Human behaviour may not abide by scientific rules, and therefore results and theories may be based off inferences rather than empirical data, making it unreliable.
Nisbett and Wilson (1977) argued that we have little knowledge of the causes of, and processes underlying our behaviour. People may have implicit views, which cannot be picked up by introspection.
Introspection is still useful in scientific psychology
Csikszentmihalyi and Hunter (2003) used introspection to make 'happiness' a measurable phenomenon. This means introspection may be used to improve people's quality of life.