-4 Detailed decomposition results
For outcome variable under-advice By parental education
By parental income
By migration background
By urbanicity By gender Group 1
Group 2 Difference Explained Unexplained
0.198***
0.125***
0.0734***
0.0251***
0.0483***
0.177***
0.140***
0.0371***
0.0369***
0.000213
0.183***
0.167***
0.0160 -0.00406 0.0201
0.182***
0.119***
0.0631***
0.0737 -0.0106
0.167***
0.144***
0.0235**
-0.0152 0.0387***
Explained Parental educ 1 Parental educ 2 Parental educ 3 Migration 2 Migration 3 Migration 4 parental income gender
Household type Care record birthyear
mother age at birth father age at birth kids in household parental involvement urbanicity 2
urbanicity 3 urbanicity 4 urbanicity 5 Drenthe Flevoland Friesland Gelderland Groningen Limburg Noord-Brabant Overijssel Utrecht Zeeland Zuid-Holland Schoolscore 2 Schoolscore 3 Schoolscore 4 Schoolscore 5 Final test 2 Final test 3 Final test 4 Denomination 2 Denomination 3
-0.000390 0.00197 0.00195**
0.00658 0.000234 -0.00103 0.000169 -0.00381**
0.00213 1.31e-06 0.000189 0.00583 0.000355 -3.37e-06 -0.000690 0.000112 0.000445 0.00104 0.000227 5.96e-05 -8.90e-05 0.000894 5.36e-05 0.000631 0.000657 0.000245 -0.000670 0.000877 -0.000323 -0.00196 -0.00151 5.49e-05 0.000428 -0.000661 8.78e-05 -0.000909
-3.38e-05 0.0125***
0.00563***
0.000391 0.00149 0.00412**
0.00113 -0.00255 0.000229 -0.00246**
0.00283 0.000411 3.60e-05 0.00660*
-0.00164 -0.000422 -0.00208 0.000284 0.00132 0.00163 0.000626 0.000655 0.000175 0.000528 3.41e-05 3.20e-05 0.00173 0.000278 -0.000417 0.000379 0.000407 -0.000920 -0.00106 1.95e-05 0.000375 -0.000211 -1.82e-05 -0.000659
-3.95e-05 0.0187***
-0.00483***
0.00770 0.000501 -0.00154 -0.000240 -0.000584 0.00149 3.50e-05 0.000804 0.00640 -0.00244 -2.90e-05 -0.00744 0.000751 -0.00157 0.00225 -0.00259 0.000301 -0.000733 0.00295 0.000101 -0.000945 0.00114 -0.000717 -0.00229 -0.00708 -0.000604 -0.00579 -0.00646 0.000280 -0.00136 -0.000461 -0.000388 -0.00383
-2.51e-05 0.00545**
-0.00405**
-0.000927 -0.00138 -0.00409**
-0.00122 -0.00155 0.00165 -0.000325 0.000293 -0.000642 0.000142 -3.18e-05 -0.00252
0.00318 -0.000726 0.00303 -0.00330 0.000568 0.00113 0.000139 1.03e-05 -0.00143 -0.000445 0.00390 0.00289 -0.000564 0.00159 0.00142 9.30e-05 0.000710 0.000205 0.000561 0.000724
2.69e-05 0.00185*
-0.000975 -0.000538 5.25e-05 0.000979 0.000261 -0.000125 -0.000771 0.000853 0.000303 6.16e-05 4.86e-06 -0.000225 -0.00108 -0.000810 -0.000283 -0.000130 5.58e-05 0.000321 9.95e-05 -0.000693 9.46e-05 -0.000662 7.88e-05 -1.14e-05 0.00131 -3.81e-05 0.00126 -0.000268 8.53e-06 3.97e-05 -0.000153 1.56e-06 0.000164 -2.20e-05 5.71e-05 -0.000150
60 Denomination 4
Denomination 5 Denomination 6 Nscct- Figure comp.
Nscct-exclusion Nscct-numbers Nscct-categories Nscct-Analogies Reading compreh.
