Dr. QuantMaster MCP Server

Instructions

Implementation Reference

• src/tools/index.ts:1144-1191 (handler)

  The handler function executing the core logic for calc_sample_size tool. Computes required sample size for various tests (t-test, ANOVA, regression, etc.) using approximate formulas based on effect size, alpha, and power.

  function handleCalcSampleSize(args: Record<string, unknown>) {
    const testType = args.test_type as string;
    const effectSize = args.effect_size as number;
    const alpha = (args.alpha as number) || 0.05;
    const power = (args.power as number) || 0.80;
  
    // Simplified sample size formulas
    const zAlpha = alpha === 0.05 ? 1.96 : (alpha === 0.01 ? 2.576 : 1.645);
    const zBeta = power === 0.80 ? 0.84 : (power === 0.90 ? 1.28 : 0.52);
  
    let n: number;
    let formula: string;
  
    switch (testType) {
      case "t_test_two":
        n = Math.ceil(2 * ((zAlpha + zBeta) / effectSize) ** 2);
        formula = "n per group = 2 × ((z_α + z_β) / d)²";
        break;
      case "t_test_paired":
        n = Math.ceil(((zAlpha + zBeta) / effectSize) ** 2);
        formula = "n = ((z_α + z_β) / d)²";
        break;
      case "anova":
        const groups = (args.groups as number) || 3;
        n = Math.ceil((groups * (zAlpha + zBeta) ** 2) / (effectSize ** 2));
        formula = "n per group = k × (z_α + z_β)² / f²";
        break;
      case "regression":
        const predictors = (args.predictors as number) || 5;
        n = Math.ceil((8 / (effectSize ** 2)) + predictors);
        formula = "n ≈ 8/f² + k (Green, 1991 rule)";
        break;
      default:
        n = Math.ceil(2 * ((zAlpha + zBeta) / effectSize) ** 2);
        formula = "Generic formula used";
    }
  
    return {
      test_type: testType,
      effect_size: effectSize,
      alpha,
      power,
      required_n: n,
      formula,
      interpretation: `${power * 100}% 검정력으로 효과크기 ${effectSize} 탐지에 필요한 최소 표본: ${n}`,
      note: "실제 계산은 G*Power 또는 pwr 패키지 사용 권장"
    };
  }

• src/tools/index.ts:108-126 (registration)

  Tool registration entry in the exported tools array, defining the name, description, and input schema for validation.

    name: "calc_sample_size",
    description: "필요 표본크기 계산 (t-test, ANOVA, regression, proportion)",
    inputSchema: {
      type: "object",
      properties: {
        test_type: {
          type: "string",
          enum: ["t_test_two", "t_test_paired", "anova", "regression", "proportion", "chi_square"],
          description: "검정 유형"
        },
        effect_size: { type: "number", description: "효과크기 (Cohen's d, f, f²)" },
        alpha: { type: "number", description: "유의수준 (기본: 0.05)" },
        power: { type: "number", description: "검정력 (기본: 0.80)" },
        groups: { type: "number", description: "집단 수 (ANOVA)" },
        predictors: { type: "number", description: "예측변수 수 (회귀)" },
      },
      required: ["test_type", "effect_size"],
    },
  },

• src/tools/index.ts:795-796 (registration)

  Registration in the handleToolCall switch statement, dispatching calls to the specific handler function.

    return handleCalcSampleSize(args);
  case "calc_power":

• src/tools/index.ts:110-125 (schema)

  Input schema definition for the calc_sample_size tool, specifying parameters like test_type, effect_size, alpha, power with types, enums, and required fields.

  inputSchema: {
    type: "object",
    properties: {
      test_type: {
        type: "string",
        enum: ["t_test_two", "t_test_paired", "anova", "regression", "proportion", "chi_square"],
        description: "검정 유형"
      },
      effect_size: { type: "number", description: "효과크기 (Cohen's d, f, f²)" },
      alpha: { type: "number", description: "유의수준 (기본: 0.05)" },
      power: { type: "number", description: "검정력 (기본: 0.80)" },
      groups: { type: "number", description: "집단 수 (ANOVA)" },
      predictors: { type: "number", description: "예측변수 수 (회귀)" },
    },
    required: ["test_type", "effect_size"],
  },