Top Common Mistakes in Pre-U Cambridge Statistics and How to Correct Them | 剑桥Pre-U统计常见误区与纠正方法

📚 Top Common Mistakes in Pre-U Cambridge Statistics and How to Correct Them | 剑桥Pre-U统计常见误区与纠正方法

Statistics in Pre-U Cambridge can be challenging, and students often fall into common traps that cost them marks. Misunderstandings around p-values, confidence intervals, assumptions, and probability can lead to inaccurate conclusions even when calculations are correct. This article identifies these frequent pitfalls and provides clear corrections, helping you build robust statistical reasoning and aim for the highest grades.

剑桥Pre-U统计课程颇具挑战，学生常常落入一些常见陷阱而失分。即使计算正确，对p值、置信区间、假设和概率的误解也可能导致不准确的结论。本文指出这些常见误区并提供清晰的纠正方法，帮助你建立扎实的统计思维，争取最高等级。

1. Misunderstanding p-values and Statistical Significance | 对p值和统计显著性的误解

Many students incorrectly believe that a p-value below 0.05 confirms the research hypothesis or indicates a large effect size. Some even interpret it as the probability that the results occurred by chance. A related error is treating ‘p < 0.05’ as a magic threshold, ignoring effect magnitude and context.

许多学生错误地认为，p值低于0.05就证实了研究假设或表明效应量很大。一些人甚至将其解释为结果由偶然因素造成的概率。一个相关错误是把“p < 0.05”当作魔法阈值，忽略效应大小和背景。

In reality, the p-value is the probability of observing a test statistic at least as extreme as the one obtained, assuming the null hypothesis is true. It does not give the probability that the null hypothesis is true, nor does it measure the size of an effect. A tiny p-value can arise from a trivially small effect if the sample size is huge. Always accompany p-values with effect size measures (such as Cohen’s d or confidence intervals) and interpret them in the context of the study’s design and practical importance.

事实上，p值是在零假设为真的前提下，观测到至少与当前一样极端的检验统计量的概率。它并不给出零假设为真的概率，也不度量效应的大小。如果样本量巨大，即使效应微乎其微也可能产生极小的p值。应始终将p值与效应量指标（如Cohen’s d或置信区间）一起报告，并结合研究设计和实际重要性进行解释。

2. Confusing Confidence Intervals with Prediction Intervals | 混淆置信区间与预测区间

A widespread mistake is thinking that a 95% confidence interval for the mean implies a 95% probability that the true population mean falls within that specific interval. Students then misuse this interval to predict a future single observation, not realising that a much wider prediction interval is required.

一个普遍错误是认为均值的95%置信区间意味着有95%的概率真实总体均值落在该特定区间内。学生随后误用该区间来预测未来单个观测值，却未意识到需要宽得多的预测区间。

Confidence intervals are a frequentist concept: if we repeated the sampling process many times, approximately 95% of the constructed intervals would capture the true parameter. For any one interval, we cannot say there is a 95% probability it contains the parameter (unless adopting a Bayesian viewpoint). To capture a single future observation, you must use a prediction interval, which accounts for both the uncertainty in estimating the parameter and the natural variability of individual data points. Prediction intervals are always wider than the corresponding confidence intervals and should be used for forecasting.

置信区间是一个频率学派概念：如果我们重复抽样许多次，大约95%构造出的区间会捕获真实参数。对于任何一个区间，我们不能说它有95%的概率包含参数（除非采用贝叶斯观点）。要捕捉单个未来观测值，必须使用预测区间，它同时考虑了参数估计的不确定性和个体数据点的自然变异性。预测区间始终比对应的置信区间更宽，应用于预测。

3. Neglecting to Check Assumptions of Tests | 忽略检验前提条件的检查

Many candidates blindly apply the two-sample t-test without examining whether the data are approximately normal or whether the groups have equal variances. When sample sizes are small and distributions are skewed, this can seriously inflate the Type I error rate and lead to false conclusions.

许多考生盲目使用两样本t检验，却不检查数据是否近似正态或各组方差是否相等。当样本量小且分布偏斜时，这样做可能严重增大第一类错误率，导致错误结论。

Before performing a t-test, assess normality through histograms, Q–Q plots or the Shapiro–Wilk test. If normality is questionable, consider a non-parametric alternative such as the Mann–Whitney U test. For unequal variances, use Welch’s t-test, which adjusts the degrees of freedom. Remember that test assumptions are not optional extras; they are essential for valid inference. Always verify independence, normality and equal variances (or adjust accordingly) and report checks in your answer.

进行t检验前，通过直方图、Q–Q图或Shapiro–Wilk检验评估正态性。如果正态性可疑，考虑非参数替代方法如Mann–Whitney U检验。对于方差不齐，使用调整自由度的Welch t检验。记住，检验前提条件并非可选附加项；它们是有效推断的关键。务必验证独立性、正态性和方差齐性（或相应调整），并在答案中报告这些检查。

4. Misinterpreting Correlation and Causation | 错误解读相关与因果

Students frequently assert that a high correlation coefficient, say r = 0.9, proves that one variable causes the other. This leads to spurious causal claims and ignores the possibility of confounding variables or reverse causation.

学生频繁断言，高相关系数（例如r=0.9）证明一个变量导致另一个变量。这导致了虚假的因果论断，并忽略了混淆变量或反向因果的可能性。

Correlation merely measures the strength and direction of a linear association. It does not imply causation. A strong correlation could be driven by a third lurking variable (e.g., ice cream sales and drowning rates both increase in summer, but ice cream does not cause drowning). To establish causality, controlled randomised experiments or advanced causal inference methods are needed. When interpreting correlational data, always use careful language such as ‘is associated with’ and discuss potential confounders.

相关系数仅仅衡量线性关联的强度和方向，并不蕴含因果关系。强相关可能由第三个潜在的变量驱动（例如，冰淇淋销量和溺水率在夏季都上升，但冰淇淋不会导致溺水）。要确立因果关系，需要控制随机实验或高级因果推断方法。在解读相关性数据时，务必使用审慎的语言，如“与……相关”，并讨论潜在的混淆因素。

5. Mishandling the Chi-squared Test of Association | 关联性卡方检验的误用

A frequent error is applying the Pearson χ² test when expected frequencies are too low. The rule of thumb is that no expected frequency should be below 1, and no more than 20

Published by TutorHao | Pre-U 统计 Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）