Technical reading Math
Vocabulary Performance Behavior
Working attitude Popularity Self-efficacy Motivation Relation-teacher Relation-others independence No conflict Closeness
9.12e-06 0.00568**
0.000742 -0.000210 -3.72e-05 -0.000117 -0.000380 -0.00114*
-0.000690 0.000837 0.00234 -0.000708 0.00235**
-0.00292*
0.00367**
0.000332 -0.000277 0.000165 0.000308 0.000134 0.00157 -0.000341 -0.000394
-2.43e-05 0.00212*
0.00179 0.000277 -5.49e-05 -0.000533 -0.000402 -0.000288 -4.29e-05 0.000281 -0.000284 -0.000532 0.000888 -0.00151 0.00256**
0.000981 -0.000540 0.000268 0.000195 -8.08e-05 0.000823 -0.000480 -0.000303
5.95e-05 0.0356**
0.000275 -0.00356**
-0.00271**
-0.00167*
-0.00292 -0.00340**
-0.00614**
0.00162 -0.0130***
-0.00420 0.00494***
-0.00320*
0.00139 0.000289 -0.00264 0.000561 0.000178 -0.000238 0.00120 -0.000446 -0.000887
-0.000953 -0.00275*
0.00426 0.00131 0.000432 0.000410 0.00178 0.00140 0.00254*
-0.000132 0.00958***
0.00349 -0.00258*
0.00139 -0.00125 -3.93e-05 0.00245 -0.000647 -0.000207 0.000113 -0.000386 0.000226 0.000348
7.43e-05 0.000257 0.000186 -0.000417 0.000427 -0.000731 0.000871 0.00335*
0.00535**
0.000147 -0.0179***
-0.00156 -0.00507**
0.00741*
-0.0125**
-0.000737 0.00103 0.000197 0.00119 -8.81e-05 -4.67e-05 0.00127 0.000932 Note: we do not report standard errors to make the table easier to read.
For outcome variable persistent under-advice By parental
education
By parental income
By migration background
By urbanicity By gender Group 1
Group 2 Difference Explained Unexplained
0.150***
0.0812***
0.0686***
0.0173**
0.0513***
0.132***
0.0959***
0.0366***
0.0259***
0.0106
0.127***
0.114***
-0.0130 -0.00607 -0.00694
0.135***
0.0781***
0.0570***
0.0288 0.0282
0.120***
0.102***
0.0178 -0.0111 0.0289*
Explained Parental educ 1 Parental educ 2 Parental educ 3 Migration 2 Migration 3 Migration 4 parental income gender
Household type Care record birthyear
mother age at birth father age at birth kids in household parental involvement urbanicity 2
-1.31e-05 0.00181 0.000816 0.00692 0.000144 -0.000173 0.000133 -0.00200 0.00128 -0.000566 0.000452 0.00452 0.000312
-7.40e-05 0.0117***
0.00644***
0.000111 0.00245 0.00178 0.000836 -0.000544 0.000131 -0.000887 0.000703 -0.000145 0.000446 0.00420 -0.000826
-0.000123 0.0179***
-0.00531***
0.00799 0.000309 -0.000179 -0.000185 -0.000308 0.000875 0.000567 0.00337 0.00500 -0.00215
-5.54e-05 -0.00509***
0.00466***
-0.000257 0.00332 -0.00173 -0.00236 -0.00116 0.000615 -0.000188 0.000107 -0.000110 -4.34e-05 0.000483 -0.00157
5.95e-05 0.00173*
-0.00112 -0.000149 -0.000126 -0.000414 0.000505 -4.64e-05 -0.000445 0.000311 5.17e-05 -1.88e-05 -7.40e-05 -0.000140 -0.000545
61 urbanicity 3
urbanicity 4 urbanicity 5 Drenthe Flevoland Friesland Gelderland Groningen Limburg Noord-Brabant Overijssel Utrecht Zeeland Zuid-Holland Schoolscore 2 Schoolscore 3 Schoolscore 4 Schoolscore 5 Final test 2 Final test 3 Final test 4 Denomination 2 Denomination 3 Denomination 4 Denomination 5 Denomination 6 Nscct- Figure comp.
Nscct-exclusion Nscct-numbers Nscct-categories Nscct-Analogies Reading compreh.
