statistics block presentation slides day 4

184
Statistics of Measurements and Reliability Kristiaan Schreve Stellenbosch University [email protected] January 26, 2015 Kristiaan Schreve (SU) Stats Block January 26, 2015 1 / 181

Upload: stephan-zeelie

Post on 16-Nov-2015

215 views

Category:

Documents


1 download

DESCRIPTION

Engineering Statistivs Presentation

TRANSCRIPT

  • Statistics of Measurements and Reliability

    Kristiaan Schreve

    Stellenbosch University

    [email protected]

    January 26, 2015

    Kristiaan Schreve (SU) Stats Block January 26, 2015 1 / 181

  • Overview I

    1 Introduction

    2 Some Important Concepts

    3 Excel Demonstration

    4 Graphing DataChoosing the right type of graphGuidelines for creating good scientific graphs

    5 Calculating Averages with Excel

    6 Standard Deviation and Variance

    7 Z Scores

    8 Higher Order Distribution Descriptors

    9 Frequency and Histograms

    10 Box-and-whisker Plots

    11 The Normal Distribution

    12 Confidence LimitsSampling distributions

    Kristiaan Schreve (SU) Stats Block January 26, 2015 2 / 181

  • Overview II

    Central limit theoremLimits of confidencet-distributionNormal distribution and t-distribution confidence limits compared

    13 One-sample Hypothesis TestingSome revisionHypothesis testingSummary of one-sample hypothesis tests

    14 Two-sample Hypothesis TestingHypotheses for two-sample means testingHypotheses for two-sample variance testingSummary of two-sample hypothesis tests

    15 Analysis of Variance - Part OneIntroduction to ANOVASingle factor ANOVA

    Kristiaan Schreve (SU) Stats Block January 26, 2015 3 / 181

  • Overview III

    After the F-test

    16 RegressionLinear regressionTesting hypotheses about regressionExcels R-squaredExcel functions for regressionMultiple regressionGuidelines

    17 CorrelationPearsons correlation coefficientCorrelation and regressionTesting hypotheses about correlation

    18 Uncertainty of MeasurementEvaluation of standard uncertainty

    Type A evaluation of standard uncertaintyType B evaluation of standard uncertainty

    Kristiaan Schreve (SU) Stats Block January 26, 2015 4 / 181

  • Overview IV

    Law of propagation of uncertainty for uncorrelated quantitiesLaw of propagation of uncertainty for correlated quantitiesDetermining expanded uncertaintyReporting uncertaintyExample

    19 Selecting the Right Method

    Kristiaan Schreve (SU) Stats Block January 26, 2015 5 / 181

  • Some Important Concepts I

    Samples and Populations

    Kristiaan Schreve (SU) Stats Block January 26, 2015 6 / 181

    [7]: pp. 10-17

  • Some Important Concepts II

    Probability

    Pr(event) =Number of ways the event can occur

    Total number of possible events

    Conditional Probability

    Pr(event|condition)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 7 / 181

  • Some Important Concepts III

    Hypothesis

    A statement of what you are trying to prove.What is the probability of obtaining the data, given that this hypothesis iscorrect?Can only be rejected.

    Null hypothesis

    H0

    Alternate hypothesis

    H1

    Kristiaan Schreve (SU) Stats Block January 26, 2015 8 / 181

  • Some Important Concepts IV

    Type I error

    Rejecting H0 when you should not.

    Type II error

    Not rejecting H0 when you should.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 9 / 181

  • Excel Demonstration I

    Accessing statistical functions (pp. 37)

    Array functions (pp. 38)

    Just remember to press Ctrl+Shift+Enter to complete the function

    Naming cells or arrays (pp. 42)

    Data analysis tools (pp. 51)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 10 / 181

    [7]: pp. 37-55

  • Graphing Data IChoosing the right type of graph

    Column graphs

    E.g. show percentage change over time for nominal values

    Discrete data: open space between columns

    Continuous data: no space between columns

    Kristiaan Schreve (SU) Stats Block January 26, 2015 11 / 181

    [7]: pp. 65-96

  • Graphing Data IIChoosing the right type of graph

    Avoid 3D. Here it works to show a zero value.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 12 / 181

  • Graphing Data IIIChoosing the right type of graph

    Pie graph

    E.g. show percentages that make up one total

    Avoid 3D effects, it can distort the ability to distinguish between sizesof the slices

    As few slices as possible

    Kristiaan Schreve (SU) Stats Block January 26, 2015 13 / 181

  • Graphing Data IVChoosing the right type of graph

    Kristiaan Schreve (SU) Stats Block January 26, 2015 14 / 181

  • Graphing Data VChoosing the right type of graph

    Line graph

    E.g. show trends, or relationships between parameters

    Figure: Global Temperature

    Kristiaan Schreve (SU) Stats Block January 26, 2015 15 / 181

  • Graphing Data VIChoosing the right type of graph

    Figure: Global Temperature

    Kristiaan Schreve (SU) Stats Block January 26, 2015 16 / 181

  • Graphing Data VIIChoosing the right type of graph

    Bar graph

    E.g. make a point about reaching a goal

    Good if the labels on the horizontal axis take too much space

    Arrange in ascending/descending order whenever appropriate

    Kristiaan Schreve (SU) Stats Block January 26, 2015 17 / 181

  • Graphing Data VIIIChoosing the right type of graph

    Kristiaan Schreve (SU) Stats Block January 26, 2015 18 / 181

  • Graphing Data IXChoosing the right type of graph

    Linear regression

    E.g. show relationship between parameters

    Use with great care!

    Kristiaan Schreve (SU) Stats Block January 26, 2015 19 / 181

  • Graphing Data XChoosing the right type of graph

    Figure: Regression example

    Kristiaan Schreve (SU) Stats Block January 26, 2015 20 / 181

  • Graphing Data IGuidelines for creating good scientific graphs

    Avoid colour graphs

    Black & white printersColour blindness: up to 10% of male population suffer from red-greedcolour blindness (www.colour-blindness.com)Using colour in presentations is OK.

    Dont wear out the viewers eyes

    Pie graphs: avoid too many slicesLine graphs: avoid too many series/lines

    Avoid unnecessary junk - it distracts from the main message (gridlines, 3D effects, etc.)

    Include all information (axis labels, units, appropriate legends)

    Excels Smooth scatter plots are almost always a bad idea

    Independent variable on horizontal axis

    Dependent variable on vertical axis

    Kristiaan Schreve (SU) Stats Block January 26, 2015 21 / 181

    Not in textbook

  • Graphing Data IIGuidelines for creating good scientific graphs

    Use regression with great care

    The order of the regression must be appropriate for the number of datapoints and trend in the data, e.g. dont fit a quadratic polynomial toonly 3 data points.In general, dont extrapolate beyond the data range.Give an indication of the goodness of fit, see Figure 3.Give the confidence limits, see Figure 18.Check that the regression curve gives a valid prediction, e.g. a curvefitted to data that predicts temperature in Kelvin, cannot give negativevalues.Too large samples can be bad (see pp. 417 in textbook)

    When plotting experimental data, use markers, with no lines betweenthem, see Figure 17.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 22 / 181

  • Graphing Data IIIGuidelines for creating good scientific graphs

    Whenever appropriate, include variability in your graphs (errorbars...). Also indicate what the error bars mean (95% confidence,min/max range, standard deviation, etc.), see Figure 17.

    Graph a categorical (discrete) variable as though it is a quantitativevariable is just wrong (see Fig 19-1 in the textbook).

    Choose the range of the variables appropriately, see Figure 5.

    When the dependent and independent variable have the same unit,make sure that the axes have the same scale, see Figure 3.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 23 / 181

  • Graphing Data IVGuidelines for creating good scientific graphs

    Figure: An example of how NOT to plot categorical data.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 24 / 181

  • Graphing Data VGuidelines for creating good scientific graphs

    Figure: Use appropriate vertical range [4]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 25 / 181

  • Graphing Data VIGuidelines for creating good scientific graphs

    Table: Data set A, Running Times. [3]

    Name Time [s]

    Thomas 19Anthony 26Emma 18Jaspal 19.6Lisa 21Meena 22Navtej 27Nicola 23Sandeep 17Tanya 23

    Kristiaan Schreve (SU) Stats Block January 26, 2015 26 / 181

  • Graphing Data VIIGuidelines for creating good scientific graphs

    Figure: Charts based on data in Table 1 [3]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 27 / 181

  • Graphing Data VIIIGuidelines for creating good scientific graphs

    Horizontal bars useful for large number of bars

    Also useful if there is too much text for the horizontal axis

    Rank of each athlete is clearly visible on bottom graph

    None of the graphs shows the distribution of the data

    Kristiaan Schreve (SU) Stats Block January 26, 2015 28 / 181

  • Graphing Data IXGuidelines for creating good scientific graphs

    Figure: Pie chart based on data in Table 1 [3]

    Pie graphs generally OK for showing discrete data

    Must show parts of a hole - not in this case!

