Appariement statistique par imputation fractionnaire
4. Plan de sondage à questionnaire scindéAppariement statistique par imputation fractionnaire
4. Plan de sondage à questionnaire scindé
À la section 3, on examine le cas où l’échantillon A et
l’échantillon B sont deux échantillons indépendants de la même population
cible. Nous allons maintenant examiner un autre cas, celui d’un plan de sondage
à questionnaire scindé en vertu duquel l’échantillon initial
S
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGtbaaaa@3881@
est sélectionné à partir d’une population cible, puis l’échantillon A
et l’échantillon B sont sélectionnés au hasard de sorte que
A
∪
B
=
S
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGbbGaey
OkIGSaamOqaiaai2dacaWGtbaaaa@3C75@
et
A
∩
B
=
ϕ
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGbbGaey
ykICSaamOqaiaai2dacqaHvpGzcaGGUaaaaa@3E15@
On observe
(
x
,
y
1
)
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaqadaqaai
aadIhacaaISaGaamyEamaaBaaaleaacaaIXaaabeaaaOGaayjkaiaa
wMcaaaaa@3CD4@
dans l’échantillon A et
(
x
,
y
2
)
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaqadaqaai
aadIhacaaISaGaamyEamaaBaaaleaacaaIYaaabeaaaOGaayjkaiaa
wMcaaaaa@3CD5@
dans l’échantillon B. On
souhaite créer des données entièrement augmentées avec observation de
(
x
,
y
1
,
y
2
)
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaqadaqaai
aadIhacaaISaGaamyEamaaBaaaleaacaaIXaaabeaakiaaiYcacaWG
5bWaaSbaaSqaaiaaikdaaeqaaaGccaGLOaGaayzkaaaaaa@3F7A@
dans
S
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGtbGaai
Olaaaa@3933@
De tels plans de sondage à questionnaire scindé gagnent en
popularité parce qu’ils réduisent le fardeau de réponse (Raghunathan et Grizzle 1995;
Chipperfield et Steel 2009). Des plans de sondage à questionnaire scindé
ont notamment été explorés dans le cadre de la Consumer Expenditure Survey (Gonzalez et Eltinge 2008) et de la National Assessment of Educational Progress (NAEP) Survey aux États-Unis. Les analystes qui utilisent les
résultats des enquêtes à questionnaire scindé peuvent s’intéresser à des
paramètres multiples, comme la moyenne pour
y
1
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaigdaaeqaaaaa@398E@
et la moyenne pour
y
2
,
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaikdaaeqaaOGaaiilaaaa@3A49@
en plus du coefficient de la
régression de
y
2
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaikdaaeqaaaaa@398F@
sur
y
1
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaigdaaeqaaOGaaiOlaaaa@3A4A@
Nous avons examiné un plan de sondage où l’échantillon initial
S
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGtbaaaa@3881@
est divisé en deux sous-échantillons :
A
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGbbaaaa@386F@
et
B
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGcbGaai
Olaaaa@3922@
On suppose que
x
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG4bWaaS