Technical reading Math
Vocabulary Performance Behavior
Working attitude Popularity Self-efficacy Motivation Relation-teacher Relation-others independence No conflict Closeness
-1.12e-05 -0.000661 -0.000146 0.000492 -3.68e-05 -8.03e-05 -0.000658 -6.49e-05 0.000369 6.73e-05 0.000747 0.000524 -0.000136 -0.000721 0.000579 0.000763 -0.000668 0.00245 4.85e-06 0.000236 -0.000935 0.000275 -0.00108 -3.46e-05 0.00222 -0.000771 -0.000179 -2.93e-05 -6.09e-05 -0.000512 -0.000867 -0.000381 0.000451 0.00186 -0.000496 0.00186**
-0.00314**
0.00257 0.000117 -0.000327 0.000104 0.000363 7.54e-05 0.00172 0.000567 -0.000805
-0.000690 -0.00267 -0.000773 0.00166 -0.000106 -0.000535 -0.000634 0.000148 0.000171 -3.17e-05 0.000370 0.00105 -0.000114 -0.000185 0.000200 0.00139 0.000609 0.00249 6.85e-06 4.86e-06 -0.000271 -5.90e-05 -0.000910 -9.54e-05 0.000814 0.000312 0.000195 -4.62e-05 -0.000266 -0.000421 -0.000257 -1.93e-05 4.29e-05 -0.000219 -0.000386 0.000542 -0.00146 0.00219*
0.000226 -0.000420 7.82e-05 9.27e-05 -2.61e-05 0.00141 0.000372 -0.000927
-0.000287 -0.00718 -0.00128 -0.00175 -9.60e-05 0.000796 -0.00283 -0.000542 0.00126 0.000121 -0.00112 0.000917 0.000400 -0.00264 -0.00467 0.000854 -0.00212 0.00796 1.74e-05 -0.000736 -0.000601 -0.00152 -0.00444 -0.000193 0.0139 -0.000365 -0.00304**
-0.00215*
-0.000866 -0.00389**
-0.00257**
-0.00334 0.000832 -0.0104***
-0.00321 0.00390**
-0.00347**
0.000988 9.96e-05 -0.00326*
0.000328 0.000222 -0.000150 0.00136 0.000632 -0.00184
0.00409 5.03e-05 -0.00285 0.00293 0.000477 0.000375 -0.000114 0.000213 -0.000872 -0.000184 0.00175 0.00152 -0.00227 -0.000766 -0.00283 3.91e-05 -1.65e-05 0.000264 0.00177 0.000999 -0.00376 -0.00104 0.000713 0.000921 0.000363 0.000206 0.00186*
0.00124 0.00113 -9.96e-06 0.00737***
0.00256 -0.00157 0.00135 -0.00107 -6.91e-06 0.00190 -0.000191 -9.92e-05 3.60e-05 -0.000665 -0.000160 0.00109
-0.00132 -0.000363 0.000345 7.19e-05 -2.23e-05 -9.36e-05 0.000614 7.94e-05 -0.000219 -6.44e-05 -0.000236 0.000803 1.58e-05 0.000564 -0.000141 3.43e-05 -1.91e-05 0.000306 6.53e-07 -3.81e-06 -2.84e-05 0.000180 -0.000207 0.000293 9.72e-05 3.11e-05 -0.000293 0.000359 -0.000368 0.000910 0.00297*
0.00238 1.11e-05 -0.0138***
-0.00114 -0.00309 0.00717*
-0.0107**
-0.000130 0.000796 5.81e-05 0.000568 -2.80e-05 -8.04e-05 -0.000895 0.00291
Groups:
By parental education:
- Group 1: pupils of low-educated parents - Group 2: pupils of high-educated parents
62 By parental income:
- Group 1: pupils of low-income parents - Group 2: pupils of high-income parents By migration background:
- Group 1: pupils with a non-western migrant background - Group 2: pupils without a non-western migrant backgorund By urbanization:
- Group 1 : pupils in low-urbanized areas - Group 2: pupils in high-urbanized areas By gender:
- Group 1: girls - Group 2: boys
63 A-5 Results with PCA
Table 5.1: results of regression with PCA
64 Table 5.2: twofold Blinder-Oaxaca decomposition by parents’ education, with PCA
Table 5.3: twofold Blinder-Oaxaca decomposition by parents’ income, with PCA
65 Table 5.4 : twofold Blinder-Oaxaca decomposition by migrant background with PCA
Table 5.5 : twofold Blinder-Oaxaca decomposition by gender, with PCA
66 Table 5.6: twofold Blinder-Oaxaca decomposition by urbanization, with PCA
67 A-6 Results with Lasso
Table 6.1: Regression with lasso
Under-advice Persistent under-advice
Lasso group1 (1)
Lasso group 2 (2)
Lasso group1 (3)
Lasso group2 (4)
Background Gender
Parent’s education
Migrant background Parents’ income
Household type Special care
Birth year
Mother’s age at birth
Father’s age at birth
Kids in household Parents’ involvement
0.0447***
(0.0164) 0.0544***
(0.0163) 0.0136 (0.0218) 0.0130 (0.0162)
-0.00201 (0.00151)
0.0160 (0.0155) 0.0486***
(0.0161) 0.0234 (0.0211) -0.0110 (0.0168) 0.0187 (0.0211)
0.0215 (0.0148)
0.00499 (0.00740) -0.0155 (0.00987)
0.0279**
(0.0142) 0.0586***
(0.0144) -0.00501 (0.0201) 0.0209 (0.0144)
0.0194 (0.0149) 0.00775 (0.0131) -0.00108 (0.00134)
0.0101 (0.0139) 0.0547***
(0.0140) 0.0141 (0.0171) -0.00930 (0.0140)
-0.0137 (0.00860) Degree of urbanicity
Very urban
Moderately urban Slightly urban
Not urban Regions
Drenthe Flevoland
Friesland Gelderland
Groningen Limburg
Noord-Brabant Overijssel
Utrecht Zeeland
-0.0362**
(0.0181)
0.0213 (0.0188)
0.0473 (0.0566) 0.0508 (0.0450) 0.0331 (0.0371)
0.157**
(0.0692) -0.0826***
(0.0305)
0.0516*
(0.0293)
-0.0613 (0.0477)
-0.0280 (0.0180)
0.0355**
(0.0179)
0.149**
(0.0603) 0.0958**
(0.0416) 0.0673*
(0.0356)
0.0329*
(0.0184)
-0.109***
(0.0398)
-0.0322**
(0.0160)
0.0156 (0.0172)
0.101**
(0.0505)
0.0597***
(0.0203) 0.109*
(0.0614)
0.0834***
(0.0259)
-0.0323**
(0.0153)
0.0286*
(0.0155)
0.120**
(0.0527)
0.0220 (0.0158)
-0.0802**
(0.0346)
68
Zuid-Holland -0.0229
(0.0204)
-0.0242 (0.0208)
-0.0326*
(0.0179)
School School score
105-109 110-119
120-139 140-220
Final test Route8 ICE
Other tests Denomination
Protestant-Chris Roman-Catholic
Other Christian Other religions
ABO (algemeen bijz.)
0.0497***
(0.0184)
-0.0272 (0.0265)
0.121***
(0.0351) -0.0326 (0.0351)
0.0419**
(0.0180)
0.152***
(0.0374)
0.0274*
(0.0163)
-0.000231 (0.0245) 0.0269 (0.0312)
0.0523 (0.0596)
0.0122 (0.0182) 0.0319*
(0.0168) -0.101**
(0.0425) 0.0695**
(0.0350) 0.0196 (0.0325) Cognitive skills
Vocabulary
Technical reading Reading compreh.
Math
Nscct- Figure comp.
Nscct-exclusion Nscct-numbers
Nscct-categories Nscct-Analogies
Nscct-total
0.00485*
(0.00264) 0.00322 (0.00277) 0.0168***
(0.00287) 0.0121***
(0.00407) 0.00718*
(0.00379) 0.00748**
(0.00354) 0.00584 (0.00378) 0.0134***
(0.00477) -0.0112 (0.00889)
0.00905***
(0.00265) 0.0107***
(0.00281)
0.00611**
(0.00284) 0.00379 (0.00271) 0.00465 (0.00283)
0.00346 (0.00240) 0.0103***
(0.00251) 0.00791***
(0.00255) 0.00418 (0.00261)
0.00325 (0.00248) 0.00620**
(0.00249)
0.0128***
(0.00245)
0.00482*
(0.00291)
0.00477*
(0.00288) Non-cognitive skills
Performance
Behavior Working attitude
Popularity
-0.0327***
(0.0105) 0.0239*
(0.0129) -0.0197*
(0.0108)
0.0129 -0.0199**
-0.0344***
(0.00878) 0.0148*
(0.00875)
0.0230**
(0.00926) -0.0187**
(0.00871) -0.0191**
69 Self-efficacy
Motivation
Relation-teacher Relation-others
independence No conflict
Closeness
(0.0118) -0.0231*
(0.0128) -0.00263 (0.0139) 0.0251*
(0.0139) 0.00834 (0.0125) -0.0137 (0.0110) -0.00790 (0.0143) 0.0216 (0.0136)
(0.00990)
-0.0165 (0.0102)
0.0266**
(0.0115)
-0.0122 (0.00895)
0.0266**
(0.0112)
(0.00932) -0.0128 (0.00994)
-0.0110 (0.0115) 0.00857 (0.0106)
Constant -0.115
(0.0999)
-42.98 (29.58)
-15.67 (26.25)
0.119*
(0.0646)
Number of pupils 2599 2598 2599 2598
R-squared 0.096 0.061 0.084 0.053
70 Table 6.2: Blinder-Oaxaca decomposition by parental education, Lasso
Under-advice
Sample 1 Sample 2
Coefficient (1)
Share of total difference (2)
Coefficient (3)
Share of total difference (4) Group 1: pupils of low-educated
parents
0.204***
(0.0142)
0.192***
(0.0137) Group2 : pupils of high-educated
parents
0.124***
(0.0125)
0.125***
(0.0121)
Difference 0.0798***
(0.0158)
0.067***
(0.0151) Overall
Explained 0.028***
(0.00998)
35% 0.0237***
(0.00857)
35%
Unexplained 0.052***
(0.0168)
65% 0.0433***
(0.0159)
65%
Explained
Background 0.0143*
(0.00733)
18% 0.011
(0.00696)
16%
Regional 0.00204
(0.00386)
3% 0.00319
(0.00336)
5%
School 0.00764
(0.00486)
9% 0.004*
(0.00229)
6%
Relative cognitive skills 0.0008 (0.00414)
1% 0.000
(0.00272)
0%
Non-cognitive skills 0.003
(0.00332)
4% -0.005***
(0.00196)
8%
Persistent under-advice
Sample 1 Sample 2
Group 1: pupils of low-educated parents
0.154***
(0.0132)
0.145***
(0.0123) Group2 : pupils of high-educated
parents
0.0843***
(0.0103)
0.0781***
(0.00986)
Difference 0.0701***
(0.0140)
0.0671***
(0.0129) Overall
Explained 0.0157
(0.00977)
22% 0.0182**
(0.00818)
27%
Unexplained 0.0544***
(0.0146)
78% 0.0489***
(0.0136)
73%
Explained
Background 0.0121*
(0.00702)
17% 0.0144**
(0.00701)
21%
Regional 0.000368
(0.00303)
0.5% 0.0005
(0.00261)
1%
School 0.00380
(0.00716)
5.5%
Relative cognitive skills -0.0019 (0.00294)
-3% 0.0012
(0.00238)
2%
Non-cognitive skills 0.00131
(0.00267)
2% 0.0020
(0.00228)
3%
71 Table 6.3: Blinder-Oaxaca decomposition by parental income, Lasso
Under-advice
Sample 1 Sample 2
Coefficient (1)
Share of total difference (2)
Coefficient (3)
Share of total difference (4) Group 1: pupils of low-income
parents
0.194***
(0.0145)
0.177***
(0.0144) Group2 : pupils of high-income
parents
