The percentage of people that have landlines is not the problem as the pollsters can weigh certain demographics more heavily. The problem with that is that when the pollster finally does reach the 28 year old single white male with a landline making 75k a year, his answers may be outliers for his cohort. His answers end up having too much weight.
While that’s possible (I’d say probable even), it shouldn’t cause too much of a skew as long as the sample is large enough. Not every 28 single white male making 75k+ that uses a landline will be an outlier