A study is conducted to measure the effect of an educational method in a group of 4 samples. After experiment, 24 observations was done for each sample. What is the appropriate analysis test for this study? In this case, what is best comparing cases or observations?
Dear Dr Abdolghani Abdollahimohammad
Hi and greeting.
I hope you have a good time not only in professional work but also in your life.
Usually, this situation we used Repeated measures analysis of variance (rANOVA). Repeated measures analysis of variance (rANOVA) is a commonly used statistical approach to repeated measure designs.
Please visit this site: https://statistics.laerd.com/statistical-guides/repeated-measures-anova-statistical-guide.php
and this: https://statistics.laerd.com/spss-tutorials/one-way-anova-repeated-measures-using-spss-statistics.php
This site is the best site for SPSS learning.
Sincerely, Morteza
Dear Dr Abdolghani Abdollahimohammad
Hi and greeting.
I hope you have a good time not only in professional work but also in your life.
Usually, this situation we used Repeated measures analysis of variance (rANOVA). Repeated measures analysis of variance (rANOVA) is a commonly used statistical approach to repeated measure designs.
Please visit this site: https://statistics.laerd.com/statistical-guides/repeated-measures-anova-statistical-guide.php
and this: https://statistics.laerd.com/spss-tutorials/one-way-anova-repeated-measures-using-spss-statistics.php
This site is the best site for SPSS learning.
Sincerely, Morteza
Dear Dr Abdolghani Abdollahimohammad
as i think the statistical method depend on type of data if its in interval scale we use t test or u can use mann witnny test as non parametric test
Dear Dr Mamun Albnaa
Hi
Excuse me, use of the T Test in repeated situation may increase Type 2 error. We cant use a T test in several condition and compare these situation with another situation. For example:
I assess anxiety 1 day(1), 12 hours(2) and 1 hours(3) before surgery and 1 day(4), 5 days(5), 30 days(6), and 60 days(7) after surgery. It is completely mistake that compare 1 with 2, 2 with 3 or 4 with 5, 6 with 7. We must examine 1,2,3,4,5,6 and 7 overall.
with best Regards.
have a nice night!!
in my country, time is 00:55 !
A type II error, also known as an error of the second kind, occurs when the null hypothesis is false, but erroneously fails to be rejected. It is failing to assert what is present, a miss. A type II error may be compared with a so-called false negative (where an actual 'hit' was disregarded by the test and seen as a 'miss') in a test checking for a single condition with a definitive result of true or false. A Type II error is committed when we fail to believe a truth.[4] In terms of folk tales, an investigator may fail to see the wolf ("failing to raise an alarm"). Again, H0: no wolf.
The rate of the type II error is denoted by the Greek letter β (beta) and related to the power of a test (which equals 1−β).
What we actually call type I or type II error depends directly on the null hypothesis. Negation of the null hypothesis causes type I and type II errors to switch roles.
The goal of the test is to determine if the null hypothesis can be rejected. A statistical test can either reject or fail to reject a null hypothesis, but never prove it true.
Dear Dr
In fact I didn't get wt u want to say
But if you want to compare between four samples we are using one way anova if our data have the conditions of parametric statistics or croscal wills
Dear Dr Abdolghani Abdollahimohammad,
are you looking for growth in performance? I think mixed linear models for repeated measures is the best way to do it.
Regards
Dear Enrique,
The research objective is to evaluate the effect of an educational method on learning. In this regard, 4 trainers was selected nd 24 times their skills were assessed.
Dear Dr Abdolghani Abdollahimohammad ·
i discussed with my friends in measurement dept and we think if used repeated measures analysis it will be good .
More important than knowing what test to use is knowing what the result of such a test (a p value) actually tells you. It does NOT tell you the probability that the null hypothesis is true, or that your results are due to random error, or that your results will replicate--in other words, the probability that the null hypothesis (no difference) is true, given your results. These are extremely widespread misinterpretations, found in many textbooks. Instead, it tells you the probability that you would get the results you did, or ones more extreme, assuming that the null hypothesis IS true. These are not the same, and you cannot (except in very rare circumstances) calculate the first from the second. Statisticians have been pointing this out for decades (see uploaded paper), but the misinterpretation is so persistent, and its results so damaging to science, that some statisticians have called for banning statistical significance tests from publications. A p value is useful evidence for assessing the likelihood that your results are due to chance variation alone, but it doesn't tell you how likely this is. The more important issues are, is the actual mean difference practically relevant enough to pursue, and do the results replicate?
Two popular choices are running a regression on the post scores taking into account pre scores and having dummy variables for each group. e.g.
post_i = B0 + B1 pre_i + B2 group2_i + B3 group3_i + B3 group4_i + e_i
where B2--B4 compare those groups with group 1 (so, an ANCOVA). An alternative is subtract post_i - pre_i and run the ANOVA on these (so setting B1 = 1 in the equation above). This problem is essentially Lord's ``paradox'' and the choice of which to use depends on several factors, but unless you have a reason to suppose B1 = 1, then the ANCOVA method is probably the better of the two (see attached for some situations where each is preferred). But, as Paul notes there are other alternatives, and as Joseph notes, whatever you do it is important to consider the actual meaning (or lack of meaning) of things like p values (I'd be more interested in the B values from above).
Dan
Article Comparing groups in a before-after design: When t test and A...
1. one way ANOVA is essential to see if the groups are equal at the beginning.
2. If there is no significant difference among the groups then after the pretest two way ANOVA for repeated measures on one factor only,
3. If there is a significant difference between any two groups or more then ANALYSIS of Covariance to ensure headstart.
A study is conducted to measure the effect of an educational method in a group of 4 samples. After experiment, 24 observations was done for each sample. What is the appropriate analysis test for this study? In this case, what is best comparing cases or observations?
A study of a group subject to four measures, can use the difference in means between the different assessments. May alternatively first calculate the average of the scores and calculate the mean differences. Then you can use the test Kruskal Wallis (four groups, same four measures) to analyze diferences obtained by the group in the 4 measurements.
Paired Samples t Test is another option (SPSS or Excel-Data Analysis).
Hello, I think you can do with ANOVA, but complemented with structural equation modeling (SEM) using the AMOS statistical program. This program uses the database SPSS, yet gives the possibility of designing models with the variables under study. It is a more complex, but much more real methodology. Please see some examples on my profile here at RG. Academic regards
I think ANOVA, but in my case (4 schools and 252 surveys) and 4 questions with open answer I use SPSS (for the quantitative answers) and atlas.ti, because this second program analyses words and creates trees with the words.
Dichotomous variable: Cochran's Q test
Ordinal variable: Friedman test
Continuous variable (normally distributed): Repeated measure ANOVA
Continuous variable (not normally distributed):Friedman test
I think you can do with ANOVA and assisted with structural equation modeling (SEM) using SPSS . It gives the possibility of designing models with the variables under study. It is a real methodology , even more complicated.
I would do a MANOVA, followed with an ANOVA post-hoc t-test analysis. If the data is ordinal or continuous use Friedman test.
What is the appropriate analysis test for this study? In this case, what is best comparing cases or observations?
You should investigate both - individuals AND cases - HLM analysis
It would depend on the level of measurement of the variables. Most likely, ANOVA is appropriate, but remember that analysis is driven first by the research question, then the level of measurement of the variables.
I would analyse in 2 steps:
Step #1: Effect of pre/post in each sample considering paired or not t-Student test in order to evaluate significance of differences inside each sample. Paired would be better but depends if this feature was considered in research plan. Segmentation would bring perception about behaviors.
Step #2: ANOVA as mentioned above to have a widely perception of pre/post effect.
First - I need to know about your dependent measure - If your DV is interval or better a parametric set of analysis may be appropriate if your DV is ordinal or less a non-parametric approach might be more appropriate. Second - I need to know the number of IVs - if more than one IV then multiple IV approach such as ANOVA may be appropriate. So in short, the answer to your question depends on your research question(s), the level of your dependent measures, and whether your data meets the statistical assumptions for the test you hope utilize. More importantly, your statistical approach should tell the story about your research question and your study findings. Your statistical analysis must conform to minimum assumptions in order for them to be useful. Your statistical analysis should add clarity to your findings.
I think that ( ANOVA with repeated measures ) test is more appropriate for your study.
- Badges
- Science topic
- Similar topics
- Education