baaSqaaiaadMgaaeqaaaaa@39C0@
est observé pour
i
∈
S
,
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGPbGaey
icI4Saam4uaiaacYcaaaa@3BA3@
que
y
1
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaigdacaWGPbaabeaaaaa@3A7C@
est recueilli pour
i
∈
A
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGPbGaey
icI4Saamyqaaaa@3AE1@
et que
y
2
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaikdacaWGPbaabeaaaaa@3A7D@
est recueilli pour
i
∈
B
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGPbGaey
icI4SaamOqaiaac6caaaa@3B94@
La probabilité de sélection
dans
A
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGbbaaaa@386F@
ou
B
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGcbaaaa@3870@
peut dépendre de
x
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG4bWaaS
baaSqaaiaadMgaaeqaaaaa@39C0@
mais ne dépend pas de
y
1
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaigdacaWGPbaabeaaaaa@3A7C@
ni de
y
2
i
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaikdacaWGPbaabeaakiaac6caaaa@3B39@
En conséquence, le plan de
sondage utilisé pour sélectionner les sous-échantillons
A
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGbbaaaa@386F@
et
B
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGcbaaaa@3870@
est non informatif pour le
modèle spécifié (Fuller 2009, chapitre 6). Soit
w
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG3bWaaS
baaSqaaiaadMgaaeqaaaaa@39BF@
le poids d’échantillonnage associé
à l’échantillon complet
S
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGtbGaai
Olaaaa@3933@
On suppose qu’il existe une procédure pour estimer la variance d’un
estimateur de la forme
Y
^
=
∑
i
∈
S
w
i
y
i
,
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaaceWGzbGbaK
aacaaI9aWaaabeaeqaleaacaWGPbGaeyicI4Saam4uaaqab0Gaeyye
IuoakiaaykW7caWG3bWaaSbaaSqaaiaadMgaaeqaaOGaamyEamaaBa
aaleaacaWGPbaabeaakiaacYcaaaa@4513@
et on désigne l’estimateur de
la variance par
V
^
s
(
∑
i
∈
S
w
i
y
i
)
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaaceWGwbGbaK
aadaWgaaWcbaGaam4CaaqabaGcdaqadaqaamaaqababeWcbaGaamyA
aiabgIGiolaadofaaeqaniabggHiLdGccaaMc8Uaam4DamaaBaaale
aacaWGPbaabeaakiaadMhadaWgaaWcbaGaamyAaaqabaaakiaawIca
caGLPaaacaGGUaaaaa@4702@
Décrivons maintenant une procédure pour obtenir un ensemble de
données entièrement imputées. D’abord, on utilise la méthode décrite à la
section 3 pour obtenir les valeurs imputées
{
y
1
i
*
(
j
)
:
i
∈
B
,
j
=
1,
…
,
m
}
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaGadaqaai
aadMhadaqhaaWcbaGaaGymaiaadMgaaeaacaaIQaWaaeWaaeaacaWG
QbaacaGLOaGaayzkaaaaaOGaaGOoaiaadMgacqGHiiIZcaWGcbGaaG
ilaiaadQgacaaI9aGaaGymaiaaiYcacqWIMaYscaaISaGaamyBaaGa
ay5Eaiaaw2haaaaa@4A88@
et une estimation
θ
^
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacuaH4oqCga
qcaaaa@396F@
du paramètre de la distribution
f
(
y
2
|
y
1
,
x
;
θ
)
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGMbWaae
WaaeaadaabcaqaaiaadMhadaWgaaWcbaGaaGOmaaqabaGccaaMc8oa
caGLiWoacaaMc8UaamyEamaaBaaaleaacaaIXaaabeaakiaaiYcaca
WG4bGaaG4oaiabeI7aXbGaayjkaiaawMcaaiaac6caaaa@4788@
On obtient l’estimation
θ
^
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacuaH4oqCga
qcaaaa@396F@
en résolvant
∑
i
∈
B
w
i
∑
j
=
1
m
w
i
j
*
S
2
(
θ
;
x
i
,
y
1
i
*
(
j
)
,
y
2
i
)
=
0,
(
4.1
)
MathType@MTEF@5@5@+=
feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaaeqbqabS
qaaiaadMgacqGHiiIZcaWGcbaabeqdcqGHris5aOGaaGPaVlaadEha
daWgaaWcbaGaamyAaaqabaGcdaaeWbqabSqaaiaadQgacaaI9aGaaG
ymaaqaaiaad2gaa0GaeyyeIuoakiaaykW7caWG3bWaa0baaSqaaiaa
dMgacaWGQbaabaGaaGOkaaaakiaadofadaWgaaWcbaGaaGOmaaqaba
GcdaqadaqaaiabeI7aXjaaiUdacaWG4bWaaSbaaSqaaiaadMgaaeqa
aOGaaGilaiaadMhadaqhaaWcbaGaaGymaiaadMgaaeaacaaIQaWaae
WaaeaacaWGQbaacaGLOaGaayzkaaaaaOGaaGilaiaadMhadaWgaaWc
baGaaGOmaiaadMgaaeqaaaGccaGLOaGaayzkaaGaaGypaiaaicdaca
aISaGaaGzbVlaaywW7caaMf8UaaGzbVlaaywW7caGGOaGaaGinaiaa
c6cacaaIXaGaaiykaaaa@6B7D@
où
S
2
(
θ
;
x
,
y
1
,
y
2
)
=
∂
log
f
(
y
2
|
y
1
,
x
;
θ
)
/
∂
θ
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGtbWaaS
baaSqaaiaaikdaaeqaaOWaaeWaaeaacqaH4oqCcaaI7aGaamiEaiaa
iYcacaWG5bWaaSbaaSqaaiaaigdaaeqaaOGaaGilaiaadMhadaWgaa
WcbaGaaGOmaaqabaaakiaawIcacaGLPaaacaaI9aWaaSGbaeaacqGH
ciITciGGSbGaai4BaiaacEgacaWGMbWaaeWaaeaadaabcaqaaiaadM
hadaWgaaWcbaGaaGOmaaqabaGccaaMc8oacaGLiWoacaaMc8UaamyE
amaaBaaaleaacaaIXaaabeaakiaaiYcacaWG4bGaaG4oaiabeI7aXb
GaayjkaiaawMcaaaqaaiabgkGi2kabeI7aXbaacaGGUaaaaa@5BCD@
Sachant
θ
^
,
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacuaH4oqCga
qcaiaacYcaaaa@3A1F@
on génère les valeurs imputées
y
2
i
*
(
j
)
∼
f
(
y
2
|
y
1
i
,
x
i
;
θ
^
)
,
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaa0
baaSqaaiaaikdacaWGPbaabaGaaGOkamaabmaabaGaamOAaaGaayjk
aiaawMcaaaaakiablYJi6iaadAgadaqadaqaamaaeiaabaGaamyEam
aaBaaaleaacaaIYaaabeaakiaaykW7aiaawIa7aiaaykW7caWG5bWa
aSbaaSqaaiaaigdacaWGPbaabeaakiaaiYcacaWG4bWaaSbaaSqaai
aadMgaaeqaaOGaaG4oaiqbeI7aXzaajaaacaGLOaGaayzkaaGaaiil
aaaa@50DC@
pour
i
∈
A
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGPbGaey
icI4Saamyqaaaa@3AE1@
et
j
=
1,
…
,
m
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGQbGaaG
ypaiaaigdacaaISaGaeSOjGSKaaGilaiaad2gacaGGUaaaaa@3E4C@
Si l’on suppose que le modèle est identifié, l’estimateur de
paramètre
θ
^
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacuaH4oqCga
qcaaaa@396F@
généré par la résolution de (4.1)
est entièrement efficace au sens où la valeur imputée de
y
2
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaikdacaWGPbaabeaaaaa@3A7D@
pour l’échantillon A ne
donne lieu à aucun gain d’efficacité. Pour le voir, notons que l’équation de
score utilisant la valeur imputée de
y
2
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaikdacaWGPbaabeaaaaa@3A7D@
se calcule comme suit :
∑
i
∈
A
w
i
m
−
1
∑
j
=
1
m
S
2
(
θ
;
x
i
,
y
1
i
,
y
2
i
*
(
j
)
)
+
∑
i
∈
B
w
i
∑
j
=
1
m
w
i
j
*
S
2
(
θ
;
x
i
,
y
1
i
*
(
j
)
,
y
2
i
)
=
0.
