Are there any other caps on other switches in that row? If so, try cutting them as well.

Also, try removing the connector in the back box for the switches and shorting the pins for the row and column for the saucer to see if the problem happens. If so, it is a board issue. If not, it is likely on the playfield. Keep checking for shorts to ground along that row and column.

EDIT: Sorry, didn’t notice that you said switching the board doesn’t fix it. Ignore the point about shorting the pins.

If both boards give the same results it is more likely to be a playfield problem. Sounds a bit like this: When a column wire is shorted to ground, and any switch in that column is closed, the switch test will show ALL switches in the ROW of the closed switch as being closed. If no switches are closed, the switch test will show no switches closed. To find the location of the short, go to the end of the switch column wire on the playfield (the switches are "daisy chained" together for an entire column or row). Then break the daisy chain one switch at a time until the short no longer shows in the switch test.

Are the diodes in the switch matrix, particularly in the problem rows/columns the correct way around?

There should be a piece of insulating paper in the saucer switch preventing the contacts from shorting against the metal frame of the saucer. If that insulation has worn away there could be a frame short to the switch when the ball is present. Inspect the saucer switch assembly carefully to insure the ball is not causing a short. There should be no metal on metal contact with the ball.