Transcript Chapter 2: Describing Location in a Distribution

2.1 Describing Location in a Distribution
Do You Sudoku?
 Each of you get a Sudoku puzzle; your work alone; no talking;
no consulting each other
 Get your phone out so you may time yourself (use the
stopwatch); total time will be rounded to the nearest 10th of a
minutes (i.e., 4.3 minutes)
 Your total time includes you checking your Sudoku to make
sure it is correct & complete
 Let’s make sure we all understand how to do a sudoku puzzle;
https://www.youtube.com/watch?v=OtKxtvMUahA
Your Sudoku times…
 Write your times on the board; copy all data on your
Sudoku; be sure to indicate which time is YOURS (we will
use these more tomorrow)
 What can we say about our distribution of times?
 What type of graphical representation should we choose to
display our data?
 SOCS (review)
 Where do YOU stand within the distribution?
Review/Preview
 mean, standard deviation (what are these?) 
primarily histograms (or dot plots or stem plots)
 median, Q1, Q3 (what are these?)  (primarily) box
plots
 What is meant by Q1? By Q3?
 PREVIEW: mean & standard deviation  density
curves  z-scores, measures of location of data
relative to mean (# of SD’s from mean)
Percentile
 The pth percentile of a distribution is the value with p
percent of the observations (= to or) less than it.
 Q1  25th percentile  25% of the distribution is
equal to or less than that given value within the
distribution
 Q3  75th percentile  75% of the distribution is
equal to or less than that given value within the
distribution
Our sudoku times…
 Q1 and Q3 are percentiles that are used often
 However we can calculate any percentile; i.e., 18th
percentile, 79th percentile, 99th percentile, etc.
 What is your percentile (comparing your time with
everyone else’s time)? What is your guess?
 In this case if a low percentile good or is a high
percentile good?
Sudoku percentiles…
 How many sudoku times do we have?
 Count where yours is; then divide by total
 Count all times below yours; then divide by total
 Either calculation is acceptable as your percentile
 Is it good to be in a low percentile or a high percentile
for this distribution? How about for test scores?
Times
0 – 59 seconds
60 – 119 seconds
Frequency
Sudoku frequency table…
120 – 179 seconds
180 – 239 seconds
240 – 299 seconds
300 – 359 seconds
360 – 419 seconds
420 – 479 seconds
480 – 539 seconds
540 – 599 seconds
600 – 659 seconds
660 – 719 seconds
720 + seconds
Sudoku frequency table…
Times
0 – 59 seconds
60 – 119 seconds
120 – 179 seconds
180 – 239 seconds
240 – 299 seconds
300 – 359 seconds
360 – 419 seconds
420 – 479 seconds
480 – 539 seconds
540 – 599 seconds
Frequency
Now let’s expand to include relative frequency,
cumulative frequency and cumulative relative frequency
Times
0 – 59 seconds
60 – 119 seconds
120 – 179 seconds
180 – 239 seconds
240 – 299 seconds
300 – 359 seconds
360 – 419 seconds
420 – 479 seconds
480 – 539 seconds
540 – 599 seconds
600 – 659 seconds
Frequency
Relative
Frequency
Cumulative
Frequency
Cumulative
Relative
Frequency
Cumulative Relative Frequency
Graph (or “ogive”)
 … example from text, age of U. S. presidents on day of
inauguration
Ogive for our Sudoku data…
Measuring position: z-scores…
Are you happy with your test score
in Physics?
 You earned a 78/100 on your test. Are you satisfied?
 How about if everyone else earned 98/100, 99/100, and
100/100?
 How about if everyone else earned 54/100, 61/100,
45/100, etc.?
 It’s all about how YOU did compared to others who
took the Physics test.
