- This topic has 3 replies, 2 voices, and was last updated 2 years ago by
.
- Posts
Greetings Sir, can you please help me with the question below?
Which TWO of the following are underlying assumptions of forecasts made using regression analysis?
(1) A curvilinear relationship exists between the two variables.
(2) The value of one variable can be predicted or estimated from the value of one other variable.
(3) A perfect linear relationship between the two variables.
(4) What has happened in the past will provide a reliable guide to the future.
A (1) and (2)
B (1) and (4)
C (2) and (4)
D (3) and (4).The answer that has been provided is option C. Without any explanation.
I know reason for as to Why Statement 1 is incorrect and Statement 4 is correct.
However I’m unable to understand why Statement 2 has been mentioned as correct statement and statement 3 is not.
Because, we indeed assume that there is a linear relationship between 2 variables, which is the reason as to why we form the general equation of y= a+bx.
Aren’t statement 2 and 3 interconnected?
Thank you so much sir 🙂
2 is correct because we are assuming that one variable is affected by the other variable
3 is not correct because we are not assuming that the relationship is perfectly linear – we are finding the line that best fits the data. (If it was perfectly linear then the correlation coefficient would be + or – 1, which is very unlikely as illustrated in the example that I work through in my lecture.)
Have you watched the lectures on this? They are in the Paper MA section of our website, because it is revision from Paper MA.
Yes sir i have watched your Lecturers, sir in case it was given that there is linear relationship between two variables, rather than a perfect linear relationship.
Would that statement be correct?And sir is this the very reason as to why we calculate the Coefficient of correlation and determination?
To test as how much can we rely upon the estimate/forecast as there is no perfect linear relationship between the variables?
- You must be logged in to reply to this topic.