(
4.2
)
MathType@MTEF@5@5@+=
feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaaeqbqabS
qaaiaadMgacqGHiiIZcaWGbbaabeqdcqGHris5aOGaaGPaVlaadEha
daWgaaWcbaGaamyAaaqabaGccaWGTbWaaWbaaSqabeaacqGHsislca
aIXaaaaOWaaabCaeqaleaacaWGQbGaaGypaiaaigdaaeaacaWGTbaa
niabggHiLdGccaaMc8Uaam4uamaaBaaaleaacaaIYaaabeaakmaabm
aabaGaeqiUdeNaaG4oaiaadIhadaWgaaWcbaGaamyAaaqabaGccaaI
SaGaamyEamaaBaaaleaacaaIXaGaamyAaaqabaGccaaISaGaamyEam
aaDaaaleaacaaIYaGaamyAaaqaaiaaiQcadaqadaqaaiaadQgaaiaa
wIcacaGLPaaaaaaakiaawIcacaGLPaaacqGHRaWkdaaeqbqabSqaai
aadMgacqGHiiIZcaWGcbaabeqdcqGHris5aOGaaGPaVlaadEhadaWg
aaWcbaGaamyAaaqabaGcdaaeWbqabSqaaiaadQgacaaI9aGaaGymaa
qaaiaad2gaa0GaeyyeIuoakiaaykW7caWG3bWaa0baaSqaaiaadMga
caWGQbaabaGaaGOkaaaakiaadofadaWgaaWcbaGaaGOmaaqabaGcda
qadaqaaiabeI7aXjaaiUdacaWG4bWaaSbaaSqaaiaadMgaaeqaaOGa
aGilaiaadMhadaqhaaWcbaGaaGymaiaadMgaaeaacaaIQaWaaeWaae
aacaWGQbaacaGLOaGaayzkaaaaaOGaaGilaiaadMhadaWgaaWcbaGa
aGOmaiaadMgaaeqaaaGccaGLOaGaayzkaaGaaGypaiaaicdacaaIUa
GaaGzbVlaaywW7caaMf8UaaiikaiaaisdacaGGUaGaaGOmaiaacMca
aaa@8EA6@
Comme
y
2
i
*
(
1
)
,
…
,
y
2
i
*
(
m
)
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaa0
baaSqaaiaaikdacaWGPbaabaGaaGOkamaabmaabaGaaGymaaGaayjk
aiaawMcaaaaakiaaiYcacqWIMaYscaaISaGaamyEamaaDaaaleaaca
aIYaGaamyAaaqaaiaaiQcadaqadaqaaiaad2gaaiaawIcacaGLPaaa
aaaaaa@4612@
sont générées à partir de
f
(
y
2
|
y
1
i
,
x
i
;
θ
^
)
,
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGMbWaae
WaaeaadaabcaqaaiaadMhadaWgaaWcbaGaaGOmaaqabaGccaaMc8oa
caGLiWoacaaMc8UaamyEamaaBaaaleaacaaIXaGaamyAaaqabaGcca
aISaGaamiEamaaBaaaleaacaWGPbaabeaakiaaiUdacuaH4oqCgaqc
aaGaayjkaiaawMcaaiaacYcaaaa@49A8@
p
lim
m
→
∞
∑
i
∈
A
w
i
m
−
1
∑
j
=
1
m
S
2
(
θ
;
x
i
,
y
1
i
,
y
2
i
*
(
j
)
)
=
∑
i
∈
A
w
i
E
{
S
2
(
θ
;
x
i
,
y
1
i
,
Y
2
)
|
y
1
i
,
x
i
;
θ
^
}
.
MathType@MTEF@5@5@+=
feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGWbWaay
buaeqaleaacaWGTbGaeyOKH4QaeyOhIukabeGcbaGaciiBaiaacMga
caGGTbaaaiaaykW7daaeqbqabSqaaiaadMgacqGHiiIZcaWGbbaabe
qdcqGHris5aOGaaGPaVlaadEhadaWgaaWcbaGaamyAaaqabaGccaWG
TbWaaWbaaSqabeaacqGHsislcaaIXaaaaOWaaabCaeqaleaacaWGQb
GaaGypaiaaigdaaeaacaWGTbaaniabggHiLdGccaaMc8Uaam4uamaa
BaaaleaacaaIYaaabeaakmaabmaabaGaeqiUdeNaaG4oaiaadIhada
WgaaWcbaGaamyAaaqabaGccaaISaGaamyEamaaBaaaleaacaaIXaGa
