Overview Correlation Regression -Definition -Deviation Score Formula, Z score formula -Hypothesis Test Regression Intercept and Slope Unstandardized Regression Line Standardized Regression Line Hypothesis Tests
Associations among Continuous Variables Early developments Sir Francis Galton was very interested in these issues in the 1880’s Galton was the cousin of Darwin and thus he became interested in evolution and heredity Galton had an intuition that the heredity could be understood in terms of deviations from means He began to measure characteristics of plants and animals including people Lots of normality Galton found many characteristics that were normally distributed Height in humans (by gender) Weight in humans (by gender) Length of animal bones Weights of seeds of plants Sweat peas The next step that Galton took was to look at distributions of measurements of parents and offspring together He first looked at sweet pea plants because the female plants can self fertilize This makes examining the data easier because only one parent influences the characteristics of the offspring He plotted mother seed sizes against daughter seed sizes
3 characteristics of a relationship Direction Positive(+) Negative (-) Degree of association Between –1 and 1 Absolute values signify strength Form Linear Non-linear
Direction Positive Negative Large values of X = large values of Y, Small values of X = small values of Y. - e.g. IQ and SAT Large values of X = small values of Y Small values of X = large values of Y -e.g. SPEED and ACCURACY
Degree of association Strong (tight cloud) Weak (diffuse cloud)
Form Linear Non- linear
Regression & Correlation Early developments Sir Francis Galton was very interested in these issues in the 1880’s Galton was the cousin of Darwin and thus he became interested in evolution and heredity Galton had an intuition that the heredity could be understood in terms of deviations from means He began to measure characteristics of plants and animals including people Lots of normality Galton found many characteristics that were normally distributed Height in humans (by gender) Weight in humans (by gender) Length of animal bones Weights of seeds of plants Sweat peas The next step that Galton took was to look at distributions of measurements of parents and offspring together He first looked at sweet pea plants because the female plants can self fertilize This makes examining the data easier because only one parent influences the characteristics of the offspring He plotted mother seed sizes against daughter seed sizes
What is the best fitting straight line? Regression Equation: Y = a + bX How closely are the points clustered around the line? Pearson’s R
Correlation Early developments Sir Francis Galton was very interested in these issues in the 1880’s Galton was the cousin of Darwin and thus he became interested in evolution and heredity Galton had an intuition that the heredity could be understood in terms of deviations from means He began to measure characteristics of plants and animals including people Lots of normality Galton found many characteristics that were normally distributed Height in humans (by gender) Weight in humans (by gender) Length of animal bones Weights of seeds of plants Sweat peas The next step that Galton took was to look at distributions of measurements of parents and offspring together He first looked at sweet pea plants because the female plants can self fertilize This makes examining the data easier because only one parent influences the characteristics of the offspring He plotted mother seed sizes against daughter seed sizes
Correlation - Definition Correlation: a statistical technique that measures and describes the degree of linear relationship between two variables Scatterplot Obs X Y A 1 1 B 1 3 C 3 2 D 4 5 E 6 4 F 7 5 Dataset Y X
Pearson’s r A value ranging from -1.00 to 1.00 indicating the strength and direction of the linear relationship. Absolute value indicates strength +/- indicates direction
The Logic of Correlation MEAN of X Below average on X Above average on X Above average on Y Below average on Y MEAN of Y For a strong positive association, the cross-products will mostly be positive Cross-Product =
The Logic of Correlation MEAN of X Below average on X Above average on X Above average on Y Below average on Y MEAN of Y For a strong negative association, the cross-products will mostly be negative Cross-Product =
The Logic of Correlation MEAN of X Below average on X Above average on X Above average on Y Below average on Y MEAN of Y For a weak association, the cross-products will be mixed Cross-Product =
Pearson’s r Deviation score formula SP (sum of products) = ∑ (X – X)(Y – Y) Deviation score formula
Deviation Score Formula Femur Humerus A 38 41 B 56 63 C 59 70 D 64 72 E 74 84 mean 58.2 66.00 SSX SSY SP
Deviation Score Formula Femur Humerus A 38 41 -20.2 -25 408.04 625 505 B 56 63 -2.2 -3 4.84 9 6.6 C 59 70 0.8 4 .64 16 3.2 D 64 72 5.8 6 33.64 36 34.8 E 74 84 15.8 18 249.64 324 284.4 mean 58.2 66.00 696.8 1010 834 SSX SSY SP = .99
Pearson’s r Deviation score formula SP (sum of products) = Below average on X Above average on X Below Average on Y Above average on Y Below average on Y SP (sum of products) = ∑ (X – X)(Y – Y) Deviation score formula For a strong positive association, the SP will be a big positive number
Pearson’s r Deviation score formula SP (sum of products) = Below average on X Above average on X Below Average on Y Above average on Y Below average on Y Deviation score formula For a strong negative association, the SP will be a big negative number ∑ (X – X)(Y – Y) SP (sum of products) =
Pearson’s r Deviation score formula SP (sum of products) = Below average on X Above average on X Below Average on Y Above average on Y Below average on Y Deviation score formula For a weak association, the SP will be a small number (+ and – will cancel each other out) ∑ (X – X)(Y – Y) SP (sum of products) =
Pearson’s r Z score formula Standardized cross-products
Z-score formula Femur Humerus ZX ZY ZXZY A 38 41 B 56 63 C 59 70 D 64 72 E 74 84 mean 58.2 66.00 s 13.20 15.89
Z-score formula Femur Humerus ZX ZY ZXZY A 38 41 -1.530 -1.573 B 56 63 -0.167 -0.189 C 59 70 0.061 0.252 D 64 72 0.439 0.378 E 74 84 1.197 1.133 mean 58.2 66.00 s 13.20 15.89
Z-score formula Femur Humerus ZX ZY ZXZY A 38 41 -1.530 -1.573 2.408 B 56 63 -0.167 -0.189 0.031 C 59 70 0.061 0.252 0.015 D 64 72 0.439 0.378 0.166 E 74 84 1.197 1.133 1.356 mean 58.2 66.00 ∑=3.976 s 13.20 15.89 r = .99
Formulas for R Z score formula Deviations formula
Interpretation of R A measure of strength of association: how closely do the points cluster around a line? A measure of the direction of association: is it positive or negative?
