AQA GCSE数学 统计 箱线图 直方图 相关性
统计在 AQA GCSE Mathematics 8300 里不是单独一张卷，它散在三份笔试卷的题目中间，属于 Specification 第 3.6 节的内容。这意味着统计题常常和比例、单位换算、代数运算混在一起出现，读题时先要认出它考的是哪一条要求。8300 的统计内容按 S1 到 S6 编号，Foundation 与 Higher 的差别集中在四分位数、箱线图、累积频率图和不等组距直方图这几块。很多学生把统计当成背公式的模块，实际上它的分数大多来自读图、写比较和说明理由。
Statistics in AQA GCSE Mathematics 8300 does not have a paper of its own; it is assessed inside the three written papers, under section 3.6 of the specification. Questions usually mix statistical work with ratio, unit conversion or algebra, so the first task is to identify which strand of S1 to S6 is being tested. The tier difference sits in quartiles, box plots, cumulative frequency diagrams and histograms with unequal class intervals. Treating statistics as a set of formulas to memorise misses where the marks actually sit, which is in reading graphs and justifying comparisons.
图表选择：分类数据与离散数据的画法
8300 的 S2 要求先判断数据类型，再决定用哪一种图。分类数据用条形图、饼图或象形图，未分组的离散数值数据用竖线图，两类图不能混着画。饼图的角度按各类频数占总数的比例乘 360 度算出来，圆心角的和必须回到 360 度。条形图的纵轴是频数，横轴是类别，条与条之间要留空隙；直方图的横轴是连续的组距，柱子紧贴在一起。很多学生以为条形图和直方图只是名字不同，实际上横轴是否连续、面积是否代表频数才是分界。阅卷时常见到把分组频数直接当高度画出的条形图，后面所有读数都会跟着错。动笔之前先写一句数据属于什么类型，能省掉大半的画图错误。
S2 asks for a decision about the data before any drawing starts. Categorical data belongs in bar charts, pie charts or pictograms, while ungrouped discrete values belong in vertical line charts, and the two families should not be swapped. Pie chart angles come from each category's share of the total multiplied by 360 degrees, and the angles must add back to a full turn. In a bar chart the bars stay apart because the categories are separate, whereas histogram bars touch because the horizontal axis is continuous. Many students lose the first mark by drawing a grouped frequency table as if the frequencies were heights, which spoils every later reading from the diagram.
时间序列：折线、趋势与缺失值
时间序列是 S2 里唯一按顺序排列数据的部分，AQA 8300 把它列在 additional foundation 一栏。横轴是时间，纵轴是观测量，相邻两点用折线连接，因为前后两期之间存在连续的变化。读这种图先看整体方向，再看有没有季节性的起伏，最后找出明显偏离趋势的点。计算相邻两期的差值能量化变化速度，正负号本身说明了上升还是下降。常见的错误有两处：把间隔不同的时间点画成等距，以及在数据缺失的那一期直接连过去。缺失值应该在图上留出断口，凭直觉补一个点等于发明数据。描述趋势时把年份和具体数值一起写进句子，比只写上升或下降更容易拿到分。
Time series is the one part of S2 where the order of values carries meaning, and AQA lists it in the additional foundation column. Time sits on the horizontal axis and the measurement on the vertical axis, with successive points joined because change between them is continuous. Reading such a graph means separating the overall direction from seasonal rises and falls, then spotting points that sit away from the pattern. A difference between two consecutive readings quantifies the speed of change and its sign shows the direction. Two mistakes are common: spacing irregular time points evenly, and joining a line straight across a missing reading, which invents data the table never supplied.
平均数、中位数与众数的选择
平均数、中位数和众数同属 S4，三者回答的是同一个问题的不同版本。平均数用到每一个数值，所以一个极端值就能把它拉走；中位数只看位置，数据偏斜时更稳；众数是出现次数最多的值，分类数据只能用众数。分组表里的众数变成众数组，答案要写组区间而不是一个数字。从分组表算平均数时用组中值代替每一个观测，得到的是估计值，答案里应该说明这一点。很多学生以为平均数一定比中位数更能代表数据，其实要看分布形状和有没有异常值。题目问哪一个平均量更合适时，理由必须落到这组数据的具体特征上，背定义拿不到解释分。
Mean, median and mode answer slightly different versions of the same question, and S4 expects a deliberate choice between them. The mean uses every value, which makes it sensitive to one extreme reading, while the median depends only on position and survives skew better. A mode is the only sensible average for categorical data, and in a grouped table it becomes the modal class, written as an interval rather than a single number. Averaging a grouped table means using midpoints in place of the observations, so the result is an estimate and the answer should say so.
极差、四分位距与异常值
极差是最大值减最小值，算得快，但它只用到两个数，对中间的数据完全不敏感。S4 要求考虑异常值，常用的判断是超过上四分位数 1.5 倍四分位距，或者低于下四分位数同样的距离。删掉异常值之前要说明理由，因为有些极端值是真实情况，删了会让描述失真。改动前后应该把平均数和中位数都重算一遍，看结论有没有变。常见的错误是把极差当成离散程度的唯一指标，遇到一个离群点就宣布数据很分散。比较两组数据时写清楚极差和四分位距各自说明了什么，比只丢出两个数字完整得多。
The range is the largest value minus the smallest, quick to compute but built on only two numbers, so it says nothing about the middle of the distribution. S4 asks for outliers to be considered, and the usual test places them beyond 1.5 interquartile ranges above the upper quartile or the same distance below the lower one. Removing an outlier needs a stated reason, because some extreme readings are genuine and deleting them distorts the description. Recomputing both the mean and the median afterwards shows whether the conclusion actually changes.
四分位数、四分位距与箱线图
四分位数把排好序的数据切成四段，下四分位数之下约有四分之一的数据，上四分位数之下约有四分之三。四分位距等于上四分位数减下四分位数，它丢掉了首尾各四分之一，所以比极差稳定。计算位置时先把数值从小到大排好，用数据的个数乘以四分之一或四分之三，得到的位置落在两个数之间就取平均。箱线图画五件事：最小值、下四分位数、中位数、上四分位数、最大值，箱子的长度就是四分位距。把两组数据的箱线图放在同一把尺子上比，比对着两个平均数看直观得多。箱线图不显示每一个点的频数，两组形状完全不同的数据也可能画出相似的箱子。
Quartiles cut an ordered data set into four parts, with roughly a quarter of the values below the lower quartile and three quarters below the upper one. The interquartile range is the distance between them, and because the outer quarters are discarded it resists extreme readings far better than the range. Positions are found by ordering the values and multiplying the count by one quarter or three quarters, averaging the two neighbouring values when the position falls between them. A box plot records five numbers, its box length is the interquartile range, and two box plots on one scale compare distributions more honestly than two means.
累积频率图与中位数的读法
累积频率图只在 Higher tier 出现，横轴用每一组的上边界，纵轴是累积到这一组的频数。第一个点从第一组的上边界起画，最后一点必须落在总频数上，这一步顺便检查有没有抄错数据。中位数在累积频率等于总数一半的位置读出，下四分位数和上四分位数分别在四分之一和四分之三处读出。读图时先水平画到曲线上，再垂直落到横轴，两个方向都要在图上留下痕迹。常见的错误是把组的下边界当横坐标，整条曲线会整体左移一组。曲线要平滑连接而不是画成折线，因为累积频率被当作连续增长的量。
Cumulative frequency diagrams appear only at Higher tier, with upper class boundaries on the horizontal axis and running totals on the vertical one. The first point sits at the upper boundary of the first class, and the last point must land on the total frequency, which doubles as a check that no value was copied wrongly. The median is read at half the total, the quartiles at one quarter and three quarters, and each reading needs both a horizontal line to the curve and a vertical drop to the axis. Plotting from class lower boundaries instead of upper ones shifts the whole curve one class to the left.
直方图与频率密度
不等组距的直方图同样属于 Higher 内容，它的关键是频率密度而不是频数。频率密度等于该组的频数除以组距，这样画出来每一根柱子的面积才等于频数。组距相同的题目可以直接拿频数当高度，一旦组距不同就必须换成频率密度。要求某个区间的频数时，把该区间覆盖的面积加起来，只覆盖半组的情况按比例取。常见的错误是把最高的一根柱子当成频数最多的那一组，组距宽的时候这个判断经常反过来。画图时柱与柱之间没有空隙，横轴标注的是组的边界而不是组中值。
Histograms with unequal class intervals are Higher tier material, and their currency is frequency density rather than frequency. Frequency density is the class frequency divided by the class width, which is what makes the area of each bar equal to the frequency it represents. Equal widths let a student use frequency as the height, but as soon as widths differ the density has to be calculated first. Finding the frequency for part of a bar means taking that share of its area, not reading the height off the axis.
散点图与相关性的四种描述
散点图属于 S6，两个变量各占一个轴，每一个点代表一次观测。描述相关性要同时给出方向和强弱，正向、负向、无相关，再加上强、弱、中等这类程度词。相关性只说两个变量有没有一起变化的趋势，不说明谁造成了谁，AQA 8300 的 S6 把这句话写在 additional foundation 的要求里。看两轴刻度是否从零开始，压缩或拉伸会改变视觉上的陡峭程度。常见的错误是把所有点依次连起来，散点图不需要连线，除非题目要求画趋势线。判断强弱时看点的分散程度，贴着一条假想线的是强相关，散得开的是弱相关。
A scatter graph belongs to S6, with one variable on each axis and each point standing for a single observation. Describing correlation means giving both direction and strength, so positive, negative or no correlation is paired with strong, moderate or weak. Correlation records that two variables move together, not that one causes the other, a point AQA writes into the additional foundation content of S6. Checking whether either axis starts at zero matters too, since compressing or stretching a scale changes how steep the pattern looks.
最佳拟合线、插值与外推
最佳拟合线穿过点群的中心，让两侧的点数大致相当，没有必要强行经过任何一个点。用拟合线做预测分两种情况：落在已有数据范围内的叫插值，可靠性较高；伸出数据范围一端的叫外推，风险随距离变大。题目要求评价一个预测时，答案要落在数据范围和趋势能否延续上，只说可能不准拿不到分。相关不等于因果，两个变量一起变化可能由第三个因素造成，也可能只是巧合。常见的错误是把外推的数值当成确定结论，还给它加上一位小数。写答案时说明这是在一定假设下得到的估计，比一个光秃秃的数字稳妥。
A line of best fit runs through the middle of the cloud, with roughly equal numbers of points on either side, and it does not have to pass through any single point. Prediction inside the range of the data is interpolation and carries reasonable confidence, while prediction beyond the end of the data is extrapolation and its risk grows quickly with distance. When a question asks for an evaluation, the answer has to address the range of the data and whether the trend can be assumed to continue. A correlation is not a cause, since a third factor or simple coincidence can make two variables move together.
抽样、样本与总体推断的限制
抽样属于 S1，位置在 additional foundation 一栏，核心是样本能提供总体的信息，但存在限制。随机抽样让每个个体被抽到的机会相同，能减少某一类人被过度代表；样本越大估计通常越稳定，但更大的样本无法修复一个有偏的抽取方式。判断题目里的抽样是否可靠，要看抽样框有没有漏掉一部分人，只在学校门口发问卷就只能覆盖走读生。常见的错误是把样本统计量直接当成总体参数，中间少了估计这一步。描述结论时用大约、估计这类词，比给出一个确定值更符合统计的语境。
Sampling sits in S1, in the additional foundation column, and its theme is that a sample informs a population while carrying limits. Random selection gives each individual the same chance of being chosen and reduces the chance that one group dominates the sample. A larger sample usually gives a steadier estimate, yet size cannot repair a biased method of selection. Judging reliability means asking whether the sampling frame leaves out part of the population, for instance a survey outside one school gate that only reaches day pupils.
比较两组分布与描述总体
比较两组数据的分布，S5 要求用统计量描述一个总体，落到题目上就是三件事：比较位置、比较离散程度、回到情境里说结论。位置用中位数或平均数，离散程度用四分位距或极差，两个指标要配对使用，不能一边说中位数一边说标准差。写比较时必须带上具体数值和单位，只写甲组更高拿不到比较分。两组中位数接近时，四分位距的差别可能才是真正的区别所在。常见的错误是只描述一组，另一组完全不提。结论最后要回到题目问的对象上，比如哪种电池更稳定，而不是停在数字层面。
Comparing two distributions under S5 comes down to three moves: compare location, compare spread, then state a conclusion in the context of the question. Location is carried by the median or the mean, spread by the interquartile range or the range, and the pair of measures should match so that a median is not compared against a standard deviation. A comparison needs actual values with units, because saying one group is higher earns nothing on its own. When the medians are close, the difference in interquartile range is often the real finding.
三份卷子里的统计题与评分口径
统计题分布在三份卷子上，每份 1 小时 30 分钟、80 分，Paper 1 不允许使用计算器，统计内容照样会出现在里面。非计算器卷的平均数题一般配好除得尽的数，遇到除不尽的情况，答案可以保留分数，或者按要求保留一位小数。AQA 8300 的评分权重在 Higher 是 AO1 四成、AO2 与 AO3 各三成，Foundation 的 AO1 占到五成，读题时先判断题目要的是套用方法还是解释推理。解释类问题只有结论往往只得一半分，结论后面要跟一句理由。常见的错误是算式写对却在单位或情境收尾上丢分。阅卷时很容易看出哪些答案把题干里的数字抄了一遍却没做运算。
Statistical questions are spread across all three papers, each lasting 1 hour 30 minutes and worth 80 marks, and Paper 1 forbids calculators without excluding the statistics content. Mean questions on the non-calculator paper tend to use numbers that divide cleanly, so a remainder can be left as a fraction or rounded to one decimal place as instructed. AQA 8300 weights the objectives at 40 per cent for AO1, 30 for AO2 and 30 for AO3 at Higher, with AO1 rising to 50 per cent at Foundation, which tells a student whether the question wants a routine method or a reasoned argument. An explanation that stops at the conclusion usually collects half the marks.
海外直订Higher Mathematics for AQA GCSE 高等数学AQA GCSE¥645.0📱 去京东购买京东精选
需要一对一辅导？Need One-on-One Tutoring?
A-Level / GCSE / IB / KS2-KS3 全科辅导，统计与数学专项可安排试听
16621398022 同微信
Visit aleveler.com
或截图后用「扫一扫 → 相册」选择本图