Unformatted text preview: 175 STA 2023 c B.Presnell & D.Wackerly  Lecture 15 STA 2023 c B.Presnell & D.Wackerly  Lecture 15 176 Recall from last time: If we plan to take a random
from a population with mean £ ¡ standard deviation ¥§
¦¤ (p. 255) ¡ #!£ £ ¤
" ¨
¢
¢ ¥ § ©¢ ¤
¨ Today: P. 265 – 271 . So . (p. 266, 261) of For Tuesday: Exercises 6.15, 6.21, 6.24, 6.28, is called its sampling distribution. is an unbiased estimator (p. 266), so dist. of ¢ concentrated around for larger sample sizes. ¥§ 6.33, 6.37, 6.38, 6.41–46 is more is called the standard error of .(p. 266) ¤ For Wednesday: P. 280 – 285
For Thursday: Exercises 7.1, 7.3–5, 7.10, 7.11, New terms ¨
$£ ¤ Distribution of Assignments : and is a random variable. ¥§ hair! , ¥§ run the country are busy driving taxicabs and cutting ¥§ Thought: Too bad that all the people who know how to ¢ sample of size – Parameter – meaningful number assoc. with a
Population. (p. 254) 7.15–20 – Statistic – meaningful number assoc. with a
Sample. (p. 254) 177 STA 2023 c B.Presnell & D.Wackerly  Lecture 15 STA 2023 c B.Presnell & D.Wackerly  Lecture 15 178 Know the mean and standard error of ,
HOW ABOUT THE DISTRIBUTION ? % ¤ If the population has a normal distribution, the § sampling distribution of is normal, with mean ¡"
£ ¢ £¢ and standard deviation (standard error)
distribution of x when n = 50. . ¡ ¤ Central Limit Theorem : (p. 267) For large , regardless of the shape of the population dist. § the sampling dist. of is : & approximately normal & standard error 3)
4120¡
(© ¡
¡"
£ ' £
¢ ' ¢
& with mean ¤ Larger sample size, For most populations, . better approximation.
is “large enough”. original population distribution distribution of x when n = 35. ¤ 179 STA 2023 c B.Presnell & D.Wackerly  Lecture 15 STA 2023 c B.Presnell & D.Wackerly  Lecture 15 180 Ex. : Time spent at the cashier at a movie theater : 1£ ¡
¢ seconds, with is “large” has an approximately distribution.
¢ & Standard Error : ©§
¨ score for 30 : ¤& & ¥§ Sampling Dist. of £ Want Sample Size on ¤ larger n 3 1 ¦§
¥ ¤ Effect of Increasing & ¥§ probability. ¢ are close to tickets for all 36 buyers is less than 30 seconds? £ Values of . ¡ ¤ Smaller standard error : What is the probability that the average time to buy smaller n . £ : ¡"
£ £ ¡ Bigger §( ¤
¢ buyers. seconds. Thirty six ticket µ 181 ¥§ STA 2023 c B.Presnell & D.Wackerly  Lecture 15 STA 2023 c B.Presnell & D.Wackerly  Lecture 15 182 Ex. : (#6.41, p. 275) Study relating IQ and juvenile
delinquency. For ALL juveniles, IQ : 3 ¢
a. What shape for the sampling distribution of ? £ has an approximately distribution. ¡ £ " ' £
&
¢ & Standard Error : Does your answer depend on the shape of the
distribution of IQ scores for juveniles? £ ¤ sec min ¡ min is “large” §( time until all 36 have tickets
¥
¦§ ¤
¥¦§ ¤
§ ¤
¥
¦ min juveniles given IQ test. ¥§ ¡ #
¢ before the movie starts? £ the probability that all 36 ticket buyers get tickets !