0.156***
(0.0131)
0.155***
(0.0131)
Difference 0.0381**
(0.0157)
0.0218 (0.0165) Overall
Explained 0.0283**
(0.0123)
75% 0.0332***
(0.00919)
152%
Unexplained 0.0097
(0.0161)
25% -0.0114
(0.0167)
-52%
Explained
Background 0.0283***
(0.00714)
75% 0.0273***
(0.00666)
125%
Regional -0.0022
(0.00412)
-6% -0.0029
(0.00414)
-13%
School 0.0091
(0.00756)
24% 0.0103*
(0.00615)
47%
Relative ognitive skills -0.0058 (0.00429)
-15% -0.0065**
(0.00290)
-30%
Non-cognitive skills -0.0010 (0.00327)
-3% 0.0052***
(0.00198)
23 Persistent under-advice
Sample 1 Sample 2
Group 1: pupils of low-income parents
0.145***
(0.0135)
0.127***
(0.0126) Group2 : pupils of high-income
parents
0.112***
(0.0120)
0.112***
(0.0116)
Difference 0.0336**
(0.0151)
0.0152 (0.0147) Overall
Explained 0.0120
(0.0124)
36% 0.0243***
(0.00824)
159%
Unexplained 0.0216
(0.0149)
64% -0.0090
(0.0145)
-59%
Explained
Background 0.0170**
(0.00818)
51% 0.0318***
(0.00803)
209%
Regional -0.0055
(0.00339)
-16% -0.0051
(0.00349)
-34%
School 0.00676
(0.0110)
20%
Relative ognitive skills -0.0063**
(0.00293)
-19% -0.0039*
(0.00231)
-26%
Non-cognitive skills 0.0001
(0.00287)
0.3% 0.00160
(0.00181)
10%
72 Table 6.4 : Blinder-Oaxaca decomposition by migrant background, Lasso
Under-advice
Sample 1 Sample 2
Coefficient (1)
Share of total difference (2)
Coefficient (3)
Share of total difference (4) Group 1: pupils with a migrant
background
0.182***
(0.0245)
0.183***
(0.0213) Group2 : pupils without a migrant
background
0.172***
(0.0124)
0.161***
(0.0119)
Difference 0.0101
(0.0266)
0.0218 (0.0231) Overall
Explained 0.0028
(0.0242)
27% 0.00451
(0.0192)
21%
Unexplained 0.0073
(0.0223)
73% 0.0173
(0.0202)
79%
Explained
Background 0.0274***
(0.00903)
271% 0.0242***
(0.00869)
111%
Regional -0.0094
(0.0103)
-93% -0.0184*
(0.0100)
-84%
School 0.0162
(0.0198)
160% 0.0225
(0.0161)
103%
Relative cognitive skills -0.0296***
(0.00687)
-293% -0.0306***
(0.00578)
-140%
Non-cognitive skills -0.0019 (0.00536)
-18% 0.0067***
(0.00231)
31%
Persistent under-advice
Sample 1 Sample 2
Group 1: pupils with a migrant background
0.116***
(0.0244)
0.111***
(0.0204) Group2 : pupils without a migrant
background
0.132***
(0.0108)
0.122***
(0.0104)
Difference -0.0152
(0.0262)
-0.0108 (0.0221) Overall
Explained -0.0078
(0.0281)
51% -0.0183*
(0.0109)
169%
Unexplained -0.0074
(0.0215)
49% 0.0075
(0.0235)
-69%
Explained
Background 0.0257***
(0.00844)
-169% 0.0255***
(0.00687)
-236%
Regional -0.0208***
(0.00794)
137% -0.0174**
(0.00807)
161%
School 0.0135
(0.0278)
-88%
Relative cognitive skills -0.0264***
(0.00505)
173% -0.0231***
(0.00434)
213%
Non-cognitive skills 0.0002
(0.00440)
-2% -0.00342
(0.00251)
31%
73 Table 6.5: Blinder-Oaxaca decomposition by level of urbanicity, Lasso
Under-advice
Sample 1 Sample 2
Coefficient (1)
Share of total difference (2)
Coefficient (3)
Share of total difference (4)
Group 1: Less urban 0.202***
(0.0173)
0.183***
(0.0146)
Group2 : Urban 0.145***
(0.0165)
0.147***
(0.0161)
Difference 0.0565**
(0.0237)
0.0361*
(0.0217) Overall
Explained 0.0390
