Assuming they called back every month for 6mo and always got a rep in that 3-5%, that's some supremely bad luck...
...unless call center support is a biased sample.
I think it’s more likely that most of the call centre staff were reasonable people who followed the process they were trained to do, but the process is broken. Some of them may have known it was broken, but either not known how to work around that, or faced barriers to doing so such as having to talk to an a-hole boss about it.
No this should not be correct (that is if you are assuming the 3-5% rules is correct and I'm not saying it is). The people who work for companies are vetted by the company so in theory the percentage of people who are bad would have to in theory be less simply because of that filter. The people a company (or someone is dealing with) is unfiltered hence the probability in theory again should be higher.
Some percentage of managers, including hiring managers, will also suck, and these can make crappy employees out of people who otherwise have potential to be good.
I think it's more than 5% on the service personnel side. Customers are a more reliable random sample than service personnel, who are likely to not like their jobs in the first place.