Dear researchers.
In my research, I have two groups of participants (i.e., control and experimental group). I administered a Likert scale questionnaire ( consists of 24 questions and five cultural items) for both groups as pretest and posttest. I instructed experimental group for two months. the purpose was to know whether the instruction had any effect on experimental group.
My research questions were:
1. Does instruction through tasks improve EFL learners' cultural understanding?
2. Are there any differences between male and female EFL learners regarding the development of cultural understanding through tasks?
Please let me know appropriate statistical tests (T-test, Anova, ...) related to my research questions.
Thank you in advance
واضح است که این مطالعه، یک تحلیل آنالیز کوواریانس است. آنالیزهایی مانند تی تست و یا آنالیز واریانس برای آن چندان صحیح نیست.
اینکه مطالعه آنالیز کوواریانس یا آنالیز کوواریانس چندگانه باشد با دیدن دیتا شما مشخص می شود. همچنین یک طرفه یا چند طرفه بودن آنالیزها. به هر حال اس پی اس اس بهترین نرم افزار برای این کار است
Clearly, this study is an analysis of covariance analysis (ANCOVA or MANCOVA). Analyzes such as t-test or analysis of variance are not very accurate.
Whether the study of analysis of covariance (ANCOVA) or multivariate analysis of covariance (MANCOVA) is determined by looking at your data. Also, one-way or multi-way analyzes. However, SPSS is the best software for this study.
Hello Seyyedehzahra,
In order to give a constructive recommendation, I would need to know more about your survey. When you say, for example, five cultural items, do you mean that you wish to compare the groups five times, using each item as a single outcome variable? Or, do you intend to create a summated or averaged score over the five items, and use that for comparing groups?
Similarly, what about the other 24 items? One at a time as DVs, or some linear combination of them into a smaller set of subscores?
If you're contemplating combining of responses into subscores, do you have information to support that the respective item sets are unidimensional (e.g., each represents a different, common factor)?
Good luck with your work.
In addition to David Morse 's very useful points, I should mention your null hypothesis.
Frankly, there is no believable hypothesis that says that after two months of instruction, there was no observable effect. If nothing else, two months of instruction tells people the socially desirable answers, even if you haven't affected them in any other way.
Your null hypothesis is that the change was too small to justify two months of work. For two months of work I would want to see a pretty significant change. For example, when I was in a music curriculum project, after one year of classes the percentage of students in the bottom quintile of test scores had gone from 20% to zero. That's meaningful, and you could justify your curriculum on that basis alone.
So you need to think about the kind of change that would justify the intervention. Your null hypothesis is that the effect of your intervention is at least as large as the smallest meaningful effect.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA