Comments on: Tues Feb 14 http://www.stat.cmu.edu/~kass/smnp/?p=85 Thu, 26 Apr 2012 14:07:02 +0000 hourly 1 http://wordpress.org/?v=3.0.4 By: Matt Bauman http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-109 Matt Bauman Tue, 14 Feb 2012 14:00:51 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-109 In the definition of S_pooled, I was curious where the -2 term came from. When pooling n data sets, does this therm become -n? In the definition of S_pooled, I was curious where the -2 term came from. When pooling n data sets, does this therm become -n?

]]> By: Jay Scott http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-108 Jay Scott Tue, 14 Feb 2012 13:59:00 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-108 Terminology clarifications on resampling (jackknifing, bootstrapping) when it concerns pseudo-data: 1. Are bootstrap values considered a kind of pseudo-data? 2. Is sampling / resampling of pseudo-data sets such as those discussed on pp. 295-297 a kind of bootstrapping? or does it have its own nomenclature? Terminology clarifications on resampling (jackknifing, bootstrapping) when it concerns pseudo-data:

1. Are bootstrap values considered a kind of pseudo-data?

2. Is sampling / resampling of pseudo-data sets such as those discussed on pp. 295-297 a kind of bootstrapping? or does it have its own nomenclature?

]]> By: Amanda Markey http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-107 Amanda Markey Tue, 14 Feb 2012 13:55:31 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-107 If you can create null hypotheses of distributions, why do we so often test whether means are different (e.g. ANOVA) instead of testing whether two distributions are different? If you can create null hypotheses of distributions, why do we so often test whether means are different (e.g. ANOVA) instead of testing whether two distributions are different?

]]> By: David Zhou http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-106 David Zhou Tue, 14 Feb 2012 13:50:33 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-106 Can you go over some non-paradigmatic cases for null hypothesis formulation? What types of samples are they to be used for - all samples where mean is of little concern? Can you go over some non-paradigmatic cases for null hypothesis formulation? What types of samples are they to be used for – all samples where mean is of little concern?

]]> By: Rich Truncellito http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-105 Rich Truncellito Tue, 14 Feb 2012 13:38:24 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-105 What happens to the pooled variance when the standard deviations of the two samples are not assumed to be equal? What happens to the pooled variance when the standard deviations of the two samples are not assumed to be equal?

]]> By: Shubham Debnath http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-104 Shubham Debnath Mon, 13 Feb 2012 21:12:55 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-104 If p-values are calculated after sampling randomly under certain conditions, what kind of variability is there in the p-values? Do they follow a particular distribution? If p-values are calculated after sampling randomly under certain conditions, what kind of variability is there in the p-values? Do they follow a particular distribution?

]]> By: Yijuan http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-103 Yijuan Mon, 13 Feb 2012 19:30:14 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-103 When we use t test for the null hypothesis H0 : μ1 = μ2, we assume σ1 = σ2. But in actual data, the two deviations may not be equal. Do we need to transform data to make them equal? Is there a test to compare σ1 and σ2? When we use t test for the null hypothesis H0 : μ1 = μ2, we assume σ1 = σ2. But in actual data, the two deviations may not be equal. Do we need to transform data to make them equal? Is there a test to compare σ1 and σ2?

]]> By: Rob Rasmussen http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-102 Rob Rasmussen Sun, 12 Feb 2012 21:36:21 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-102 I am still having trouble after reading the sections with knowing when/how to choose between the different statistical metrics presented. I am still having trouble after reading the sections with knowing when/how to choose between the different statistical metrics presented.

]]> By: Eric VanEpps http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-101 Eric VanEpps Sun, 12 Feb 2012 17:19:32 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-101 In explaining how p-values are calculated, you mention the use of pseudo-data generated by simulating data from the null hypothesis's distribution. In practice, how do we simulate this data? Can you provide an example other than the binomial case in class? Thanks. In explaining how p-values are calculated, you mention the use of pseudo-data generated by simulating data from the null hypothesis’s distribution. In practice, how do we simulate this data? Can you provide an example other than the binomial case in class? Thanks.

]]> By: Scott Kennedy http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-100 Scott Kennedy Sat, 11 Feb 2012 18:53:58 +0000 http://www.stat.cmu.edu/~kass/smnp/?p=85#comment-100 In section 10.3.2, what does it mean that T_n is an asymptotically normal estimator? In section 10.3.2, what does it mean that T_n is an asymptotically normal estimator?

]]>