Why is reliabililty important?

Reliability is fundemental in the world of statistics, during this blog i hope to go some way to explaing why.

Relaibility is the consistency of a set of measurements and is often used to describe a test. Reliability is often confused with vailidity and i must stress that it does not imply valid research. A reliable measure is measuring something consistently, but you may not be measuring what you intend to measure. For example, if an individual, who weighed 150lbs, was to step onto some bathroom scales and took 5 readings which all came out at 125, then the scale is reliable but not vaild. To be reliable and valid the scale would of had to produce 5 readings all coming out at 150lbs.  

What’s more, one of the key questions that people want answering when research is release is, “is this research reliable?” And why not? It is reasonable for people to expect reliable data. After all, if the research is reliable then the results should be substantially similar. When research suffers from poor reliability, any consequent theories and results will be widely discredited. One of the main reasons for this is due to reliabilitys threat to vailidity, most importantly internal validity (IV); without internal validity you cannot have external validity (EV). A poorly executed study and badly recorded results (IV) are not applicable to any population (EV) and therefore cannot be generalised.

Furthermore, we learnt this week how outliers can occur from errors in taking measurements, such errors can lead to poor reliability and the consequent outliers. Such discrepancies in data can royally screw up your results if not handled with correctly. For example, outliers can greatly affect the mean of your results and can unfortunately lead to inaccurate assumptions and an incorrect anaylsis of the data. However, please note that reliability is not the sole cause of outliers, its just it can merely be contributing factor.

I believe it is fair to say that reliability is essential in everyday life but its importance in the world of research and statistics could not be stressed more. For research to be reliable, scientific researchers need to take great care, ensuring they maintain a high level of reliability in their work at all times. It is evident that poor reliability can greatly hinder and individual’s research, in more than one way.

Advertisements

10 responses to this post.

  1. You’ve written a good description and discussion of the topic and seem to have grasped it well. Do you think reliability is more or less important than validity? I think if you were to write this again then you could expand more on reliability itself rather than discussing it alongside validity. x

    Reply

  2. I agree that reliability is important because if research is reliable it can be replicated by other researchers, due to the controlled conditions, and similar findings should be found. If similar findings are found in many replicated versions of the research then the results can be generalised to the target population with more supporting evidence. However, I think that you could have gone into more depth and explained a couple different kinds of reliability, such as inter-rater reliability which is where there is a high degree if agreement between researchers. For example, when studying aggressive behaviour displayed in children more than one researcher is used to observe behaviour to determine what behaviour constitutes aggression.

    Also, you mentioned validity – is reliability more important than validity? Or are they equally important? It’s good that research can be replicated, that measurement tools are consistant, such as IQ tests, but what use is this if the study and the tools used are not really measuring what it is claiming to measure? For instance, in the past some have argued that research conducted, in laboratory settings, into eyewitness testimony (e.g. Loftus and Palmer 1974) is criticised because although it has high reliability (it takes place in an controlled condition), it lacks validity where the situation is artificial and the participants know that their answers will not have any detrimental effects on anyone. Therefore is it an accurate measure of eyewitness memory? The reason I mention this is because it would have been nice to see you thoughts on whether researchers should give up some reliability to maintain validity and maybe an example.

    (I enjoyed reading your blog as it started me thinking about reliability and validity, and how difficult it can be to balance both.)

    Reply

  3. […] https://blc25.wordpress.com/2011/10/14/why-is-reliabililty-important/ (comment awaiting moderation; 22:08 14/10/11) Advertisement Eco World Content From Across The Internet. Featured on EcoPressed A cheat sheet of peer-to-peer car sharing players Share this:TwitterFacebookLike this:LikeBe the first to like this post. comments Comments TA ← How aware should we be of participant reactivity in our research? […]

    Reply

  4. I do agree with you ,that realability is etreamly important when carrying out the reserch study. Like you said if the findings are not consistent over the time , that means there has to be some kind of error occuring. But would be the measurment error or other confunding variables. Either of this it threets internal validty of the study. This means that there is no control over variables . If there is no control over variables them there is no point of carrying out the study , as the results could be affected by confunding variable. If the study is affected by confunding variable them the resercher is not answearing the question that he asked and aimed to measure. Therefore he can conclude his null hypothesis are true. If the reserch has got low internal validty them it is hard to genaralize findings outside the experiment setting, as not representive finisgs cannot be generalized to general public and situations. Therefore in order to maintain validity , relability has to be maintained too.

    I also agree with previous comment stating that , there are different assesments of relability that you could have explained a bit more. I will give another example; test retest relability- is used when the questionaire is assed on its consistency , a ”certan time” later. If the teacher carry out the test them it can see if its realiable week later, But here come the stairs. The scores could not be the same , not because of the results being not consitent but the confunidng variables. Some people could be in different, mood, setting, ability to carry out the test. Them the test retest is not really measuring relability of the exam but the confunidng variables. Like you can see there are some weakness of the importance of realability too. Another example includes the size sample. If aimed to measure construct on a large sample them the results would be diffuclt of being assed week later due to drops out etc. (especially if one of the study ethical principles is to maintain confidentaility and ,for ex; questionaire cannot be assed week later). Therefore due to the sort of the reasons ,the findings (which could be perfect) would be affected.

    I like your point that you made about outliers, hovewer if the resercher is clever enough to carry out the statistcal test them , it will look for outliers, and will eliminate them too. But yes you right if the outliers would not be noticed they would greatly affect the relability.
    Overall , some of the points that you made , made me think and realise stuff about relability.

    Reply

  5. I think you’ve written a good post on reliability, and that you’ve mentioned the connection between validity and reliability. I think you could have mentioned some research studies that could have problems with reliability such as people with degenerative diseases or when using qualitative research methods.

    Reply

  6. Great blog.

    When it comes to reliability, I always ask my self this – when do I know whether a research result is reliable or not, I mean where do we draw the line of reliability?

    Liam Healy pointed out that “Some argue that an acceptable reliability coefficient is 0.70” whereas “short-term equivalence (two week test-retest) between alternate-forms of the same measure should be in the range 0.65 – 0.80, the same-form retest should be in the range 0.75 to 0.90”. 🙂

    You have briefly mentioned validity here, I would like to expand on this point a little more, the two ( reliability and validity ) in a way are linked together, they influence on each other. You probably know this by now but I just want to state it again as you have not brought this up, before a researcher can conclude their method/procedure is valid, the data has to be reliable first, but bear in mind reliability do not cause validity, vice versa. Without reliability you cannot state that the research is valid, leads to that the results may not look as good and polished. So it is hard accept the fact that if your results are unreliable, nobody would believe your data and research findings at all, and whether it is valid or not.

    Thank you !

    Reply

  7. […] 2) https://blc25.wordpress.com/2011/10/14/why-is-reliabililty-important/#comment-57 […]

    Reply

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: