今天答辩了，pass！持续了两个星期的考前抑郁症应该会得到缓解。

 

昨天晚上还周期性发一些考前怪梦，记得两个，一个是b1和b2怎么减都抵消不走，弄得u1-u2无办法计算。这种梦比较难对付，因为不断repeat，就算偶然一次解了它又repeat，有解的概率总是低过无解的。

还有就是语文考试，交卷的时候才发现有一整面没有做，作文也没有写，然后觉得难怪做得这么慢都够时间。

不过对这种梦我是身经百战。所以交卷的时候想，觉得没写作文就没写，无所谓，一点都不紧张。

我的梦世界经验值都在提高的。原地复活，满血满魔！

 

答辩的过程就skip。上年8月8日在科罗拉多的presentation场面比这个大呢。我现在想起来还好似昨天发生的一样。

如果我安排下个星期三其实就可以凑到9月9，和上年8月8呼应。我是个喜欢摘日的人啊。然后10月10。。。

 

还有就是最近在打一个游戏，微软的，叫做几何战争（看我attach的vedio），就是在一个有限平面空间射击，里面有黑洞等等乱七八糟的，然后空间都被扭曲了，然后子弹就因为空间扭曲的原因不飞直线了，敌人还被黑洞吸进去，黑洞吸收太多就会爆炸，黑洞还会吸另一个黑洞。特别过瘾。

 

好，mark down今天这个日子。9月2日。2009。星期三。热浪

 

SDSU
presents
a thesis defense for
Master of Science degree in Statistics

Wednesday, September 2, 2009
9:00 am  GMCS 405

Tao Lin

Power analysis of statistical models
for testing treatment effect with covariates

Abstract:

A complete random block design with cut-offs and baseline measurement
is commonly used in clinical trials. Since cut-offs build up blocks
for later treatment assignment, the block factor must be included in
analysis model under FDA regulatory. If we put the baseline
measurement into the analysis model, the redundant information of the
baseline measurement and block factor which generate from the
baseline value may reduce the power and affect point estimates. If we
ignore baseline measurement however, we lost information and may
finally lost power of the test.

The objective of this thesis is to evaluate three optional models for
a data structure defined by a block design with baseline
measurements. They are the conservative randomized block design
model, a modified change from baseline model and a randomized block
design with covariate model. The primary objective is to develop and
compare the power of these models and draw conclusions on model
selection for future study under similar conditions.The type I error
and biasness are reviewed as a prerequisite for power comparison
between models. Previous researches demonstration,algebra proof and
Monte Carlo simulation are performed to investigate the power of
these statistical models.

Thesis Committee:

Chii-Dean Lin, Thesis Chair, Department of Mathematics & Statistics
Kung-Jong Lui, Department of Mathematics & Statistics
John Alcaraz, Department of Public Health