In chapter 1...
 we described entire sets of data (SOCS)
 now focus on individual observations (and how that
observation compares to the entire distribution)
 Similar to percentiles; how you did on your Sudoku
time versus how everyone else did
_____ earned an 86%...
All the scores for that Physics test were:
67
75
80
79
81
77
73
83
74
93
72
77
82
77
83
90
79
85
83
89
73
80
78
86
84
Input data into list, do 1-var stats
Mean? Median? SD?
mean (
x ) = 80
median = 80
s = 6.07
So, 86 is above average, but by how much?
≈ 1 standard deviation
Another measure relative standing (besides percentile)
that is used very often in statistics  z-scores
Z-Scores
 z-scores are one way to describe a particular data point
in a distribution via the # of SD’s above or below mean
(remember what standard deviation is ...how much
variation exists from the mean; measure of spread)
Standardizing a Value
 When we ‘standardize’ a data value, we convert the raw
data or original data to standard deviation units
 So 86% (raw, original data) on Physics test once
standardized converts to a standardized value of ≈ 1
(number of SD’s away from mean)
Z-Score Formula
x  mean
z
SD
# of SD’s (+ or -) away from mean; directional
86% on Physics test ...
 Remember ...
x = 80
s = 6.07
 So ≈ 1 SD above mean
 Now calculate exactly

86  80
z
 0.99
6.07

x  mean
z
SD
Z-Scores, Standardizing, # of SD’s
Away From Mean
 Got 93/100. Calcuate z-score.
• Got 72%. Calculate z-score
x  mean
z
SD
Z-Scores, Standardizing, # of SD’s
Away From Mean
93/100 raw score 
72% raw score

93  80
z
 2.14
6.07
72  80
z
 1.32
6.07
Now on to AP Statistics...
 82% on an AP Stats test. Disappointed? Happy?
(relative to 86 on Physics test)
 AP Stats distribution of test scores was fairly
symmetric with 𝑥 = 76, s = 4
 Calculate z-score for 82% on AP Stats test.
AP Stats Test Score
82  76
z
 1.5
4
 Raw score: lower than Physics test score
(82% vs. 86%)
 Relative to everyone else who took each test: better; z-
score of 1.5; 1 ½ SDs above mean
Z-Scores, Standardizing, # of SD’s
Away From Mean
 Standardize (raw score changes to z-score)
observations from distributions to express relative
standing in a distribution
 Standardize (change to z-scores) observations to
express values from two or more observations on a
common scale (i.e., 8-year old with height 5 foot vs. 18year old with height 5 foot)
 ACT vs. SAT scores; different scales
REVIEW…Measuring Relative
Standing: Percentile
 Can also describe performance on Physics test or AP
Stats test by using percentiles.
 pth percentile of a distribution as a value with p
percent of observations at or below it.
Measuring Relative Standing:
Percentile
 Go to list of physics test scores; sort ascending
 Score of 86/100 on Physics test is where on list?
 22nd on list; so 22/25 = 88th percentile
 88th percentile; 88% of students are at or below a test
score of 86/100
Measuring Relative Standing:
Percentile
 How about test score of 72/100?
 Only 2 at or below 72/100; so 2/25 = .08
 8th percentile; 8% of students are at or below a test
score of 72/100
Measuring Relative Standing:
Percentile
Note: Some textbooks define the pth percentile of a
distribution as the value with p percent of observations
BELOW the given value.
If this is the case, it is never possible for an individual to
fall at the 100th percentile
AP readers & me: either acceptable
Transforming Data…
Transforming data…
 Think about the last time you went out to dinner. How
much was the bill?
 Write on the board (round to nearest dollar)
 Input into lists
 Do 1-var stats; What is 𝑥 ? What is s? How about other
key values, like Q1, Q3, etc.?
 Graph it (let’s do a histogram)
Transforming data…
 The chef prepared such a delicious meal we want to
‘tip’ her. We are going to tip the chef $10
 Input new values into other list (+$10 each entry)
 1-var stats; What is 𝑥 ? What is s? How about other
key values like Q1, Q3, etc.?
