Bias
Sources of error
Validity
Reliability
Corroboration
Confirmation Bias
Cherry Picking
Bias is where there is is a tendency towards a certain outcome. For example, if you complete all of one type of sediment first in your liquefaction investigation, your procedure may change slightly in later measurements, which could favour a particular type of sediment. Human judgement is always subject to bias, and you need to try and eliminate all sources of bias from your investigation.
Some
times sources of error, such as inaccurate measurements can affect your results. A good example of this is parralax error, shown in the diagram.
Validity is a development on the concept of a "fair test". For an investigation to be valid, you must be sure that the conclusion is correct. Evidence for this includes a good trend line, control of variables, and repeatable results. If you conclusion is valid, you are certain that the independent variable is (or isn't) affecting the dependent variable (or not). However, you can never be absolutely certain in science, as you cannot test a theory in every possible situation.
If your test is reliable, it means that someone else will be able to follow your method and get similar results. Evidence for a reliable investigation is included in a good method, as well as repeating measurements more than once to show that they are similar.
If you are unsure of how reliable a source of information is, you should find a more reliable source of information and see if it says the same thing.
Only accepting evidence that supports your existing belief.
Misrepresenting data by only referring to the parts which support your argument.
An interesting article about making a test fair.
http://undsci.berkeley.edu/article/fair_tests_01