I am analyzing the data using two instruments i.e. a Questionnaire and Observation (Checklist). A questionnaire used to asked teaching strategies used by the teachers. They rated themselves using Likert scale. But I need to confirm whether they are using teaching strategies for real in classroom. I used observation checklist to confirm it.
Here I have two data set. One is questionnaire in the form of Likert scale and second one is observation checklist.
What statistical tool I use to compare the results and produce the concrete findings of my research.
I have read literature and came into conclusion that chi-square test would be appropriate. I need valuable suggestions that what I need to do?
Dear Humair Akhtar ,
Thanks for the question.
A "Chi-square test" is a very general term. Many different tests use a Chi-square test.
One approach that is used in Likert-type of questions, which I generally disagree, is to perform a Chi-square test in contingency tables. Why do I disagree with this approach? A contingency table should be analogous to a correlation analysis. In this case, there is no direction between the variables. I disagree with this because, in most cases, we are interested on how a given variable (for example, one type of teaching strategies) is associated with a given Likert-type answer.
Thus, it would be appropriate to use a regression analysis, in which the response variable (i.e., dependent variable) is a question from the Likert-scale questions, and the explanatory variable(s) (i.e., independent variable) comes from the questionnaire. Since Likert-scale is categorical and ordered (e.g., "very likely" > "likely"), you should use an ordinal (i.e., cumulative) logistic regression.
I do not use SPSS, but here is an easy example in R that you could follow: https://stats.idre.ucla.edu/r/dae/ordinal-logistic-regression/
Please let me know if you have questions. Thnkas, Nick
Chi-square is not just one thing!
you need to define your variables (strategy and the likert scale)
based on this you can use correlation, regression, and maybe even more complicated methods such as CFA and SEM (CB or PLS)
but you need to initially define your variables in terms of how you are looking at them.
i am trying to make it simple as possible.
when you have defined variables, you have scales for each one. then you can measure how each scale is linked to the other(s) or how they impact each other.
using logistic regression is when you have dichotomous variables so I would not jump to analysis before having a clear definition of my variables. you can use SPSS, R, PLS or any other tool you have. however, for such analysis SPSS would suffice.
hope this helps.
For better understanding and getting clear significance association of each and every predictors or variables with dependent variable is application of regression than Chi-square test as chi-square test has limitation in clarifying strength of association in the two variables(dependent vs independent) irrespective of type of different tools used by researchers.
For example, if you find that age categories is found as factor in chi-square test using Pearson X2 and p value, unlike regression, X2 test does not explain which age category/ ies is/are factor in research.
Dear Getasew Amogne Aynalem Pouya Zargar Nick VL Serão
I have explained and made clear understanding of the key variables of study. I need to know whether teachers said truth about using teaching strategies in classroom. They quietly rated best of themselves. But I used alternative approach to see whether they were telling the truth. I used an observation checklist to see they are using teaching strategies in classroom. Now I have the data of observation. Here which analysis would be appropriate that would show both the responses of questionnaire and observation from where I can conclude that they were telling a truth or lie.
I need your suggestions using SPSS.
Regards
your analysis seems simpler than initially thought. you can report your observation depending on its format. your questionnaire can be tested as you have your variable defined. descriptive statistics, correlations and if necessary regression can show the relation between your observation results and your questionnaire.
- Badges
- Science topic
- Similar topics
- Education