If anyone here has done statistics at univerity level, can you please explain to me what the formula for LSD does? I have an exam tomorrow and I'm rather lost, can't find it in my text book either.
Statistics
- Thread starter joshhem
- Start date
- Tagged users None
🥰 Love BigFooty? Join now for free.
- Joined
- Jul 11, 2001
- Posts
- 2,212
- Reaction score
- 45
- Location
- Around here.....
- AFL Club
- Hawthorn
- Other Teams
- Hawthorn
Not too sure about 'univerity' level.....
Least significant difference.... (for Comparing > two Means)
This is simply a t-test on various pairs of treatments. Instead of calculating s2 for each pair of treatments, the Error Mean Square from the ANOVA table is used as an overall mean value. Note that this assumes that the variances are the same for each treatment, but this is also one of the assumptions of analysis of variance. This results in t2(n-1) = d/sd, where d = difference between two treatments, sd = [MSerror * (2/n)] and n = number of observations per treatment. Therefore to be significant we need d > t2(n-1) . sd
The LSD test should only be used for comparisons which have been decided upon before the experiment is performed (i.e. a priori). It is sometimes said that only (n-1) comparisons are permissible, where n is the number of treatments
Copied and Pasted from here....
(I don't know if this stuff is right...but still....)
Least significant difference.... (for Comparing > two Means)
This is simply a t-test on various pairs of treatments. Instead of calculating s2 for each pair of treatments, the Error Mean Square from the ANOVA table is used as an overall mean value. Note that this assumes that the variances are the same for each treatment, but this is also one of the assumptions of analysis of variance. This results in t2(n-1) = d/sd, where d = difference between two treatments, sd = [MSerror * (2/n)] and n = number of observations per treatment. Therefore to be significant we need d > t2(n-1) . sd
The LSD test should only be used for comparisons which have been decided upon before the experiment is performed (i.e. a priori). It is sometimes said that only (n-1) comparisons are permissible, where n is the number of treatments
Copied and Pasted from here....
(I don't know if this stuff is right...but still....)
- Joined
- Feb 14, 2002
- Posts
- 17,797
- Reaction score
- 6,869
- Location
- South-West Gippsland
- AFL Club
- Brisbane Lions
- Other Teams
- Lexton, Northcote Park
Translation please.
- Joined
- Oct 19, 2001
- Posts
- 2,120
- Reaction score
- 8
- Location
- Brussels, Belgium
- AFL Club
- Essendon
- Other Teams
- wtf
I went to uni, and I'd love to explain what LSD does, but it's much better if you just try it. Turn on, tune in, drop out.
Log in to remove this Banner Ad
- Joined
- Jul 11, 2001
- Posts
- 2,212
- Reaction score
- 45
- Location
- Around here.....
- AFL Club
- Hawthorn
- Other Teams
- Hawthorn
Chong chong chingy chuuung chopstick!
- Joined
- Apr 20, 2001
- Posts
- 28,272
- Reaction score
- 11,281
- Location
- Adelaide Oval / Anfield
- AFL Club
- Adelaide
- Other Teams
- Adelaide Utd,Liverpool, WWT Eagles.
I nearly failed Economic Statistics 2 - I got 50.
Though some would say that is the perfect score
Just enough study - Just enough drinking
Though some would say that is the perfect score
Just enough study - Just enough drinking
Thats what I thought LSD meant too Pete
- Joined
- Dec 7, 2000
- Posts
- 11,504
- Reaction score
- 8,301
- Location
- Victoria home of football
- AFL Club
- West Coast
- Other Teams
- Eagles
Ohh man, Uni statistics. I hate the stuff, luckily I have my exam on monday and won't have to deal with it any more. Unfortunately I will have 2 more Stats subjects to do over the next 2 yrs. Bugger 
They try and disguise the subjects by giving them elaborate names such as "Statistics for Business and Marketing" and "Risk Analysis", when in actual fact they mean to say "Stats" and "Even harder Stats".
They try and disguise the subjects by giving them elaborate names such as "Statistics for Business and Marketing" and "Risk Analysis", when in actual fact they mean to say "Stats" and "Even harder Stats".
- Thread starter
- #9
Originally posted by Stucey
Not too sure about 'univerity' level.....
Least significant difference.... (for Comparing > two Means)
This is simply a t-test on various pairs of treatments. Instead of calculating s2 for each pair of treatments, the Error Mean Square from the ANOVA table is used as an overall mean value. Note that this assumes that the variances are the same for each treatment, but this is also one of the assumptions of analysis of variance. This results in t2(n-1) = d/sd, where d = difference between two treatments, sd = [MSerror * (2/n)] and n = number of observations per treatment. Therefore to be significant we need d > t2(n-1) . sd
The LSD test should only be used for comparisons which have been decided upon before the experiment is performed (i.e. a priori). It is sometimes said that only (n-1) comparisons are permissible, where n is the number of treatments
Copied and Pasted from here....
(I don't know if this stuff is right...but still....)
Thanks for that, a lot of it was hard to understand, but i've pretty much got it under contol now. I'm aiming for my first 7!! (but we wont get too optimistic.)
ANOVA and regression have to be the worst processes to ever use!
I somehow managed to scrape through Business Statistics at Uni last year. Heaps of people found it fairly hard, although if you studied hard and knew how all the formulas you'll do alright. I got a 61 - Credit, so I was happy with that.