Class Notes (837,487)
United States (325,080)
Psychology (127)
PSYC 51A (16)
Lee Yoona (16)
Lecture 16

PSYC 51A Lecture 16: Psyc 51A: Ch.16: Regression

5 Pages
Unlock Document

Lee Yoona

Correlation ● Describe association b/w two variables ○ The order of association does not matter ● One statistics (r) combines slope and scatter ● r statistics is independent of the scale of the variables ● Order of the two variables does not matter Regression ● Predict exact value of one variable Y from other variable X ○ Order matters ● Three statistics (b, a anY.X quantify slope (b), y-intercept (a), and scY.X,r (s separately ● Regression statistics are dependent on the scale of variables ● Order of two variables matter Questions from Univariate Statistics vs. Regression ● Is there any mean difference on weight loss depending on different treatments? ○ Within subject design ○ Correlation: Consider all individual score ■ Positive/negative correlation ● Univariate statistics don’t indicate anything except that the groups don’t differ. Mean differences (not all scores taken into consideration). Each statistic (Mean and SD) summarizes information about just one variable ● Need a summary statistic that is computed using information from two variables. What if we want to ask a question, “if someone lost 3 lbs using drug, what is the predicted range of their weight loss with behavioral therapy? (Here, we want to predict the possible changes in Y by changes in X. In this example, Y=behavioral therapy & X=drug). ● PS. Correlation: X variable is positively/negatively correlated with Y variable ● Curvilinear relationship ● Linear relationship - straight line relationship ○ We will only deal with this type of relationship ○ Need a statistic that captures the information in a scatter plots like this ● The general form of a predictive equation for a straight line relationship is: (weight loss with Therapy) = b(weight loss with No Drug) + a ● Ŷ = bX +a ● Where ○ Y - dependent variable X ○ Ŷ - estimate of dependent variable (values predicted by the line equation) ○ X - independent variable ○ b - slope ○ a - y-intercept ● Slope = b = ΔY = Y1-Y2 ΔX = X1-X2 ● slope is positive or negative direction; magnitude = +∞ ● y-intercept = a ; Y value where X= 0 (where the line crosses the Y axis) How do we calculate b and a? 1. Scatter (error from the predicted line) ● For univariate distributions we computed variability using squared deviation between sampled values and the mean, SS = Σ(X-M ) and SS = Σ(Y-M ) 2 X X Y Y ● For bivariate distributions we use the squared deviation between the sampled values and the value predicted by the straight line equation, residual(Y- 2 Ŷ) ● The criterion for the best fitting line is to minimize the sum of squared
More Less

Related notes for PSYC 51A

Log In


Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.