Week 7 (Correlation & Regression) Flashcards
Which of the 3 goals of science does correlational research have?
Describe and predict. Doesn’t and can’t involve explanation
Why is internal validity low in correlational research?
Because of the way data is collected, there is a lack of control
Why is external validity high in correlational research?
No strong control like other studies. Control removes external validity
When is correlational research used?
When the manipulation of the independent variable is not possible, feasible or ethical.
For establishing reliability and validity
(Test-retest, convergent, discriminant)
Exploratory research
Is research setting important in correlational research?
No. Even research done in a lab is correlational when no variables are manipulated or controlled
What variables are involved in correlational research?
Both continuous or categorical variables can be involved
What is the correlation coefficient?
Pearson’s r
Ranges from -1.00 to +1.00
Indicates the strength of relationship and sign indicates the direction
What is a scatterplot?
A visual presentation of the relationship between two variables
Two measurements for each individual is represented by a single point
How do we interpret a strong correlation?
It doesn’t mean there is a causal relationship
There is a possible third variable
Might be reversed- might not be in the direction that you think
Might reflect the influence of outliers
How do we interpret a weak correlation?
Does not mean there isn’t a causal relationship
Might reflect curvilinear relationship (exponential growth)
Might reflect outliers
Might reflect truncated range (haven’t measured the variables over a big enough range to capture)
What is a spurious correlation?
Relationship between two variables seems causal but is not, often is a potential third variable involved
What is truncated range?
Occurs when haven’t captured enough data and only looking at a part of the range
Masks relationships or having what appears to be a relationship when there isn’t one at all
What are the assumptions for Pearson’s r?
Linearity - straight line, not curved
Interval or ratio data - two continuous variables
Univariate normality - both variables should be normally distributed and have no outliers
Bi-variate normal - distribution of variables in combination should be normal
Correlation coefficient sand their strengths?
0-.10 trivial
.10-.29 weak
.30-.49 moderate
.50-1.00 strong
How to calculate a confidence interval for Pearson’s r
r needs to be converted to a z score
r and N can be used to calculate CI in the excel worksheet
How do you summarise correlation?
r(N=x) = (correlation coefficient), p = x, 95% CI [x,y]
Plain English about strength of relationship
Statistics
Interpret confidence intervals
What these findings mean
What is regression?
Quantifies the strength of association among one or more predictable variables and one criterion or outcome variable
What is the intercept in simple regression?
It is the value of the outcome variable when the predictor variable is 0
By centring the predictor, it makes the intercept equal to the mean of the outcome variable
What is the simple regression equation?
Y = (B coefficient)x + (intercept)
What is the b coefficient (slope)?
How much the outcome variable (Y) changes for each unit of change in the predictor variable (Y)
What is R-squared?
It is how much of the variance in the outcome variable is accounted for by the predictor variable
x100 to convert to percentage of how much variation in outcome variable
What does it mean if R-squared = 1?
If you are given the value of the predictor you can calculate the exact value of the outcome
What does it mean if R-squared = 0?
The predictor is not helpful at all
How do we summarise simple regression? (Statistics)
F(df, df) = (F), p = x, R = x, R^2 = x
F = name of statistic
How do we summarise simple regression in text?
Plain English telling about statistical significance
Stats
% from R^2 and explain
B coefficient and increase/decrease every…
How else can univariate normality be tested other than skewness tests etc?
Visual inspection. Using histograms, box plots and other graphical method such as stem and leaf plots
How do we check bivariate normality?
Scatterplots, should be no regions of large empty space and points should be evenly distributed
What did Cohen (1988) define a strong correlation as?
One in which the association is clearly vis able to the naked eye
What is conditional probability?
A p-value form of probability.
Probability of an event occurring conditional on some other specific situation or event being true