+ Reply to Thread
Results 1 to 14 of 14

Thread: Likert Scale Survey_regression analysis

  1. #1
    Points: 52, Level: 1
    Level completed: 4%, Points required for next Level: 48

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Likert Scale Survey_regression analysis




    Dear all,

    I am new in this, so please help - I am conducting a survey for my thesis using Likert method. Through this research survey I would like to attain following things:

    - Varaibles (predictors) that affect students dropout behavior, and draw some conclusion.

    I am not sure what kind of regression model be used, did some research and found few options:

    1. Linear regression (ANOVA)
    2. Logistic regression, and
    3. Descriptive statistics

    Could any of you guide me through, please?

    Have a good day ahead

  2. #2
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: Likert Scale Survey_regression analysis

    You have to decide what form your DV will take before anyone can help (or you need to let us know what it is anyhow). Is the thing you are trying to predict measured on interval, ordinal, or nominal levels?

    Different methods require different measurements so if poster's don't know that they can not answer your question.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  3. #3
    Points: 52, Level: 1
    Level completed: 4%, Points required for next Level: 48

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Likert Scale Survey_regression analysis

    Thanks for your reply Noetsi.

    Fair enough, I will try. Just an example: I am trying to understand how students’ integrate in their academic environment ( dependent variable or predictor), there are many aspects (several independent variables here) to look at to understand students’ academic integration behavior. Through survey questions I am trying to extract their perception/view, and this is why I have used Liker items.
    Now, Likert items are like strongly disagree, disagree, undecided, agree and strongly agree – the difference between each of these items is unknown - this would be ORDINAL. However, if I code them like strongly disagree=1, disagree=2, undecided=3, agree=4 and strongly disagree=5 then differences between each item stays same – perhaps this would be INTERVAL scale.

    May I ask which kind of regression model would be a good fit for above, and why?

    Thanks in advance

  4. #4
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: Likert Scale Survey_regression analysis

    What is critical is how your DV is measured not your predictors. That said, there is a huge dispute in the literature if likert can be treated as ordinal or interval [assuming they can be ordered and normally they can be]. Formally likert is ordinal, but various authors argue that you can treat it like interval if you can make the following assumption: The distance between each level of your scale [say between 1 and 2 as compared to 2 and 3] is the same. That is a judgment call, there is no way IMHO you can ever know this. Note that technically interval data requires many levels which likert will never meet, but some argue that as long as you have five distinct levels at least this is not a major issue [12 levels is a lot better, but you will rarely if ever see that many meaningful levels in likert data].

    If the dependent variable is likert and you can make this assumption than formally you can use linear methods like linear regression. If you consider this assumption not to be met than the DV is ordinal and you have to use something like logistic regression [you can not use ANOVA which assumes interval DV].

    Again it matters what your dependent variable is. Period. Not your predictors.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  5. #5
    Points: 52, Level: 1
    Level completed: 4%, Points required for next Level: 48

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Likert Scale Survey_regression analysis

    Thanks again - your explanation is sensible. Can one do the following:

    I have already collected some data ( idenpendent variables) with Likert like responses i.e. strongly disagree=1, disagree=2, undecided=3, agree=4 and strongly disagree=5.

    Can I use logistical regression ? How would I know that my DV (=academic integration) outcome is within 0 (not-integrated), 1( integrated)?

    I think I am missing something in here!
    Last edited by Monjurul; 01-26-2016 at 09:08 AM.

  6. #6
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: Likert Scale Survey_regression analysis

    I think you are missing that it does not matter what form the predictors are, likert is fine. It only matters what form the DV is.

    I don't understand what you mean by this:
    How would I know that my DV (=academic integration) outcome is within 0 (not-integrated), 1( integrated)?
    If you DV is likert or 2 levels (1 integrated 2 non-integrated) then you can use logistic regression. If it has two distinct levels you use regular logistic regression. If it has 3 or more levels and is ordinal (as likert certainly is) you can use ordered logistic regression.

    If it has 5 or more levels and the distance between the levels is similar you probably could get away with doing linear regression although there are disputes on this...
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  7. #7
    Points: 52, Level: 1
    Level completed: 4%, Points required for next Level: 48

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Likert Scale Survey_regression analysis

    I think I have made a mistake on my survey - I have collected data only on indepdent variables using Likert scale!

    I wasn't sure that I needed to collect reponses on DV as well! Or do I convert one of the parameter using any sort of equation?

  8. #8
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: Likert Scale Survey_regression analysis

    If you don't measure your DV how can you run any statistic? I think I am missing something here.

    You generate parameters you don't convert them. And you can't generate any parameter without a measured DV.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  9. #9
    Points: 52, Level: 1
    Level completed: 4%, Points required for next Level: 48

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Likert Scale Survey_regression analysis

    I can not ask students directly whether they are academically integrated into the university environment , and ask them to answer in Likert form.

    I was hoping, I could calclulate or establish a relationship from Indep. variables.

  10. #10
    Points: 52, Level: 1
    Level completed: 4%, Points required for next Level: 48

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Likert Scale Survey_regression analysis

    I think I have messed up the entire thing!

  11. #11
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: Likert Scale Survey_regression analysis

    If you have no independently measured DV you can not run any statistic. Well you can, but it won't correspond to reality in the real world except by blind chance. But you don't necessarily have planned it as a DV. For example we ran a series of motivation likert scale questions. By chance one was on overall motivation. So we were able to use that as the DV, with other forms of motivation being predictors of that overall value.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  12. #12
    TS Contributor
    Points: 22,339, Level: 92
    Level completed: 99%, Points required for next Level: 11
    spunky's Avatar
    Location
    vancouver, canada
    Posts
    2,135
    Thanks
    166
    Thanked 537 Times in 431 Posts

    Re: Likert Scale Survey_regression analysis

    Quote Originally Posted by Monjurul View Post
    I think I have messed up the entire thing!
    without a DV? yeah, it kind of looks like it.
    for all your psychometric needs! https://psychometroscar.wordpress.com/about/

  13. #13
    Points: 52, Level: 1
    Level completed: 4%, Points required for next Level: 48

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Likert Scale Survey_regression analysis

    Time out! I think I did things right, perhaps my way of thinking got mixed up.

    I have attached a XLSX file, just to provide some insights - I hope, you can guide me from here on.

    I look forward to your feedback

    Have a good day ahead
    Attached Thumbnails Attached Thumbnails Click image for larger version

Name:	How_it_Looks.jpg‎
Views:	9
Size:	10.1 KB
ID:	5769  
    Attached Files
    Last edited by Monjurul; 01-27-2016 at 07:58 AM.

  14. #14
    Points: 52, Level: 1
    Level completed: 4%, Points required for next Level: 48

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Likert Scale Survey_regression analysis


    Is my understanding of the concept is totally wrong?

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats