The unpaired t-test | Independent samples t-test
HTML-код
- Опубликовано: 10 окт 2024
- In this video, we will see how the unpaired t-test is calculated. We will also discuss how the corresponding confidence interval is calculated.
1. Study designs (0:56)
2. The example data (01:44)
3. How to calculate the t-statistic (04:33)
4. Confidence interval (09:28)
5. Assumptions (11:54)
6. Welch's t-test (14:49)
For more videos in a logical order, go to:
www.tilestats.com
Thank you for explaining so clearly. Just one question- did you meant in the video that If the two sample sizes are different, then formula for calculating standar error will be different when we are creating confidence interval as compared to when we are calculating t statisic
When the two groups have different sample size, one needs to use the more complicated equation (the weighted average of the pooled variance) to compute the SE in both the t-test and the confidence interval.
First of all, thanks for the nice video.
At 13:05 You mentioned that you made another video for analyzing normality, but at least for the Shapiro-wilk test I could not find a video.
Thanks for reminding me of this missing video. I will try to create such a video after the summer.
@@tilestats You are welcome. And best greeting to Sweden :).
how about comparing the mean difference of groups 1 vs the mean difference of group 2? for example, two groups as in two independent variables, with two measurement points (i.e. pre-test & post-test) for one dependent variable (e.g. BP).
Not sure I understand, do you mean a study design as in this video
ruclips.net/video/2o4UDqCVMGg/видео.html
@@tilestats not quite, that why I am confused. I am going to conduct a study: intervention A vs intervention B, with Blood pressure as the dependent variable outcome. For each individual in each group, BP will be measured at baseline and at 3 months, to see the reduction in BP. So, will be comparing the mean difference of intervention A & the mean difference of intervention B between these two groups.
is this still a unpaired t-test?
Yes, you compare the difference between the differences before and after, right?
@@tilestats yes. But when it comes to sample size calculation using G power or SPSS, there is missing standard deviation values for the mean difference of either intervention. (from what I see in a research paper). Is this normal to not have a SD for the mean difference within group?