The following LOGISTIC procedure output analyzes the relationship between a binary response and an ordinal predictor variable, wrist_size Using reference cell coding, the analyst selects Large (L)…

Question

Options

• A0.0819
• B0.5663
• C-3.7727
• D-1.0415

• A
  4% (1)
• B
  4% (1)
• C
  9% (2)
• D
  83% (19)

Explanation

Option D (-1.0415) is correct because, with reference cell coding, the reference category (Large) has all indicator variables set to zero. The estimated logit is therefore equal to the intercept alone: logit(Large) = β₀ = -1.0415. No other terms contribute when every dummy variable equals 0.

Why the distractors are wrong:

• A (0.0819) is a parameter estimate (β coefficient) for one of the non-reference wrist size categories, not the logit for Large.
• B (0.5663) is similarly a beta coefficient for another non-reference level - adding it to the intercept would give that category's logit, not Large's.
• C (-3.7727) is the estimated logit for a non-reference category (intercept + its beta coefficient), not for the reference level.

Memory tip: Think of the reference level as the "baseline" - it carries no indicator weight, so its logit is the intercept alone. A quick check: if you see a value that matches the Intercept row in the LOGISTIC output, that's always the answer for the reference category's logit.

No community discussion yet for this question.