It is common to see research papers that are locked in to an experimental design from very early stages - even when it becomes apparent that the results will not answer the questions that end users would want to know.
This is largely due to the academic approval process witch require committees of committees and where the researcher must be made to feel suitably indulged that his request for approval is being considered at all - let alone in a timely manner.
As a result most papers follow their initial concrete path to a conclusion that could have been far more useful if some flexibility were shown along the way.
Another reason that the path is followed is that the researcher mostly just wants it all to end - eventually and even if a much more useful thesis were discovered in the process of the research - nobody working under the conditions of most graduate students and many post-docs would ever derail what they have completed by suggesting it be changed partway through. (Not to say that many professors would be only to happy to keep their minions toiling away for a few extra months or years - but even they get their rankings based on numbers - since quality is anybody's guess.)