Imperfect reliability and validity of research tools is a big reason that different studies examining the same research question come up with different conclusions. Some measurement procedures are just more reliable and valid than others, and for some phenomena there is no 100% reliable/valid measurement tool, so inferences/extrapolations must be made from the data available.

Reliability refers to the repeatability or consistency of a measure. If a variable is measured more than once using the same tool, will the results be the same? For example, if you are measuring vitamin C content of a urine sample, do repeated measures using the same sample yield the same vitamin C content? If so, the measurement technique is reliable.

To determine if a measure is reliable, you can compare your measurement technique to a "gold standard" or repeat the measure to determine if the same results are obtained. And, if more than one person is performing the measurement, make sure they obtain the same results with the same measure.

Validity refers to measuring what you thing you are measuring. For example, is your nutrition knowledge test a true indicator of nutrition knowledge? If you ask questions about car engines only, then your nutrition knowledge test is certainly not a valid indicator of true nutrition knowledge.

Types of validity include:

FACE VALIDITY: Does the measurement instrument seem relevant to the study to the study subjects?

CONTENT VALIDITY: Does the measurement instrument a good way to assess according to experts in the field?

CRITERION VALIDITY: Does your measuring instrument predict or agree with parallel indicators of the same phenomenon?

CONSTRUCT VALIDITY: Do two tests of the same measure agree?

A study can be reliable (repeatable) without being valid. When you are evaluating a research article, it is important to evaluate the reliability/validity of the measurement tools. Hopefully, the authors of the article address these issues in the article text.