    Kristiaan Schreve (SU) Stats Block January 26, 2015 29 / 181

  • Graphing Data XGuidelines for creating good scientific graphs

    Figure: Histogram showing distribution of data in Table 1 [3]

    Histograms show continuous data - no spaces between the bars.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 30 / 181

  • Graphing Data XIGuidelines for creating good scientific graphs

    Figure: Correct histogram showing distribution of data in Table 1 [3]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 31 / 181

  • Graphing Data XIIGuidelines for creating good scientific graphs

    Table: Data set B: Wind in January [3]

    Wind type Days

    Strong wind 10Calm 5Gale 7Light breeze 9

    Total 31

    Kristiaan Schreve (SU) Stats Block January 26, 2015 32 / 181

  • Graphing Data XIIIGuidelines for creating good scientific graphs

    Figure: Bar chart based on data in Table 2 [3]

    Discrete data should have spaces between columns

    Sequence of wind categories is not helpful

    Kristiaan Schreve (SU) Stats Block January 26, 2015 33 / 181

  • Graphing Data XIVGuidelines for creating good scientific graphs

    Figure: Bar chart based on the data in Table 2 [3]

    Meaningless to compare Total to wind categories. Looks likeanother category.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 34 / 181

  • Graphing Data XVGuidelines for creating good scientific graphs

    Figure: Bar chart based on the data in Table 2 [3]

    Note discontinuity at start of Y-axis. This distorts the effect of thecolumns.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 35 / 181

  • Graphing Data XVIGuidelines for creating good scientific graphs

    Figure: Correct bar chart based on the data in Table 2 [3]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 36 / 181

  • Graphing Data XVIIGuidelines for creating good scientific graphs

    Figure: This is how you show a discontinuity in an axis.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 37 / 181

  • Graphing Data XVIIIGuidelines for creating good scientific graphs

    Figure: Pie chart based on the data in Table 2 [3]

    Data in Table 2 is ideal for pie charts.

    Including the Total makes no sense in the pie graph since it representscomponents of the total.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 38 / 181

  • Graphing Data XIXGuidelines for creating good scientific graphs

    Figure: Correct pie chart based on the data in Table 2 [3]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 39 / 181

  • Graphing Data XXGuidelines for creating good scientific graphs

    Figure: Graphing experimental data. Error bars show the measurement errorrange.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 40 / 181

  • Graphing Data XXIGuidelines for creating good scientific graphs

    Figure: Graphing regression curves

    Kristiaan Schreve (SU) Stats Block January 26, 2015 41 / 181

  • Graphing Data XXIIGuidelines for creating good scientific graphs

    Figure: Example of a bad graph

    Kristiaan Schreve (SU) Stats Block January 26, 2015 42 / 181

  • Graphing Data XXIIIGuidelines for creating good scientific graphs

    Figure: Example of a bad graph

    Kristiaan Schreve (SU) Stats Block January 26, 2015 43 / 181

  • Graphing Data XXIVGuidelines for creating good scientific graphs

    Figure: Example of a bad graph

    Kristiaan Schreve (SU) Stats Block January 26, 2015 44 / 181

  • Calculating averages with Excel I

    Mean (Excel: AVERAGE, AVERAGEA, AVERAGEIF, AVERAGEIFS,TRIMMEAN)

    (We dont do geometric mean or harmonic mean on pp 106-107)

    Median (Excel: MEDIAN)

    Mode (Excel: MODE.MULT, MODE.SNGL)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 45 / 181

    [7]: pp. 97-112

  • Standard Deviation and Variance I

    Population variance

    2 =

    (X X )2

    N

    Excel function: VAR.P and VARPA

    Kristiaan Schreve (SU) Stats Block January 26, 2015 46 / 181

    [7]: pp. 113-123

  • Standard Deviation and Variance II

    Sample variance

    s2 =

    (X X )2

    N 1

    Excel functions: VAR.S and VARA

    Why divide by (N 1)? Calculating the average of the sample, X ,effectively takes away one degree of freedom.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 47 / 181

  • Standard Deviation and Variance III

    Standard deviation of a population

    =2 =

    (X X )2

    N

    Excel function: STDEV.P and STDEVPA

    NOTE: the standard deviation has the same unit as the originalmeasurements

    Kristiaan Schreve (SU) Stats Block January 26, 2015 48 / 181

  • Standard Deviation and Variance IV

    Standard deviation of a sample

    s =

    s2 =

    (X X )2N 1

    Excel function: STDEV.S and STDEVA

    NOTE: whenever presenting a mean, always provide a standarddeviation as well

    Kristiaan Schreve (SU) Stats Block January 26, 2015 49 / 181

  • Z Scores I

    How do you compare scores in one year to another year for, say,Mechatronics 424?Z scores take the mean as a zero point and the standard deviation as aunit of measure. Therefore, for a sample

    z =X X

    s

    and for a population

    z =X

    Kristiaan Schreve (SU) Stats Block January 26, 2015 50 / 181

    [7]: pp. 131-145

  • Z Scores II

    IQ scores are typically transformed Z scores

    IQ = 16z + 100

    The implication of this formula: mean IQ score is 100, standard deviationof IQ scores is 16.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 51 / 181

  • Z Scores III

    Excel function related to Z scores

    STANDARDIZE

    PERCENTILE.EXC, PERCENTILE.INC

    PERCENTRANK.EXC, PERCENTRANK.INC

    QUARTILE.EXC, QUARTILE.INC

    Kristiaan Schreve (SU) Stats Block January 26, 2015 52 / 181

  • Higher Order Distribution Descriptors I

    Descriptors

    Variance: Describes the spread in the data.

    Skewness: Describes how symmetrically the data is distributed.

    Kurtosis: Describes whether or not there is a peak in the distributionclose to the mean.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 53 / 181

    [7]: pp. 152-156

  • Higher Order Distribution Descriptors II

    SkewnessExcel function: SKEW

    skewness =

    (X X )3

    (N 1)s3

    Kristiaan Schreve (SU) Stats Block January 26, 2015 54 / 181

  • Higher Order Distribution Descriptors III

    KurtosisExcel function: KURT

    kurtosis =

    (X X )4

    (N 1)s4 3

    Kristiaan Schreve (SU) Stats Block January 26, 2015 55 / 181

  • Frequency and Histograms I

    Frequency: Excel function: FREQUENCY - Remember: it is an arrayfunction.

    Histogram: Use the Data Analysis Tool

    Kristiaan Schreve (SU) Stats Block January 26, 2015 56 / 181

    [7]: pp. 156-160

  • Frequency and Histograms II

    Histrogram: Shows the number of items in a certain category.

    Frequency distribution: Shows the percentage of the total in a certaincategory, i.e. the histogram number for the category isdivided by the total number of samples in the histogram.

    Histograms and frequency distributions are good to study centraltendencies, i.e. the tendency of all values in a sample of random variablesto be scattered around a certain value.The following is a guideline for the number of intervals K (from [2])

    K = 1.87(N 1)0.4 + 1

    As N, the number of measurements, becomes large, choose K

    N [2]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 57 / 181

  • Box-and-whisker Plots I

    Figure: Box-and-whisker plot generated with Python

    Kristiaan Schreve (SU) Stats Block January 26, 2015 58 / 181

    Not in textbook

  • Box-and-whisker Plots II

    Gives an indication of the distribution of the data

    Compare with histogram

    Useful to compare different distributions

    Matlab and Python both have useful tools to create these tools. Moredifficult with Excel.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 59 / 181

  • Box-and-whisker Plots III

    Figure: Box-and-whisker plot generated with Python

    Kristiaan Schreve (SU) Stats Block January 26, 2015 60 / 181

  • Box-and-whisker Plots IV

    Example (Showing results of robot movement - Table)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 61 / 181

  • Box-and-whisker Plots V

    Example (Showing results of robot movement - Box-and-whisker plot)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 62 / 181

  • The Normal Distribution I

    f (x) =1

    2e

    (x)2

    22

    f (x) Probability density

    Standard deviation

    Mean

    Kristiaan Schreve (SU) Stats Block January 26, 2015 63 / 181

    [7]: pp. 173-183

  • The Normal Distribution II

    Properties of the normal curve [8], pp. 141

    Point where curve reaches its maximum is at x =

    Curve is symmetric about a vertical line through x =

    Points of inflection at x = . It is concave downward if < x < + , concave upwards otherwise.Approaches horizontal axis asymptotically in both directions awayfrom x =

    Total area under the curve above the horizontal axis is 1.

    Other names for the normal curve

    Gaussian curve

    Bell curve

    Kristiaan Schreve (SU) Stats Block January 26, 2015 64 / 181

  • The Normal Distribution III

    Standard Normal Distribution

    = 0

    = 1

    If Z scores are normally distributed, it will fit the standard normaldistribution.