 Graph it (again); let’s do a histogram
Transforming data…
 So, what do you think happens when we add (or
subtract) an amount to a distribution?
 Shape unchanged
 Spread unchanged
 Everything (all/each data point) just moved to right or
left, including the center
Transforming data…
 Now, instead of dining in Santa Clarita, we go to exact
restaurant but in Beverly Hills
 They charge 20% more for their menu offerings
 We order the same thing
 Take original distribution and multiply by 20%
Transforming data…
 1-var stats
 What is 𝑥 ? What is s? How about other key values
like Q1, Q3, etc.?
 Graph new data (histogram)
Transforming data…
 So, what do you think happens when we multiply (or
divide) a distribution?
 Shape unchanged
 Everything (each/all data points) else is
increased/decreased, including center AND spread
Linear Transformation
 A linear transformation changes the original
variable x into the new variable xnew given by an
equation of the form xnew = a + bx
 Adding/subtracting the constant a shifts all values of x
upward or downward by the same amount.
 Multiplying/dividing by the positive constant b
changes the size of the unit of measurement.
Linear Transformation Effects
 Multiplying (or dividing) each observation by a
positive number b multiplies (or divides) both measures
of center (mean and median) and measures of spread
(interquartile range and standard deviation) by b.
 Everything changes except the shape of the distribution
 Adding (or subtracting) the same number a (either
positive, zero, or negative) to each observation adds (or
subtracts) a to measures of center and location (mean,
median, quartiles, percentiles) but does not change
measures of spread (IQR, standard deviation) nor shape
of distribution.
Other examples of linear transformations…
 Consider a data set of summer temperatures (in
Fahrenheit) of all 50 U. S. states
 Decided we want those temperatures in Celsius (not
Fahrenheit)
 Conversion: subtract 32, then divide by 9
 Z-scores… how are these linear transformations?
What is the formula to convert to a z-score?
 Review… what are z-scores?
Homework…
Page 99, #1, 3, 5, 9 (skip part ‘c’), 11, 13, 15, 19, 21, 23
Page 99, #25- #30 MC
SECTION 2-1 HW QUIZ TOMORROW …
2.2 Density Curves & Normal
Distributions…
Key Strategies Used for Exploring
Uni-variate Distributions
 Always plot data, make a graph (usually box plot, stem
plot, histogram, dot plot)
 Look for overall patterns (SOCS)
 Calculate numerical summary (1-var stats)
New Step for Uni-variate Data…
• Sometimes the overall pattern of a large number of
observations is so regular we can describe it by a
smooth curve (density curve) model
Density Curves (in general)
 Fairly symmetric, uni-modal, no gaps/outliers
 Good description of data, good mathematical model




for distribution
Often easier to use and fairly accurate
Idealized description
Ignore minor irregularities
Fill in bars
density curve
Density Curves Model…
 NOTE: all ‘models’ are wrong, but some are very
helpful
 Density curves fits this description well
Essential Characteristics of Density
Curves
Density curves are always…
• On or above horizontal axis
• Have an area of exactly 1 under it
• Describe an overall pattern of a distribution.
• And …
Characteristics of Density Curves
 Area under density curve and above any interval of
values on the horizontal axis is the proportion of all
observations that fall in that interval.
Side Note on Density Curves …
• This material is introductory only
• Density curves can be many shapes (uniform, bimodal,
etc.)
• Real power of density curves comes when the
function/distribution is Normal or approximately
Normal
• Normal (or fairly Normal) distributions are KEY for
the entire course
Density Curves & Mean/Median
 Don’t worry about locating mean and median by eye
on density curve
 Will use mathematical methods to locate
 Just know, in general:
Sample vs. Population
(statistic)
Sample
Mean
𝑥
(parameter)
Population
µ
Standard
Deviation
s
σ
Density Curves…
Usually for density curves we will use µ and σ
Normal Distributions
 Important type of density curves
 Good descriptor/model of many real-life data sets,
such as test scores, biology, heights, weights, outcomes
of chance, inference, etc.
