Hi everyone;
Hope you're doing well!!
I'm conducting a PhD dissertatind in educational research, I made a test to compare some students performance but I've got lost since am very bigginer with statistics !
So here is the problem:
I've made a 2 tests to comapre a control and a treatment group in pre and post treatment!!
so is this an Independent t-test or dependent one? or ANCOVA?? am really confused.
I wonder whether I can use both independent to compare control and treatment groups in a pre study; then an independent t-test to compare the control group in a post study.
_________________________
then I will use a dependent one to compare pre and post treatment group performance and the pre and post control treatment!!
Is this valid in statistics???
thanks for your replies in advance

Hi, as far as I understand: You have two Groups (control and treatment) and in each Group you have pre and post treatment. And now you want to find out if there is an effect of the treatment in the treatment group versus the control group. What you can do here is a dependent (or repeatet measures) ANOVA where you have to specifiy the ID variable. Here, the effect measure is the outcome, and predictors are Group (control versus treatment), Timepoint (before versus after) and the interaction tern Group:Timepoint. And the significance of the interaction term tells you if the treatment group reacts differently on the treatment compared to the conrol group

thank you for your answer.
yes this is what I want to do... so!! hummmmmm: it is better to use the repeated measure ANOVA!! okey!!
but I want to knw whether the use of depedent and independent t-test is wrong???
sorry for me insisting but i was asked to make a t-test by my supervisor !!!!!
thanks a lot for your help

Hi, the ANOVA is the generalization of the T-test in case you have more groups or combinations then two. In principle you can perform multiple T-tests as well: You can compare the after-treatment-data of the two groups via an independent t-test, and for each group you can compare the before versus after data with a dependent t-test. However, it is recommended to choose an ANOVA over multiple t-tests for two reasons: 1.) with multiple t-tests you increase the familiwise error rate, i.e., you increase the chance to get a false positive result. 2.) If you do not analyze all data at the same time (e.g. comparing only the after-treatment-data of the two groups via a T-test) your test may loose power because you don't use all informaton you have. But what you can do is: For each group and each participant you calculate a new variable which is:
outcome after - outcome before
So you take the difference between before and after for each participant, let's call this variable D. And subsequently you compare the D-values of the treatment group versus the control group via an independent t-Test. But the result should be similar compared to the above mentioned interaction term in the ANOVA.

I understand the matter of power and am convinced,,
..humm... but after reading about repeated measure anova, I found that it recommends to repeat the testing for more than three time points,, and this isn't my aim neither my study???

Hi, repeated measures means only that you measure the same individuals more than one time, i.e., your measurements are not anylonger independent. In your case, you have participants which have been measured two times each. Usually you have an ID variable in your dataframe for each participants, and you specify this variable in your ANOVA function. Which statistical software do you use? if you tell me the rough structure of your data frame, I can provide you a scetch of the corresponding code in R

hi again,
okey!!
am using spss.
alright; look I have made a tests to see whether my students' oral performance will be improved if I introduce to them some interactive listening activities. thus before introducing the interactive activities >I tested the learners oral performance of course i v got 16 of the students are the experimental and other 16 are control.. the students are randomly chosen.
then after 19 weeks of interactive listening sessions (only for experimental group) the students retook the test.

Hi, this sounds ideal for a mixed ANOVA. Unfortunately, I am not used to SPSS, so I can't help you with setting up the model. But the rough structure is: You have the outcome variable "oral performance = Y", and as predictors you use the categorical variable "treatment versus control = X_1" and the categorical variable "before versus after = X_2" and the interaction term X_1:X_2. Furthermore, you have to tell your ANOVA function the ID variable for each participlant, so that the program knows the repeated measures structure. Finally, the significance of the interaction term X_1:X_2 tells you if control and treatment group differ in before versus after performance.

But, as mentioned above, you can also use an independent T-test: But here, you firstly have to calculate the difference "performance after - performance before" for each participant, and with this new variable you can make an independent T test comparing the groups control versus treatment.