<h4>母比率の差の検定</h4><p>2つの独立な標本から得られた比率の差について、正規近似を用いた検定を行います。</p><h4>検定統計量の理論的基礎</h4><p class='step'><strong>Step 1: 標本比率の計算</strong></p><p>各標本から標本比率を計算:</p><div class='formula'>$\hat{p}_1 = \frac{x_1}{n_1} = \frac{18}{50} = 0.36
lt;/div><div class='formula'>$\hat{p}_2 = \frac{x_2}{n_2} = \frac{12}{40} = 0.30
lt;/div><p class='step'><strong>Step 2: 帰無仮説の下での共通比率の推定</strong></p><p>帰無仮説$H_0: p_1 = p_2 = p$の下で、共通比率$p$を最尤推定:</p><div class='formula'>$\hat{p} = \frac{x_1 + x_2}{n_1 + n_2} = \frac{18 + 12}{50 + 40} = \frac{30}{90} = \frac{1}{3} \approx 0.3333
lt;/div><p class='step'><strong>Step 3: 標準誤差の計算</strong></p><p>比率の差の標準誤差:</p><div class='formula'>$SE(\hat{p}_1 - \hat{p}_2) = \sqrt{\hat{p}(1-\hat{p})\left(\frac{1}{n_1} + \frac{1}{n_2}\right)}
lt;/div><div class='formula'>$= \sqrt{\frac{1}{3} \times \frac{2}{3} \times \left(\frac{1}{50} + \frac{1}{40}\right)}
lt;/div><div class='formula'>$= \sqrt{\frac{2}{9} \times \left(\frac{1}{50} + \frac{1}{40}\right)}
lt;/div><p class='step'><strong>Step 4: 分母の詳細計算</strong></p><p>分母の計算:</p><div class='formula'>$\frac{1}{50} + \frac{1}{40} = \frac{4}{200} + \frac{5}{200} = \frac{9}{200} = 0.045
lt;/div><div class='formula'>$SE = \sqrt{\frac{2}{9} \times 0.045} = \sqrt{\frac{2 \times 0.045}{9}} = \sqrt{\frac{0.09}{9}} = \sqrt{0.01} = 0.1
lt;/div><h4>検定統計量の計算</h4><p class='step'><strong>Step 5: Z統計量の算出</strong></p><p>検定統計量$Z$は以下のように計算:</p><div class='formula'>$Z = \frac{\hat{p}_1 - \hat{p}_2}{SE(\hat{p}_1 - \hat{p}_2)} = \frac{0.36 - 0.30}{0.1} = \frac{0.06}{0.1} = 0.6
lt;/div><div class='key-point'><div class='key-point-title'>母比率の差の検定の特徴</div><ul><li><strong>正規近似の条件</strong>:$n_1\hat{p}, n_1(1-\hat{p}), n_2\hat{p}, n_2(1-\hat{p}) \geq 5
lt;/li><li><strong>独立性の仮定</strong>:2つの標本が独立に抽出されること</li><li><strong>共通比率の推定</strong>:帰無仮説の下で最尤推定を使用</li><li><strong>連続性補正</strong>:小標本の場合はイエーツの補正を適用</li><li><strong>片側・両側検定</strong>:研究仮説に応じて選択</li></ul></div><