Likert Scale – Continuous Data

Six Sigma – iSixSigma Forums General Forums Methodology Likert Scale – Continuous Data

Viewing 16 posts - 1 through 16 (of 16 total)
• Author
Posts
• #53570

Doa
Participant

I did not find an answer within the discussion forum. For those industry, can or is Likert Scale data treated as continuous data? For example when running an Attribute Agreement Analysis with a 1-10 rating scale.

Marty

0
#190704

Severino
Participant

An attribute agreement analysis is meant for data which is non-continuous so there is no need for you to treat it as though it is continuous. If you are looking to perform other types of analysis, I recommend you evaluate your likert scale using procedures which focus on ranks rather than treating the values as if they’re continuous.

0
#190707

Brar
Participant

data on a scale for 1-10 can be treated as continous data. Even though there is always a debate but for all practical purposes you will get the same results as treating the data as discrete; however if the scale if 1-3 or 1-5; then be careful in treating the data as continous.

0
#190720

Doa
Participant

yes this seems to be the concensus, stick to 1-10.

0
#190722

Stanley
Participant

This will look like contrarianism but no Likert scale can be used as continuous data, at least in a pure mathematical sense. Reason: There is no assurance that a “10” (or a “2” or a “6”) is the same for everyone taking the survey. To be continuous, there would have to be a hard-and-fast definition for each score, fractional scores would have to be possible and the score would have to mean the same thing to everyone taking the test, much as a temperature of 32 degrees is the same no matter who does the measurement.

0
#190723

Doa
Participant

Contrarian or not…..this is why I use the site. Truly constructive discourse is how decisions are reached. You are correct, you thoughts were my first thoughts as well, which is what precipitated the post.

The temperature example points out my initial dilemma; yet in my specific case I have a series of data used to grade or rate the quality of printed graphics 1 = poor, 10= best. My thinking is that when the instructions are given to the appraisers, they are instructed that fractional ratings are not allowed.

It appears that in this case, whole numbers provide enough precision to perform my analysis.

Thanks for the input…how is business since Jeppesen was acquired by Boeing?

Marty

0
#190724

Stanley
Participant

Marty, we’re still here…. Another point I’d make on a Likert scale. Don’t remember where I read this but you’ll get better results with an odd number of points. When I’m using Likert scales, I like to keep the choices to 5 – reduces the amount of inner debate when answering the questions (Is this a 6 or a 7?). For the same reason, I’m trying to take yellow out of stoplight charts….
On the other hand, you’re probably right about the precision. I’d make the same compromise between precision and information gained, after all, the only purpose of any measurement is to decrease uncertainty….

0
#190725

Stanley
Participant

Marty, we’re still here…. Another point I’d make on a Likert scale. Don’t remember where I read this but you’ll get better results with an odd number of points. When I’m using Likert scales, I like to keep the choices to 5 – reduces the amount of inner debate when answering the questions (Is this a 6 or a 7?). For the same reason, I’m trying to take yellow out of stoplight charts….
On the other hand, you’re probably right about the precision. I’d make the same compromise between precision and information gained, after all, the only purpose of any measurement is to decrease uncertainty….

0
#190726

Doa
Participant

Stephen,

Good point regarding odd number scale…..first I’ve heard of this. I’ll keep it in mind.

Thanks….for the input.

Marty

0
#190728

Severino
Participant

[email protected] wrote:

This will look like contrarianism but no Likert scale can be used as continuous data, at least in a pure mathematical sense. Reason: There is no assurance that a “10” (or a “2” or a “6”) is the same for everyone taking the survey. To be continuous, there would have to be a hard-and-fast definition for each score, fractional scores would have to be possible and the score would have to mean the same thing to everyone taking the test, much as a temperature of 32 degrees is the same no matter who does the measurement.

From a pure mathematical sense the only data that is continuous have no units of measure. Any data from a measurement device cannot be continuous, but it can be good enough. There is no universal answer to the question of whether or not your data is close enough to continuous. It really depends on what you are trying to do with it. That is why we have things like tests for normality and gage R&Rs.

0
#190729

Rohatgi
Participant

Lickert Scale is always applied on Discrete variables. Not on Continuous

0
#248508

Nicola Brammer
Participant

You can always get several people to measure/assess according to your scale and average the results, this will increase the number of possible data points and make the data more ‘ continuous’.  Also you can then analyse the s.d. between the measurement repeats to make an assessment of the measurement system.

0
#248509

Robert Butler
Participant

@KatieBarry – is this resurrect old threads month?  This isn’t a complaint – just wondering.

0
#248510

Darth
Participant

@rbutler  Glad it isn’t just me.  People must be so bored that they are getting back onto isicksigma.com (LOL) and so desperate for something to do they are responding to decade old posts.  Next might be responses to test questions or dissertation topics from people who graduated a decade ago.  We might even see someone pop in to comment on the upcoming debate between Mikel Harry and Stan about the 1.5 shift.  Hope all is well with you Robert.

0
#248512

Katie Barry
Member

@rbutler and @darth — If you look at the newsletter, I’ve been sharing “Classic iSixSigma” discussions the past couple of weeks. This is one of them.

As for any others, one was spam that I deleted as soon as I saw it.

0
#248514

Robert Butler
Participant

@KatieBarry Ah Ha! So it IS resurrect old threads month!!!  :-)

I’m sorry to have to admit this but I don’t think I’ve ever checked the newsletter. Thanks for the heads up.

0
Viewing 16 posts - 1 through 16 (of 16 total)

You must be logged in to reply to this topic.