How to ensure validity and reliability in research?

classicdame, MSN, EdD · Aug 2, 2012

good textbook or instructor question. Your instructor may feel strongly about one method over another, so tell him/her "I am thinking of doing this because---" then see what reaction you get.

Double-Helix, BSN, RN · Aug 2, 2012

In a retrospective study, you're looking at data previously reported or collected. Therefore, in order for your research to be accurate, you need to make sure the studies you are examining are reliable.

What qualities/information will you look for in these studies to ensure that they are valid/reliable before you put their data into your research?

RNbethy · Aug 2, 2012

I'm currently conducting a research specifically a retrospective study. However I'm not using questionnaire or interview as my instrument to gather data. But I gather data through charts review of the patients. How will I measure or ensure the validity and reliability of my research?

To answer your question, I need more information. What is your research question and what is it that you are actually trying to measure?

Validity means that you are actually measuring what you intend to measure. For example, if you wanted to measure stress and used blood pressure as a measure of stress, the validity of this measure would depend on how much BP is KNOWN to be related to stress. There are different kinds of validity - content validity, construct validity, and so on. It's best to include a few different kinds of validity to demonstrate validity.

Reliability means that you consistently measure the same thing. For example, if you are measuring blood pressure, is the machine consistently correct? Or is it calibrated one moment and uncalibrated the next? Or were several BP machines used with different calibrations...? There are also different kinds of reliability... and most researchers use several measures of reliability.

Usually, it's best to use a measurement that has already been proven to be reliable and valid... This will add strength to your study. Search the literature to find if others have used the measurement and if the tool is valid and reliable. The higher the coefficients, the better... Don't quote me on this, but > 0.7 is approximately the minimum requirement; > 0.9 is ideal...

nurseprnRN, BSN, RN · Aug 3, 2012

i don't see how you can do any kind of study without a systematic data collection tool. otherwise you'll find yourself forgetting to write some things down, and then writing down other things from another source, and then suddenly realized that one of those past charts probably had that information too... this way lies madness. trust me, you will not remember all your data.

make a data collection tool that includes everything you want to know, plus demographics (because these often become confounding variables later, and you don't want to go back and check them all again). that way you can help to keep your data collection consistent, and consistent is the key to reliable.

if you are doing a meta-analysis, that is, combining data from other studies to do one big one, your faculty can help you figure out how to evaluate the studies to include. many a meta-analysis starts out with three hundred possible studies, and whittles them down to twenty -- or fewer-- when they look at comparability and validity.

all that said, i'm not sure what kind of "research" you're doing. sometimes people confuse research with data collection. sample size is a big factor in research, too, because you have to have a big enough sample to have your conclusions be valid to have applicability. can you give us a clue?

RNbethy · Aug 4, 2012

I think it's important to distinguish between a retrospective study and a systematic review as there seems to be a question about what the OP is actually doing here....

A systematic review compiles results from lots of studies. Generally, a systematic review looks at the best available evidence... Retrospective studies (which are considerably lower in terms of rigor) are not generally included here unless they are the only type of research available on the subject. Generally, a retrospective study measures an effect (i.e. lung disease) and looks at a variable (not an entire study) that occurred previously (i.e. smoking, exposure to asbestos or whatever) to determine correlation. Sometimes this may involve reviewing variables in data of large studies that occurred previously... but a retrospective study is not a systematic review.