One reason that replication of research results is so difficult in psychology may be that, as reversal theory research has shown, people are changing frequently between different motivational states, and the individual may be in different states even in the same situation at different times.
For example, a participant in a study may experience the study as a kind of game on one occasion (in the paratelic state) and as serious on another (in the telic state).Laboratory atmosphere, which is difficult to measure, may play a part in this. Likewise for the other three pairs of motivational states posited in the theory.
This is the kind of subtle-seeming difference that might easily throw off results. It might also explain some experimental replications of classic experiments that produce different results from the originals.
If the motivational state is indeed an independent variable that is likely to affect the outcome, then it needs to be controlled. But subjects cannot be used as their own controls if they might be in different motivational states when data collection is repeated. And creating matched groups does not guarantee that these groups will still be matched when data collection is carried out, especially if significant time passes from matching to performance.
At the very least, motivational state ought to be measured before and after performance. Fortunately reversal theory state measures are available for doing exactly that.