I am conducting a research, and I am proposing a new format of lesson plan. I want to show that the new lesson plan that i am proposing is much more effective than the existing ones. I want to measure the efficacy of this lesson plan through gathering the data from the "classroom observation rating tool of teachers" during classroom observation of teachers and compare their rating to the previous rating they had when they used previous lesson plans.
1. Should i use paired sample t-test?
2. structural equation modelling?
3. What other statistical test should i perform?
4. Could i hypothesize on correlation, efficacy, and what else?Is there any other hypothesis to formulate from this study?
You cannot prove anything with a statistical test. You can demonstrate that your data are sufficient to conclude the sign of the difference (if A is better than B or the other way around), but the analysis is about your data, not about any true state of nature you like to "prove". And even this conclusion refers to the population that is effectively sampled (who had a chance to be part of your study?). Be aware that different schools have different policies, they have different pupils (e.g. from either rural or urban regions, from religious or secular regions etc.), countries have different educational systems, and there are different cultures involved. So be crystal clear to what part of the school/social/cultural system your conclusion applies.
Your study design is suboptimal because you will compare recent results (of the proposed method) with older results (of the previous methods). You may see a time effect. You won't be able to distinguish if the differences you see is due to the time or due to your new method.
Jochen Wilhelm 's answer is absolutely correct.
Another point in order to better understand and choose the right model. Statistical Analysis You need to write about Measure, your independent and dependent variables (Scale, Ordinal, Nominal or Interval).
However, with this information, I think it is sufficient to use a test-retest analysis.
However, by knowing other information from your study or seeing a picture of the data file, more analysis can be used.
I think this would be a quasi-experimental and an "interrupted time series design". the rating sheet is this one https://www.google.com/search?q=cot+deped&sxsrf=ALeKk03xG9HXwW67NAnPq8HGZ57SjsshXw:1616590922617&source=lnms&tbm=isch&sa=X&ved=2ahUKEwie38G9_sjvAhXF7WEKHVYoCwkQ_AUoAXoECA0QAw&biw=1366&bih=600#imgrc=Cjr6C47tIwhc1M
Well done Ray van
Now answer a very important question.
Do you want to test the same teachers? That is, to compare the past performance of a number of identical teachers with their performance today? Or that the teachers studied in the past and today are different from each other.
I want to compare the past performance of a number of identical teachers with their performance today using my intervention. Maybe it would be a retrospective way perhaps? and the cofounding variables would be "time" or (perhaps some professional advancement like trainings, seminars before my proposed "new format of lesson plan exist". Is it correct to call it a dichotomous moderator?im not sure though. I dont have enough confidence to perform my test. Help please
This is the whole point.
" I don't have enough confidence to perform my test."
You have a simple task ahead of you and you do not need to be stressed.
Do you have something called a factor or Independent Variables in your study? For example, gender or income level or age groups
If so, you can use GLM analysis.
If you do not have a factor (or Independent Variables) , then a simple Paired t-test is enough to compare before and after the same people.
Of course, you can still use the test-retest as a basic analysis. Especially since you are dealing with the same people.
If you do not find someone around you to help with statistical analysis, email me your data file in SPSS format.
Ray van Note another point.
The best graphs for studies similar to yours are called Bland-Altman. Try to draw them too.
I would like to conduct a research and this is related to my previously asked question in this link: https://www.researchgate.net/post/What_statistical_test_should_i_use5
Previously, i presumed to be using interrupted time series by proposing an intervention and then compare the result of it to the previous one. It would be a quasi experimental because of lockdown and limited social interaction.
As i research deeper, i think my research design would be causal-comparative research?
I found out that causal-comparative (retrospective) have three types
a. exploration of the cause
b. exploration of the effects
c. exploration of the consequence.
and i think that :
1. it would be better if i do exploration of the effects by comparing the results of two groups: my school who has not yet done this proposed intervention and the other school who have been doing this already for some time.
2. Is my question asked in this link :
https://www.researchgate.net/post/What_statistical_test_should_i_use5
can be considered as exploration of the consequence ?Because, after i gather the data in exploration of effects, i want to apply the intervention in my own school.
Is it okay to do both explorations in one research?What are the suggested statistical tool should i use if i do both explorations?
- Badges
- Science topic
- Similar topics
- Chemistry
- Sulfites
- Inorganic Chemistry
- Inorganic Chemicals
- Salts