How to Calculate Confidence Interval of Fitted Values

In regression analysis, the confidence interval of fitted values provides a range of plausible values for the predicted outcome. This guide explains how to calculate and interpret these intervals, with practical examples and an interactive calculator.

What is a Confidence Interval of Fitted Values?

A confidence interval of fitted values in regression analysis represents the range within which we expect the true value of a predicted outcome to fall, with a specified level of confidence (typically 95%). This interval accounts for both the uncertainty in the estimated regression coefficients and the variability of individual data points.

Key Points:

Fitted values are the predicted values from a regression model
Confidence intervals provide a measure of prediction uncertainty
Common confidence levels are 90%, 95%, and 99%

The confidence interval for a fitted value is calculated by considering both the standard error of the regression estimate and the standard error of the individual prediction. This results in a wider interval than the confidence interval for the mean response at a given point.

How to Calculate Confidence Interval of Fitted Values

The formula for calculating the confidence interval of a fitted value is:

Confidence Interval = Fitted Value ± t_{α/2, n-p-1} × SE_prediction

Where:

Fitted Value = Predicted value from the regression model
t_{α/2, n-p-1} = Critical t-value from the t-distribution
SE_prediction = Standard error of the prediction
α = Significance level (1 - confidence level)
n = Number of observations
p = Number of predictors

The standard error of the prediction (SE_prediction) is calculated as:

SE_prediction = √(SE_residual² + SE_estimate²)

Where:

SE_residual = Standard error of the residuals
SE_estimate = Standard error of the regression estimate

Step-by-Step Calculation

Fit your regression model to obtain the fitted values and residuals
Calculate the standard error of the residuals (SE_residual)
Calculate the standard error of the regression estimate (SE_estimate)
Combine these to get the standard error of the prediction (SE_prediction)
Determine the critical t-value based on your desired confidence level and degrees of freedom (n-p-1)
Calculate the confidence interval using the formula above

Note: The degrees of freedom for the t-distribution is n-p-1, where n is the number of observations and p is the number of predictors in your model.

Worked Example

Let's calculate the 95% confidence interval for a fitted value in a simple linear regression model with 10 observations and 1 predictor.

Step	Calculation	Value
1	Fitted Value	5.2
2	Standard Error of Residuals (SE_residual)	1.2
3	Standard Error of Estimate (SE_estimate)	0.8
4	Standard Error of Prediction (SE_prediction)	√(1.2² + 0.8²) = 1.56
5	Critical t-value (95% CI, df=8)	2.306
6	Confidence Interval	5.2 ± 2.306 × 1.56 = [1.7, 8.7]

This means we are 95% confident that the true value falls between 1.7 and 8.7.

Interpreting the Results

When interpreting confidence intervals for fitted values:

The interval provides a range of plausible values for the predicted outcome
A narrower interval indicates more precise predictions
Wider intervals occur when there's more uncertainty in the prediction
Confidence intervals should not be interpreted as probabilities that the true value falls within the interval

Practical Considerations:

Always check the assumptions of your regression model before interpreting confidence intervals
Consider the context of your data when interpreting the width of the intervals
For multiple predictions, the intervals may overlap or not overlap depending on the data

FAQ

What is the difference between a confidence interval for the mean and a confidence interval for a fitted value?: The confidence interval for the mean provides a range of plausible values for the average response at a given point, while the confidence interval for a fitted value provides a range for an individual prediction.
How does sample size affect the width of the confidence interval?: Larger sample sizes generally result in narrower confidence intervals, as there is less uncertainty in the estimates. However, the relationship between sample size and interval width depends on the specific regression model and data.
Can confidence intervals for fitted values be negative?: Yes, confidence intervals for fitted values can be negative if the predicted value itself is negative. The interpretation remains the same - the interval represents the range within which we expect the true value to fall.
What if my confidence interval is very wide?: A wide confidence interval indicates high uncertainty in your prediction. This could be due to a small sample size, high variability in your data, or a weak relationship between your predictors and the outcome variable.
How do I know if my confidence interval is appropriate for my data?: You should check the assumptions of your regression model (linearity, normality of residuals, homoscedasticity) and ensure that the model is appropriate for your data before interpreting confidence intervals.