    Normal distribution of IQ scores

    Kristiaan Schreve (SU) Stats Block January 26, 2015 65 / 181

  • The Normal Distribution IV

    Cumulative Normal DistributionGives the cumulative area under the normal distribution.

    F (x) =1

    2

    x

    e(x)2

    22

    Figure: Cumulative Normal Distribution

    Kristiaan Schreve (SU) Stats Block January 26, 2015 66 / 181

  • The Normal Distribution V

    Vertical axis gives area under normal distribution to the left of x .

    Asymptotically approaches 1.

    Areas under the normal distribution is used to calculate probabilities asfollows:Probability of an event between two values:

    P(x1 < x < x2) =1

    2

    x2x1

    e(x)2

    22

    Kristiaan Schreve (SU) Stats Block January 26, 2015 67 / 181

  • The Normal Distribution VI

    Figure: Probability of event between x1 and x2

    Grey area is the probability of an event, x , between x1 and x2, i.e.P(x1 < x < x2)

    F (x1) is probability of an event, x , less than x1, i.e. P(x < x1). Thisis found from the cumulative distribution function.

    Therefore P(x1 < x < x2) = F (x2) F (x1)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 68 / 181

  • The Normal Distribution VII

    In Excel: P(x1 < x < x2) =NORM.DIST(x2,mean,standarddeviation,TRUE) - NORM.DIST(x1,mean,standard deviation,TRUE)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 69 / 181

  • The Normal Distribution VIII

    Probability of an event less than a value:

    P(x < x1) =1

    2

    x1

    e(x)2

    22

    Figure: Probability of event less than x1

    Kristiaan Schreve (SU) Stats Block January 26, 2015 70 / 181

  • The Normal Distribution IX

    Grey area is the probability of an event, x , less than x1, i.e.P(x < x1) = F (x1)

    In Excel: P(x < x1) =NORM.DIST(x1,mean,standarddeviation,TRUE)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 71 / 181

  • The Normal Distribution X

    Probability of an event more than a value:

    P(x > x1) =1

    2

    x1

    e(x)2

    22

    Figure: Probability of event more than x1

    Kristiaan Schreve (SU) Stats Block January 26, 2015 72 / 181

  • The Normal Distribution XI

    Grey area is the probability of an event, x , more than x1, i.e.P(x < x1) = 1 F (x1)Note: the cumulative distribution gives the area to the left of x1.Since we are interest in the area to the right, we must subtract F (x1)from 1.

    In Excel: P(x < x1) = 1NORM.DIST(x1,mean,standarddeviation,TRUE)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 73 / 181

  • The Normal Distribution XII

    Excel functions

    NORM.DIST, NORM.S.DIST

    NORM.INV, NORM.S.INV

    Use NORM.DIST(x,mean,standard deviation,TRUE) for thecumulative distribution function

    Use NORM.DIST(x,mean,standard deviation,FALSE) for theprobability density function

    Kristiaan Schreve (SU) Stats Block January 26, 2015 74 / 181

  • The Normal Distribution XIII

    Example (Interpreting the normal curve [3], pp. 291)

    The example refers to the distribution of normal IQ scores.

    What proportion of the population measures an IQ less than 105?

    90% of the population will have an IQ below what value?

    The top 1% of the population will have an IQ above what value?

    What range of IQs define the 95% interval?

    Someone with a measured IQ in excess of 140 is considered eligiblefor MENSA. What is the probability that a randomly chosen personfalls in this category?

    Kristiaan Schreve (SU) Stats Block January 26, 2015 75 / 181

  • Confidence Limits ISampling distributions

    A sampling distribution is the distribution of all possible values of astatistic for a given sample size.

    Remember, the statistic, can be anything, e.g. the mean or thestandard deviation.We are talking about a statistic, because we are talking about samples,not populations, which would have parameters.In other words, if we repeatedly take samples from the samepopulation, we would get a slightly different statistic, say the mean,each time. The sampling distribution is the description of all thepossible values that the statistic can have.

    The sampling distribution therefore has its own mean and standarddeviation.

    The mean of the sampling distribution of the mean is x .

    The standard deviation of the sampling distribution is called thestandard error.

    The standard error is denoted as x .

    Kristiaan Schreve (SU) Stats Block January 26, 2015 76 / 181

    [7]: pp. 187-189

  • Confidence Limits ICentral limit theorem

    Theorem (Central limit theorem)

    If X is the mean of a random sample of size n taken from a populationwith mean and finite variance 2, then the limiting form of thedistribution of

    Z =X

    n

    as n, is the standard normal distribution with = 0 and = 1. [8]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 77 / 181

    [7]: pp. 189-195

  • Confidence Limits IICentral limit theorem

    Implications of the central limit theorem.

    Sampling distribution of the mean is approximately a normaldistribution if sample size is large enough (i.e. 30 or more samples).

    The mean of the sampling distribution mean is the same as thepopulation mean, = x .

    The standard error (or standard deviation of the sampling distributionmean) is equal to the population standard deviation, divided by thesquare root of the sample size, x =

    N

    .

    The population does not have to be a normal distribution.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 78 / 181

  • Confidence Limits ILimits of confidence

    Theorem (Confidence interval of ; known)

    If x is the mean of a random sample of size n from a population withknown variance 2, a (1 )100% confidence interval for is given by

    x z/2n< < x + z/2

    n

    where z/2 is the z value leaving an area of /2 to the right. [8]

    Note: for non-normal populations, n > 30, still give good resultsthanks to the central limit theorem.

    Work through the example on pp. 195-198.

    Excel function: CONFIDENCE.NORM, CONFIDENCE.T

    Note: only use CONFIDENCE.NORM when n > 30 and if thepopulation is normally distributed.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 79 / 181

    [7]: pp. 195-199

  • Confidence Limits It-distribution

    What if the sample size is < 30 or the distribution is not normal?t-distribution works better.

    t =x s/

    n

    Kristiaan Schreve (SU) Stats Block January 26, 2015 80 / 181

    [7]: pp. 199-201

  • Confidence Limits IIt-distribution

    Shape of the distribution depends on the degrees of freedom or df.

    Figure: From [6]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 81 / 181

  • Confidence Limits IIIt-distribution

    Theorem (Confidence interval for ; unknown)

    If x and s are the mean and standard deviation of a random sample from anormal population with unknown variance 2, a (1 )100% confidenceinterval for is given by

    x t/2sn< < x + t/2

    sn

    where t/2 is the t value with n 1 degrees of freedom, leaving an area of/2 to the right. [8]

    Excel functions:

    T.INV, T.INV.2T

    T.DIST, T.DIST.2T, T.DIST.RT

    Repeat example on pp. 195-198, but use t-scores.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 82 / 181

  • Confidence Limits INormal distribution and t-distribution confidence limits compared

    Figure: Comparison of 90% confidence limits for the normal and t-distributions

    Kristiaan Schreve (SU) Stats Block January 26, 2015 83 / 181

    Not in textbook

  • Confidence Limits IINormal distribution and t-distribution confidence limits compared

    Note: the range of for the t-distribution is much larger than for the normaldistribution.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 84 / 181

  • One-sample Hypothesis Testing ISome revision

    Hypothesis Essentially a guess about the way the world works.

    Null hypothesis H0 The data wont show anything new or interesting.Any deviation from the norm, is strictly due to chance.

    Alternative hypothesis H1 Explains the world differently.

    H0 Can only reject or not reject. Can never accept a hypothesis.

    Type I error Incorrectly rejecting H0

    Type II error Not rejecting H0 when it should have been rejected.

    Hypothesis testing is about setting criteria for rejecting H0. This sets theprobability of making a Type I error. The probability is called .

    Kristiaan Schreve (SU) Stats Block January 26, 2015 85 / 181

    [7]: pp. 203-204

  • One-sample Hypothesis Testing IHypothesis testing

    Figure: From [6]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 86 / 181

    [7]: pp. 205-209

  • One-sample Hypothesis Testing IIHypothesis testing

    and are areas that show the probabilities of making decisionerrors.

    is typically 0.05. This corresponds to a 5% chance of making aType I error. It also represents the likelihood that the sample mean x ,is in that shaded region.

    represents the likelihood that x is in the H1 distribution.

    is never set beforehand. It depends on the distributions and where is set.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 87 / 181

  • One-sample Hypothesis Testing IIIHypothesis testing

    Example on pp 207-209

    Kristiaan Schreve (SU) Stats Block January 26, 2015 88 / 181

  • One-sample Hypothesis Testing IVHypothesis testing

    Guidelines for writing the hypotheses [8], pp. 299

    For a simple direction such as more than, less than, superior to,inferior to, etc., state H1 as an appropriate inequality (< or >). H0will be stated with the = sign.

