Timeline for Hypothesis Testing with ANOVA tables
Current License: CC BY-SA 4.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 17, 2020 at 17:15 | comment | added | Stochastic | I would suggest doing multiple correlation i.e. getting an ANOVA table. When it comes to which variables to exclude, look at the p-values in the ANOVA table. Alternatively you can also use adjusted $R^2$, AIC, BIC or Mallow's Cp for the exclusion of variables. Here are some notes on variable selection. | |
| May 17, 2020 at 16:34 | comment | added | MrUsername12 | I have been told to keep using the multivariate model when testing the Hypothesis "The soil being clay or not does not make a difference". When I do this, i still get different answers from the p-value in the ANOVA of the soil column and from using t.test(yield~ soil), but using the t.test function only considers those 2 and not the whole model, so I should use the ANOVA in this case? | |
| May 17, 2020 at 16:19 | history | edited | Stochastic | CC BY-SA 4.0 |
added 377 characters in body
|
| May 17, 2020 at 16:14 | comment | added | Stochastic | Remember it will be different when you model your response (yield) with only one predictor variable. Maybe I was not clear earlier, but for the each variable by itself you will use a t-test while I all are included, you will use a f-test. Sometimes you might find that using a predictor alone to model a response could be significant, but then when fitting it together with other predictors it is not significant at all. | |
| May 17, 2020 at 14:23 | comment | added | MrUsername12 | Don't worry, I have solved this part using ANOVA tables I believe. I have updated my question to include another issue I have encountered. | |
| May 16, 2020 at 19:20 | comment | added | Stochastic | Firstly, are you using dummy variables for the rainfall? It also depends on how R sees the rainfall variable - is it a factor according to R? You can use just the means and compare them, but I would suggest that you use an ANOVA table since you can set up the hypothesis test then. It would be a more "formal" way of testing for the significance of the rainfall. If you are able to provide some data, I might be able to help more | |
| May 16, 2020 at 17:56 | comment | added | MrUsername12 | I am using R. The rainfall is categorised in 3 levels of "None", "Light" and "Heavy" but when I use the t.test it says I need to only have 2 levels of rainfall. Would it suffice to compare each of the means of a specific variety to the rest, or do I need to do it another way like using ANOVA tables? How would I do that? | |
| May 16, 2020 at 11:56 | vote | accept | MrUsername12 | ||
| May 16, 2020 at 10:25 | history | answered | Stochastic | CC BY-SA 4.0 |