Normal Distributions
 Symmetric, single-peaked (uni-modal)
 Bell-Shaped
 Mean ≈ Median (very close in value) & represents
highest point on density curve
 All are same basic shape
 More criteria later …
Normal Distributions
 Exact density curve for particular Normal distribution
is described by its mean µ and its standard deviation σ.
Investigating Normal
Distributions…
 See your activity in your data & questions called
“Investigating Normal Distributions”
68-95-99.7 Rule (Empirical Rule)
If Normal distribution with mean µ and standard
deviation σ, then:
68% of observations fall within 1 SD’s of µ
95% of observations fall within 2 SD’s of µ
99.7% of observations fall within 3 SD’s of µ
68-95-99.7 Rule (Empirical Rule)
For Normal Distributions Only
Normal Distributions
Notation: If a distribution is considered Normal, then
notation to identify that distribution as Normal is:
N ( µ, σ )
Normal distribution with mean µ and SD σ.
Girls… who considers themselves
short?
Females in class who consider themselves short (names
and heights in inches)
Actress you think is short? (height in inches)
Athlete you consider short? (height in inches)
Who considers themselves tall? Average height?
“Short” Girls
• Distribution of heights (in inches) of young women
aged 18 to 24 is approximately Normal.
• N (64.5, 2.5)
• So how ‘short’ are these females?
“Short Girls”
The Standard Normal Distribution
 Many Normal density curves/distributions, depending
on µ and σ.
 But ALL standard Normal density curves are the
SAME because they are STANDARDIZED to z-scores.
𝑥−𝜇
𝑧=
𝜎
Standard Normal Distribution:
N (0, 1)
Standard Normal Distribution
Note:
If variable/distribution we standardize has a Normal
distribution, then so does the new variable, z (linear
transformation)
Standard Normal Calculations
Four cases for finding area under a curve
Area to right, area to left, area ‘inside’ and area ‘outside’
(see drawings)
There is a proportion of observations that lie in some
range of values. How do we find this area under the
curve?
Standard Normal Table (Table A)
Area to LEFT of value
Standard Normal Table (Table A)
Practice: Find the area under the curve such that
…(sketch each curve)
• z < 1.5
• z < -2.17
• z<0
• z > -1
• z > 3.04
• z > -2.42
Standard Normal Table (Table A)
More practice… sketch curves.
• -1 < z < 1
• -1.4 < z < 1.07
• 0.49 < z < 0.99
• 0 < z < 3.49
• z < -1 or z > 1.9
• z < 1.55 or z > 3.01
• z < -3 or z > 3
Is cholesterol a problem for boys?
The level of cholesterol in the blood is important
because high cholesterol levels may increase the risk of
heart disease.
The distribution of blood cholesterol levels in a large
population of people of the same age and gender is
roughly Normal. For 14-year old boys, the mean is µ =
170 milligrams of cholesterol per deciliter of blood
(mg/dl) and the standard deviation is σ = 30 mg/dl.
Is cholesterol a problem for boys?
µ = 170 mg/dl
σ = 30 mg/dl
Draw density curve and label with raw data /information.
Is cholesterol a problem for boys?
• Table A is standardized (not raw data).
• Now draw a standardized density curve (change to z-
scores).
• Think about how we are changing the scale on this
density curve… input µ and σ …
x  mean
z
SD
Is cholesterol a problem for boys?
What proportion of cholesterol levels are at or below 170
mg/dl?
Use regular density curve then use the standardized
density curve. Shade in area.
Answer in context, always. Draw and mark up the
density curve, always.
Is cholesterol a problem for boys?
What proportion of cholesterol levels are at or below 140
mg/dl?
Sketch the density curve and shade in area.