amyAaaqabaGccaaISaGaamyEamaaDaaaleaacaaIYaGaamyAaaqaai
aaiQcadaqadaqaaiaadQgaaiaawIcacaGLPaaaaaaakiaawIcacaGL
PaaacaaI9aWaaabuaeqaleaacaWGPbGaeyicI4Saamyqaaqab0Gaey
yeIuoakiaaykW7caWG3bWaaSbaaSqaaiaadMgaaeqaaOGaamyramaa
cmaabaWaaqGaaeaacaWGtbWaaSbaaSqaaiaaikdaaeqaaOWaaeWaae
aacqaH4oqCcaaI7aGaamiEamaaBaaaleaacaWGPbaabeaakiaaiYca
caWG5bWaaSbaaSqaaiaaigdacaWGPbaabeaakiaaiYcacaWGzbWaaS
baaSqaaiaaikdaaeqaaaGccaGLOaGaayzkaaGaaGPaVdGaayjcSdGa
aGPaVlaadMhadaWgaaWcbaGaaGymaiaadMgaaeqaaOGaaGilaiaadI
hadaWgaaWcbaGaamyAaaqabaGccaaI7aGafqiUdeNbaKaaaiaawUha
caGL9baacaaIUaaaaa@8FAA@
Ainsi, en vertu de la
propriété de la fonction de score, le premier terme de (4.2) évalué à
θ
=
θ
^
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacqaH4oqCca
aI9aGafqiUdeNbaKaaaaa@3BEC@
est proche de zéro et la solution de
l’équation (4.2) est essentiellement la même que celle de l’équation (4.1),
c’est-à-dire qu’on ne gagne pas en efficience en utilisant la valeur imputée de
y
2
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaikdacaWGPbaabeaaaaa@3A7D@
pour calculer l’EMV pour
θ
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacqaH4oqCaa
a@395F@
dans
f
(
y
2
|
y
1
,
x
;
θ
)
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGMbWaae
WaaeaadaabcaqaaiaadMhadaWgaaWcbaGaaGOmaaqabaGccaaMc8oa
caGLiWoacaaMc8UaamyEamaaBaaaleaacaaIXaaabeaakiaaiYcaca
WG4bGaaG4oaiabeI7aXbGaayjkaiaawMcaaiaac6caaaa@4788@
Toutefois, les valeurs imputées de
y
2
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaikdacaWGPbaabeaaaaa@3A7D@
peuvent améliorer
l’efficience des inférences pour les paramètres de la distribution conjointe de
(
y
1
i
,
y
2
i
)
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaqadaqaai
aadMhadaWgaaWcbaGaaGymaiaadMgaaeqaaOGaaGilaiaadMhadaWg
aaWcbaGaaGOmaiaadMgaaeqaaaGccaGLOaGaayzkaaGaaiOlaaaa@4055@
À titre d’exemple simple,
prenons l’estimation de
μ
2
,
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacqaH8oqBda
WgaaWcbaGaaGOmaaqabaGccaGGSaaaaa@3B01@
la moyenne marginale de
y
2
i
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaaS
baaSqaaiaaikdacaWGPbaabeaakiaac6caaaa@3B39@
En vertu d’un échantillonnage
aléatoire simple, l’estimateur imputé de
μ
=
E
(
Y
2
)
