I see no reason why it could not happen. If the first dimer was the 3'end of one primer annealing in he middle of the other primer to give a product of about 30 bases and then there was some annealing of the same primer nearer to the end of the dimer product it would reach 50-60 bases, Hard to say without seeing the sequences of the primers The expected band with just dimerization is small so thermal diffusion will spread the band so it will look larger than its real size and also being a small product it may trap ETBR differently from your ladder bands. DNA with ETBR intercalated moves slower in an electric field than non stained dna so I would expect that a band that traps etbr differently ( in amount and position, perhaps AT rich) may move at a different speed therefore look like a different size so at small sizes the ladder may not be completely accurate. The way to test if it is PD would be to use a hot start polymerase when there will be very little dimerization unless you have contamination.

It might even be possible if the 2 primers annealed together with a mismatched base giving a 30 mer with a bubble in the middle and heteroduplex bands always move slowly ( looks larger) in agarose gels

I agree with Paul

'1. 'primer oligomers' as they are called are a recognised phenomenon 

2. As explained by Paul correct size exact movement presupposes a linear conformation.  Secondary structure as explained by Paul could result in retardation and thus an apparent increase in actual size

3. If secondary structure is complicating sizing then you might want to try adding 10% formamide to your loading buffer and denaturing for 2 min followed by snap cooking on ice for 5 min before loading on your gel 

I meant 'cooling' of course not 'cooking'. Predictive text late in the evening !!