Psych Exam 4 Class Slide Questions Flashcards
What is a correlation?
Relationship between 2 variables (x and y)
What are the two pieces of information that correlations provide?
Magnitude and direction
What is magnitude?
Degree of relationship between two variables
What is direction?
How the variables are related
What is a positive direction?
As x increases y also increases
What does a negative direction indicate?
As x increases y tends to decrease
What do correlation coefficients assume?
Linear relationships
What is a linear relationship?
Relation represented by a straight line
What do scatterplots graph?
Two variable distribution by placing dots at the value of paired scores
What does the pattern point reveal in a scatterplot?
Direction of the correlation
How is magnitude determined?
How closely data points hug the hug line
What happens with a stronger relationship?
The closer the data points will be to the line
What happens with the variables if there is no correlation?
Value of one variable doesn’t predict value of other variable
What is Pearson’s r?
Single statistic indicating direction and strength of correlation
What does Pearson’s r require?
Multiplying deviations from mean X by mean of Y (cross products)
What is the second step in finding Pearsons r?
Divide by standard deviations and number of pairs
What is the equation for Pearsons r?
r= E (X-X-)(Y-Y-)/ noXoY
What is the value of a perfect positive correlation?
r= +1
What is the value of a perfect negative correlation?
r= -1
What is the value of no correlation?
r= 0
What is a small, moderate, and large correlation strength value?
r=.1 (small), r=.3 (m), r=.5 (l)
What do strong correlations not provide?
Causation
What do non-linear effects do to Pearsons r?
Make it look small
What reduces the effects of Pearsons r?
Restriction of range/talent
What artificially increases the size of Pearsons r?
Discontinuous distributions
What happens to correlations as sample size increases?
They become more accurate
What are correlations determined by?
Specific measures representing the variables/conditions of measurement
What Is the first issue with hypothesis testing?
Nulls are point estimates almost always false
What is the second issue with hypothesis testing?
No “grey area” between significance and non-significance
What does failing to reject the null provide?
Little info for conclusions
What are confidence intervals?
Interval estimates
What do confidence intervals represent?
Range of values where we are likely to find pop value
What is the first reason to use confidence intervals?
Evidence on likely range of possible values of parameters
What do CI’s provide a clear indication of?
How sample size influences accuracy of guessing
What is the second reason to use a confidence interval?
Tool for determining importance of result
What is the basic structure of a confidence interval?
CI= Sample mean +- Margin of error
What do we decide in confidence intervals?
The range we want
What does the range of a confidence interval determine?
Z or T scores we will use
What does a larger interval include?
Larger range of values
What is a higher confidence coefficient?
Confidence pop value falls within boundary
What is the equation for confidence intervals for single means when O^2x is known?
C = X- (+-) z(ox-)
What does a 95% CI mean?
95% confident that what we want falls between the two values
What does a confidence interval of 95% not mean?
Probability of ux being in interval is 95% (its either there or it isn’t)
What does a confidence interval of 95% spelled out mean?
Based on ways intervals are constructed 95% will contain uX
What equation do you use when o^2x is not known?
X- (+-) tp (sX-)
What is tp?
t value with prob p of obtaining a value so deviant (determined by interval width)
What is the formula for independent samples t-test?
C= (X-(-) Y-) (+-) tpSX-(-)Y-
What is regression?
Use correlations to determine strength and direction of X and Y relationship
What do we use a regression line to predict?
Person’s score on Y (Y’) based on x score
What does the regression line give?
Predicted Y value (Y’) for every value of X (we regress Y onto X)
What happens when r=0?
z’ y will always be zero, Y’ will always be the mean of Y
What does the line of best fit minimize?
Distance between itself and observed values of Y
What is Pearsons least squares criterion?
Sum of squared deviations from line is minimized
What does Pearsons least squares criterion have the same idea as?
Sum of squares being minimized around mean (Y’ is the mean this time)
What does all of these things in regression assume?
Linear relation, normal distribution of Y, variability of Y around Y’ must be same for value of X
What term has this definition: variability of Y around Y’ must be same for value of X
Homoscedasticity
What is the equation for standard error of the estimate of Y on X?
Syx= Square root of E (Y-Y’)^2 divided by n
What is a useful alternative form of Syx?
Syx = Sy square root 1-r^2
What is the symbol r^2?
Variability Y can be explained by X (coefficient of determination)
What does b mean?
Unstandardized regression coefficient
What does B mean?
Standardized regression coefficient
What is the coefficient of determination?
Proportion of variability in Y related to X
What is the coefficient of non-determination (k^2)?
Proportion of variability Y not related to X
r^2 and k^2 are just proportions of what?
S^2y
What is regression towards the mean?
Any time r<1 value of z’y will be less extreme than zx
What does regression toward the mean not predict?
Values will all end up at the mean of Y over time
What does regression towards the mean not imply?
That you can get extreme scores on x and y (less likely when correlation isn’t perfect)
What happens in post-test only?
Designs measure the DV after IV is administered
What is a dependent samples design?
Measurements in one sample are related to measurements in other samples
What are the statistical assumptions for independent samples?
Random sampling, sampling w/ replacement, normal distribution, homogeneity of variance
What are the statistical assumptions of correlated samples?
Random sampling, sampling w/replacement, normal distribution
What is the equation for degrees of freedom in independent samples?
(nx-1) + (ny-1)
What is the equation for df in correlated samples?
n-1 (pairs we have in sample is n)`
What is a repeated measures design?
Same participants measured under two or more conditions, each subject his or her own control
In a within-subjects design what do repeated measures designs expose?
Participants to all levels of Iv
What do matched groups designs do to participants before random assignment?
Rank them
What are matched pairs designs?
Participants naturally paired based on a characteristic
What could you also calculate the dependent t using?
Difference scores D (X-Y)
What do you calculate for every pair in the sample?
Difference between X and Y scores
For this look at the alternative T formulas please
For correlated samples what is the new way to measure hypothesis?
u= 0
What are the advantages of dependent sample t-tests?
Fewer participants, eliminates difference possibilities between levels of IV due to individual differences
What are the disadvantages or dependent samples t-tests?
Sequence and order effects
What are sequence and order effects?
Order participants are exposed to IV levels influences performance
What are independent samples power considerations?
Alpha, effect size, sample size, pop SD, relative n of X and Y samples
What are the correlated samples power considerations?
Alpha, effect size, sample size, pop SD, correlation of X and Y