Currently, I am working on how to relate Extension officers education level or years of experiences on their decision on selecting contact farmer. eight variables were used to measures officers´ perception towards selection of best contact farmer. likert scale is ranging highly important to least important (+2 to -2 ).
years of experience has measured using years and four level of education level were used. ( secondary school, diploma, bachelor, Master).
my hypothesis will be like
more experiences Extension officer has best choice of selecting contact farmers or more educated farmers has selected best farmer
1. can I use Pearson correlation ( or any bi-variate ) to analysis?
2.What should I do for the likert scale data? should I need to take single ranking value for analysis?
or is there a better choice/analysis I could apply?
Dear Kodikarage,
I suggest that you need to start with a reliability analysis with you Likert variable data to see whether they can be taken to be describing the same thing. The way to do this is to start with a Principal Component Analysis and check the loadings on the first factor. I have written a brief guide on this process: https://www.researchgate.net/publication/280936182_Advice_on_Reliability_Analysis_with_Small_Samples
In terms of the rest of the analysis, if the reliability analysis works it should allow you to compare years of experience and level of education with your single scale variable. For the years of experience, I suggest you start with a scatter plot, look at the dispersion of point and decide whether Pearson correlation is appropriate (see our guide: https://www.researchgate.net/publication/274635640_Pearson_Correlation), otherwise use Spearmank rank correlation. For the education level, I suggest you use Jonkheere-Tepstra - a nonparametric test of trends in multiple groups as you only have a few catgories for this ordinal variable.
Technical Report Advice on Reliability Analysis with Small Samples
Technical Report Pearson Correlation
thank you very much. I have another issue of using likert scale data, can I use row data of eight variables as it is for analysis or do I need to do any alteration for analysis. could you please see the attachment for my data and suggest a better option.
thanks
In order to use Pearson product moment correlation coefficient, the data should be in interval or ratio scale. Since, our data is in ordinal scale, Pearson correlation is out of the option. You could use Spearman's rank correlation coefficient which measures the correlation between two ordinal variables. However, Spearman's correlation does not detect the linearity of the relationship hence be careful when interpreting the results.
I forgot one thing. Since one variable of the two is interval/ratio, first it has to be ranked and ranked variable is used with the Likert scale variable to calculate the Spearman's correlation.
thank you very much for your detail explanation. I will try for spearman´s correlation
If you have several Likert-scored items, you can add them together to form a scale, and then assess the reliability of that scale using Cronbach's alpha.
Because there have been so many questions here about using Likert-scored items, I have compiled a set of resources on that topic:
https://www.researchgate.net/post/How_can_one_help_people_create_simple_scales_from_Likert-scored_items
Hi Kodikarage,
I wanted to know how you did your analysis finally. It will help me to lead my analysis in a more generous way. For example i am trying to analyse how unemployment rate (continuous) impacts fear of failure(Scale) in different regions.
Hi mam nirosha.. I see the use of Pearson as the more appropriate correlation test to apply (likert responses when averaged ill give you a continuous data, thus validating the use of Pearson). May I know the findings of your analysis, what correlation test dis you finally apply?
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis