Regression and Standard Error Flashcards
What is Regression?
It identifies the best fitting/regression line through a data set
What should be represented on each axis?
x axis should be the independent variable
y axis should be the dependent variable
Examples of what we can use regressions for?
Workplace productivity (x: manual dexterity, y: units produced)
Crime (x: money cut from local budget, y: crimes reported)
Effectiveness of training (x: hours trained, y: job performance)
Health (x: age first smoked, y: lifespan)
How can you pretest in regressions?
When looking at the graph, if you wanted to work out roughly what a score on the x axis would affect the y axis. You draw up to the regression line and then straight across at a right angle to the y axis.
What is the regression equation and what does each component stand for?
Y = a + (b x X)
Y = predicted score a = intercept b = slope X = known score (the score we want to find out the prediction for)
How do we calculate the regression line?
Sum of least squares/least squares solution.
It is the sum of squared deviations (from the line)
What does the slope tell us?
Direction of relationship
Nature of relationship.
When you calculate the slope, it will display how much of an increase on the y axis there will be for every unit increase on the x axis.
This is an average
What is the intercept?
This is where the regression crosses the y axis.
What are the issues with making predictions?
There could be other variables that haven’t been looked at that could be having an effect.
Maybe the wrong variable was looked at.
What is standard error?
The average amount by which our estimate is likely to be wrong by.
What are the sources of error in regression?
Estimated Y value
Slope of regression line, b
Point of intercept, a
What is the confidence interval?
Confidence Intervals are found by using the standard error. This is the range with which we can be fairly sure the score will fall within.
