A man (Brad) born in 1670 had six daughters . The six daughters produced one son each.
Brad also had five sons. And the five sons produced one daughter each.

Brad was a guest at six weddings between 1720 and 1735. All six of his grandsons married. One married an Irish girl, and the others married one of the five daughters. ( Their first cousin)
So, these six marriages produce twelve children born after 1775.
These twelve are the great grandchildren of Brad.

Now, take a look at each of these twelve grandchildren. They are ALL related to Brad. But two are descended from the Irish girl and the other ten are not.

Would the DNA reflect that the ten are more closely related to one another than they are to the descendants of the Irish girl? Even though the paper trail is clear that all twelve are the great grandchildren of Brad, the figures produced by gedmatch software predict that they are closer to grandchildren.

The children of the Irish girl would carry X % of Brad's DNA.
The other grandchildren would carry (X + .04) % of Brad's DNA.
In other words, Brad would be represented a bit more because of the marriages of the cousins.

Of the twelve descendants born after 1775, ten would have more of Brad's DNA than the other two would have.

Is this an example of "skewing " the DNA results?