Can you use your raw-score density curve to determine
this proportion? Or do you need to use your
standardized density curve?
Is cholesterol a problem for boys?
You must standardize your data value.
µ = 170 mg/dl
σ = 30 mg/dl
x  mean
z
SD
Is cholesterol a problem for boys?
140  170
z
 1
30
Now that we have a z-score, we can use Table A
(remember reads to left)
So, proportion of cholesterol levels at or below 140 is
about 15.87%.
Is cholesterol a problem for boys?
What proportion of cholesterol levels are at or below 110
mg/dl?
What proportion of cholesterol levels are at or below 187
mg/dl?
Share answers and sketches. Include context always!
Sketch always!
Is cholesterol a problem for boys?
What proportion of cholesterol levels are at or above 200
mg/dl?
What proportion of cholesterol levels are at or above 115
mg/dl?
Share answers and sketches. Include context always!
Sketch always!
Is cholesterol a problem for boys?
What proportion of cholesterol levels are between 137
mg/dl and 224 mg/dl?
Share answers and sketches. Include context always!
Sketch always!
Calculator Time 
• Need to know how to do these calculations using Table A.
Definitely.
• But, calculator can also do calculations. Easier, faster.
• Calculator vs. Table A: Are they the same?
• normalcdf (low, high, µ, σ );
second-vars, 2
• If you use your calculator, then you MUST: draw, shade, label
density curve; calculate (define input), answer in context
(always!)
• Let’s try some of these problems with calculator now
Working Backwards…
Given a proportion, find the x, the raw specific value in
your distribution.
x  mean
z
SD
Working Backwards…
What is the cholesterol level for a 14-year old boy for him
to be in the top 25% of population?
µ = 170 mg/dl
σ = 30 mg/dl
Draw density curve, labeled with µ and σ; area shaded.
Working Backwards…
• Remember, Table A is value at or below
• So want to look up 0.75 (not 0.25)
• Look in body of Table A (body is area under density
curve; that’s what we have)
• Area of 0.75 to left corresponds to z = 0.67
Working Backwards …
x  170
0.67 
30
A 14-year old boy would have to have a cholesterol level
of 190.1 mg/dl for him to be in the top 25% of
population.
invnorm (area, µ, σ)
Normal Distribution Criteria
Criteria for a distribution to be considered Normal?
symmetric
Uni-modal
mean ≈ median
68-95-99.7
… and …. one more thing…
Normal Distribution Criteria
 Unless problem situation clearly states that the
distribution is Normal or fairly Normal…
 It is risky to assume a distribution is (fairly) Normal
without creating a Normal Probability Plot (NPP)
 Note: You need raw data to create a NPP
 Often you will be given a NPP and you must interpret
it
Normal Distribution Criteria
If points on NPP lie close to a straight line, then data is
Normal (or fairly Normal).
If points on a NPP do not lie close to a straight line, data
is not Normal.
Outliers appear far away from the overall pattern in NPP.
Normal Probability Plot
Normal Probability Plots
Normal Probability Plot
 Write on board number of pets you own
 Enter into L1
 Create NPP (together)
 According to our NPP, is our data Normal? Why or
why not?
Caution: Normal vs. Fairly Normal
• If problem states Normal, then data is Normal; Exactly.
• If problem doesn’t state Normal: check for … discuss
with partner for 1 minute
• Unimodal, symmetric, mean ≈ median, 68-95-99.7,
and NPP… means FAIRLY NORMAL
• If Normal, mean it. If fairly Normal, mean it.
Homework…
Page 128 #33, 35, 37, 39, 41, 43, 45, 47, 49, 51, 53, 55, 57, 61,
63, 65, 67
MC: #69 – 74
Case Closed…
Pg 126 & 127
On line MC
FRQ’s
Read chapter review pg 134
Chapter 2 ap stats practice test (?)
Chapter 2 review exercises
FRAPPY