Aorus
07-09-2009, 11:41 PM
Hi All,
I'm not sure how to approach this problem.
I have a data set which is clearly linear, but the error in the x-variable is significant relative to the error in the y-variable (i.e. the errors are the same). The x and y values are in the range of 50 to 110, with a constant standard error of 0.34. I have 90 data points. Since the least squares linear regression method assumes a negligible error in the x-variable, I'm not sure that it's appropriate to this problem.
Ultimately what I'm after is a statistically sound way of predicting the y-variable and determining the uncertainty in this prediction, i.e. the 95% confidence interval for the predicted y (yhat). If simple linear regression is not suitable, is there another method?
Thanks!
I'm not sure how to approach this problem.
I have a data set which is clearly linear, but the error in the x-variable is significant relative to the error in the y-variable (i.e. the errors are the same). The x and y values are in the range of 50 to 110, with a constant standard error of 0.34. I have 90 data points. Since the least squares linear regression method assumes a negligible error in the x-variable, I'm not sure that it's appropriate to this problem.
Ultimately what I'm after is a statistically sound way of predicting the y-variable and determining the uncertainty in this prediction, i.e. the 95% confidence interval for the predicted y (yhat). If simple linear regression is not suitable, is there another method?
Thanks!