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacqaH8oqBca
aI9aGaamyramaabmaabaGaamywamaaBaaaleaacaaIYaaabeaaaOGa
ayjkaiaawMcaaaaa@3E49@
est
μ
^
I
,
m
=
1
n
{
∑
i
∈
A
(
m
−
1
∑
j
=
1
m
y
2
i
*
(
j
)
)
+
∑
i
∈
B
y
2
i
}
,
(
4.3
)
MathType@MTEF@5@5@+=
feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacuaH8oqBga
qcamaaBaaaleaacaWGjbGaaGilaiaad2gaaeqaaOGaaGypamaalaaa
baGaaGymaaqaaiaad6gaaaWaaiWaaeaadaaeqbqabSqaaiaadMgacq
GHiiIZcaWGbbaabeqdcqGHris5aOWaaeWaaeaacaWGTbWaaWbaaSqa
beaacqGHsislcaaIXaaaaOWaaabCaeqaleaacaWGQbGaaGypaiaaig
daaeaacaWGTbaaniabggHiLdGccaaMc8UaamyEamaaDaaaleaacaaI
YaGaamyAaaqaaiaaiQcadaqadaqaaiaadQgaaiaawIcacaGLPaaaaa
aakiaawIcacaGLPaaacqGHRaWkdaaeqbqabSqaaiaadMgacqGHiiIZ
caWGcbaabeqdcqGHris5aOGaaGPaVlaadMhadaWgaaWcbaGaaGOmai
aadMgaaeqaaaGccaGL7bGaayzFaaGaaiilaiaaywW7caaMf8UaaGzb
VlaaywW7caaMf8UaaiikaiaaisdacaGGUaGaaG4maiaacMcaaaa@6E84@
où
les valeurs de
y
2
i
*
(
1
)
,
…
,
y
2
i
*
(
m
)
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWG5bWaa0
baaSqaaiaaikdacaWGPbaabaGaaGOkamaabmaabaGaaGymaaGaayjk
aiaawMcaaaaakiaaiYcacqWIMaYscaaISaGaamyEamaaDaaaleaaca
aIYaGaamyAaaqaaiaaiQcadaqadaqaaiaad2gaaiaawIcacaGLPaaa
aaaaaa@4612@
sont générées à partir de
f
(
y
2
|
y
1
i
,
x
i
;
θ
^
)
.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGMbWaae
WaaeaadaabcaqaaiaadMhadaWgaaWcbaGaaGOmaaqabaGccaaMc8oa
caGLiWoacaaMc8UaamyEamaaBaaaleaacaaIXaGaamyAaaqabaGcca
aISaGaamiEamaaBaaaleaacaWGPbaabeaakiaaiUdacuaH4oqCgaqc
aaGaayjkaiaawMcaaiaac6caaaa@49AA@
Pour des valeurs de
m
,
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGTbGaai
ilaaaa@394B@
suffisamment grandes, on peut écrire
μ
^
I
,
∞
=
1
n
{
∑
i
∈
A
y
^
2
i
+
∑
i
∈
B
y
2
i
}
=
1
n
{
∑
i
∈
A
E
(
y
2
|
y
1
i
,
x
i
;
θ
^
)
+
∑
i
∈
B
y
2
i
}
.
MathType@MTEF@5@5@+=
feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaafaqaaeGaca
aabaGafqiVd0MbaKaadaWgaaWcbaGaamysaiaaiYcacqGHEisPaeqa
aaGcbaGaaGypamaalaaabaGaaGymaaqaaiaad6gaaaWaaiWaaeaada
aeqbqabSqaaiaadMgacqGHiiIZcaWGbbaabeqdcqGHris5aOGaaGPa
VlqadMhagaqcamaaBaaaleaacaaIYaGaamyAaaqabaGccqGHRaWkda
aeqbqabSqaaiaadMgacqGHiiIZcaWGcbaabeqdcqGHris5aOGaaGPa
VlaadMhadaWgaaWcbaGaaGOmaiaadMgaaeqaaaGccaGL7bGaayzFaa
aabaaabaGaaGypamaalaaabaGaaGymaaqaaiaad6gaaaWaaiWaaeaa
daaeqbqabSqaaiaadMgacqGHiiIZcaWGbbaabeqdcqGHris5aOGaaG
