1.15
実験結果の比較: スチューデントの T 検定
t 検定は、サンプル平均と母集団平均を比較したり、2 つのデータセットからの 2 つの平均を比較したりするために使用される統計手法です。検定統計量は、選択した信頼区間でのデータセット内の標準偏差、平均、測定値の数から計算され、この信頼水準での臨界値の表と比較されます。検定統計量が臨界値より小さい場合…
メソッド、サンプル、または分析者の変更が分析結果に与える影響は、2つの実験の1つだけを変更することによって調査されます。
t検定は、1つの平均の統計的有意性を別の平均値または既知の値と比較します。
t検定では、対応のないデータは、同じソースからの2つの独立したデータセットです。対応のあるデータとは、関連するデータの 2 つのセットです (たとえば、一連のサンプル上の 2 つの異なるメソッドの 2 つのセット)。
検定統計量は、既知の値に対して選択した信頼区間から導き出されます。選択した有意水準における特定の自由度の表形式の検定統計量と比較すると、前者が後者の範囲内にある場合、帰無仮説が受け入れられます。
片側検定では、増加のみまたは減少のみを探しますが、両側検定では、いずれかの方向の変化が検索されます。
View the full transcript and gain access to JoVE Core videos
Q1: What is the difference between paired and unpaired data in a t-test?
Unpaired data are two independent sets of replicate measurements from the same source, while paired data refer to measurements taken on the same samples from two different methods or at different time points. Paired data allow direct comparison of methods or conditions on identical subjects, whereas unpaired data compare separate groups.
Q2: How does a t-test determine if differences between means are statistically significant?
The t-test calculates a test statistic from the mean, standard deviation, and number of measurements at a selected confidence interval. This statistic is compared to a critical value from a table. If the test statistic is smaller than the critical value, the null hypothesis is accepted, indicating the difference comes from random errors, not systematic causes.
Q3: What does it mean when a t-test rejects the null hypothesis?
When the test statistic exceeds the critical value, the null hypothesis is rejected, indicating the difference between means is statistically significant and cannot be explained by random errors. The difference may arise from systematic error methodological and sampling errors, analyst variation, or true phenomenological differences in the data.
Q4: What is the difference between one-tailed and two-tailed t-tests?
A one-tailed test examines only one side of the normal distribution curve, looking for either an increase or a decrease in a specific direction. A two-tailed test uses both sides of the distribution, testing for any change in either direction. Two-tailed tests are more conservative and commonly used when the direction of difference is unknown.
Q5: How does changing the method, sample, or analyst affect t-test results?
The t-test studies the influence of these variables by altering only one factor in a pair of experiments, isolating its effect on results. By comparing means from experiments with controlled single-variable changes, researchers can determine whether differences are statistically significant or arise from random variation in the measurement process.
Q6: What role does the confidence interval play in calculating the t-test statistic?
The test statistic is derived from the selected confidence interval, which determines the critical value threshold used for comparison. The confidence interval establishes the significance level at which the test operates, influencing whether observed differences are deemed statistically significant or attributable to random measurement error.
Q7: Why is the degree of freedom important when comparing a t-test statistic to critical values?
The degree of freedom determines which critical value table row to use for comparison at a chosen significance level. Different sample sizes produce different degrees of freedom, affecting the threshold for accepting or rejecting the null hypothesis. This ensures the t-test appropriately accounts for sample size when evaluating statistical significance.