    If the claim suggests an equality and direction such as at least, equalto or greater, at most, no more than, etc., then state H0 using (6 or>). State H1 with the opposite inequality (< or >)sign.

    If no direction is claimed (two-tailed tests), state H1 with 6= and H0with =.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 89 / 181

  • One-sample Hypothesis Testing VHypothesis testing

    One-sided (or one-tailed) tests are stated as

    H0 : = (or )0H1 : > 0

    or

    H0 : = (or )0H1 : < 0

    Two-sided (or two-tailed) tests are stated as

    H0 : = 0

    H1 : 6= 0

    Kristiaan Schreve (SU) Stats Block January 26, 2015 90 / 181

  • One-sample Hypothesis Testing VIHypothesis testing

    Reject H0, with variance known, if x > b or x < a, where

    a = 0 z/2n

    b = 0 + z/2n

    Figure: From [8]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 91 / 181

  • One-sample Hypothesis Testing VIIHypothesis testing

    The above is for a two-tailed test. A similar test can be formulated for aone-tailed hypothesis.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 92 / 181

  • One-sample Hypothesis Testing VIIIHypothesis testing

    Tests on a single mean (variance unknown)Rejection of H0 at significance level for

    t =x 0s/

    n

    when

    t > t/2,n1 or t < t/2,n1

    Excel function: T.DIST

    Kristiaan Schreve (SU) Stats Block January 26, 2015 93 / 181

  • One-sample Hypothesis Testing IXHypothesis testing

    Hypotheses involving variancesWhat if the hypothesis uses a variance rather than a mean?

    H0 : 2 = (or )20

    H1 : 2 > 20

    or

    H0 : 2 = (or )20

    H1 : 2 < 20

    Kristiaan Schreve (SU) Stats Block January 26, 2015 94 / 181

  • One-sample Hypothesis Testing XHypothesis testing

    Two-sided (or two-tailed) tests are stated as

    H0 : 2 = 20

    H1 : 2 6= 20

    Kristiaan Schreve (SU) Stats Block January 26, 2015 95 / 181

  • One-sample Hypothesis Testing XIHypothesis testing

    Hypotheses involving variancesThe chi-square distribution is used in the hypothesis test

    Like the t-distribution, it also involves the degrees of freedom in thesample (df=n-1).

    2 =(N 1)s2

    2

    Kristiaan Schreve (SU) Stats Block January 26, 2015 96 / 181

  • One-sample Hypothesis Testing XIIHypothesis testing

    Kristiaan Schreve (SU) Stats Block January 26, 2015 97 / 181

  • One-sample Hypothesis Testing XIIIHypothesis testing

    H0 is rejected at significance level under the following conditionsOne-tailed hypothesis

    For H1 : 2 < 20

    2 < 21

    For H1 : 2 > 20

    2 > 2

    Two-tailed hypothesis

    2 < 21/2 or 2 > 2/2

    Excel functions

    CHISQ.DIST, CHISQ.DIST.RT

    CHISQ.INV, CHISQ.INV.RT

    CHISQ.TEST

    Kristiaan Schreve (SU) Stats Block January 26, 2015 98 / 181

  • One-sample Hypothesis Testing ISummary of one-sample hypothesis tests

    H0 Value of Test Statistic H1 Critical Region

    = 0 or 0 z = x0/n known < 0 z < z = 0 or 0 > 0 z > z = 0 6= 0 z < z/2

    and z > z/2 = 0 or 0 t = x0s/n unknown < 0 t < t = 0 or 0 df = n 1 > 0 t > t = 0 6= 0 t < t/2

    and t > t/2

    2 = 20 or 2 20 2 =

    (n1)s22

    2 < 20 2 < 2,df

    2 = 20 or 2 20 df = n 1 2 > 20 2 > 2,df

    2 = 20 2 6= 20 2 < 2/2,df

    and 2 > 2/2,df

    Kristiaan Schreve (SU) Stats Block January 26, 2015 99 / 181

    Not in textbook

  • Two-sample Hypothesis Testing IHypotheses for two-sample means testing

    Objective: does the two samples come from two different populations ornot?

    Null hypothesis: Difference between the two samples are strictly due tochance. They come from the same population.

    Alternative hypothesis: There is a real difference between the samples.They come from different populations.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 100 / 181

    [7]: pp. 219-235

  • Two-sample Hypothesis Testing IIHypotheses for two-sample means testing

    One-tailed tests

    H0 : 1 2 = 0H1 : 1 2 > 0

    or

    H0 : 1 2 = 0H1 : 1 2 < 0

    Two-tailed tests

    H0 : 1 2 = 0H1 : 1 2 6= 0

    Kristiaan Schreve (SU) Stats Block January 26, 2015 101 / 181

  • Two-sample Hypothesis Testing IIIHypotheses for two-sample means testing

    Hypothesis testing procedure

    1 Write the hypotheses, H0 and H1

    2 Select the probability for making a Type I error

    3 Calculate 1, 2, 1 and 2

    4 Compare the test statistic to a sampling distribution of test statistics(see next slides)

    5 Reject (or do not reject) H0

    Kristiaan Schreve (SU) Stats Block January 26, 2015 102 / 181

  • Two-sample Hypothesis Testing IVHypotheses for two-sample means testing

    For this type of testing, the sampling distribution of the difference betweenmeans is needed.The sampling distribution of the difference between means is thedistribution of all possible values of differences between pairs of samplemeans with the sample sizes held constant from pair to pair.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 103 / 181

  • Two-sample Hypothesis Testing VHypotheses for two-sample means testing

    Figure: From [6]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 104 / 181

  • Two-sample Hypothesis Testing VIHypotheses for two-sample means testing

    NOTE:

    All samples from population 1 must have the same size.

    All samples from population 2 must have the same size.

    The two sample sizes are not necessarily equal.

    Characteristics of the sampling distribution of the difference betweenmeans according to the Central Limit Theorem

    For large samples, it is approximately normally distributed.

    For normally distributed populations, it is normally distributed.

    The mean is the difference between the population meansx1x2 = 1 2The standard deviation (or standard error of the difference between

    means) is x1x2 =

    21N1

    +22N2

    Kristiaan Schreve (SU) Stats Block January 26, 2015 105 / 181

  • Two-sample Hypothesis Testing VIIHypotheses for two-sample means testing

    Tests on two means (variance known). Rejection of H0 at significancelevel for

    z =(x1 x2) (1 2)

    21N1

    +22N2

    when H1 : 1 2 < 0 (one tailed tests)

    z < z

    or H1 : 1 2 > 0 (one tailed tests)

    z > z

    or (two tailed tests)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 106 / 181

  • Two-sample Hypothesis Testing VIIIHypotheses for two-sample means testing

    z > z/2 or z < z/2

    Kristiaan Schreve (SU) Stats Block January 26, 2015 107 / 181

  • Two-sample Hypothesis Testing IXHypotheses for two-sample means testing

    Tests on two means (variance unknown, but equal)Central Limit Theorem no longer applicable. Now, rather use thet-distribution.Calculate the pooled estimate of the standard error of the differencebetween means.

    s2p =(N1 1)s21 + (N2 1)s22

    (N1 1) + (N2 1)df = (N1 1) + (N2 1)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 108 / 181

  • Two-sample Hypothesis Testing XHypotheses for two-sample means testing

    Rejection of H0 at significance level for

    t =(x1 x2) (1 2)

    sp

    1N1

    + 1N2

    when H1 : 1 2 < 0 (one tailed tests)

    t < t,dfor H1 : 1 2 > 0 (one tailed tests)

    t > t,df

    or (two tailed tests)

    t > t/2,df or t < t/2,df

    Kristiaan Schreve (SU) Stats Block January 26, 2015 109 / 181

  • Two-sample Hypothesis Testing XIHypotheses for two-sample means testing

    Tests on two means (variance unknown, and unequal)Same test as the previous test (Two means, variance unknown), but thedegrees of freedom will be adjusted as follows [5], pp. 356:

    df =(s21/n1 + s

    22/n2)

    2[(s21/n1)

    2

    n11 +(s22/n2)

    2

    n21

    ]df will in general not be an integer. Round down to nearest integer to uset table.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 110 / 181

  • Two-sample Hypothesis Testing XIIHypotheses for two-sample means testing

    Rejection of H0 at significance level for

    t =(x1 x2) (1 2)

    s21N1

    +s22N2

    when H1 : 1 2 < 0 (one tailed tests)

    t < t,dfor H1 : 1 2 > 0 (one tailed tests)

    t > t,df

    or (two tailed tests)

    t > t/2,df or t < t/2,df

    Kristiaan Schreve (SU) Stats Block January 26, 2015 111 / 181

  • Two-sample Hypothesis Testing XIIIHypotheses for two-sample means testing

    Hypothesis testing of paired samples [5], pp. 359One-tailed test

    H0 :(1 2) = D0H1 :(1 2) > D0

    [or H1 : (1 2) < D0]

    Two-tailed test

    H0 :(1 2) = D0H1 :(1 2) 6= D0

    t =d D0sd/

    n; df = n 1

    Assumptions

    The relative frequency distribution of the population of differences isapproximately normal.

