การออกแบบวิจัยและการตีความสถิติ
GRE Subject Test · GRE Psychology หัวข้อ 11 4:42 การบรรยายภาษาอังกฤษ · คำบรรยายภาษาอังกฤษ + 中文 ลอยตัวบนภาพ
บท
Transcript
Choose a test from the design, scale and assumptions, not merely the number of groups.
根据设计、尺度和假设选择检验方法，而不仅仅是依据组数。
Paired observations require a paired or repeated-measures analysis; independent groups require an independent-group analysis.
配对观测需要配对或重复测量分析；独立组别需要独立组分析。
An interaction can be meaningful even if averaged main effects cancel.
即使平均主效应相互抵消，交互作用仍可能具有意义。
Statistical significance is evidence under a model, not a condition for describing an observed difference.
统计显著性是在模型下的证据，而非描述观测差异的条件。
Report estimates and uncertainty and qualify population claims.
报告估计值和不确定性，并对总体主张加以限定。
A true experiment manipulates an independent variable with random assignment of participants to conditions; between-subjects designs give different people to each condition, within-subjects designs measure the same people in every condition and need counterbalancing.
真实实验通过随机分配参与者到不同条件来操纵自变量。 被试间设计为每个条件提供不同的被试。
A quasi-experiment lacks full assignment, and a correlational study does not manipulate an independent variable.
被试内设计对同一被试在所有条件下进行测量，需要平衡处理。 准实验缺乏完全随机分配，而相关研究不操纵自变量。
Correlation alone does not establish causation or temporal order.
仅凭相关性无法确立因果关系或时间顺序。
Random assignment supports internal validity (causal inference); random sampling supports external validity (generalisability).
随机分配支持内部效度，即因果推断。 随机抽样支持外部效度，即可推广性。
They answer different questions.
它们回答的是不同的问题。
Match analysis to design, measurement scale and assumptions.
分析需与实验设计、测量尺度及假设相匹配。
Two independent quantitative groups may suit an independent-samples t-test; paired data require a paired analysis.
两个独立的定量组可能适用独立样本t检验。 配对数据需要配对分析。
Factorial designs often use analysis of variance; categorical counts may suit a chi-square test when expected-count assumptions hold.
析因设计通常使用方差分析。 当期望频数假设成立时，分类计数可能适用卡方检验。
Correlation or regression examines specified associations; neither alone establishes causation.
相关或回归用于检验特定的关联。 但两者单独均不能确立因果关系。
A main effect averages one factor across the other factor.
主效应是在另一个因子的水平上对某一因子取平均。
An interaction means one effect depends on the other factor.
交互作用是指一个效应依赖于另一个因子。
Nonparallel sample means suggest an interaction pattern, but variability and the design are needed for population inference.
非平行的样本均值暗示了交互作用模式，但需要变异性与设计才能进行总体推断。
Interactions can exist when averaged main effects cancel; lack of significance is not proof of no effect.
当平均后的主效应相互抵消时，交互作用仍可能存在；不显著并非无效应的证明。
Mean age 26 with most students 25 or younger is compatible with right skew, but does not uniquely determine distribution shape.
平均年龄为26岁且大多数学生年龄在25岁或以下，与右偏分布相容，但不能唯一确定分布形状。
Statistical significance is not effect size.
统计显著性不等于效应量。
A p-value is a probability of results at least as extreme under the specified null model and assumptions, not the probability the null is true.
p值是在指定的零模型和假设下，出现至少同样极端结果的概率，而非零假设为真的概率。
Type I and Type II errors depend on design and decision thresholds.
I型错误和II型错误取决于研究设计和决策阈值。
Fictional mean scores are; task A, novices 4 and experts 8; task B, novices 8 and experts 4.
虚构的平均分如下：任务A，新手为4，专家为8；任务B，新手为8，专家为4。
Compute the marginal means and describe the pattern.
请计算边际均值并描述该模式。
The novice marginal mean is 4 plus 8, over 2, which is 6; the expert marginal mean is 8 plus 4, over 2, which is 6.
新手边际均值为4加8除以2，等于6；专家边际均值为8加4除以2，等于6。
The task A marginal mean is 6, and the task B marginal mean is also 6.
任务A的边际均值为4加8除以2，等于6；任务B的边际均值为8加4除以2，等于6。
The task contrast reverses by expertise, an interaction pattern despite equal marginal means.
任务对比随专业水平反转，这是一种交互作用模式，尽管边际均值相等。
Means alone do not establish a population interaction; variability, design and inference are needed.
仅凭均值无法确立总体中的交互作用，还需要变异性、设计和推断。
Parallel lines mean B adds +2 at both levels of A: main effects and no interaction.
两条线平行，说明 B 在 A 的两个水平上都加 2：只有主效应，没有交互作用。
Read the labels before drawing a conclusion.
先读清标签，再下结论。
The same 24 people complete two tasks.
同样的24人完成两项任务。
Another analyst treats the 48 scores as independent.
另一位分析师将这48个分数视为独立数据。
Explain the problem and a suitable analysis family.
请解释存在的问题以及合适的分析家族。
Pause before the model.
在查看模型前暂停。
Scores within a person are related.
同一个人的得分之间存在关联。
Ignoring pairing uses an inappropriate independence assumption and loses information about within-person differences.
忽略配对会采用不恰当的独立性假设，并丢失关于个体内差异的信息。
For suitable quantitative difference data and assumptions, a paired t-test may fit; a justified non-parametric alternative may be appropriate otherwise.
对于合适的定量差值数据和假设，配对t检验可能适用；否则，合理的非参数替代方法可能更为恰当。
Randomized or counterbalanced order also addresses sequence effects.
随机化或平衡顺序也能解决序列效应问题。
Drawing a causal conclusion from a correlational design, or reporting a significant interaction as if it were a main effect of one factor.
从相关设计得出因果结论，或将显著交互作用报告为某一因素的main effect。
State the source and the limits.
请说明来源与局限性。
A small p-value accompanies a tiny estimated effect.
较小的p值伴随着微小的估计效应。
Separate the statistical claim from practical importance.
请将统计声明与实际重要性区分开来。
Pause and write a justified answer.
请先暂停，完成这道原创练习再继续。