Using the TI 84 Calculator to Perform an ANOVA test.

[@imviz777]

You saved me so much time, he didn't even tell us this was on the calculator. He was making us do it by formulas.

[@keithhudson3138]

Thank you, sir!!! This is the last thing we learn in my stats class. You made it so much easier.

[@emilyford4529]

Oh my gosh. I was absent during this lesson and I couldn't figure out the calculator commands! Thank you so much!

[@Andreia01]

Thank you so much!

[@carlossoto7083]

This should be renamed to say "One-Way ANOVA" not simply ANOVA

[@CesarClouds]

Thank you.

[@Donferish]

Awesom video. You are the best! thanks.

[@juliocandiaguilar]

haha I forgot how to use anova and I needed for my final for stats! well you deserve that like!

[@CPaulsem]

Great video, thanks a bunch!

[@redbonekh]

Sorry, but the decision part is kind of double negative confusing to me. Do you mean "Accept H0" also? Or is the essence different from "Failed to reject the H0"?

[@larrygreen7328]

Hypothesis testing can never result in accepting the null hypothesis. Either you reject the null hypothesis or you state that there was not enough evidence to make a conclusion (fail to reject).

[@victoriam9356]

thank god I thought I had to do this all by hand ;_;

[@kalun1256]

My college just taught us to use the F-ratio and C.V . Excuse me, what is the P value stands for ?

[@woodchuk1]

P-value just stands for probability value...it's the chance that you would get a test statistic for F as extreme or more extreme than the one you obtained from the sample data, assuming the null hypothesis was true. In this problem, it means that even if the means of the 4 groups were in fact statistically equal, you'd get an F statistic as large as this one or larger about 22% of the time. Since that chance is larger than the 5% chance you adopted as your likelihood of wrongly concluding that the means were not equal when they in fact were, you don't have enough evidence to claim that the means are different based on this data. Make sense?

[@woodchuk1]

The CV you are familiar with is the value at which the p-value equals the significance level. So, at this "critical value," you have only a 5% chance (assuming an alpha level of 0.05) of getting a test statistic for F as large as the CV. Any F lower than the CV will have a higher p value than the significance level, and anything higher will be less likely to come up...that is, the p value will be lower than the alpha level. So, if you get a p value less than 0.05, either 1) the null is true, the means are equal, and you obtained a test statistic that should happen less than 5% of the time, or more likely, 2) the null hypothesis isn't true and the means aren't equal.

[@RoseLightRS]

thanks :) but i wonder why my professor teaches a different assumption, that the smallest Std. Dev. times 2 is greater than the largest standard deviation? I dont even know how to find that on my ti-84.. :/

[@woodchuk1]

Sounds like your professor was using a general guideline for variance equality based on something called the Hartley's F-max test. While it can be useful, it's just a general guideline. Plus it wastes information, since it only uses information about two of the groups involved. More sophisticated tests for homogeneity of variance exist that take into account all the groups, such as the Levene, Brown-Forsythe, O'Brien, Bartlett, and Fligner-Killeen tests.