Simple Statistics Question
Got a simple stats question, that I'm sure all you analysts can answer.
If Corr(A,B)=0, and Corr(B,C)=0, then does Corr(A,C)=0? If so, how do your prove this using the correlation/covariance formulas
Thanks
Got a simple stats question, that I'm sure all you analysts can answer.
If Corr(A,B)=0, and Corr(B,C)=0, then does Corr(A,C)=0? If so, how do your prove this using the correlation/covariance formulas
Thanks
+630 | Xxxxxxxxxxxxxx | 20 | 5d | |
+65 | Investment Banking Interns - Learn to Abuse the Office Printer. Here Are 5 Tips to Make Sure You Land The Return Offer This Summer. | 7 | 12h | |
+46 | In Light of BofA Events Remember What's Important In Life | 1 | 3h | |
+40 | Did someone hack WSO on LinkedIn? | 12 | 2d | |
+38 | Dream Wife | 36 | 1h | |
+31 | Best Golf Trip locations | 26 | 3h | |
+30 | BAN Bronzbomba from WSO | 5 | 23h | |
+26 | Thoughts on Stifel’s Sex Scandal? | 17 | 3h | |
+19 | Footy in NYC this summer | 1 | 2d | |
+19 | Firms with a strong Physical fitness culture | 9 | 3d |
Career Resources
Corr(A,B)=0 -> Cov(A,B)=0 Corr(B,C)=0 -> Cov(B,C)=0
Cov(A,C)=E(AC)-E(A)E(C) Cov(A,B)=E(AB)-E(A)E(B)=0 -> E(AB)=E(A)E(B) Cov(B,C)=E(BC)-E(B)E(C)=0 -> E(BC)=E(B)E(C)
To prove that Corr(A,C)=0 the variables should be independent. While independence implies uncorrelatedness, uncorrelatedness does not imply independence.
If the variables are independent, then it naturally follows that Corr(A,C)=0
So assuming the variables are independent, can you still prove it mathematically?
I understand the formulas and logic you provide, but isnt there a way to prove it using the formulas?
If X and Y are independent, then the following holds for expectations
E(XY) = E(X)E(Y)
Hence, it follows directly from the covariance formula that Cov(A,C)=0.
Alternatively, Cov(A,B)=E(AB)-E(A)E(B)=0 -> E(AB)=E(A)E(B) -> E(A) = E(AB)/E(B) Cov(B,C)=E(BC)-E(B)E(C)=0 -> E(BC)=E(B)E(C) -> E(C) = E(BC)/E(B)
Cov(A,C)=E(AC)-E(A)E(C)=E(AC) - [E(AB)/E(B)]*[E(BC)/E(B)]=E(AC)-E(AB)E(BC)/(E(B))^2
Given that that the variables are independent, the expression reduces to Cov(A,C)=E(AC)-E(AC)=0
Since Corr(A,C) = Cov(A,C)/st.dev(A)*st.dev(C), it follows that Corr(A,C)=0
The alternative route is redundant, though, once you know that the variables are independent.
Hope it helps.
...how are you coming to the conclusion that A is independent of C? The fact that cov(A,B) = 0 and cov(B,C)=0 only tells you that A is independent of B and C is independent of B but you don't necessarily know the relationship between A and C. The relationship could actually be A=C. You're assuming that step it seems, and as a result you can't really say whether cov(AC) would or wouldn't be 0
Come on, NY, read the math! You can derive the result from Cov(A,B)=0 and Cov(B,C)=0 (see the alternative route). However, you need to make the assumption of independence.
The fact that Cov(A,B)=0 does not mean that A is independent of B! However, if two variables are independent, it implies zero covariance.
corr(a,b)=corr(b,c)=0 Set correlation equations equal to each other (A-E(A))(B-E(B))/(s.d(A))(s.d(B))=((B-E(B))(C-E(C))/(s.d(b))(s.d(c))) Multiply both sides by s.d(B)/(B-E(B)) and get: (A-E(A))/s.d(A)=(C-E(C))/s.d(C) Multiply both sides by (A-E(A))/s.d(A)) and you get (A-E(A))^2 / (s.d. (A))^2 =((A-E(A))(C-E(C))/(s.d(A))(s.d(C))) Which is equal to Var(A)/Var(A) = corr(A,C) Therefore corr(A,C)=1
Someone please tell me if you see a flaw, but it looks good to me
Qui vero deserunt voluptas sunt. Similique numquam similique nisi tenetur molestiae. Id odio quam temporibus consectetur explicabo non quas.
Ad possimus id odit perferendis id omnis ullam. Maiores molestias animi qui quam occaecati velit.
Qui fuga doloremque explicabo aut fugit id qui. Ex odit eveniet exercitationem eos qui et. Quia nobis minus voluptas facere et.
Qui accusantium quia ut vel et totam. Enim officiis est quidem eos modi sunt. Qui est dolor illum molestiae qui itaque occaecati. Et facilis non animi.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...