(0.0297)
69% 0.0376
(0.0268)
104%
Unexplained 0.0175
(0.0364)
31% -0.0015
(0.0331)
-4%
Explained
Background -0.0147
(0.00946)
-26% -0.0125**
(0.00531)
-35%
Regional 0.0299
(0.0247)
53% 0.0452*
(0.0232)
125%
School -0.00380
(0.0114)
-7% -0.0107
(0.00924)
-30%
Relative ognitive skills 0.0269***
(0.00584)
47.5% 0.0209***
(0.00501)
58%
Non-cognitive skills 0.000815
(0.00402)
1.5 -0.00526**
(0.00214)
-14%
Persistent Under-advice
Sample 1 Sample 2
Group 1: Les urban 0.154***
(0.0138)
0.142***
(0.0129)
Group2 : Urban 0.101***
(0.0150)
0.0941***
(0.0135)
Difference 0.0539***
(0.0203)
0.0479**
(0.0187) Overall
Explained 0.0592*
(0.0309)
110% 0.0465*
(0.0251)
97%
Unexplained -0.0053
(0.0354)
-10% 0.00141
(0.0331)
3%
Explained
Background -0.0035
(0.00895)
-6.5% -0.0113
(0.00819)
-24%
Regional 0.0474**
(0.0228)
88% 0.0387*
(0.0225)
81%
School -0.00530
(0.0162)
-10%
Relative cognitive skills 0.0217***
(0.00432)
40.5% 0.0167***
(0.00375)
35%
Non-cognitive skills -0.00110
(0.00324)
-2% 0.00246
(0.00232)
5%
74 Table 6.6: Blinder-Oaxaca decomposition by gender, lasso
Under-advice
Sample 1 Sample 2
Coefficient (1)
Share of total difference (2)
Coefficient (3)
Share of total difference (4)
Group 1: girls 0.192***
(0.0153)
0.167***
(0.0131)
Group2 : boys 0.157***
(0.0119)
0.166***
(0.0135)
Difference 0.0351**
(0.0156)
0.001 (0.0144) Overall
Explained -0.0102
(0.0105)
-29% -0.0133*
(0.00730)
-1330%
Unexplained 0.0453***
(0.0169)
129% 0.0138
(0.0153)
1380%
Explained
Background 0.0003
(0.00242)
1% 0.0026
(0.00212)
260
Regional -0.0006
(0.00219)
-2% 0.0011
(0.00222)
110
School 0.0022
(0.00190)
6% 0.0005
(0.00123)
50%
Relative cognitive skills -0.0122*
(0.00637)
-34% -0.0084
(0.00582)
-840
Non-cognitive skills 3.62e-05
(0.00671)
0% -0.0090***
(0.00265)
-900 Persistent under-advice
Sample 1 Sample 2
Group 1: girls 0.144***
(0.0139)
0.115***
(0.0112)
Group2 : boys 0.111***
(0.0106)
0.124***
(0.0118)
Difference 0.0331**
(0.0138)
-0.0087 (0.0126) Overall
Explained 0.00548
(0.00865)
16% -0.0177**
(0.00762)
203%
Unexplained 0.0276*
(0.0147)
84% 0.0089
(0.0139)
-103%
Explained
Background 0.0005
(0.00209)
1.5% 0.0009
(0.00189)
-11%
Regional -0.0007
(0.00191)
-2% 0.0006
(0.00173)
-7%
School 0.0017
(0.00197)
5%
Relative cognitive skills -0.0044 (0.00534)
-13% -0.0164***
(0.00460)
188%
Non-cognitive skills 0.0083*
(0.00507)
25% -0.0029
(0.00492)
33%
75 A-7 Logistic decomposition results
Table 7.1: Logistic Blinder-Oaxaca decomposition by parental education
Table 7.2: Logistic Blinder-Oaxaca decomposition by income
76 Table 7.3: Logistic Blinder-Oaxaca decomposition by urbanization
Table 7.5: Logistic Blinder-Oaxaca decomposition by gender
77 A-8 Results with sample weights
Table 8.1: Blinder-Oaxaca decomposition by parental education, with sample weights
Table 8.2: Blinder-Oaxaca decomposition by parental income, with sample weights
78 Table 8.3: Blinder-Oaxaca decomposition by migrant background, with sample weights
Table 8.4: Blinder-Oaxaca decomposition by urbanization, with sample weights