PaVlaadweadaqadaqaamaaeiaabaGaamyEamaaBaaaleaacaaIYaaa
beaakiaaykW7aiaawIa7aiaaykW7caWG5bWaaSbaaSqaaiaaigdaca
WGPbaabeaakiaaiYcacaWG4bWaaSbaaSqaaiaadMgaaeqaaOGaaG4o
aiqbeI7aXzaajaaacaGLOaGaayzkaaGaey4kaSYaaabuaeqaleaaca
WGPbGaeyicI4SaamOqaaqab0GaeyyeIuoakiaaykW7caWG5bWaaSba
aSqaaiaaikdacaWGPbaabeaaaOGaay5Eaiaaw2haaiaai6caaaaaaa@7E2E@
En
vertu du scénario de l’exemple 2.1, on peut écrire
y
^
2
i
=
β
^
0
+
β
^
1
y
1
i
+
β
^
2
x
2
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaaceWG5bGbaK
aadaWgaaWcbaGaaGOmaiaadMgaaeqaaOGaaGypaiqbek7aIzaajaWa
aSbaaSqaaiaaicdaaeqaaOGaey4kaSIafqOSdiMbaKaadaWgaaWcba
GaaGymaaqabaGccaWG5bWaaSbaaSqaaiaaigdacaWGPbaabeaakiab
gUcaRiqbek7aIzaajaWaaSbaaSqaaiaaikdaaeqaaOGaamiEamaaBa
aaleaacaaIYaGaamyAaaqabaaaaa@4AB8@
où
(
β
^
0
,
β
^
1
,
β
^
2
)
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaqadaqaai
qbek7aIzaajaWaaSbaaSqaaiaaicdaaeqaaOGaaGilaiqbek7aIzaa
jaWaaSbaaSqaaiaaigdaaeqaaOGaaGilaiqbek7aIzaajaWaaSbaaS
qaaiaaikdaaeqaaaGccaGLOaGaayzkaaaaaa@4284@
satisfont
∑
i
∈
B
(
y
2
i
−
β
^
0
−
β
^
1
i
y
^
1
i
−
β
^
2
x
2
i
)
=
0
MathType@MTEF@5@5@+=
feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaaeqbqabS
qaaiaadMgacqGHiiIZcaWGcbaabeqdcqGHris5aOGaaGPaVpaabmaa
baGaamyEamaaBaaaleaacaaIYaGaamyAaaqabaGccqGHsislcuaHYo
GygaqcamaaBaaaleaacaaIWaaabeaakiabgkHiTiqbek7aIzaajaWa
aSbaaSqaaiaaigdacaWGPbaabeaakiqadMhagaqcamaaBaaaleaaca
aIXaGaamyAaaqabaGccqGHsislcuaHYoGygaqcamaaBaaaleaacaaI
YaaabeaakiaadIhadaWgaaWcbaGaaGOmaiaadMgaaeqaaaGccaGLOa
GaayzkaaGaaGypaiaaicdaaaa@55E7@
et
y
^
1
i
=
α
^
0
+
α
^
1
x
1
i
+
α
^
2
x
2
i
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaaceWG5bGbaK
aadaWgaaWcbaGaaGymaiaadMgaaeqaaOGaaGypaiqbeg7aHzaajaWa
aSbaaSqaaiaaicdaaeqaaOGaey4kaSIafqySdeMbaKaadaWgaaWcba
GaaGymaaqabaGccaWG4bWaaSbaaSqaaiaaigdacaWGPbaabeaakiab
gUcaRiqbeg7aHzaajaWaaSbaaSqaaiaaikdaaeqaaOGaamiEamaaBa
aaleaacaaIYaGaamyAaaqabaaaaa@4AB0@
où
(
α
^
0
,
α
^
1
,
α
^
2
)
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaqadaqaai
qbeg7aHzaajaWaaSbaaSqaaiaaicdaaeqaaOGaaGilaiqbeg7aHzaa
jaWaaSbaaSqaaiaaigdaaeqaaOGaaGilaiqbeg7aHzaajaWaaSbaaS
qaaiaaikdaaeqaaaGccaGLOaGaayzkaaaaaa@427E@
satisfont
∑
i
∈
A
(
y
1
i
−
α
^
0
−
α
^
1
x
1
i
−
α
^
2
x
2
i
)
=
0.
