I'm in a graduate level stats class and my final is Monday. I was pretty sure I was going to flunk the ANOVA section even though I tried tutoring and office hours to get help with it. Sometimes you just need a friendly, patient RU-vid guy to carefully and slowly explain with a practical, real-world example. I can't express how relieved I feel now that I understand this concept. Thank you so much, Math and Science guy, you're an Earth angel!
You're just awesome ! I thought this as the hardest subject I have ever had in our bio-statistics course but you made it like piece of a cake. Thank you so very much .
That you so much sir!! This helps me a lot. I'm studying masters now and we have statistics subject and this is my report all about finally I can deliver it well. Thank you God bless!!!
I have seen that the best way to learn these quantitative research techniques are here. I have been confused on a very simple thing to understand. Thank you
I have to agree with the comments before me, just watching this first video I want to watch and learn more from you. Great job on the explanation and on keep me engaged!!
I’m think you get the basic understanding of concepts or get to see the terminology during your professor’s lecture and when you watch this video it completes the picture. This happens to most of the students and it is not your professor is bad at teaching or this video is the best of the world. It is how our brains work!
First of all this is really good explanation. Query : Mean would only be effective only if distribution score for all school score follow normal distribution. What if any one of the group dose't follow normal distribution in that case should we consider median . pls clarify what should be the approach. Thank you
I came to know what's ANOVA and why it is used but I want to know what are one way and two-way ANOVA. can someone tell me the difference between one-way-ANOVA and two-way-ANOVA?
Data Nerd One way ANOVA means that there is only one independent variable used in constructing the groups under comparison. Two way ANOVA introduces a second independent variable into the analysis. For example, one way ANOVA could be used to determine if the means of three or more age groups differ in a set of test scores. If we want to break down the results further into age group and gender, we can use a two way ANOVA to determine if the means of the groups differ by either age group or gender, or both.
Azazel fall The F ratios used in both one and two-way ANOVA are always right tailed. This is because the between group variance is always in the numerator and the within group variance in the denominator. If the F ratio is less than 1, the denominator is bigger than the numerator, so there’s more variation within each group than between the individual groups...that is, there’s no point in conducting a hypothesis test since there’s too much overlap between the samples for them to be distinguishable. Since we only care if the F ratio is larger than 1 regardless of whether the ANOVA is 1 way or 2 way, the test is always right tailed.
Celia Lin I agree that the name "analysis of variance" seems like a misnomer since we're comparing means...but think of it this way. Every sample we are comparing not only has its own mean, but it's own distribution, and this distribution has a variance associated with it. If the samples have variances that are small, then the values in the samples are not widely spread out. Therefore, if you made each sample into its own bell curve to represent its distribution about the sample mean, it is likely that these distributions would have little overlap. As such, it is likely that if the means are far enough apart, we will say that the F ratio is significant. If, however, the samples have large variances, there is a good chance that there may be a fair amount of overlap between the bell curves representing their distributions. Given enough overlap, these samples become indistinguishable and basically form one large distribution "in the background." As such, we can't really say that the means are really different in this case. So, we are basically using the spread of the data - the variance - to quantify how much variation is coming from the "signal" (whatever effect we are measuring) as opposed to the "noise" in the background from the random variation. If this ratio is small, we have too much "noise" to conclude that our "signal" makes any true difference in our samples - that is, the means of the samples are not statistically different. Make sense?
Anteater23 he just explain from the begining of the video why they call analysis of variance just lister..in short we just call it because we want to see how sample mean one vary from other means
milimo mashini I understand it. If the variation is mostly due to variation between the sample means and the grand mean instead of the variation within the samples then the sample means must be quite different to the grand mean - implying that they're not all the same. So when you take the F statistic which is sum of squares between / sum of squares within you'd expect to get a significantly large value if the variation was mostly due to the variation between the samples. If the F statistic is large enough then you're confident the means are not all equal.
Your videos are good but you spent so much time repeating the same over and over again, prolonging the vital lesson to learn in the video. Take note of this.
I thought this video was gonna give me more explanation and calculations to determine power and how and by how much outliers can change the end probability (area under the probability curve)
