What are "Valid" Measures?

​

They are very important, that's what they are :)  Saying that a measurement is valid means that it is measuring what you think it is.  When you get on the scale and it tells you your weight, you expect it to be correct.  In other words, valid.  When you take your temperature, you also expect it to be correct--or valid.  The same goes for other types of measures, say an instrument on anxiety. You expect the instrument to be valid and give you a correct "reading" of a person's anxiety level. Typically, that type of instrument is a questionnaire. It might ask about the frequency, duration, and severity of symptoms ("No matter what I do I cannot get my mind off of my problems." Answer Choices: Always, Sometimes, Never, Rarely).  Notice that I am calling it an "instrument". That's because it is a tool that measures something.  The format is a questionnaire or survey; but the actual mechanism is an instrument. It's measuring. 

​

How do we know that these types of instruments are valid?  How do we calibrate them? There is a whole field of psychometrics that addresses these questions and has developed some handy ways to do it. I will not go into that detail here (here's a good book to check out if you are interested: Measuring Humans). Instead, let me show you how the types of questions we chose can influence validity and then give you some advice on how to find good valid measures. 

​

A quick side note: you will probably hear people talk about reliability and validity in tandem. That's because they go together. A valid measure will be reliable--it will give you similar results each time you use it (just like the thermometer) unless of course something has happened to change things (you took some Tylenol). 

​

I will keep with my example of anxiety.  Let's reconsider the question "No matter what I do I cannot get my mind off of my problems." Certainly this could be a symptom of anxiety, but it is not the only one.  This means that we would need additional questions on the instrument to measure anxiety. How about "I find it difficult to relax."  Sure.  This also could be a symptom of anxiety.  The answer scale gives us some sense of validity as well.  The scale I mentioned above helps us see how often the individual feels this way. If, someone is not able to stop thinking about their problems all the time AND they always find it difficult to relax, we might start thinking that they are in fact experiencing clinical anxiety.  Together, multiple questions with a good rating scale give us a more accurate picture.  Using just one of these questions, or a yes/no scale, would not help us as much and would likely result in an in-valid instrument. For example, if you ask "I find it difficult to relax" and someone says "yes", well, why is that?  Is it a chronic issue or is it due to a difficult day?  The language we use could also be misleading.  For example, how might you interpret the questions "I find it difficult to relax" vs. "I find it difficult to calm down".  Are they implying different things?  To find out we would have to test them out on a large group of individuals. This is the process psychometricians are engaged in to try to discern the best combination of questions and respective answer scales to use.  It is a process, a long one, to develop a strong valid instrument.  This means that it is not a good idea to create your own questions for research.  Not unless you are going to go through the full development process to determine validity.  There are many options out there to find good valid methods.  Here are some places to look:

1.  Research articles.  Researchers have likely already studied the constructs you are interested in.  What instruments did they use?

2. PsychTests. This is a database available through a University library that gives you all types of instruments. Check it out!

3. Research Methods text books. They use examples on almost every page.  Look in the index for your key words and see if you can find something.

4. Ask others.  If you are surrounded by other students and faculty, they may have seen something you are looking for, or at least and article that might get you on the right track. 

​

It takes a little digging, but it is much shorter a process than developing your own valid instrument from scratch, and it will ensure the best quality results (at least in terms of measurement error!)

​

​