Hierarchical Regression Analysis

Hierarchical regression analysis is a multiple regression procedure in which predictors are entered in researcher-specified blocks to test how much additional variance each block explains beyond earlier ones.

위계적 회귀분석은 연구자가 정한 순서대로 예측변수를 블록 단위로 투입하여 각 블록이 이전 단계보다 추가로 설명하는 분산을 검정하는 다중회귀 절차이다.

How do I test whether new predictors explain variance beyond control variables?

통제변수 외에 새로운 변수가 추가로 설명력을 갖는지 어떻게 검정하나요?

Hierarchical (sequential) regression enters predictors in theory-driven blocks, such as controls first and focal variables second, and tests the change in R² at each step. It fits when the question is incremental validity: whether a set of variables adds explanatory power over variables already in the model.

When to use Hierarchical Regression Analysis

• Testing focal predictors after controlling for demographics
• Assessing incremental validity of a new scale over established measures
• Adding an interaction term in a final block to test moderation

When to use something else

• The order of entry is chosen by the software from the data: that is stepwise regression, which is generally discouraged for inference
• The outcome is binary: use hierarchical logistic regression

Assumptions and requirements

• Linear relationships between predictors and outcome
• Independent, homoscedastic, approximately normal residuals
• No severe multicollinearity (check VIF)
• Block order fixed in advance on theoretical grounds

What the analysis reports

• R² and adjusted R² for each model
• ΔR² with F-change test and p-value for each block
• Unstandardized and standardized coefficients with standard errors
• Collinearity diagnostics such as VIF

How to report it

Adding [block 2 predictors] in Step 2 explained an additional [value]% of the variance in [outcome], ΔR² = [value], ΔF([df1], [df2]) = [value], p = [value].

Related methods

• Multiple Linear Regression
• Moderation Analysis
• Stepwise Regression

Run Hierarchical Regression Analysis on Quorant