Transcript z-table
Chapter 9
Hypothesis tests with the t statistic
• 當母體為未知時(我們通常不知),用樣本s
來取代
• 因為用s來估計,所呈現出來的分佈已不
是z distribution,而是 t distribution
The Normal Distribution
• Bell shaped, symmetric, & unimodal
• Notation: X~N(,2)
– 學生身高(X) X~(135, 102)
• Characteristics:
–
–
–
–
–
Symmetrical
Mean=median
大部分分數落在mean,少部分分數落在兩尾
兩尾向兩端無限延伸
常態分配曲線下的面積總合=1
The Standard Normal Distribution
標準常態分配
• Notation: Z~N(0, 1)
• Characteristics:
– The standard normal distribution has a mean of 0 and
standard deviation of 1
– The original scores need to convert to z score!
– Areas under the curve has fixed probabilities
associated with z-scores
• These areas are presented in normal curve table or z-table.
“標準化”的概念
• 標準化 standardization
• 為何要將原使分數標準化?
• raw scores z-scores
•
z
X
or
xx
z
s
• 所有的z-score distribution 皆為µ=0,σ=1
的分佈
• z-score distribution=standardized
distribution
• 抽樣分配的 z
z
x
x
抽樣分配 z formula
(母體已知)
z
x
x
抽樣分配t formula
(母體未知)
x
t
sx
The estimated
standard error
Xi X
2
n 1
df
Sum of
Square
d (SS)
2
s
n
• z formula
• t formula
t statistic t 統計量
z statistic z 統計量
z
x
x
x x
x
t
2
sx
ss / df
s
n
n
x
2
n
Sample Variance
樣本變異數
• Sample variance
(S2)
:
Xi X
2
n 1
Degree of
freedom (df)
Sum of
Squared (SS)
Xi X =
2
• S2=
n 1
概念公式
n Xi Xi
2
n(n 1)
計算公式
2
• 當母體已知,每一sample 都可計算z score
• 從母體裡抽取許多樣本(n固定),則我們會有許多
z scores
• 所有z scores集合起來 →抽樣分配的 z
distribution
z
• 當母體未知,每一sample 都可計算t score
• 從母體裡抽取許多樣本(n固定),則我們會有許多t
scores
• 所有t scores集合起來 →抽樣分配的 t distribution
t
t distribution
• 當n小時,t distribution 為一非常態的分配
• 當n大時,df 亦增大,t statistic 會趨近
z statistic,而t distribution 會趨近常態
Why??
• n愈大,s2會愈接近2
Why?
n要多大distribution 才會接近
z distribution?
t distribution 特質
• Bell shape
• 分佈較z distribution扁平,
• 有較長的tails
• When n ≈
,s ≈ ,t ≈ z
t table
Assumption
•
•
•
•
Independent observation
population is normal (when n is small)
Random sample
Variable need to be ordinal, interval, or
ratio in nature
• Hypothesis
Ho: =k
H1: >, <, or ≠k
•
•
•
•
決定 tc (df; 單雙尾; level; 查表)
計算 to (公式)
判斷 tc & to 比大小
Reject or fail to reject Ho??