MathType@MTEF@5@5@+=
feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaadaaeqaqabS
qaaiaadMgacqGHiiIZcaWGbbaabeqdcqGHris5aOGaaGPaVpaabmaa
baGaamyEamaaBaaaleaacaaIXaGaamyAaaqabaGccqGHsislcuaHXo
qygaqcamaaBaaaleaacaaIWaaabeaakiabgkHiTiqbeg7aHzaajaWa
aSbaaSqaaiaaigdaaeqaaOGaamiEamaaBaaaleaacaaIXaGaamyAaa
qabaGccqGHsislcuaHXoqygaqcamaaBaaaleaacaaIYaaabeaakiaa
dIhadaWgaaWcbaGaaGOmaiaadMgaaeqaaaGccaGLOaGaayzkaaGaaG
ypaiaaicdacaGGUaaaaa@5553@
Ainsi, en ignorant les termes d’ordre plus
faible, on obtient
V
(
μ
^
I
,
∞
)
=
1
n
V
(
y
2
)
+
(
1
n
b
−
1
n
)
V
(
y
2
−
y
^
2
)
,
MathType@MTEF@5@5@+=
feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn
hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr
4rNCHbGeaGqiFu0Je9sqqrpepC0xbbL8F4rqqrpipeea0xe9Lqpe0x
e9q8qqvqFr0dXdHiVc=bYP0xH8peuj0lXxdrpe0=1qpeeaY=rrVue9
Fve9Fve8meaabaqaciGacaGaaeqabaWaaeaaeaaakeaacaWGwbWaae
WaaeaacuaH8oqBgaqcamaaBaaaleaacaWGjbGaaGilaiabg6HiLcqa
baaakiaawIcacaGLPaaacaaI9aWaaSaaaeaacaaIXaaabaGaamOBaa
aacaWGwbWaaeWaaeaacaWG5bWaaSbaaSqaaiaaikdaaeqaaaGccaGL
OaGaayzkaaGaey4kaSYaaeWaaeaadaWcaaqaaiaaigdaaeaacaWGUb
WaaSbaaSqaaiaadkgaaeqaaaaakiabgkHiTmaalaaabaGaaGymaaqa
aiaad6gaaaaacaGLOaGaayzkaaGaamOvamaabmaabaGaamyEamaaBa
aaleaacaaIYaaabeaakiabgkHiTiqadMhagaqcamaaBaaaleaacaaI
YaaabeaaaOGaayjkaiaawMcaaiaacYcaaaa@55B8@
qui
est inférieure à la variance de l’estimateur direct
μ
^
b
=
n
b
−
1
∑
i
∈
B
y
2
i
.
ISSN : 1712-5685
Politique de rédaction
Techniques d ’enquête publie des articles sur les divers aspects des méthodes statistiques qui intéressent un organisme statistique comme, par exemple, les problèmes de conception découlant de contraintes d’ordre pratique, l’utilisation de différentes sources de données et de méthodes de collecte, les erreurs dans les enquêtes, l’évaluation des enquêtes, la recherche sur les méthodes d’enquête, l’analyse des séries chronologiques, la désaisonnalisation, les études démographiques, l’intégration de données statistiques, les méthodes d’estimation et d’analyse de données et le développement de systèmes généralisés. Une importance particulière est accordée à l’élaboration et à l’évaluation de méthodes qui ont été utilisées pour la collecte de données ou appliquées à des données réelles. Tous les articles seront soumis à une critique, mais les auteurs demeurent responsables du contenu de leur texte et les opinions émises dans la revue ne sont pas nécessairement celles du comité de rédaction ni de Statistique Canada.
Présentation de textes pour la revue
Techniques d ’enquête est publiée en version électronique deux fois l’an. Les auteurs désirant faire paraître un article sont invités à le faire parvenir en français ou en anglais en format électronique et préférablement en Word au rédacteur en chef, (statcan.smj-rte.statcan@canada.ca , Statistique Canada, 150 Promenade du Pré Tunney, Ottawa, (Ontario), Canada, K1A 0T6). Pour les instructions sur le format, veuillez consulter les directives présentées dans la revue ou sur le site web (www.statcan.gc.ca/Techniquesdenquete).
Note de reconnaissance
Le succès du système statistique du Canada repose sur un partenariat bien établi entre Statistique Canada et la population, les entreprises, les administrations canadiennes et les autres organismes. Sans cette collaboration et cette bonne volonté, il serait impossible de produire des statistiques précises et actuelles.
Normes de service à la clientèle
Statistique Canada s'engage à fournir à ses clients des services rapides, fiables et courtois. À cet égard, notre organisme s'est doté de normes de service à la clientèle qui doivent être observées par les employés lorsqu'ils offrent des services à la clientèle.
Droit d'auteur
Publication autorisée par le ministre responsable de Statistique Canada.
© Ministre de l'Industrie, 2016
L'utilisation de la présente publication est assujettie aux modalités de l'Entente de licence ouverte de Statistique Canada .
N° 12-001-X au catalogue
Périodicité : Semi-annuel
Ottawa
Date de modification :
2016-06-22