    The paired differences are randomly selected from the population ofdifferences.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 112 / 181

  • Two-sample Hypothesis Testing IHypotheses for two-sample variance testing

    Comparing the variances of two samplesTwo-tailed hypothesis

    H0 :21 =

    22

    H1 :21 6= 22

    To compare variances of two samples, the F-test is used.The test statistic is the F-ratio

    F =s2as2b

    where s2a > s2b

    To draw a conclusion, the F-distribution is needed.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 113 / 181

    [7]: pp. 239-248

  • Two-sample Hypothesis Testing IIHypotheses for two-sample variance testing

    Figure: From [6]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 114 / 181

  • Two-sample Hypothesis Testing IIIHypotheses for two-sample variance testing

    NOTE

    The distribution depends on two dfs, dfa and dfb.

    dfa = na 1dfb = nb 1

    Kristiaan Schreve (SU) Stats Block January 26, 2015 115 / 181

  • Two-sample Hypothesis Testing IVHypotheses for two-sample variance testing

    Rejection of H0 at significance level when

    F > F1/2(dfa, dfb) and F < F/2(dfa, dfb)

    The F-test can be used to see if the variances of two samples differsignificantly before deciding which t-test to use for testing the differencebetween the means. In this case, we are not looking for small differencesbetween the variances, therefore it is desirable to choose a higher , say0.2 for the variance test.Excel functions

    F.TEST

    F.DIST, F.DIST.RT

    F.INV, F.INV.RT

    Data analysis tool: F-test two sample for variances

    Kristiaan Schreve (SU) Stats Block January 26, 2015 116 / 181

  • Two-sample Hypothesis Testing ISummary of two-sample hypothesis tests

    H0 Value of Test Statistic H1 Critical Region

    1 2 = 0 z = (x1x2)(12)2

    1N1

    +2

    2N2

    1 2 < 0 z < z

    1 and 2 known 1 2 > 0 z > z1 2 6= 0 z < z/2

    and z > z/21 2 = 0 t = (x1x2)(12)

    sp

    1N1

    + 1N2

    1 2 < 0 t < t,df

    1 and 2 unknown 1 2 > 0 t > t,dfbut equal 1 2 6= 0 t < t/2,dfdf = N1 + N2 2 and t > t/2,df

    Kristiaan Schreve (SU) Stats Block January 26, 2015 117 / 181

    Not in textbook

  • Two-sample Hypothesis Testing IISummary of two-sample hypothesis tests

    H0 Value of Test Statistic H1 Critical Region

    1 2 = 0 t = (x1x2)(12)s21

    N1+

    s22

    N2

    1 2 < 0 t < t,df

    1 and 2 unknown 1 2 > 0 t > t,dfand unequal 1 2 6= 0 t < t/2,dfdf =

    (s21/n1+s22/n2)

    2[(s2

    1/n1)

    2

    n11+

    (s22/n2)

    2

    n21

    ] and t > t/2,df1 2 = D0 t = dD0sd/n 1 2 < D0 t < t,df

    df = n 1 1 2 > D0 t > t,df1 2 6= D0 t < t/2,df

    and t > t/2,df

    Kristiaan Schreve (SU) Stats Block January 26, 2015 118 / 181

  • Two-sample Hypothesis Testing IIISummary of two-sample hypothesis tests

    H0 Value of Test Statistic H1 Critical Region

    21 = 22 F =

    s2as2b

    21 < 22 F < F(dfa, dfb)

    dfa = na 1 21 > 22 F > F1(dfa, dfb)dfb = nb 1 21 6= 22 F < F/2(dfa, dfb)

    andF > F1/2(dfa, dfb)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 119 / 181

  • Analysis of Variance - Part OneIntroduction to ANOVA

    Example (Based on Table 12-1, [6])

    Table: Data from Three Training Methods

    Method 1 Method 2 Method 395 83 6892 89 7589 85 7990 89 7499 81 7588 89 8196 90 7398 82 7795 84

    80

    Mean 93.44 85.20 75.25Variance 16.28 14.18 15.64Standard Deviation 4.03 3.77 3.96

    Kristiaan Schreve (SU) Stats Block January 26, 2015 120 / 181

    [7]: pp. 251-253

  • Analysis of Variance - Part One IIntroduction to ANOVA

    Example (Continued...)

    Hypothesis

    H0 :1 = 2 = 3

    H1 :Not H0

    = 0.05

    Performing multiple t-tests possibly sets us up for a disaster. Lets see why:

    Chance of NOT making a Type I error with one comparison, with asignificance level of = 0.05 is 95%.

    So, for 3 samples, 3 tests must be done: Method 1 Method 2,Method 1 Method 3 and Method 2 Method 3.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 121 / 181

  • Analysis of Variance - Part One IIIntroduction to ANOVA

    Each test will have a probability of NOT making a Type I error ofpi = 95%.

    The combined probability of NOT making a Type I error is therefore

    p(p1 p2 p3) = 0.95 0.95 0.95 = 0.86

    Therefore, the combined chance (note, this is covered in chapter 16)of making a Type I error is

    1 p(p1 p2 p3) = 0.14 or 14%

    In general, the chance of making a Type I error increases as1 (1 )N where N is the number of t-tests.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 122 / 181

  • Analysis of Variance - Part One IIIIntroduction to ANOVA

    Table: Increasing chance of making a Type I error for multiple t-tests, from [6]

    Number of samples t Number of tests Pr(at least one significant t)3 3 0.144 6 0.265 10 0.406 15 0.547 21 0.668 28 0.769 36 0.84

    10 45 0.90

    Kristiaan Schreve (SU) Stats Block January 26, 2015 123 / 181

  • Analysis of Variance - Part One IVIntroduction to ANOVA

    The idea with ANOVA is to separate the total variability into the followingcomponents [8]

    1 Variability between samples, measuring systematic and randomvariation.

    2 Variability within samples, measuring only random variation.

    3 Finally, determine if component 1 is more significant than component2.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 124 / 181

  • Analysis of Variance - Part One VIntroduction to ANOVA

    The idea can also be illustrated with the following plots.The figure shows a single factor experiment at two levels, i.e. two treatments.

    Figure: From [5] pp. 627

    Is there sufficient evidence to indicate a difference between the populationmeans?

    Kristiaan Schreve (SU) Stats Block January 26, 2015 125 / 181

  • Analysis of Variance - Part One VIIntroduction to ANOVA

    How about these two plots?

    Figure: From [5] pp. 627

    What statistics of the two samples in these plots did we intuitively use tomake a decision on the difference between the population means?

    Kristiaan Schreve (SU) Stats Block January 26, 2015 126 / 181

  • Analysis of Variance - Part One ISingle factor ANOVA

    Recall the definition of the sample variance

    s2 =

    (x x)2

    N 1

    This is often called the Mean Square, because it is almost a mean ofsquared deviations.

    Numerator: sum of squares=

    (x x)2

    Denominator: degrees of freedom, df

    Kristiaan Schreve (SU) Stats Block January 26, 2015 127 / 181

    [7]: pp. 253-265

  • Analysis of Variance - Part One IISingle factor ANOVA

    We can calculate the following variances (or mean squares) (alternativedefinitions are derived from [8], pp 472).

    MST =SSTdfT

    =

    ki=1

    nij=1 y

    2ij

    (ki=1

    nij=1 yij

    )2(ki=1 ni

    )1(k

    i=1 ni) 1

    Mean Square for all the data.

    Subscript T is for total data.

    Numerator: Total sum of squares

    Denominator: Total degrees of freedom. All the data - 1.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 128 / 181

  • Analysis of Variance - Part One IIISingle factor ANOVA

    In the second equation

    k is the number of samples or treatmentsni is the number of data points in the i

    th sampleyij is the j

    th data point, from the i th sample.

    MSW =SSWdfW

    =

    ki=1

    (nij=1 yij

    )2ni

    (k

    i=1

    nij=1 yij

    )2(ki=1 ni

    )1k

    i=1(ni 1)

    Mean squares within samples. It is a pooled estimate of thepopulation variance.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 129 / 181

  • Analysis of Variance - Part One IVSingle factor ANOVA

    Indication of variances within samples.

    Subscript W stands for within

    Numerator: Within samples sum of squares

    Denominator: Sum of degrees of freedom of each sample

    MSB =SSBdfB

    =SST SSWdfT dfW

    Mean squares between samples. Indicates how the means differ.

    Subscript B stands for between

    Numerator: Between samples sum of squares

    Kristiaan Schreve (SU) Stats Block January 26, 2015 130 / 181

  • Analysis of Variance - Part One VSingle factor ANOVA

    Denominator: Number of samples - 1

    Note that

    SSB + SSW = SST anddfB + dfW = dfT

    Note that both MSW and MSB are estimates of the population variance.If there is a meaningful difference between the variances, then the samplescannot all come from the same populations and therefore there is ameaningful difference between the samples that cannot be attributed justto random errors.ANOVA translates

    H0 :1 = 2 = . . . = k

    H1 :Not H0

    Kristiaan Schreve (SU) Stats Block January 26, 2015 131 / 181

  • Analysis of Variance - Part One VISingle factor ANOVA

    into

    H0 :2B 2W

    H1 :2B >

    2W

    Variances are compared with the F-distribution.The test statistic is therefore

    f =MSBMSW

    Reject H0 at significance level if f > f

    Kristiaan Schreve (SU) Stats Block January 26, 2015 132 / 181

  • Analysis of Variance - Part One IAfter the F-test

    If H0 is rejected, how can you find where the differences lie?Planned comparisons

    Also called a priori tests

    Essentially it is t-tests comparing means of different samples.

    The test statistic is

    t =x1 x2

    MSw [1n1

    + 1n2 ]

    The hypotheses are:

    H0 :1 2H1 :1 > 2

    The rest of the test is a standard t-test with df = dfW .Kristiaan Schreve (SU) Stats Block January 26, 2015 133 / 181

    [7]: pp. 258-261

  • Analysis of Variance - Part One IIAfter the F-test

    Unplanned comparisonsThere may be some situations where the conditions for the t-testmentioned above are not met. This is then called a unplanned comparison.Also known as a posteriori or post hoc tests.Numerous tests are available...

    Kristiaan Schreve (SU) Stats Block January 26, 2015 134 / 181

  • Regressions ILinear regression

    Kristiaan Schreve (SU) Stats Block January 26, 2015 135 / 181

    [7]: pp. 293-299

  • Regressions IILinear regression

    Figure: Left: Scatter plot. Right: With linear trend line.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 136 / 181

  • Regression ITesting hypotheses about regression

    Residual variance of estimate

    s2yx =

    (y y )2

    N 2

    =

    (y y )2

    N n 1

    n is the degree of the polynomial fitted to the data. In the linear case,n = 1.

    N is the number of data points.

    y y is the difference between the measured and predicted value.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 137 / 181

    [7]: pp. 299-306

  • Regression IITesting hypotheses about regression

    Standard error of estimate

    syx =

    s2yx =

    (y y )2N 2

    Hypothesis

    H0 :No real relationship

    H1 :Not H0

    Similar to ANOVA, the hypothesis will compare variances. Therefore,rewrite

    Kristiaan Schreve (SU) Stats Block January 26, 2015 138 / 181

  • Regression IIITesting hypotheses about regression

    H0 :2Regression 2Residual

    H1 :2Regression >

    2Residual

    To find the variances, we need the sums of squares and their correspondingdegrees of freedom.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 139 / 181

  • Regression IVTesting hypotheses about regression

    Figure: Deviations in a scatter plot, from [6]

    Kristiaan Schreve (SU) Stats Block January 26, 2015 140 / 181

  • Regression VTesting hypotheses about regression

    SSResidual =

    (y y )2

    This represents the variability around the regression curve.

    SSRegression =

    (y y)2

    This represents the gain in prediction by using a regression curve ratherthan just the average of the data.

    SSTotal =

    (y y)2

    This represents the total variance.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 141 / 181

  • Regression VITesting hypotheses about regression

    The following identities hold

    SSResidual + SSRegression = SSTotal

    dfResidual + dfRegression = dfTotal

    dfResidual = N 2dfTotal = N 1

    Kristiaan Schreve (SU) Stats Block January 26, 2015 142 / 181

  • Regression VIITesting hypotheses about regression

    Similar to ANOVA, we use mean squares for the variances

    MSRegression =SSRegressiondfRegression

    MSResidual =SSResidualdfResidual

    MSTotal =SSTotaldfTotal

    Test the hypothesis with an F test

    F =MSRegressionMSResidual

    Reject H0 at significance level if F > F

    Kristiaan Schreve (SU) Stats Block January 26, 2015 143 / 181

  • Regression VIIITesting hypotheses about regression

    Testing the slope(Note, this is a different approach from the textbook on pp. 267)Is the slope different from zero? Or, is the mean an equally goodpredictor?Hypotheses

    H0 : = 0

    H1 : 6= 0

    This is a standard one-sample, two tailed, t-test. In what follows, = 0The test statistic is

    t =b

    sb; df = N 2

    Kristiaan Schreve (SU) Stats Block January 26, 2015 144 / 181

  • Regression IXTesting hypotheses about regression

    Denominator estimates the standard error of the slope

    sb =syx

    sx

    N 1

    syx =

    (y y )2N 2

    sx =

    (x x)2N 1

    Kristiaan Schreve (SU) Stats Block January 26, 2015 145 / 181

  • Regression XTesting hypotheses about regression

    Testing the interceptIs the intercept not zero?Hypotheses

    H0 : = 0

    H1 : 6= 0

    This is a standard one-sample, two tailed, t-test. In what follows, = 0The test statistic is

    t =a

    sa; df = N 2

    sa =syx

    sx

    1N +

    x2

    (N1)s2x

    Kristiaan Schreve (SU) Stats Block January 26, 2015 146 / 181

  • Regression IExcels R-squared

    Coefficient of Determination

    R2 =SSRegression

    SSTotal

    When R2 1, there is a good correlation.When R2 0, not so!

    Kristiaan Schreve (SU) Stats Block January 26, 2015 147 / 181

    Not in textbook

  • Regression IExcel functions for regression

    SLOPE

    INTERCEPT

    STEYX

    FORECAST

    TREND

    LINEST

    Data analysis tool: Regression

    Kristiaan Schreve (SU) Stats Block January 26, 2015 148 / 181

    [7]: pp. 307-319

  • Multiple regression I

    Regression for more than one dependent variable.E.g. a plane:

    y = a + b1x1 + d2x2

    Any number of dependent variables are possible.

    y = a +

    bixi

    Other types of fitting is also possible in Excel (logarithmic, exponential,higher order polynomials, etc.). Make careful decisions about the trend inthe data and choose an appropriate model. Use hypothesis testing to testyour assumptions.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 149 / 181

    [7]: pp. 320-327

  • RegressionGuidelines

    Give an indication of the goodness of fit.

    Report the range of the dependent variable(s) for which theregression was done and therefore the range for which the goodnessof fit test is valid.

    Check the validity of the prediction of the regression result over therange of the dependent variable. E.g. sometimes the predicted resultmust be a positive value (e.g. the score of the tut test). If theregression result allows the possibility of predicting a negative value inthis case, the result must be reconsidered.

    Fit the lowest order curve possible.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 150 / 181

    Not in textbook

  • Correlation IPearsons correlation coefficient

    Correlation is an alternative to regression for looking at relationshipsbetween parameters. With regression it is possible to makepredictions. With correlation it is easier to say that relationships arestronger than others.

    Positive correlation means that as one parameter increases, the otheralso increases.

    Negative correlation means that as one parameter increases, the otherdecreases.

    Note that correlation does not imply causality. (The same is true forregression.)

    Kristiaan Schreve (SU) Stats Block January 26, 2015 151 / 181

    [7]: pp. 331-334

  • Correlation IIPearsons correlation coefficient

    Pearsons product-moment correlation coefficient

    r =

    [1

    N1]

    (x x)(y y)sxsy

    =cov(x , y)

    sxsy

    Numerator: covariance represents how x and y vary together.

    Denominator: Standard deviations of x and y variables.

    r = 1 implies perfect negative correlation (minimum value r canhave)

    r = 1 implies perfect positive correlation (maximum value r can have)

    r = 0 implies no correlation.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 152 / 181

  • Correlation ICorrelation and regression

    r =

    r 2 =

    SSRegression

    SSTotal

    r 2 is just Excels Coefficient of Determination

    R2 = 0.667 implies SSRegression is 66.7% of SSTotal . To find out ifthat is significant, do a hypothesis test...

    Kristiaan Schreve (SU) Stats Block January 26, 2015 153 / 181

    [7]: pp. 334-337

  • Correlation ITesting hypotheses about correlation

    Correlation coefficient greater than zero?Sample statistic is r .Test for positive correlation

    H0 : 0H1 : > 0

    Test statistic (N 2) degrees of freedom.

    t =r

    srWhere

    = 0

    sr =

    1r2N2

    Reject H0 at significance level if t > t.Kristiaan Schreve (SU) Stats Block January 26, 2015 154 / 181

    [7]: pp. 338-340

  • Correlation IITesting hypotheses about correlation

    Example (Too much data for regression? [6] pp. 371)

    Say, N = 102 and = 0.05.Say r = 0.195Is it a significant correlation?

    t = rN2

    1r2 = 1.988

    t = 1.984. Since t > t, reject H0. We suspect the correlation issignificant.BUTr 2 = 0.038, which implies that SSRegression is just 4% of SSTotal .

    Kristiaan Schreve (SU) Stats Block January 26, 2015 155 / 181

  • Correlation IIITesting hypotheses about correlation

    Example (Too much data for regression? Continued...)

    r t t N 2 Reject?0.195 2.178 1.980 120 Yes0.195 2.085 1.982 110 Yes0.195 1.988 1.984 100 Yes0.195 1.886 1.987 90 No0.195 1.778 1.990 80 No

    Kristiaan Schreve (SU) Stats Block January 26, 2015 156 / 181

  • Correlation IVTesting hypotheses about correlation

    Do two correlation coefficients differ?

    H0 :1 = 2

    H1 :1 6= 2

    We have to transform the r value with

    zr = 0.5[ln(1 + r) ln(1 r)]

    The test statistic is then

    z =z1 z2z1z2

    where

    Kristiaan Schreve (SU) Stats Block January 26, 2015 157 / 181

  • Correlation VTesting hypotheses about correlation

    z1z2 =

    1

    N1 3+

    1

    N2 3Reject H0 at significance level if

    z/2 > z or z > z/2

    Kristiaan Schreve (SU) Stats Block January 26, 2015 158 / 181

  • Uncertainty of Measurement I

    Based on ISO Guide 98-3[1].

    Formal standard for expression of uncertainty in measurement.

    True valuesof measurand can never be known.

    Therefore, measurement errorcan also never be known.

    Measurement results therefore should be expressed in statisticalterms, i.e. as a distribution.

    Therefore, we should report some nominal value, e.g. the mean value,with some expression of the measurement uncertainty.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 159 / 181

    Not in textbook

  • Uncertainty of Measurement II

    Example (Measuring Power Dissipated from a Resistor [1])

    If a potential difference V is applied to the terminals of atemperature-dependent resistor that has a resistance of R0 at the definedtemperature t0 and a linear temperature coefficient of resistance , thepower P (the measurand) dissipated by the resistor at the temperature tdepends on V , R0, and t according to

    P = f (V ,R0, , t) =V 2

    R0[1 + (t t0)]P is never directly measured. We will measure V and t. With enoughrepetitions, measurement uncertainties for V and t can be found.Hopefully, the uncertainty in the reference values of R0, and t0 areknown. Then we need a method to propagate the uncertainty of thesevalues to the uncertainty of the measurand P.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 160 / 181

  • Uncertainty of Measurement III

    The example illustrates a few things

    The measurand is seldom measured directly. Often it is derived froma functional relationship such as

    Y = f (X1,X2, ...,XN)

    Y is the measurand.

    The Xi is either known from measurements or from some priorknowledge (e.g. a catalogue value).

    There are two types of evaluation of standard uncertainty

    Type A is determined from statistical analysis of a set ofmeasurements.Type B is determined by any other means.

    We need a method to propagate uncertainty (see slide 164).

    Kristiaan Schreve (SU) Stats Block January 26, 2015 161 / 181

  • Uncertainty of Measurement IV

    To find the mean value of the measurand, do you take the mean ofthe input quantities or do you first calculate the measurand for eachset of measurements and then take the mean of the measurand?

    Kristiaan Schreve (SU) Stats Block January 26, 2015 162 / 181

  • Uncertainty of Measurement V

    Example (When to calculate the mean)

    The table shows voltage andtemperature readings for the powerdissipated by the resistor in theprevious example. If R0 = 4.33 , = 0.00393 and t0 = 20

    C, themean power dissipated is

    21.43545 W if P is calculatedfor each data point and then themean of the 10 power values aretaken.

    21.43568 W if the mean voltage(10.006565 V) and meantemperature (40.0563 C) isused.

    The difference is due to the nonlinearfunction for P. The GUM guide [1]states that for nonlinear relations, themeasurand for each data point mustbe calculated and then the mean ofthe set of measurands must be taken.

    Voltage [V] Temperature [C]

    10.030 39.9309.991 39.9629.971 39.916

    10.023 40.10210.000 39.94910.039 40.25010.073 40.3159.987 39.9219.935 40.124

    10.017 40.093Kristiaan Schreve (SU) Stats Block January 26, 2015 163 / 181

  • Uncertainty of Measurement IEvaluation of standard uncertainty

    From the examples it is clear that there are two types of uncertainty.

    One is based on a set of repeated measurements. (Type A.) In theexample, it is the standard uncertainty of the temperature t andvoltage V .Another is based on other information, e.g. data sheets. (Type B.) Inthe example, it is the standard uncertainty of the constants R0, andt0.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 164 / 181

    Not in textbook

  • Uncertainty of Measurement IType A evaluation of standard uncertainty

    Type A standard uncertainty is based on repeated measurements.

    It is typically estimated with

    sx =sN

    Note, it is the standard error (or standard deviation of the samplingdistribution mean).

    Can also be evaluated by other means, depending on the situation.

    It is important to always report the degrees of freedom with the TypeA standard uncertainty.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 165 / 181

    Not in textbook

  • Uncertainty of Measurement IType B evaluation of standard uncertainty

    Type B standard uncertainty is NOT based on repeatedmeasurements.

    Typical sources of information [1]

    previous measurement dataprevious experience and good engineering judgementmanufacturers specificationsdata provided in calibration and other certificatesuncertainties assigned to reference data taken from handbooks.

    If the source does not give the standard uncertainty explicitly, it maybe derived. The GUM Guide [1] gives several examples in section 4.3.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 166 / 181

    Not in textbook

  • Uncertainty of MeasurementLaw of propagation of uncertainty for uncorrelated quantities

    When the measurand is not directly measured, as in the example, thestandard uncertainty of the measurand depends on the combinedType A and Type B standard uncertainties.It can be shown, if the input quantities are independent, that thecombined standard uncertainty is

    s2c (y) =Ni=1

    (f

    xi

    )2s2(xi )

    This is called the law of propagation of uncertaintyf is the function Y = f (X1,X2, ...,XN) and xi are the estimates of Xi .Note, the partial derivatives essentially scales the input uncertainties.It is sometimes called sensitivity coefficients.If the partial derivatives cannot be calculated directly, they may beevaluated numerically, or estimated experimentally (see sections 5.1.3and 5.1.4 in the GUM Guide [1]).

    Kristiaan Schreve (SU) Stats Block January 26, 2015 167 / 181

    Not in textbook

  • Uncertainty of MeasurementLaw of propagation of uncertainty for correlated quantities

    The law of propagation of uncertainty for correlated input quantitiesis

    s2c (y) =Ni=1

    (f

    xi

    )2s2(xi ) + 2

    N1i=1

    Nj=i+1

    f

    xi

    f

    xjs(xi , xj)

    s(xi , xj) is the estimated covariance associated with xi and xj . It iscalculated as

    s(xi , xj) =1

    N(N 1)

    Nk=1

    (xi ,k xi )(xj ,k xj)

    EXCEL: Covariance is calculated with COVARIANCE.P (populations)or COVARIANCE.S (samples).How do you handle a situation where some quantities are correlatedand some not?

    Kristiaan Schreve (SU) Stats Block January 26, 2015 168 / 181

    Not in textbook

  • Uncertainty of MeasurementDetermining expanded uncertainty

    In some practical cases the combined uncertainty is insufficient tocapture the uncertainty.

    The expanded uncertainty is

    U = ksc(y)

    where k is the coverage factor.

    Typically, 2 k 3.The result of the measurement is then typically expressed asY = y U.k can be chosen to cover a certain confidence interval, in which canthe confidence level should also be given.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 169 / 181

    Not in textbook

  • Uncertainty of Measurement IReporting uncertainty

    In general, give all the information needed to repeat the evaluation.

    Rather report too much.

    What is reported should be in line with the intended use of themeasurement result, e.g. a calibration certificate for a nano-metreprecision measurement device would require a lot more informationthan a laser distance sensor you can buy at the local hardware store.

    Consider to include the following [1]

    clearly describe the methods used to calculate the measurement resultand its uncertainty from the experimental observation (Type Astandard uncertainty) and input data (Type B standard uncertainty)list all the uncertainty components and document fully how they wereevaluated.present the data analysis in such a way that each of its important stepscan be readily followed and the calculation of the reported result canbe independently repeated

    Kristiaan Schreve (SU) Stats Block January 26, 2015 170 / 181

    Not in textbook

  • Uncertainty of Measurement IIReporting uncertainty

    give all the corrections and constants used in the analysis and theirsourcesin the case of reporting expanded uncertainty report the coveragefactor.

