Hi,
I’m struggling to decide which analysis I must do for my 10 point Likert scale questionnaire.
Participants were asked to rate 12 classroom behaviours on their perceived levels of disruption for these behaviours from 0 (not disruptive) - 10 (disruptive). Before this, participants were asked if they had experienced FSM (free school meals) or not to determine socioeconomic status. This is to measure whether or not having FSM affects their perceptions.
Just because we can run a statistical test does not mean we should. Run enough tests and by chance you'll have statistical significance. Large samples can easily show statistical significance for trivial changes. Avoid p-hacking and harking. Also, don't make the mistake of some authors where each question is considered as a single-item of great importance (again, slicing and testing multiple times will find all kinds of useless information).
What is your research purpose? Practical significance versus statistical significance? Lurking/confounding variables? Sample size? How did you develop the instrument and measure validity/reliability? Remember, just mentioning Cronbach's alpha is not enough.
Lydia,
There is an excellent discussion of this on the "Statistics By Jim" website.
See < https://statisticsbyjim.com/hypothesis-testing/analyze-likert-scale-data/ >
I hope this helps.
Don P.
The intensity Index can be calculated to analyse the data. It is averaging out the score. For this, one has to make categories of the variable e.g. disruptive, somewhat disruptive, undisruptive. divide the overall score like in your case it must be between 1-10. so divide the score in these categories. Then calculate the mean. of each respondent. Calculate an overall score to know the level of disruption among overall respondents. Now, using the scores you can interpret which category showed higher percentages and discuss accordingly. Furthermore, you can check the differences among the categories in relation to independent variables like age, gender, academic performance etc. using ANOVA.
Descriptive statistics for central tendency( mean, median, and mode) and dispersion(range, variance and standard deviation):
To generate summary reports of survey data.
Methods
mean, standard deviation, regression, hypothesis testing and sample size determination
To analyze the data, the intensity index can be calculated. In this case, the variable e.g Destruptive, Undestruptive. The total scores has to be divided in this case and mean of the respondent.
To know the level of distruption among the total number of students the total scores will be calculated.
The score is used in interpretation of the data.
Researchers can also make used of descriptive statistical tools such as mean, median and mode likewise inferential statistics such as regression and correlation .
There's an interesting paper (Norman, 2010) that's sort of two or three steps ahead in answering this question: Article LIkert scales, levels of measurement adn the “laws” of statistics
. Basically, there are the arguments that common methods you could get as answers are prinicipally "wrong" but then there are also counterarguments that they might be adequate after all. But you want to be aware of the whole discussion to make an informed choice of course.- Badges
- Science topic
- Similar topics
- Medicine
- Epi
- Age Groups
- Children