Interpretation of R r = .10 very small association, not usually reliable r = .20 small association r = .30 typical size for personality and social studies r = .40 moderate association r = .60 you are a research rock star r = .80 hmm, are you for real?
Interpretation of R-squared The amount of covariation compared to the amount of total variation “The percent of total variance that is shared variance” E.g. “If r = .80, then X explains 64% of the variability in Y” (and vice versa)
Hypothesis testing with r Hypotheses H0: = 0 HA : ≠ 0 Test statistic = r Or just use table E.2 to find critical values of r
Practice alcohol tobacco A 6.47 4.03 B 6.13 3.76 C 6.19 3.77 D 4.89 3.34 E 5.63 3.47 mean SSX SSY SP
Practice alcohol tobacco A 6.47 4.03 B 6.13 3.76 C 6.19 3.77 D 4.89 3.34 E 5.63 3.47 mean 1.55 .30 .64 SSX SSY SP
Properties of R A standardized statistic – will not change if you change the units of X or Y. (bc based on z-scores) The same whether X is correlated with Y or vice versa Fairly unstable with small n Vulnerable to outliers Has a skewed distribution
Linear Regression
Linear Regression But how do we describe the line? If two variables are linearly related it is possible to develop a simple equation to predict one variable from the other The outcome variable is designated the Y variable, and the predictor variable is designated the X variable E.g. centigrade to Fahrenheit: F = 32 + 1.8C this formula gives a specific straight line
The Linear Equation F = 32 + 1.8(C) General form is Y = a + bX The prediction equation: Y’ = a+ bX Where a = intercept b = slope X = the predictor Y = the criterion a and b are constants in a given line; X and Y change
The Linear Equation F = 32 + 1.8(C) General form is Y = a + bX The prediction equation: Y’ = a + bX Where a = intercept b = slope X = the predictor Y = the criterion Different b’s…
The Linear Equation F = 32 + 1.8(C) General form is Y = a + bX The prediction equation: Y’ = a + bX Where a = intercept b = slope X = the predictor Y = the criterion Different a’s…
The Linear Equation F = 32 + 1.8(C) General form is Y = a + bX The prediction equation: Y’ = a + bX Where a = intercept b = slope X = the predictor Y = the criterion Different a’s and b’s …
Slope and Intercept Equation of the line The slope b: the amount of change in y with one unit change in x The intercept a: the value of y when x is zero
Slope and Intercept Equation of the line The slope The intercept The slope is influenced by r, but is not the same as r
When there is no linear association (r = 0), the regression line is horizontal. b=0. and our best estimate of age is 29.5 at all heights.
When the correlation is perfect (r = ± 1 When the correlation is perfect (r = ± 1.00), all the points fall along a straight line with a slope
When there is some linear association (0<|r|<1), the regression line fits as close to the points as possible and has a slope
Where did this line come from? It is a straight line which is drawn through a scatterplot, to summarize the relationship between X and Y It is the line that minimizes the squared deviations (Y’ – Y)2 We call these vertical deviations “residuals”
Regression lines Minimizing the squared vertical distances, or “residuals”
Unstandardized Regression Line Equation of the line The slope The intercept
Properties of b (slope) An unstandardized statistic – will change if you change the units of X or Y. Depends on whether Y is regressed on X or vice versa
Standardized Regression Line Equation of the line The slope The intercept A person 1 stdev above the mean on height would be how many stdevs above the mean on weight?
Properties of β (standardized slope) A standardized statistic – will not change if you change the units of X or Y. Is equal to r, in simple linear regression
Exercise X Y 1 3 4 5 6 9 7 Calculate: b = a = β = Write the regression equation: Write the standardized equation: Mean: 5 5 Stdevp: 3.27 1.41 r= 0.866025404 b= 0.375 a= 3.125 X Y Y' 1 3 3.5 1 4 3.5 5 4 5 5 6 5 9 6 6.5 9 7 6.5
Exercise X Y 1 3 4 5 6 9 7 Calculate: b = .375 a = 3.125 β = .866 Write the regression equation: Write the standardized equation: Mean: 5 5 Stdevp: 3.27 1.41 r= 0.866025404 b= 0.375 a= 3.125 X Y Y' 1 3 3.5 1 4 3.5 5 4 5 5 6 5 9 6 6.5 9 7 6.5
Regression Coefficients Table Predictor Unstandardized Coefficient Standard error Standardized Coefficient t sig Intercept a SEa - Variable X b SEb
Summary Correlation: Pearson’s r Unstandardized Regression Line
Exercise in Excel X Y 1 -1.5 2 -3 4 -4 7 -2 9 -6 Calculate: 2 -3 4 -4 7 -2 9 -6 Calculate: r = b = a = β = Write the regression equation: Write the standardized equation: Sketch the scatterplot and regression line Mean: 5 5 Stdevp: 3.27 1.41 r= 0.866025404 b= 0.375 a= 3.125 X Y Y' 1 3 3.5 1 4 3.5 5 4 5 5 6 5 9 6 6.5 9 7 6.5