    The numerical result of the uncertainty is reported in one of thefollowing four ways. (Assume a mass ms of an object weighing about100 g is being reported.) The words below in parentheses may beomitted. [1]

    ms=100,021 47 g with (a combined standard uncertainty)sc=0,35 mgms=100,021 47(35) g, where the number in parentheses is thenumerical value of (the combined standard uncertainty) sc referred tothe corresponding last digits of the quoted result.ms=100,021 47(0,000 35) g, where the number in parentheses is thenumerical value of (the combined standard uncertainty) sc expressed inthe unit of the quoted result.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 171 / 181

  • Uncertainty of Measurement IIIReporting uncertainty

    ms=(100,021 47 0,000 35) g, where the number following thesymbol is the numerical value of (the combined standarduncertainty) sc and not a confidence interval.

    Report an expanded uncertainty as

    ms=(100,021 47 0,000 79) g, where the number following thesymbol is the numerical value of (an expended uncertainty) U = ksc ,with U determined from (a combined standard uncertainty)sc=0,35 mg and (a coverage factor) k=2,26 based on thet-distribution for v=9 degrees of freedom, and defines an intervalestimated to have a level of confidence of 95 percent.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 172 / 181

  • Uncertainty of Measurement IExample

    Continue with example from beginning of section.Use

    s2c (y) =Ni=1

    (f

    xi

    )2s2(xi ) + 2

    N1i=1

    Nj=i+1

    f

    xi

    f

    xjs(xi , xj)

    to calculate the combined uncertainty for

    P = f (V ,R0, , t) =V 2

    R0[1 + (t t0)]Let

    Kristiaan Schreve (SU) Stats Block January 26, 2015 173 / 181

  • Uncertainty of Measurement IIExample

    x1 = V

    x2 = R0

    x3 =

    x4 = t

    Ignore the uncertainty contribution of t0. Assume it is a very well knownreference value with negligible uncertainty. Then

    Kristiaan Schreve (SU) Stats Block January 26, 2015 174 / 181

  • Uncertainty of Measurement IIIExample

    f

    V=

    2V

    R0[1 + (t t0)]f

    R0=

    V 2

    R20 [1 + (t t0)]f

    =

    (t t0)V 2

    ((t t0) + 1)2R0f

    t=

    V 2

    [(t t0) + 1]2R0

    Evaluate these values at mean values of V ,R0, and t, i.e.V = 10.007 V, R0 = 4.33 , = 0.00393 and t = 40.056

    C.This gives

    Kristiaan Schreve (SU) Stats Block January 26, 2015 175 / 181

  • Uncertainty of Measurement IVExample

    f

    V= 4.284

    f

    R0= 4.950

    f

    = 398.506

    f

    t= 0.078

    Assume only V and t is correlated. Hence, from EXCEL, finds(V , t)=0.00296. Also, from the data we can find

    Kristiaan Schreve (SU) Stats Block January 26, 2015 176 / 181

  • Uncertainty of Measurement VExample

    s2(V ) = 0.00149

    s2(t) = 0.02076

    Finally, lets assume that somehow we know that

    s2(R0) = 0.001

    s2() = 0.02

    Now it is straight forward to calculate s2c (P).

    Kristiaan Schreve (SU) Stats Block January 26, 2015 177 / 181

  • Selecting the Right Method I

    Method Typical Use

    Confidence interval of ; known

    Calculate confidence limits for your estimateof the population mean. You know the pop-ulation variance.

    Confidence interval of ; unknown

    Calculate confidence limits for your estimateof the population mean. You do not knowthe population variance.

    One sample hypothesistest: z-test on singlemean

    You have one sample and some guess of thepopulation mean. You want to know if theguess is right or how it differs. You knowthe population variance.

    One sample hypothesistest: t-test on singlemean

    You have one sample and some guess of thepopulation mean. You want to know if theguess is right or how it differs. You do notknow the population variance.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 178 / 181

    Not in textbook

  • Selecting the Right Method II

    One sample hypothesistest: 2-test on singlevariance

    You have one sample and some guess of thepopulation variance. You want to know ifthe guess is right or how it differs.

    Two sample hypothesistest: z-test on two means

    You have two samples and want to know ifthey are the same or not. You know thepopulation variance.

    Two sample hypothesistest: t-test on two meanswith equal variances

    You have two samples and want to know ifthey are the same or not. You do not knowthe population variance, but know that theyare equal.

    Two sample hypothesistest: t-test on two meanswith unknown, unequalvariances

    You have two samples and want to knowif they are the same or not. You have noknowledge about the population variance.

    Two sample hypothesistest: paired samples.

    Comparing two samples, but the specimensin the two samples are somehow linked. Youdo not know the population variance.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 179 / 181

  • Selecting the Right Method III

    Two sample hypothesistest: F-test

    Comparing the variances of two samples.

    Single factor ANOVA Seeing if there is a difference in the meansof more than two samples.

    Single factor ANOVA:Planned comparison

    A priori t-tests on the means of selectedsamples to find out if there is a significantdifference.

    Single factor ANOVA:Unplanned comparison

    A posteriori test on sample means. Not cov-ered in this course.

    Regression If you suspect there is a trend between thedependent and independent variables.

    Regression: F-test Test the above mentioned suspicion.

    Regression: Testing theslope

    See if the slope of the linear regressioncurve is significant, otherwise the mean isan equally good predictor.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 180 / 181

  • Selecting the Right Method IV

    Regression: Testing theintercept

    See if the intercept plays a significant role.Otherwise it could have been zero.

    Regression: Coefficientof Determination R2

    Indication of goodness of fit. Is not a hy-pothesis test. Should be combined with anF-test for the regression.

    Correlation: Pearsonscorrelation coefficient

    Similar to coefficient of determination, butdistinguishes between positive and negativecorrelation. Tests if data is correlated, butdoes not tell how. Is not a hypothesis test.

    Correlation: Is correla-tion coefficient greaterthan zero?

    Hypothesis test to evaluate correlation coef-ficient.

    Correlation: Do two cor-relation coefficients dif-fer?

    Is there a new correlation between the data?

    Kristiaan Schreve (SU) Stats Block January 26, 2015 181 / 181

  • References I

    Uncertainty of measurementpart 3: guide to the expression of uncertainty inmeasurement, 1995.

    R.S. Figliola and D.E. Beasley.

    Theory and Design for Mechanical Measurements.

    Wiley, Hoboken, 4th edition, 2006.

    A Graham.

    Statistics: A Complete Introduction.

    Hodder & Stoughton, 2013.

    D Huff and I Geis.

    How to Lie with Statistics.

    Norton, New York, 1954.

    W. Mendenhall and T Sincich.

    Statistics for Engineering and the Sciences.

    MacMillan, New York, 3rd edition, 1992.

    INBO 519.502462 MEN.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 182 / 181

  • References II

    J. Schmuller.

    Statistical Anlysis with Excel for Dummies.

    Wiley, Hoboken, 2nd edition, 2009.

    J Schmuller.

    Statistical Analysis with Excel for Dummies.

    Wiley, Hoboken, 3rd edition, 2013.

    R.E. Walpole and R.H. Myers.

    Probability and Statistics for Engineers and Scientists.

    MacMillan, New York, 4th edition, 1990.

    Kristiaan Schreve (SU) Stats Block January 26, 2015 183 / 181

  • The End

    Kristiaan Schreve (SU) Stats Block January 26, 2015 184 / 181

    IntroductionSome Important ConceptsExcel DemonstrationGraphing DataChoosing the right type of graphGuidelines for creating good scientific graphs

    Calculating Averages with ExcelStandard Deviation and VarianceZ ScoresHigher Order Distribution DescriptorsFrequency and HistogramsBox-and-whisker PlotsThe Normal DistributionConfidence LimitsSampling distributionsCentral limit theoremLimits of confidencet-distributionNormal distribution and t-distribution confidence limits compared

    One-sample Hypothesis TestingSome revisionHypothesis testingSummary of one-sample hypothesis tests

    Two-sample Hypothesis TestingHypotheses for two-sample means testingHypotheses for two-sample variance testingSummary of two-sample hypothesis tests

    Analysis of Variance - Part OneIntroduction to ANOVASingle factor ANOVAAfter the F-test

    RegressionLinear regressionTesting hypotheses about regressionExcel's R-squaredExcel functions for regressionMultiple regressionGuidelines

    CorrelationPearson's correlation coefficientCorrelation and regressionTesting hypotheses about correlation

    Uncertainty of MeasurementEvaluation of standard uncertaintyLaw of propagation of uncertainty for uncorrelated quantitiesLaw of propagation of uncertainty for correlated quantitiesDetermining expanded uncertaintyReporting uncertaintyExample

    Selecting the Right Method