Can anyone help me? I have a question with five-point Likert item:
0 = Very little; 1= Little; 2= Moderately; 3= Many; 4= Too many.
Does anybody have any idea how transform this answers on a dichotomous scale (e.g: 0 = no; 1 = yes) ? I use SPSS software for analysis.
Regards! Luciano Vitorino
It is not recommended to re-categorize Likert scale question into dichotomous because you will lose the power of the data while categorizing. But, if you really want to do it. Follow the following.
Do you want to transform one question? If many. Follow the following procedure.
You can do it using Compute new variable function of SPSS.
First: The variable you computed need to be in numeric. You can compute the values of the variable by adding each value of the likert scale questions.
Second: Check normal distribution for the newly computed variable using either Shapiro-Wilk test or Kolmogrov-Smirnov test. The use of these methods depends on the size of your sample. If you found out the distribution of the newly computed variable is normal, then define your cut point using mean else median.
Third: Transform the new variable into categorical. You can do this through Recode into different variable. Check whether the values the newly computed variable of each respondent is above or below the calculated mean/median. Then, change the values into either 0 or 1 using if/then function.
It is not recommended to re-categorize Likert scale question into dichotomous because you will lose the power of the data while categorizing. But, if you really want to do it. Follow the following.
Do you want to transform one question? If many. Follow the following procedure.
You can do it using Compute new variable function of SPSS.
First: The variable you computed need to be in numeric. You can compute the values of the variable by adding each value of the likert scale questions.
Second: Check normal distribution for the newly computed variable using either Shapiro-Wilk test or Kolmogrov-Smirnov test. The use of these methods depends on the size of your sample. If you found out the distribution of the newly computed variable is normal, then define your cut point using mean else median.
Third: Transform the new variable into categorical. You can do this through Recode into different variable. Check whether the values the newly computed variable of each respondent is above or below the calculated mean/median. Then, change the values into either 0 or 1 using if/then function.
Just one thing that I wanted to add to the previous answer: the scale you are describing is a generic ordinal scale, not a Likert-type item.
The link below may be of help, in terms of how to actually transform the variable in SPSS.
http://libguides.library.kent.edu/SPSS/RecodeVariables
Thank so much Mihiretu M. Kebede your comments were helpful.
Best regards. Luciano
Dear Luciano Magalhães Vitorino,
I agree with Mihiretu M. Kebede. It is not recommended to recode scales into different forms. In the one you provided, there are 4 scales and you want to decrease them into two. If you do this transformation, you will decrease the power of your data and this transformation will have effects on the results you obtained. Instead of this, researchers are advised to form the format of the scales or use more appropriate ones before collecting the data.
For the transformation of the scales and for the other related issues, I highly recommend SPSS Survival Manual: A Step by Step Guide to Data Analysis Using IBM SPSS (http://spss.allenandunwin.com.s3-website-ap-southeast-2.amazonaws.com/)
Best regards.
Sibel.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS