Chapter 6 — Statistics

What is statistics?
Statistics is a branch of mathematics that deals with collecting, organizing, presenting, analysing and interpreting numerical information. It helps turn raw numbers into meaningful summaries. In everyday life, statistics are used in weather reports, survey results, exam scores and business figures.

Types of data
Data can be classified in several ways. One important classification is qualitative (categorical) and quantitative (numerical). Quantitative data may be discrete (whole numbers, like number of students) or continuous (measurable on a scale, like height). Another distinction is between raw (unprocessed) data and grouped data, where values are arranged into class intervals.

Variables and observations
A variable is any characteristic that can take different values (for example, marks scored). An observation is one recorded value of a variable from an individual item or person. When many observations are collected, they form a data set. Understanding whether the variable is categorical or numerical, and if numerical whether it is discrete or continuous, determines which methods are appropriate to summarise it.

Scales of measurement
Quantitative data may be measured on an interval or ratio scale. Ratio scales have a meaningful zero (e.g., weight), while interval scales do not (e.g., temperature in Celsius). For many statistical measures such as mean and standard deviation, data should be on a numerical scale where arithmetic operations make sense.

Population versus sample
Data may represent an entire population (every member of a group) or a sample (a subset). In class 10 we often treat given datasets as complete lists; in advanced study, sampling methods and sampling error become important. Whether data form a population or a sample affects which formula (denominator n or n−1) is used for some measures, but the basic concepts remain the same.

Presentation and preliminary checks
Before computing numeric summaries, check data for errors, missing values and outliers. Organise data using tables or lists and note the range and any obvious clusters. Proper classification helps later calculations and avoids mistakes when setting class intervals or computing frequencies.

Summary
Understanding what data are and the types of variables available prepares students to use statistical measures such as mean, median and mode. Correct classification and careful recording are the first steps in any analysis and ensure that computed results are meaningful and appropriate for decision making.

Purpose of frequency distributions
Frequency distributions help summarise large data sets by grouping values into classes and counting how many observations fall in each class. This makes it easier to visualise and compute statistical measures. Good frequency tables reduce repetition and reveal patterns such as clustering or gaps in data that are not obvious from raw lists.

Choosing class intervals
For continuous data, choose class width (interval size) so that the number of classes is reasonable (usually 5–15). Intervals should be of equal width, mutually exclusive and exhaustive. For data with minimum value L and maximum U, class width h can be approximated by (U−L)/k where k is desired number of classes; then round h to a convenient value (like 5, 10 or 2) and adjust limits accordingly. Avoid overlapping intervals — write limits clearly to show which endpoint is included.

Constructing the frequency table
Start from a suitable lower class limit, add class widths to form successive intervals. For each observation, determine which interval it belongs to and add one to that interval's frequency. It helps to tally marks (||) while counting. Include columns for class marks (mid-points), relative frequency and cumulative frequency to assist later calculations like mean and median.

Class marks, boundaries and widths
Class mark (mid-point) = (lower limit + upper limit)/2 and is used as a representative value for each class in many calculations. When classes are stated with integers, be aware of class boundaries: for continuous data we might use 9.5–19.5 to avoid ambiguity. If classes are equal width, calculations simplify and comparison across classes is easier.

Relative and cumulative frequency
Relative frequency = frequency/total number of observations; this shows proportion in each class and can be converted to percentages. Cumulative frequency is the running total of frequencies up to and including a class; it is essential for finding medians, quartiles and percentiles and for drawing ogives. There are two cumulative frequencies commonly used: 'less than' cumulative frequency (up to an upper class boundary) and 'more than' cumulative frequency (from a lower class boundary downwards).

Practical tips to avoid errors
Always check that the sum of class frequencies equals the total number of observations. Make sure classes cover the entire data range and that no observation falls outside. If you have isolated values or small datasets, consider leaving data ungrouped. When grouping, try to keep class width consistent to ensure formulas for median and mode work correctly for grouped data.

Use in further analysis
A well-constructed frequency table is the foundation for drawing histograms, ogives, finding mean by class marks, estimating median and mode using formulas, and computing measures of dispersion. Investing time to set up classes clearly saves time and reduces mistakes later on.

What is the arithmetic mean?
The arithmetic mean (or average) of a set of numbers is the sum of the observations divided by the number of observations. It represents the central value of the data if they were spread evenly. The mean is sensitive to every value in the dataset, including extremes (outliers). Teachers and examiners often call it 'mean' or sometimes 'arithmetic mean'.

Formula for ungrouped data
For n observations x1, x2, ..., xn, mean x̄ = (x1 + x2 + ... + xn)/n. This simple formula is the starting point for many statistical calculations and is useful where all individual values are known.

Calculation steps and practical advice
1. List all observations and check for any obvious data entry errors. 2. Add all observations carefully to obtain the total sum. Use a calculator for large sums to avoid mistakes. 3. Count the number of observations n. 4. Divide the total sum by n to obtain the mean. Write results to appropriate decimal places and include units if values have units (e.g., cm, marks, rupees).

Properties of the mean
The mean is unique and lies between the minimum and maximum values of the data. It obeys linearity: if each value increases by a constant k, the mean increases by k; if each value is multiplied by a constant a, the mean multiplies by a. The mean minimises the sum of squared deviations Σ(xi − c)^2 with respect to c, which is the reason it appears in many optimisation and estimation problems.

When mean is not best
Because the mean uses all data, it can be strongly affected by outliers. For skewed distributions, the median may better represent a typical value. For categorical data (like colours, brands) mean is not defined; use mode or proportions instead.

Mean and subsequent measures
Knowing the mean is necessary to compute variance and standard deviation. In grouped data, mean is approximated using class marks. Remember that grouped mean is an estimate because individual values within classes are replaced by midpoints; narrower classes improve accuracy.

Real-life uses
Mean is used in exam average calculations, average income, mean temperature and many other practical measures. For class tests, the mean gives a quick idea of performance; but always check spread (SD) to understand variation among students.

Median — definition and method
The median is the middle value of an ordered data set. To find the median, first arrange data in ascending (or descending) order. If the number of observations n is odd, the median is the value at position (n+1)/2. If n is even, the median is the average of the values at positions n/2 and (n/2)+1. Median is less sensitive to extreme values than the mean and is useful for skewed distributions.

Detailed step-by-step for median
1. Sort the data from smallest to largest. 2. Count the observations n. 3. If n is odd, pick the (n+1)/2-th observation directly. 4. If n is even, take the average of the two middle values. 5. Report median with correct unit if any. For small datasets, doing this by inspection is straightforward; for larger datasets grouped into classes, use median class and interpolation formula.

Mode — definition and method
The mode is the value that occurs most frequently in the data set. A data set may be unimodal (one mode), bimodal (two modes), multimodal (more than two), or have no mode if all values occur once. Mode is particularly useful for categorical data where mean and median are not defined. For grouped data the modal class is the class with the highest frequency and an estimate of mode can be calculated using a formula.

Comparing mean, median and mode
Mean uses all values and is influenced by extremes, median is positional and robust against outliers, and mode indicates the most typical or frequent value. In symmetric distributions these three measures are close or equal. In right-skewed distributions mean > median > mode, and in left-skewed distributions mean < median < mode. Use this relation to comment on skewness in exam answers.

Special cases and examples
When there are repeated values, mode gives useful practical information (for example most common shoe size). If a dataset has two values tied for highest frequency, report both as modes. If grouped, use modal class and the grouped mode formula to estimate the modal value. Always state which measure best represents the data depending on context.

Use in interpretation
Median is widely used in income and salary statistics because it gives a better idea of typical earnings avoiding extreme high incomes. Mode is used in product design (common size) and quality control. When reporting central tendency include the chosen measure and explanation why it is appropriate.

Grouped data and need for class marks
When data are grouped into class intervals, individual values are not known. To compute the mean we use class marks (mid-points) as representative values of each class. Class mark xi for class with limits a and b is (a+b)/2. Multiply each class mark by its class frequency fi to get fi·xi; sum these products and divide by total frequency N to get mean. This is an estimate of the true mean of original ungrouped data.

Direct method
Mean for grouped data (direct) is given by x̄ = Σ(fi·xi) / Σfi where xi are class marks and fi are class frequencies. This requires computing all class marks and products, which can be time-consuming for many classes. The direct method is simple to understand and useful for checking results from shortcut methods.

Assumed mean method
To simplify arithmetic, choose an assumed mean A (often a central class mark). Compute deviations di = xi − A and use Σ(fi·di). Then mean x̄ = A + [Σ(fi·di)/Σfi]. This reduces calculation because deviations are smaller numbers and may cancel out, making addition easier and reducing rounding errors. Choose A near the centre to keep di values small.

Step-by-step calculation and table layout
1. Create frequency table with class limits and frequencies. 2. Compute class marks xi. 3. Choose assumed mean A if using shortcut method. 4. Compute di = xi − A and fi·di for each class. 5. Sum frequencies Σfi and Σ(fi·di). 6. Apply formula for mean. Laying out columns (class, fi, xi, di, fi·di) makes work neat and allows quick verification.

Accuracy and interpretation
Grouped mean is an approximation because it assumes all values in a class equal the class mark. The approximation improves with narrower classes. Always check that Σfi equals total observations and that computed mean lies between minimum and maximum class marks. Mention that grouped mean is suitable when only grouped data are available; otherwise raw data mean is exact.

Practical tips
When classes have equal widths and are symmetric around centre, mean often equals central class mark if distribution is symmetric. Use calculators for Σ(fi·xi) when numbers are large. When presenting answers in exams include the table and steps taken, and round final answer suitably with units if needed.

Median for grouped data
When data are grouped, median cannot be read directly but can be estimated using the median class. The median class is the class where cumulative frequency reaches or exceeds N/2 (N = total frequency). Use the formula: Median = l + [(N/2 − cf) / f] × h where l = lower limit of median class, cf = cumulative frequency before median class, f = frequency of median class, and h = class width. This linear interpolation assumes data are evenly distributed within the median class and gives an estimated median value inside that class.

Finding the median class and parameters
To apply the median formula first prepare the grouped frequency table and cumulative frequency column. Locate the smallest class whose cumulative frequency is at least N/2; that class is the median class. Take cf as cumulative frequency just before that class. Use the actual class width h (upper limit − lower limit). If classes are continuous, use exact limits; if classes are discrete use conventions given in the question.

Mode for grouped data
Mode estimation for grouped data uses the modal class — the class with highest frequency. The grouped mode is given by: Mode = l + [(fm − f1) / (2fm − f1 − f2)] × h where l = lower limit of modal class, fm = frequency of modal class, f1 = frequency of class before modal class, f2 = frequency of class after modal class, and h = class width. This formula assumes a smooth distribution and interpolates within the modal class to estimate where the peak lies.

Why interpolation matters
Both median and mode formulas use linear interpolation because grouping loses exact positions of individual observations. Interpolation assumes a uniform spread of observations within the class. While approximate, these formulas are accepted in exams and practical work when only grouped data are available.

Special situations and checks
If modal class is first or last, f1 or f2 might be zero; apply the formula carefully and state assumptions. If two classes tie for highest frequency, the dataset may be bimodal and using the modal formula will not give a unique result — report both modal classes and explain. Always check units and class widths for consistency.

Practical examples in exams
Students are often asked to compute grouped median and mode and compare them with grouped mean to comment on skewness. Show working: frequency table, cumulative frequencies, identification of median or modal class, substitution into formula and rounding appropriately.

Why measure dispersion?
Measures of central tendency identify a typical value but do not reveal how spread out data are. Dispersion measures quantify variability, which helps judge consistency, risk and reliability. For example, two classes may have the same mean marks but very different spreads; dispersion measures tell which class performances are more consistent.

Range — simplest measure
Range is the difference between the maximum and minimum values: Range = Maximum − Minimum. It is easy to compute and gives a quick sense of spread, but it is based on only two values and is sensitive to outliers. Range is most useful for initial checks and descriptive summaries rather than detailed analysis.

Mean deviation (average absolute deviation)
Mean deviation (MD) measures average distance of data values from a central value (often the mean or median). For ungrouped data around mean, MD = (Σ|xi − x̄|)/n. For grouped data, use class marks xi and MD = (Σfi·|xi − x̄|)/Σfi. MD uses absolute values so each deviation contributes positively; this gives an intuitive average distance from the centre.

Calculating MD step-by-step
1. Choose a centre c (mean or median). 2. For each observation compute absolute deviation |xi − c|. 3. Sum these absolute deviations. 4. Divide by the number of observations n (or Σfi for grouped). For grouped data, replace xi by class marks and multiply deviations by class frequencies. MD requires careful arithmetic but produces a robust measure when compared to variance in presence of moderate outliers.

Median versus mean as centre
MD around median often gives a smaller value than MD around mean for skewed distributions and is less affected by outliers. In many practical situations median is preferred centre for computing MD when data are non-symmetric.

Comparison and limitations
MD is less affected by extreme values than variance and SD because it does not square deviations. However, MD is less convenient algebraically and lacks some mathematical properties that variance has (like additivity under independence). Both MD and SD are useful; choose based on context and exam instructions.

Interpretation and use
Lower MD indicates data clustered tightly around centre; higher MD shows more spread. MD is a clear, interpretable average distance and is useful when communicating variability to non-technical audiences because it uses original data units without squaring.

Concept of variance
Variance measures the average of squared deviations of data values from their mean. Squaring gives positive values and emphasises larger deviations. For a population of n observations, population variance σ^2 = (Σ(xi − μ)^2)/n. Squared units (e.g., square marks or square cm) make variance less intuitive, which is why standard deviation is widely used instead.

Standard deviation
Standard deviation (σ or s) is the square root of variance and has the same unit as the data, making it easier to interpret. SD summarizes variability around the mean: most data in a symmetric distribution lie within a few SDs of the mean.

Formulas for ungrouped data and computational shortcut
Population variance σ^2 = Σ(xi − x̄)^2 / n and standard deviation σ = sqrt(σ^2). For computational ease, use: σ^2 = [Σxi^2 / n] − x̄^2. This avoids subtracting the mean from each value then squaring, often reducing rounding errors and saving time when using a calculator. Write down Σxi and Σxi^2 clearly to avoid mistakes.

Step-by-step calculation
1. Compute sum Σxi and mean x̄. 2. Compute Σxi^2 (sum of squares). 3. Apply computational formula to find variance. 4. Take square root to get SD. When showing work, list values of Σxi, Σxi^2, n, variance and SD, rounding as required in the question.

Interpretation and examples
A small SD means data are clustered near mean; a large SD means data are spread out. SD is affected by outliers more strongly than mean deviation because of squaring. Use SD with mean to compare datasets; if means differ greatly, use coefficient of variation for fair comparison.

Practical notes and exam tips
Use a calculator for accurate Σxi^2 values and show intermediate sums in your answer. Check that variance is non-negative and SD is the square root. For sample-based questions in higher classes, remember the (n−1) correction; for class 10 standard exercises treat data as full sets unless instructed otherwise.

Need for class marks and grouped formulas
In grouped data individual values are not known, so we use class marks xi as representative values. The variance and standard deviation are computed using these mid-points and class frequencies fi. The formulas are adapted to include fi, making the calculations similar to ungrouped cases but weighted by frequency.

Direct computation method
Grouped variance σ^2 = [Σfi·(xi − x̄)^2] / Σfi and standard deviation σ = sqrt(σ^2) where xi are class marks and x̄ is grouped mean. Calculation requires computing (xi − x̄)^2 for each class, multiplying by fi and summing. This direct method is straightforward but may involve larger numbers if class marks are big.

Assumed mean (shortcut) method
To simplify arithmetic choose an assumed mean A and compute deviations di = xi − A. Then σ^2 = [Σfi·di^2 / Σfi] − [Σfi·di / Σfi]^2. This decomposes variance into mean of squares minus square of mean of deviations and avoids handling large values directly. The assumed mean method often reduces the size of numbers and the chance of arithmetic error; choose A close to central class mark for best effect.

Steps for grouped calculations and table layout
1. Construct frequency table with class marks xi and frequencies fi. 2. Compute fi·xi and fi·xi^2 if using direct method, or fi·di and fi·di^2 for assumed mean method. 3. Sum columns (Σfi, Σfi·xi, Σfi·xi^2 or Σfi·di, Σfi·di^2). 4. Compute mean and variance using the chosen formula. 5. Take square root of variance to get SD. Keeping a neat table with labeled columns prevents mistakes and helps check results quickly.

Accuracy, interpretation and limitations
As with grouped mean, variance and SD are approximations due to representing each class by its mark. Narrower class widths improve accuracy. SD helps compare variability of two distributions of grouped data when expressed in same units. Remember that SD is sensitive to outliers and grouping choices; always interpret SD in context and alongside mean and other measures.

Common errors to avoid
Be careful with class boundaries and marks; use consistent units and double-check sums of frequencies and computed columns. Use Σfi as denominator and not Σfi−1 unless sample correction is specified. When rounding, retain sufficient decimal places in intermediate steps to avoid rounding errors in the final SD.

Combining two data sets
Sometimes we need the mean and standard deviation of a combined group formed by merging two separate groups. This might occur when combining test scores from two sections or merging survey results. Combined measures allow comparison across larger samples without recalculating from raw data if only group summaries are given. The idea is to use weighted sums that account for group sizes and internal variation.

Combined mean
If group 1 has mean x̄1 and size n1, and group 2 has mean x̄2 and size n2, the combined mean x̄ = (n1·x̄1 + n2·x̄2) / (n1 + n2). This weighted mean uses group sizes as weights, giving correct overall average. It reflects the contribution of each group according to its size.

