In Algeria we find significant difference between MAFM estimated by Hajnal method and MAFM estimated by promotion directly toward vital registration or surveys.
Hi there, by Hajnal method you meant the singulate mean age at marriage, right? A lot of things to be considered when you use this mean age at first marriage measurement. The wise one is to look at the formula and assumption behind this. What happens when there is actually marriage 50 in your sample or population? You also need to look at the geographical information for example, what happens when there is actually a surge of young unmarried workers coming to certain areas?
We should also notice that Hajnal's background in developing this method was based on the fact that there was no substantial data of age at first marriage given the absence of registration system at that time. Moreover, it is really difficult to apply the formula up to the lowest level of area since we need to be aware of variations and short term shifts.
Most of the population measurement requires an in-depth look in the age structure, so we need to be careful in using this measurement before we do the interpretation.
Malgré que les enquêtes et recensements en plus des données provenant de l'état civil permettent de calculer directement l'âge moyen au premier mariage et l'écart d'âge entre époux, beaucoup d'auteurs continuent à utiliser les estimations à partir de la méthode de Hajnal 'sans pour autant le préciser). Dans le cas algérien les estimations directes remettent en cause certaines conclusions généralement admises. C'est ainsi que nous n'observons pas une baisse de l'écart d'âge entre époux.
The key point when using the Hajnal method is the population structure as regards the first marriage. So if first marriage pattern is changing over time, i.e. over cohorts, this measurement is biased.
You could read a small note I wrote in journal Population, "probabilities
Peut-on affirmer que la méhode de Hajnal mesure la durée moyenne du célibat et que lorsque la célibat définitif augmente on surestime l'âge moyen au mariage?
J'ai l'impression que la référence citée dans ma réponse a été tronquée : Sardon, "Quotients, fréquences et événements", Population, 1993, Vol 48, N°2, pp. 489-495.
Vous y verrez que si l'évolution de l'intensité du phénomène au fil des générations influe, le sens de variation du calendrier du phénomène joue également.
Ainsi, si le célibat définitif augmente cela signifie que l'intensité du premier mariage diminue. Dans ce cas, la méthode de Hajnal surestime l'âge moyen au 1er mariage seulement si le calendrier du 1er mariage devient plus précoce ou reste stable. Au contraire, si le calendrier devient plus tardif la méthode de Hajnal sous-estimera l’âge moyen au 1er mariage.
Hajnals' method (SMAM) is a period measure equivalent to life expectancy (between bounds). The direct calculation of mean age at first marriage in any year is also a period measure. They give the same result in the absence of bias - essentially that marriage rates have not changed during the last 40 years and the population has not been subject to marriage-related migration or marriage-related mortality; non-response also matters.
A detailed discussion of Hajnal's singulate mean age at marriage and its biases can be found in Booth, Heather (2001) Trends in mean age at first birth and first birth intervals in the Pacific Islands. Genus, LVII(3-4), pp.165-190.
Tables quantifying bias are in found in Booth, Heather (1994) The estimation of levels and trend in age at first birth and age at fist marriage in the Pacific Islands. Working Papers in Demography No.45, Research School of Social Sciences, The Australian National University, 40pp. [full text available on ResearchGate]
Yes. Sex-specific migration and different trends in age at marriage ( perhaps related to educational changes or employment patterns) can cause different biases.
Yes, but you can try to roughly estimate bias - or rather you can say the gap is likely to be wider/narrower due to x ,y, z..... You also need to take age structure into account - past declines in fertility change the marriage market; the education factor will very likely also operate. See DOI: 10.4054/DemRes.2010.23.7