" There are 18 minutes until the movie starts. What is
¥ &
¤ ¤ 183 STA 2023 c B.Presnell & D.Wackerly  Lecture 15 STA 2023 c B.Presnell & D.Wackerly  Lecture 16 184 Thought: An honest politician is one who, when bought,
stays bought.—(Simon Cameron) b. ASSUME : Nondelinquent juveniles have same 3 ) § score for Assignments : . 3 ¤ Want ¡ nondelinquents: #
¢ mean IQ with same std. dev. Look at Today: P. 280 – 285
: For Thursday: Exercises 7.1, 7.3–5, 7.10, 7.11,
7.15–20 & ©¨
§ For Monday: P. 299–304, COMPUTER
DEMONSTRATION
For Tuesday: Exercises 7.37, 7.42, 7.44–46 nondelinquents, got Last Time: SEE FORMULA SHEET
,
Estimator
¢ STA 2023 c B.Presnell & D.Wackerly  Lecture 16 the time that an interval estimator actually encloses
a parameter when we compute a large number of
intervals (based on diff. samples). Ex., Sample etc. ¤ , Conﬁdence Level (p. 256) : conﬁdence coefﬁcient 3 ¢ ¤
¤ Point estimator (p. 261) : formula for a SINGLE £¡ expressed as percentage. Ex., ¡ Have : Large Sample from Population with ﬁxed £¡ ¤ Conﬁdence Coefﬁcient (p. 282) : the proportion of Chapter 7 : Estimation based on a Single but Unknown mean, 186 3 &
185 approximately NORMALLY ¡
¡! ¢ ¡! STA 2023 c B.Presnell & D.Wackerly  Lecture 16 ¥§ § 3 ¥§ For “large” sample size,
distributed. Standard Error of Est. ¡" Parameter £ (like for ALL juveniles)? 3 . Do you think true popn. mean IQ is
std. dev. #
¢ 1 ¢ ) § ¤ &
! c. Actually took sample of , etc. Useful new idea! NUMBER that is intended to be “close” to the ¤¥ 3 ¥¤
¡ !¤ ¦ § ©§
¨
¤ ¥§ ¢ ¢
:
¥ is a point estimator for , let ¦
§§ Notation: for any parameter value. be such that A §& Interval estimator (p. 282) : formula that tells us
how to compute TWO NUMBERS that are intended
0 to ENCLOSE the value of the parameter zA BETWEEN THEM
.5.025=.475
.025 0 z.025
§ ¤ 187 STA 2023 c B.Presnell & D.Wackerly  Lecture 16 STA 2023 c B.Presnell & D.Wackerly  Lecture 16 Conﬁdence
.5.05=.45 (alpha) Coefﬁcient ¢ © z ¢ .05 0 188 .99 .05
§ .98
.95
Note: .90 ! ¡!
α/2 § ¥ § ¥ § ©
¤ .025 α/2 1−α .025
.95 and the resulting interval the . “middle” (shaded) region? are in the & ¤ Thus, in repeated sampling, approximately ¤ If we obtain a value of in the “tails” (unshaded) & ¢ the value of § and the resulting interval
Plug that value into the formula ¡¡ ! § region. . . & What fraction of the time will this occur? £ £ £ .025 of the intervals will enclose the true value of % ¢ £ ¡¡ ¢ §
! ¤ What proportion of the values of § ¤ & Plug that value into the formula
value of with . ¥§ £
£ ¤ .95 in the “middle” (shaded) region. ¢ .025 If we obtain a value of 190 £ ¡¡ ! §
!
¡¡ ¢ © ¥§
©£ ¨§ STA 2023 c B.Presnell & D.Wackerly  Lecture 16 §
§ © 3 , use it!! If not, estimate ¡¡ ! If you know £ § § Recall not ! § ¦¤§ ©
¥£ 3 189 A 95% Conﬁdence Interval for ¦£ §
¥ STA 2023 c B.Presnell & D.Wackerly  Lecture 16 § NOTE : ¨ !
£ ¡¡ ¢ ¢
¢ ¨ !
£ ¡¡ ¢ © ¢ ¤ STA 2023 c B.Presnell & D.Wackerly  Lecture 16 ¥ ¨ ¦ ¥
©§¢¤ than steel, but brittle at high temps. Want to estimate (large sample) p. 283 36 specimens randomly selected, measured temp. ¨ ¥
£ ¦£ § ¥ § 3 # ¥ §
and is the z value that cuts off an area of $¢ the mean temp. at which metalic glass becomes brittle. ¤ ¦£ §
¥ 1¡¤ ¢¤
¢
¡ at which each became brittle : Conﬁdence Interval for Ex. : (#7.71, p. 314) Metalic Glass, 4 times stronger 192 191 STA 2023 c B.Presnell & D.Wackerly  Lecture 16 in the upper tail of a standard normal distribution true mean temp. metalic glass gets brittle is unknown, ¦£ § ©
¥
"
#! £¡ 3 ¤ £¡ 3
(
! 3 ¡ ! §
3! 1 3 #
¡ " # ¢ ¥ § £
!
¥§
¢
3 ! 1 # ¡ ¡ ! # ¢
476.98 ¡" ¨
¥ § £
: ¤ ¢
¢ © ¤ £¡ 3 3¤ £¡ – the region of “believable” values for
conﬁdence level. !
# ¢ ( conﬁdence interval for 483.02 £ ¢ ¢ interval!! at the $ Note : Lower conﬁdence coefﬁcient . $ #!
" £¡ CI conﬁdence ¦£ §
¥ ¡ ! 1 3 #
¡ " ¡¡ ¢ ¥ § £ !
¨ !
¥ § '£ ¡¡ ¢ ¥ § . and 193 table formula sheet standard errors "
#! ¤
conﬁdence $ ¡" interval for 3 !
¡¡ ¢ estimator STA 2023 c B.Presnell & D.Wackerly  Lecture 16 Ex. Back in Exercise 7.71, ﬁnd a NOTE : the basic form of this interval is formula sheet ! £ interval. HOW? 1−α ¥©£ § 3 # ¥§ Problem actually asks for a α/2 § CI α/2 £¡ 3 ¤ ...
View
Full
Document
This note was uploaded on 12/15/2011 for the course STA 2023 taught by Professor Ripol during the Spring '08 term at University of Florida.
 Spring '08
 Ripol
 Statistics, Standard Deviation

Click to edit the document details