Which measure of correlation is most suitable when data are recorded as ranks rather than precise numerical values?
Correct answer: B. Spearman's rank correlation
- A. Pearson's product-moment correlation
- B. Spearman's rank correlation
- C. The coefficient of variation
- D. The standard error of the mean
Explanation
Spearman's rank correlation measures association using the relative order of observations. Pearson's correlation is generally used with quantitative measurements and their actual distances.
Report an error
The more specific you are, the faster it gets fixed. A source beats an opinion.
Prefer email? support@testustad.com
About Correlation and Regression
Correlation describes the direction and strength of association between two variables using scatter diagrams and correlation coefficients. Regression covers fitted lines, regression equations, prediction and least squares, with a clear distinction between association and causation and between estimating a response value and reversing the roles of the variables.
Practise Correlation and Regression
30 free Correlation and Regression MCQs from Quantitative Methods, each with the correct answer and an explanation. Unlimited attempts, no account needed.
Exams that ask Quantitative Methods questions like this
Quantitative Methods is on this paper prepared for on TestUstad, and all of them draw the same bank, so this question is worth knowing for it.
More Correlation and Regression questions
If the standard deviation of X is doubled while the standard deviation of Y and the correlation remain unchanged, the regression coefficient of Y on X will:
If the regression coefficients are b_yx = 0.5 and b_xy = 0.72, the correlation coefficient is:
A regression equation used to predict X from Y is known as the regression equation of:
If every value of X is multiplied by 5 and every value of Y is multiplied by 2, the correlation coefficient will:
A correlation coefficient close to zero can still occur when two variables have:
Which condition is required for an ordinary least-squares regression slope to be uniquely defined?