In the first part of this passage, Popper says that what distinguishes between science and "pseudo-science" is whether or not its theories are falsifiable. A criterion of a good scientific experiment is that it is an attempt to test and refute the theory, rather than to confirm it: "it is easy to obtain confirmations...if we look for confirmations".
A problem with this very strict perspective is how we should respond when an experiment throws up a result that appears to go against an established theory. Should we discard the theory, or modify it, or perhaps assume that there are problems with the way we did the experiment? If we have a theory that works most of the time, and has made successful predictions, then do we really want to consider it to have been undermined if we encounter some anomalous results? It would seem more sensible to be sceptical of the experiment which threw up the anomalies. One problem Popper doesn't answer is how much contrary evidence we should allow before a theory is overturned.
Popper criticises "pseudo-science" for producing ad hoc explanations for results which appear to falsify a theory. But there are situations where "science" appears to do this as well. When anomalous evidence is obtained, the first response it typically to ignore it or to come up with some explanation as to why it fits after all.
For example, by Popper's definition, it would appear that evolutionary science is a pseudo-science: Evans (2001:34) says "Take guilt. On the face of it, it is hard to see why natural selection would have endowed us with this emotion", before proceeding to give reasons why we should have expected natural selection to have eliminated the emotion of guilt. Evans then goes on to give a series of counter-reasons why guilt is a useful feature that must have come about through natural selection: the question he never even considers is whether the theory of natural selection might be suspect.
Newtonian mechanics has been very successful in explaining the behaviour of objects at a macro level, but does not work for sub-atomic particles. This has not been reason to disregard the Newtonian model; rather, new explanations have been formulated to explain quantum behaviour - in other words the Newtonian model has been refined, and one might say that this process is not a lot different in principle from the "ad hoc" explanations given by Marxists and Freudians.