- This topic has 6 replies, 3 voices, and was last updated 8 years, 3 months ago by Anonymous.
6th September 2012 at 3:20 am #2141
I’m doing a research on motivation of farmers under the extension project.
The adapted a Motivation Model for Extension” of Lewis, 1972 using 9 items of motivators.
Under the 9 items, 37 motivation statements were answered by 90 farmers through 5-point Likert Scale.
However, after calculating each farmer’s motivation score (Mean), I would like to categorize it into three level only such as High, medium and low.
I’m not sure if it is ok to just group the first two high (3.50-5.00) into high level, put the medium (2.50-3.49) as it is and group the last two low (1.00-2.49) as the low level?
Please give advice.7th September 2012 at 6:55 am #2147AnonymousInactive
is there any particular reason as for why you’d want to do that? the bulk of the statistical literature argues against artificially discretising variables or restricting the variance of scores in general…7th September 2012 at 4:31 pm #2146Ingo RohlfingMember
Oscar is right, you should have a reason (a theoretical one) why three categories are better than five. Apart from this, it is not uncommon to reduce the number of categories in the way you describe. However, as robustness checks, you should also run your analysis with different thresholds.10th September 2012 at 11:37 am #2145
Thank you to the two advice. I did try to run my data based on 3 categories and found that most of them were in the middle. So, I agree with you not to change five into three level.
Thanks again.10th September 2012 at 11:48 am #2144
I have another question regarding Likert scale again.
This interval (0.50):
and this interval [0.80 from (Max-Min/total level)]
Is the difference between these two interval categories significant?
Which one is most suggested to use in the thesis such as ANOVA and/or simple descriptive analysis?
I found more papers used the second one, but the statistic book used the first one.
Please give advice again.12th September 2012 at 11:33 am #2143Ingo RohlfingMember
I am not sure what you mean by significant. Since these are scales, not data, statistical significance does not apply here. My response is that this should be decided on the basis of theory (or convention, when most papers use the second scheme).26th September 2012 at 3:22 pm #2142
Thank you for your reply, and sorry for my unclear question.
‘Significant’ here simply means “whether their difference is important?”
My professor also said like you.
So, I think I will follow the way most papers use (the second one).
- You must be logged in to reply to this topic.