I previously estimated that something like 2/3 to 3/4 of offspring failed to survive in historical times - and that this was the principal mechanism for elimination of deleterious mutations.
Modern child mortality rates are, by contrast, so low that it is inevitable that mutations will accumulate - and reducing intelligence is an inevitable consequence (since 'g' is a proxy measure of fitness).
Evidence for this comes from various sources including A Farewell to Alms: a brief economic history of the world, by Gregory Clark. Princeton University Press, 2007.
Also theoretical considerations:
And, further evidence on this matter is available from a pair of review/ meta-analysis papers:
A Volk and J Atkinson. Is child death the crucible of human evolution. Journal of Social, Evolutionary and Cultural Psychology. 2008; 2: 247-260.
A Volk, J Atkinson. Infant and child death in the human environment of evolutionary adaptation. Evolution and Human Behaviour. 2013; 34: 182-192.
In the 2013 paper, a review of hunter gatherer mortality found an average 48.8% child mortality rate - noting that child mortality rates are an underestimate, as not all deaths are recorded.
Historical data showed an average of 46.2% with a minimum of 35%, until modern times in developed countries, when it drops to 1% .
(However, among individuals, some will have a probability of lower, and others of higher mortality rates among their offspring, according to their health, status, child rearing abilities etc.)
So about a half of children are known to have died before adult maturity in most times and most places, and the real percentage must have been higher.
In the 2008 paper, the authors note that most women who reach adulthood will have children, but about 5% may be infertile; by contrast about 10% of men fail to find a mate and about 5% are infertile. To this can be added the fertility-reducing effect of later marriage among low status men - often to older women with less reproductive potential.
This fits the idea that selection against deleterious mutations is stronger among men than women - with the variance of reproductive success larger among men; a smallish proportion of the fittest men differentially producing most of the viable offspring selection.
This also fits the anatomical picture of sexual dimorphism, with men as considerably more massive and strong than women, as consistent with some significant degree of de facto polygyny.
So, among men at least 65% or two thirds will fail to reproduce according to direct measures from anthropological and historical data.
Rates of failure to reproduce will differ between the sexes, with mortality differentially concentrated among men (and indeed male fetuses, babies and children - who suffer greater mortality than females - http://www.huli.group.shef.ac.uk/lummaaproceeding2001.pdf)
Given that 45-50 % directly-measured child mortality rates represents a minimum level; this evidence is reasonably consistent with my previous estimate; and emphasizes the massive change in selection pressure, and presumably mutation elimination, represented by a fifty-fold decline in child mortality rates from historical to modern times.
Note: I should add that it is the number of surviving (and reproductively viable) children which is the key factor; not the proportion of children that survive. i.e. Reproductive success is about both fertility and mortality.