Sexual selection is usually a more proximate and often more rapid and powerful mechanism of natural selection than selection based on differential survival - because animals excluded from mating, or whose matings do not lead to viable offspring, have their genetic contribution cut short immediately - in the same generation.
Adverse sexual selection is reproductive death: genetic death.
Assortative mating is a mechanism of sexual selection in which similar females mate with similar males - similar in terms of some aspect of their phenotype such as 'beauty', status, or an ability such as intelligence - these aspects of phenotype being significantly correlated with genetic differences.
Thus the 'fittest' (those having highest reproductive potential and - through most of human history - the lowest deleterious mutation loads) will pair with the fittest, and vice versa.
When humans are producing a several-fold surplus of offspring, natural selection needs to have mechanisms by which as high a proportion as possible of the following generation are on average parented by those carrying the minimum load of deleterious mutations - and this requires that those carrying the greatest mutation load be mostly prevented from producing sexually mature offspring (I say 'mostly' because by chance some small proportion offspring of the population of those with highest mutation loads will - by chance - be low in mutations).
In assortative mating, then, a minority of the population who are the most 'attractive' and fittest males and females will pair-off and have (on average) a large number of (mostly) fit children (without accumulation of mutations) - while the majority of the population who are least attractive, the least fit, and the heaviest mutation-load bearers will be left-over.
This population majority of left-over males and females may not have a sexual partner (reproductive death), or the females may have a share in mating with the minority of high fitness males, or else a male and a female pair who are both low attractiveness/ low fitness/ heavy mutation load will mate - but have a very low (but not zero) probability of raising offspring to viable sexual maturity.
(The effective of sexual selection is usually greatest on the males; with a higher proportion of males than females having zero matings, zero long term sexual partners, zero viable offspring. Thus, it is one function of the male sex to be the main way in which new mutations are purged from the population.)
Therefore, assortative mating of this type - with 'the fittest' parenting almost all of the next generations offspring - has the potential, in combination with normal natural selection based on survival, to (on average) purge all newly occurred mutations from a population with each generation; so each new generation can - in effect - start afresh with a minimal mutational load.
Reflecting on the above scenario in light of human society in the developed nations over the past century plus - it can be seen that the effects of the industrial revolution has been to put assortative mating into reverse; and not merely to fail to purge all new mutations from a population with each generation - but actually to amplify the proportion of new mutations in a population with each generation.
This happens by a combination of chosen sub-replacement fertility among the fittest (instead of, as has been usual in human history, the fittest parenting almost-all the next generation); with the least-fit who carry the heaviest mutational load parenting most of the next generation (instead of almost none of them).
Furthermore, instead of a system which over-produces offspring who are then selected for optimum fitness; we now have a system which under-produces offspring - so there is no possibility of ordinary natural selection or sexual selection or any combination of the two being able to purge the new mutations from each new generation!
In such a scenario it is easy to understand how a fitness-sensitive trait such as intelligence (as objectively measured by reaction time) has declined so much and so rapidly over the past century plus.
The effect of failing to purge newly occurring mutations with each generation is that mutations accumulate and overwhelm that particular lineage with random genetic damage; each new generation will have more genetic damage and lower fitness than the previous generation; and after a number of generations, that particular lineage will lose fitness completely and become extinct.
Note: The above idea concerning the vital role of assortative mating in purging of mutation accumulation is substantially derived from my conversations with Michael A Woodley.
Note added 18 September 2014
In addition to the problem of mutation accumulation by relaxation of selection, when a population has begun shrinking, as is the case for the native populations of all Western and developed nations, there is an increasing danger of extinction due to 'mutational meltdown' (http://en.wikipedia.org/wiki/Mutational_meltdown) .
The unusual twist with modern humans is that the populations have begun falling due to chosen sub-replacement fertility, and before mutation accumulation has reached a level sufficient to prevent fertility. This will probably have accelerated the severity of mutation accumulation, and increased the risk of mutational meltdown.