BRIEFLY DESCRIBE THE CRITERIA THAT YOU WOULD USE TO DECIDE IF A
PUBLISHED RESEARCH PAPER WAS SCIENTIFICALLY SOUND.

To determine if a research paper is scientifically sound, it must
satisfy the following criteria:

Thesis
The paper must have an idea that it is trying to prove.

Hypothesis
A prediction of what the results of the experiment will be.

Manipulated Variable
There must be some sort of variable that is pulled through a range to
test or prove the thesis.

Responding Variable
Basically the results of the experiment for each value of the
manipulated variable.

Controlled Variable
All other variables (besides Manipulated Variable) that must be kept
constant to retain the validity of an experiment.

Anecdotal Evidence
A proper research paper should NOT have anecdotal evidence (ex. "I
have a friend who told me that water expands when it gets hotter")

Experimental Error
Error must be clearly defined.  EVERY experiment has some degree of
error.

Sample Size
The number of times the experiment is repeated, for a scientific
research paper, this number should be fairly high for accurate
results.

Observations
Qualitative and Quantitative observations made during the experiment. 
These must be clear, concise but must not contain interpretations of
data.  Everything must be recorded straight from the experiment

Bias
A research paper should not have bias.  Everything must be
scientifically proven.

Conclusion
Every research paper should finish with a conclusion, declaring
whether the thesis is true or false.

If the research paper satisfies the above criteria, then it is most
likely scientifically sound.

Hope that helps.

-Tox-ga

As a PhD researcher, the one method for determining if the research is
sound is by it's ability to be replicated.  If no other lab, other
than your own can perform the experiment, most scientists will have
doubts about the research protocol and the research results.

Research might look well done, when a publication of that research is
well written.  However, since it is possible to fabricate data, as
shown repeatedly today, in various research areas, the quality of any
research publication, must be determined by the ability of all capable
investigators to replicate the protocol and the research results.