problem in choosing Likert scale

[Roy FGH]

I want to apply a likert scale in my opinion research about marketing
science. but a got a little problem in choosing the scale. should I
choose a 4 Likert scale, or 5 Likert scale or maybe anyone can help
choosing another scale for me? And maybe anyone can give me the reason
or the recommended literature. Thanks for advancing.

[Art Kendall]

Bottom line: (even though I put it at the top).
Stick with conventional 5 point actual Likert response scales unless
there are reasons to do otherwise.

I don't have the literature cites here, but I am sure that some people
from the Leiden group follow this discussion list.

A few thoughts:

The number of points  on an agreement scale depends on balancing several
considerations.
On the one hand you want any variable to be measured at as high a level
as possible and as close to the underlying construct as possible in
terms of  meaning and measurement level.

On the other hand people can only make  "so many"  distinctions.  If
people are sophisticated in the subject matter,  they can usually make
more distinctions.
In an actual Likert scale there are several items that have 5 possible
responses. So with 5 items, summative scale scores can go from 5 to
25, which is certainly going to be close enough to interval.  If you
only have a few items in the scale and it is not difficult for your
respondents in your pretests, use a longer response scale.

 If you want to eliminate "neutral", go with a 6 point scale.

If the scale has been used in much previous research, go with what they did.

You can always collapse response categories afterwards.  You cannot do
the reverse.

I am sure others will be able to add more, but that is all that pops to
mind before coffee.

Art Kendall
Social Research Consultants


Roy FGH wrote:

> I want to apply a likert scale in my opinion research about marketing
> science. but a got a little problem in choosing the scale. should I
> choose a 4 Likert scale, or 5 Likert scale or maybe anyone can help
> choosing another scale for me? And maybe anyone can give me the reason
> or the recommended literature. Thanks for advancing.
>
>

[Kooij, A.J. van der]

> ... but I am sure that some people from the Leiden group follow this discussion list.
Right.
 
With the optimal scaling (Leiden) procedures in SPSS Categories (CATREG and CATPCA) you can analyze likert scale date on ordinal scaling level.

You obtain transformation plots that will reveal the relevant distinctions between

the categories. Below are some transformation curves(category quantifications on

y-axis) often observed for likert scale variables.

> You can always collapse response categories afterwards.  You cannot do the reverse.

Right. Note that the transformations indicate which categories can be collapsed.


Regards,

Anita van der Kooij

Data Theory Group

Leiden University

 

 

                        .

 

 

 

                    .

        .   .   .

    .

 

 

 

.    

1   2   3   4   5   6   7

The distinction is mainly between the 2 extremes and all the middle categories  

 

 

                        .

                .   .                         .   .   .   .   .

 

 

 

 

           

            .

    .   .                                 .

.                                     .

1   2   3   4   5   6   7             1   2   3   4   5   6   7

The variable is essentially transformed to a binary variable  

 

 

                     

                    .   .                                     .

                .

            .

                                                          .

        .

 

                                                      .

    .

                                                  .

                                              .

.                                     .   .

1   2   3   4   5   6   7             1   2   3   4   5   6   7

A linear trend for the lower or higher categories, leveling off after the

middle category


________________________________

From: SPSSX(r) Discussion on behalf of Art Kendall
Sent: Sat 04/11/2006 17:54
To: [hidden email]
Subject: Re: problem in choosing Likert scale



Bottom line: (even though I put it at the top).
Stick with conventional 5 point actual Likert response scales unless
there are reasons to do otherwise.

I don't have the literature cites here, but I am sure that some people
from the Leiden group follow this discussion list.

A few thoughts:

The number of points  on an agreement scale depends on balancing several
considerations.
On the one hand you want any variable to be measured at as high a level
as possible and as close to the underlying construct as possible in
terms of  meaning and measurement level.

On the other hand people can only make  "so many"  distinctions.  If
people are sophisticated in the subject matter,  they can usually make
more distinctions.
In an actual Likert scale there are several items that have 5 possible
responses. So with 5 items, summative scale scores can go from 5 to
25, which is certainly going to be close enough to interval.  If you
only have a few items in the scale and it is not difficult for your
respondents in your pretests, use a longer response scale.

 If you want to eliminate "neutral", go with a 6 point scale.

If the scale has been used in much previous research, go with what they did.

You can always collapse response categories afterwards.  You cannot do
the reverse.

I am sure others will be able to add more, but that is all that pops to
mind before coffee.

Art Kendall
Social Research Consultants


Roy FGH wrote:

> I want to apply a likert scale in my opinion research about marketing
> science. but a got a little problem in choosing the scale. should I
> choose a 4 Likert scale, or 5 Likert scale or maybe anyone can help
> choosing another scale for me? And maybe anyone can give me the reason
> or the recommended literature. Thanks for advancing.
>
>



**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.
**********************************************************************