In psychology, it is the role of researchers to study a range of behaviors and in order to do this, these behaviors must be measured. Sometimes these behaviors are overt and so can be measured by simply observing an individual and sometimes these behaviors are covert behaviors, which cannot be measured directly as they cannot be seen (e.g. attitudes, self-esteem, intelligence, etc).
Hypothetical constructs are variables which cannot be seen and so can’t be measured directly and so for us to be able to measure them, operational definitions are used. Many of these include physiological measures such as measuring heart rate or sweating as an indicator of something such as sexual arousal, fear,etc. However, these are not direct measures of the concept being studied and so may not actually be valid. As these are not direct measures, we may also be looking at a correlation rather than causation, meaning we cannot make conclusions about the relationships between the variables being observed.
The Beck depression inventory is an example of how an operational definition is used to measure a hypothetical construct. The cognitive and emotional areas of depression are being measured and these are taken as a measure of depression as a whole.
Behavioral psychologists argue that covert behaviors should not be studied as if they are not directly observable and therefore measurable then we can never be certain. However, many other areas of psychology rely on the study of covert behaviors. For example, in clinical psychology, If trying to diagnose a mental illness we may rely on how an individual report feelings and emotions so that we can interpret this, though this is a covert behavior.
Although we may think we are capable of measuring most things, whether that be through a direct or indirect measure, problems with measurement will always be an issue, as we may not be measuring what we claim to or may instead be measuring a variety of things. These factors should be considered, as often the concept being measured is an oversimplification of the variable being studied.
laurenpsychology 
I agree that is complicated to measure covert behaviours, and it should not be assumed that by measuring operational definitions, we are actually measuring these covert behaviours.
You mention the Beck depression inventory, and im going to continue with this, explaining how this works and why it is not an entirely valid method.
The BDI is one of the most widely used instruments for measuring the severity of depression. It is a 21 multiple choice self inventory, used for those 13years and older.
The first problem with using the Beck depression inventory, is that it is self scored. The individuals scores can easily be exaggerated or minimised, depending on the outcome the individual is hoping for. Another problem is the individuals mood. Although looking for depressive symptoms, someone may just be having a ‘bad’ or ‘off’ day when they complete the questionnaire, which would influence the outcome significantly. The BDI was developed as an assessment of the intensity of the depression, yet today Is widely used as a diagnostic tool, for efficiency. But there is no evidence that this is more valid or reliable than any other depression scale.
With all this information, we can see that such covert methods aren’t reliable. Although it measures cognitive and emotional areas, it does not truly measure depression, meaning it is not entirely valid. And this is the biggest problem with using covert measures. They do not truly measure what you are looking for, and therefore, how can you be sure, that the results are due to covert behaviour?
This post is very informative, and you explain well, the basics of measuring variables in Psychology. You could have expanded slightly on the explanation of an operational definition by saying it defines something in terms of a process or a set of tests, rather than just giving an example. I also found the argument about the covert behaviors interesting. Overall a very good read covering just the right amount of information 🙂
I agree also that not everything can be measured no matter how much work work is done. There will always be the issue of achieving ethical approval, informal consent on behalf of the participants who you may wish to obtain and also will the equipment have an effect either mentally or physically?
For example when carrying out a study on clients with clinical depression, if you simply cannot the ethical guidelines produced by the BPS then it cannot be measured.
Do you think it is possible to measure someone’s thoughts? or do we simply have to rely on Freud’s theory of dream analysis. I think overall that not everything can be measured, if everything could be measured and studied then things would be a lot clearer in today’s society.
Your examples and ideas on the topic were really interesting and related well to psychology. You could have further discussed how using statistics and manual diagnoses for mental illnesses can be harmful as well as good; for example a criteria of anorexia is that the person must be 15% below normal weight for them so we are led to believe that 100% of anorexics have this and anyone who doesn’t is not anorexic. This could lead to problems as if we followed this directly it prevents someone who is 14% and has all the other symptoms from getting the treatment they clearly need. Despite this we need somewhere to draw the line, we can’t have different doctors diagnosing the same things differently and diagnoses need as clear boundaries as we can set. It’s a complicated argument.
Also in psychology when we are looking at thoughts and behaviours it is pretty impossible to measure them, it could be argued that quantitative ways such as recording how many times a person felt sad or engaged in social behaviour do not allow a true picture or enough explanation.
One thing you failed to mention when discussing operational definition was convergent validity. This is the degree to which one measurable behaviour is similar to another behaviour. The scores of one test should correlate with scores from another, this can be used to strengthen decision to measure one variable as a representation or operation definition of the actual desired variable. The higher the convergent validity the more confidence we have in measuring a variable that it is theoretically related to. For example, when measuring aggression, an example of high convergent validity would show a correlation between self report aggression and heart rate measurement (just an example not a fact).
i agree fully with most of your points but you say that covert behaviours cannnot be asured gdirecty as thaey are deep rooted within us …. this may be true but i have in thinking surely by operationlising these internal behaviours you can envisage them externally for example if an inividuals feels lonely they may sit by them selves or feel alone in a crowded rooom but you woud only be able to tell this from oprtationalizing the subsquent behaiou5ts in order to not mistake them for a shy person . xoxo
Your blog topic was very appropriate :). I’d like to add that specifically when measuring intelligence the defintion of what you call intelligence is very questionable. An amazing boxer whose been banged in the head loads from fighting won’t seem very intelligent but his reflexes and skills are definetly intelligent. It goes the same with football players and the way they move around the pitch and coordinate such amazing passes or just have amazing game vision. That sort of intelligence is extremely difficult to measure and it’s interesting that reaction tests are mostly done to the pressing of buttons. Intelligence is generally measured as IQ and there are cases of people with high iqs who are underachievers. How would you measure the ability someone has to achieve what they want? In a biological sense it would be the ability to dominate one’s environment which isn’t very measurable.
I think you made an excellent point regarding correlation not meaning causation. However, it doesn’t mean the information we get from correlation is not valid, it just needs more investigation. The fact is, mental illness does exist, certainly to the extent that people claim to be mentally unwell. This doesn’t mean we should abandon investigation, diagnosis and treatment because we are so reliant in these cases on peoples self report and ensuing validity issues. Just because a doctor could not see the tumor in a cancer patient 150 years ago didn’t mean he didn’t believe his patient when they said they were unwell.
Inference is uncertain and therefore can be dangerous with regards to treatment however some action and recognition is better than none. If we took the behavioural approach and refused to entertain anything that is not currently directly observable not only are we ignoring people, the object of our study, but we are closing the door on any possible progress in how we study hypothetical constructs. Constructs we can observe now were not so a hundered years ago (now with EEG, fMRI etc). This is likely to be the case now. We should keep an open mind and therefore maintain the idea that yes, we can study anything.
Yes I agree operational definitions can lack validity, by not really measuring what it is meant to be measuring. For example, in my last semester’s blog I discussed Adams et al (1999) study in which physiological measures were used to measure sexual arousal in heterosexual males after watching a mixture of pornography, including gay pornography. What is to say that they were really measuring anxiety instead of sexual arousal?
One way that psychologists try to overcome this problem of operational definitions is by using more than one method of measuring a variable. So in the case of the Adams et al study, the researchers could have given participants a questionnaire so that they could rate their own sexual arousal and then compare the physiological and self-reported measure. Would this be any more reliable, as the questionnaire could just be reflecting the socially desirable answer?
Also, if a psychologist is able to show that a new measurement of a hypothetical construct produces similar scores to an older method, that has been supported numerous times, then it is said to have concurrent validity. Of course you then have to question whether the older method actual measures what it’s suppose to?
i have always struggled with the idea of hypothetical constructs, and how can someone make one up if if there is no previous construct to work from. so how can we be sure they are valid? ive always thought the only real way of measuring something is if it has a physical way of making itself apparenet like making a participants heart go faster when they feel certain emotions because the heart beats can be measured.