OLS Residuals and Model Assumptions

In OLS regression, a residual represents the discrepancy between the actual observed values of the dependent variable and the values predicted by the regression model. It indicates how well the model fits the data, with smaller residuals suggesting a better fit and more accurate predictions.

Multicollinearity among regressors is not a classical OLS assumption; rather, it is a potential issue that can arise when independent variables are highly correlated. The classical OLS assumptions focus on the relationship between the dependent variable and the parameters, as well as the properties of the error terms, not the correlations among regressors.

Homoscedasticity refers to the property of a regression model where the variance of the residuals remains constant across all levels of the independent variable (X). This consistency in variance is crucial for valid statistical inference, ensuring that the model's predictions are reliable and that standard errors are accurate.

In the presence of heteroscedasticity, Ordinary Least Squares (OLS) estimators remain unbiased because they still accurately estimate the population parameters. However, the variability of these estimators is affected, leading to biased standard errors. This can result in incorrect conclusions when conducting hypothesis tests or constructing confidence intervals.

In Ordinary Least Squares (OLS) regression, the residuals represent the differences between observed and predicted values. The inclusion of an intercept ensures that the regression line passes through the mean of the dependent variable, resulting in the sum of these residuals being zero. This property reflects the balance of positive and negative deviations around the regression line.

The Breusch-Pagan test is specifically designed to detect heteroscedasticity in regression models by examining whether the variance of the residuals is dependent on the values of the independent variables. It assesses the relationship between the squared residuals and the predictors, helping to identify if the assumption of constant variance is violated.

Autocorrelation in residuals indicates that the error terms are correlated across observations, which violates the assumption that errors are independent of each other. This correlation can lead to inefficient estimates and affect the validity of hypothesis tests, ultimately compromising the reliability of the regression model's results.

The Durbin-Watson statistic is used to detect autocorrelation in the residuals of a regression analysis. Autocorrelation occurs when the residuals are not independent from one another, which can lead to inefficient estimates and misleading statistical inferences. This statistic helps assess the presence of such patterns in the data.

In a Q-Q plot, if the points lie approximately on a straight diagonal line, it indicates that the residuals are normally distributed. This alignment suggests that the quantiles of the residuals match the quantiles of a normal distribution, confirming the assumption of normality in the data.

The assumption of 'no perfect multicollinearity' requires that the regressors in a regression model are not perfectly linearly related to each other. This ensures that each regressor contributes unique information to the model, allowing for accurate estimation of coefficients and avoiding issues with redundancy among the predictors.

A funnel pattern in a scatter plot of residuals against fitted values indicates that the variability of the residuals changes with the level of the fitted values. This phenomenon, known as heteroscedasticity, suggests that the assumption of constant variance in the errors is violated, which can affect the reliability of statistical inferences.

The Gauss-Markov theorem asserts that under certain classical assumptions, the Ordinary Least Squares (OLS) estimator is the most efficient among all linear estimators, meaning it has the smallest variance. This property ensures that OLS provides unbiased estimates of the true parameters, making it a reliable method for linear regression analysis.