Flashcards
Statistics Module 3
Flashcards do conjunto (23)
- 1A measurement of variability in scatterplots that measures the strength of a linear correlation.Correlation coefficient: "r" value
- 2What does "r" measure?Measures the strength and direction of a linear association. r is positive if there is a positive linear association and negative if there is a negative linear association.
- 3The closer the "r" value is to ONE, or NEGATIVE ONE, the closer______The closer it is to having a perfect linear association.
- 4Can an r value alone tell you if the association is linear? Or whether there is no association?The correlation coefficient r is a measure of the strength and direction of a linear association, but the value of r alone cannot tell you if an association is linear, and a value of r that is close to zero does not mean that there is no association. It is important to look at the scatterplot before interpreting a correlation coefficient.
- 5What's a lurking/confounding variable?One that isn't being measured by the study but still impacts the results of the study. Creates an association between two variables.
- 6What is extrapolation? What's the problem with it?the process of using a statistical model, like a line, to make predictions that are outside the range of the available data. The problem with extrapolation is that we don’t know whether the pattern continues in the same way beyond the data points. Because of this, extrapolation is unreliable for predicting values of a response variable.
- 7In regression statistics, what is a residual?The difference between the ACTUAL data point and the PREDICTED point based on the line of best fit.
The measure of variation of points from a line. - 8Why do we do the sum of the SQUARED RESIDUALS?Because some residuals are positive and negative, so they tend to cancel themselves out. Therefore squaring them all makes them positive.
- 9If we use the sum of squared residuals as a measure of fit, does a smaller or larger sum indicate a better fit?A smaller sum indicates a better fit. This is because when the line fits data well, the estimates are accurate, meaning that residuals are small.
- 10What is an LSR line? Other names for it?Least Squares Regression line
or line of best fit.
No other line fits the data better! - 11How do you get a residual? Explained? Unexplained?Resdidual =
UNEXPLAINED: (observed y-value) MINUS (predicted y-value) or y-yhat.
EXPLAINED: yhat-mean of y values. - 12What is an SSE in statistics?A Residual sum of squares, or
Sum of the Squares of the Errors, also known as a least squares regression line. - 13What is the TOTAL DEVIATION for a point on a scatterplot?The difference between the y-value and the MEAN of all the y-values. or
y − (mean of)y= - 14—
- 15What is TOTAL deviation, and EXPLAINED deviation?TOTAL deviation: The distance from the y-value in question to the mean of all the y-values.
EXPLAINED deviation: the part of total deviation that is explained as the difference from the mean of the y values to the mean predicted by the line. - 16What's an UNEXPLAINED (also called residual) deviation?The difference between the value of y predicted by the regression line and the actual value of y
- 17How do you compute the UNEXPLAINED (also called RESIDUAL) deviation?Subtract a predicted y-value from an observed y-value, or y-yhat
- 18What's the coefficient of determination?r squared. It explains the total variability in all the y values that is explained by the least squares regression line.
It's always between 0 and 1. - 19What's the advantage to using r squared, or the coefficient of determination?It can be used to asses NONLINEAR models, whereas r can only be used to asses LINEAR models!
- 20How do you calculate error?OBSERVED value minus PREDICTED value
- 21When is a linear model appropriate?When the residual plot displays no pattern and the residuals are scattered randomly.
- 22When should a nonlinear model be examined?When a pattern is present among the data plots, no matter how high the rsquared value is.
- 23Do I want residuals squared (rsquared) to be small and se (standard errors) to be big? or vice versa?I want rsquared to be large and se to be small.
Informações
- Descrição
- Statistics FlashCards sobre Statistics Module 3, criado por Diana Kleven em 22-06-2017.
- Criado por
- Diana Kleven
- Visualizações
- 1
- Avaliação