# Comparing unique items on a Likert scale

#### aknox001

##### New Member
Hello everybody. I conducted a survey using 3 measures (two IVs and a DV), all Likert. I found that nobody scored highly in the DV (which was burnout), so there's probably a bias of not wanting to admit to 'bad' things!
I would like to see if certain items on the DV have more of a 'skew' where people are more likely to respond in a certain way. How can I check this?

#### GretaGarbo

##### Human
I would just do a linear regression with the two IV:s as independent variables (explanatory variables) and the DV as a dependent variable. This would investigate if there is a shift in the mean of the dependent variabel when one IV changes. A lot of people in published papers do it this way.

But this is controversial. Some people say that the Likert scale is just an ordinal variable and not a ratio scale so that it is incorrect to do a regression on that. But my view is that it is OK.

#### aknox001

##### New Member
I would just do a linear regression with the two IV:s as independent variables (explanatory variables) and the DV as a dependent variable. This would investigate if there is a shift in the mean of the dependent variabel when one IV changes. A lot of people in published papers do it this way.

But this is controversial. Some people say that the Likert scale is just an ordinal variable and not a ratio scale so that it is incorrect to do a regression on that. But my view is that it is OK.
Thank you very much for your reply. My study involved regression analysis anyway. But I would still like to look at individual items and see if any have a greater likelihood of having a “not bad” response. Do you know how this could be done?

#### GretaGarbo

##### Human
I found that nobody scored highly in the DV (which was burnout), so there's probably a bias of not wanting to admit to 'bad' things!
Please note that in statistics "bias" means parameter estimate minus populations value. (bias = sample mean - population mean). So "bias" means systematic error.

(I am not English or American. What "bias" means in the every day English language I don't know. For me "bias" is a statistical or mathematical concept.) (It might surprice you that not everybody is American.)

If you have a sample of people, you must believe that they are a representative sample of the population, and that they response as the population would do (with a sampling error and all that).

But your measurement variable does not have to correspond to an underlying latent "true" feeling of burn out.

#### aknox001

##### New Member
Please note that in statistics "bias" means parameter estimate minus populations value. (bias = sample mean - population mean). So "bias" means systematic error.

(I am not English or American. What "bias" means in the every day English language I don't know. For me "bias" is a statistical or mathematical concept.) (It might surprice you that not everybody is American.)

If you have a sample of people, you must believe that they are a representative sample of the population, and that they response as the population would do (with a sampling error and all that).

But your measurement variable does not have to correspond to an underlying latent "true" feeling of burn out.
By bias I meant response bias, specifically social desirability bias. Apologies for not being specific. I am quite aware that not everyone is an American, thank you. I myself am not an American either, nor is English my mother tongue. I don’t see why your comment was necessary.

Thank you for your initial help followed by your condescending response.

Last edited:

#### GretaGarbo

##### Human
Thank you for your initial help followed by your condescending response.
I take it that the OP want to end this conversation.

Last edited:

#### ondansetron

##### TS Contributor
By bias I meant response bias, specifically social desirability bias. Apologies for not being specific. I am quite aware that not everyone is an American, thank you. I myself am not an American either, nor is English my mother tongue. I don’t see why your comment was necessary.

Thank you for your initial help followed by your condescending response.

Don't be too sensitive. You're asking for help, @GretaGarbo wasn't inappropriate to note the distinction.

#### aknox001

##### New Member

Don't be too sensitive. You're asking for help, @GretaGarbo wasn't inappropriate to note the distinction.
For which I was grateful and said thank you for. However, while I fully understand the point that was made and should have specified in the first place, I do not see the need for the sarcastic comment. It’s possible to raise a point without it. For what it’s worth, I do think this forum is fantastic.

#### ondansetron

##### TS Contributor
For which I was grateful and said thank you for. However, while I fully understand the point that was made and should have specified in the first place, I do not see the need for the sarcastic comment. It’s possible to raise a point without it. For what it’s worth, I do think this forum is fantastic.
I think we'll just have to write it off as miscommunication because I don't see a sarcastic comment. Text isn't a good medium to convey attitude and intentions.