Help understanding Microsoft Excel ANOVA output (Significance F ?)

#1
Hi,

So I created my own very simple excel table, with a set of X and Y values, and then I told Excel to run a regression analysis.

http://s22.postimg.org/41djtsab3/stats.png

I have been going through every piece of info provided by the Excel output table, attempting to understand what it all means and how each piece of info is calculated.

Sadly I am struggling to understand "Significance F".
I am really rusty with statistics and I believe I might need to look this up in a table... but I did google for "F distribution tables" and I cannot see any figures anywhere even close to 0.124027063 ...

Can someone please help me understand this?

Thanks a lot in advance.
 

hlsmith

Omega Contributor
#2
Without really looking at your output, I would say it is the p-value for the F-statistic. So, if you look up the F-statistic on the F Distribution Table, do you get the 0.124 value?
 

Mean Joe

TS Contributor
#4
Don't use that .jpg. (it is giving you the significance value for F when you have (numerator degrees of freedom, denominator degrees of freedom) -- just for your future information)

So you have F=4.5 (derived from 3.6 / 0.8), with 1 numerator degree of freedom and 3 denominator degrees of freedom.
(see how 3.6 is SS for regression, and 1 df for regression -- the regression stats are your numerator in F calculation).

In Excel, type in a cell: =fdist(4.5,1,3)
You'll get your p-value, aka the significance F.
 

hlsmith

Omega Contributor
#5
I shouldn't be helping you since I am typically all applied stats, and usually botch these things up. So here we go, your table is just for 0.05 level of significance, so all we can determine using it, is whether the value is smaller or larger than 0.05 level. The book on my shelf has 0.10 in the table, which will get us closer.

First determine your degrees of freedom http://ron.dotsch.org/degrees-of-freedom/, from a quick scan I am guessing you have 1 and 3, but I am probably off - since I don't regularly do this. Then I look them up on your table and see: 10.13. Well our number is smaller than that number so this means it has a p-value greater than 0.05. I go to my book and it has a value of 5.54 for alpha = 0.10, well your value is just a little smaller than this number, and the p-value is actually 0.12 if I did my degrees of freedom correctly and we had a F-distribution table that includes all values.

I see Mean Joe beat me to the submit button.
 
#7
I have an additional question if i may.

Basically this is a conceptual query... I struggle to understand where ANOVA sits alongside Regression.
Excel runs a regression on my selection of continuous independent and dependent varialbes, and one of the output boxes is labelled ANOVA

Yet I read that ANOVA is for analysis of a situation that involves a continuous output variable but categorical input variables? (Yeah im not 100% sure what im saying either).

To put it another way: If I assign a manufacturing process A, B , C to 3 randomly selected workers, and then i record their daily output... my continuous dependent variable is output, but the "treatments" (in this case manufacturing process A, B, C) are categorical - not continuous numbers.
So if I want to determine whether the average output over the course of a year is the same or different for A, B, C... I do ANOVA.

But if I have house value as dependent variable, and various input varialbes like floor space, bathrooms etc... then I run a regression ...... why then do i get an output box in excel labelled ANOVA? Where's the crossover?

thanks - hope this made some small bit of sense.
 

hlsmith

Omega Contributor
#8
Short answer, ANOVA is a special case of regression. Like a square and rectange are in the same family. I didn't really follow all of your question in particular.