Validity is the idea of results accurately reflecting the real world. Reliability is the idea that results can be consistently repeated. These are the two things are generally both important in the conclusion of an experiment.
There can often be a misunderstanding with the two about how they work together, but reliability can be present without valid, such as if an experiment repeatedly gives the same results which do not reflect the real world then the methods used would be reliable but not valid. There is often a correlation between the two but they do exist separately.
Reliability can be more important in research, because if another researcher was to repeat an experiment and achieve the same results as the original researcher then it appears that they are doing something right. Despite this validity is still important in research because it shows the truth of what is being studied in comparison to the truth.
psuef2 
Validity is when you measure what you were supposed to measure in an experiment and ecological validity is the extent to which results can be generalised to the real world.
Without validity, a piece of research is worthless as the results from the experiment will be false. Reliability is equally important because if the study is repeated and the results are significantly different, then there must be some other variable affecting the experiment which can be identified and controlled. When a piece of research is found to be valid and has been repeated a few times and the same result found then the findings can be trusted and applied to everyday life.
A piece of research is not worthless without validity. With studies lacking in internal validity, it simply means that findings do not represent what has been tested, but these are still findings and can be equally important to the original aim. With studies lacking in external validities (ecological, cultural etc), the issue lies in how the results can be generalised – possible not the such a high population as intended, but the results can still be used on a more niche group.
Whilst reliability is important to studies, enabling others to support or criticise methods and findings, and providing some reassurance that the results were not anomalies, studies lacking in reliability can be important. A study which is methodically unreliable is very questionable, as the results cannot simply be otherwise used, such as with validity issues, because conclusions cannot be applied to a population if they cannot be replicated and thus supported.
Reliability and validity are both equally important if the research conducted is to be considered scientific. If research is not accepted by the scientific community the findings cannot be generalised to the target population. This is because the researcher cannot be sure that the results are due directly to the manipulation of the independent variable, rather than the impact of any extraneous variables; therefore cause and effect cannot be established. In order to increase the reliability of research, the method needs to be unambiguous and clearly written; otherwise researchers who replicate the study may misinterpret aspects of the procedure and would consequently gather different results. Reliability can also be improved if research is conducted in a laboratory setting, as this allows researchers to minimise the potential bias introduced by extraneous variables. In addition, to achieve valid results the research findings must be relevant to the research hypothesis which the experimenter set out to test. The type of sample chosen can also affect validity; as larger samples have higher levels of external validity than smaller samples, because they are more likely to be representative of the target population.
Observations should be consistent; ideally multiple observers should produce the same recordings. The extent to which these observers agree is called inter-rater or inter-observer reliability. This is calculated by dividing total agreements by the total number of observations. A result of 0.80 or more suggests good inter-rater reliability. When it comes to self-report techniques such as questionnaires and interviews, there are two different types of reliability to consider. Internal reliability is a measure of how consistent something is within itself, for example, each questionnaire question should be measuring the same thing. External reliability is a measure of consistency over several different occasions. For example, the results gained for one patient interview one week should be the same as the results gained the next week by a different interviewer. One method of testing reliability is the test-retest method where a person is given a questionnaire or interview on one occasion; this is then repeated after an interval, such as a week or a month. If the measure is reliable the outcome should be the same every time.
There are two different types of validity, internal and external. Internal validity concerns what goes on inside a study, whether what was intended to be measured was measured. External validity or ecological validity concerns things outside of the study and the extent to which the results can be generalisable to other people and situations. The validity of self-report techniques also needs to be assessed. There are several ways to do this, including face validity and concurrent validity. Face validity looks at whether the test looks as if it is measuring what the researcher intended to measure. Concurrent validity can be established by comparing past and present tests that have been on the same topic.
Reliability can be present without validity, as can validity be present without reliability. However, the two cannot really increase without the other. Results cannot be consistently produced (reliable) if they do not represent real life and what the research is aiming to investigate (validity) I would argue that reliability and validity are equally important, and research which has not been developed to maximise both is problematic. To improve validity and reliability, research should be high in mundane realism and low in extraneous variables. Inter rater or inter observer reliability should be high, meaning that when two or more researchers conduct the same research, consistent results are obtained. This type of reliability also helps to reduce subjective tendencies in some research.
I agree that reliability and validity are both equally important but I disagree that they cannot increase without the other. Reliability can increase if researchers continue to produce the same results but these results will not necessarily be valid.
As you have stated, reliability can be seen without validity, where more then one experimenter gets the same results as the first meaning that the study becomes highly reliable. However, what you seem to be missing is the fact that if it is not valid, then what would be the point of having it be reliable? If it is not true to real life, then ultimately, it is not useful and should be taken with a pinch of salt, so what would be the point of having something being able to be repeated if it means absolutely nothing in the first place?
Danny
If the results are reliable then it would seem more generalisable but I do agree with you that if the results are completely invalid then they are ultimately useless, but they may have high reliability but only moderate validity then it is somewhat useful. Researchers may also repeat the experiments because they believe they could achieve results with higher validity.
Eldweeklypsych brought up a good point that reliability and validity are as important as each other and Kilsorrow brought up another good point as there would be no point in conducting reliable research if it were not valid. However with there being many different types of reliability and validity, is it possible to always be valid and reliable. Case studies are examples of studies which are valid as much detailed and in depth data has been collected, however they are not reliable as it is impossible to ethically recreate the situation, but the results still influence other research so although they are unreliable, they are still of great use.
You are correct in saying that validity is “the idea of results accurately reflecting the real world,” it refers to the legitimacy of an experiment or study, which results in whether their findings can be applied to other aspects of life, it is also about the extent to which a method or procedure measures what it is supposed to measure. Reliability refers to the consistency of a measuring instrument. Reliability can be tested in a number of ways. It can be measured in terms of whether two independent assessors give similar scores, known as inter-rater reliability, or whether the test is likely to produce the same results on two separate occasions, known as test-retest reliability.
In response to the comment about “Reliability can be more important in research” it should be pointed out that even without reliability (or even validity) experiments and studies still play an important role in debates and criticism and encouraging others to go and gather more information into the subject whether this be to support the research or disprove it.
it.