Combined variance and standard deviation
To find combined variance σ^2 for two groups, use total sum of squares. If Σx1^2 and Σx2^2 are known, combined variance = [Σx1^2 + Σx2^2] / (n1 + n2) − x̄^2. When only group means and SDs are known, compute Σx1^2 = n1·(σ1^2 + x̄1^2) and similarly for group 2, since Σ(xi − x̄1)^2 = n1·σ1^2 and Σxi^2 = n1·(σ1^2 + x̄1^2). Add these to get combined Σxi^2, then proceed to find overall variance and SD = sqrt(variance).

Derivation and careful steps
1. Compute combined mean x̄ by weighted average. 2. Compute Σxi^2 for each group using n(σ^2 + mean^2). 3. Sum Σxi^2 values for all groups to get total sum of squares. 4. Compute mean of squares = [Σxi^2_total] / (n1 + n2). 5. Subtract x̄^2 to obtain combined variance. 6. Take square root to find combined SD. This process uses algebraic identities and ensures both within-group and between-group variability are included.

Practical considerations and interpretation
Track units carefully and use consistent definitions (population formulas with denominator n are common in class 10 unless stated otherwise). This method avoids needing raw lists of all observations and is efficient when groups are large. Note that combining groups can change SD in non-intuitive ways: if group means differ, combined variance increases because of between-group differences even if individual group variances are small.

Common exam tasks
Problems often give n, mean and SD for two groups and ask for combined mean and SD. Show working steps: compute sums n·mean and n·(σ^2 + mean^2), sum these across groups, find combined mean, compute variance and SD, and present numerical answers with appropriate rounding and units.

Linear transformation of data
A linear transformation changes each data value xi to yi = a·xi + b where a and b are constants. Such transformations model unit changes (for example, converting Celsius to Fahrenheit) and shifting of data. It is important to know how central measures and dispersion change under these transformations so we can quickly adjust summaries without recomputing from raw data.

Effect on mean
The mean of transformed data ȳ = a·x̄ + b. This is because applying the transformation to every xi and averaging simply scales and shifts the mean. The shift b adds uniformly to each observation and therefore to the mean, while multiplication by a scales the mean by a. This property holds for both grouped and ungrouped data as long as the transformation is applied consistently.

Effect on variance and standard deviation
Variance transforms as σy^2 = a^2·σx^2 because squaring removes sign of a and b has no effect on spread — adding a constant shifts all values equally and does not change variability. Standard deviation transforms as σy = |a|·σx. Note the absolute value: if a is negative, the direction of scale reverses but spread remains scaled by |a|. The constant b disappears from variance and SD formulas because it does not change distances between observations.

Special cases and applications
If a = 1 and b ≠ 0, mean shifts by b but variance and SD remain unchanged. When converting units, apply correct a and b; for example, Celsius to Fahrenheit yi = (9/5)xi + 32 so a = 9/5 and b = 32. If mean in °C is 25 with SD 3, mean in °F = (9/5)·25 + 32 = 77 and SD in °F = (9/5)·3 = 5.4. This shows how both mean and SD change predictably under unit conversion.

Use in standardisation (z-scores)
Linear transformations are useful for standardising data. A common transformation is zi = (xi − x̄)/σ which sets mean of z to 0 and SD of z to 1. This is equivalent to a = 1/σ and b = −x̄/σ and helps compare different datasets on the same scale or apply probability tables in advanced study.

Exam tips and checks
Apply formulas directly to avoid recalculating from raw data; always note sign of a when reporting SD and include absolute value if needed. State clearly which measure (mean or SD) is transformed and show substitution of a and b. These rules are often tested in short numerical problems and theory questions asking for explanation.

Percentiles and quartiles — definitions
Percentiles divide ordered data into 100 equal parts; the p-th percentile is a value below which p% of observations lie. Quartiles are special percentiles: Q1 is the 25th percentile, Q2 is the 50th percentile (median) and Q3 is the 75th percentile. Quartiles summarise the spread and are robust to outliers, making them useful for describing skewed distributions.

Finding quartiles for ungrouped data
Arrange data in order. For N observations, position of Q1 can be found by (N+1)/4, Q2 by (N+1)/2 and Q3 by 3(N+1)/4. If the position is not an integer, interpolate between neighbouring values: for example if the position is 2.25, Q1 = value at position 2 + 0.25×(value at position 3 − value at position 2). This linear interpolation gives a precise value and is commonly used in exams for small datasets.

Interquartile range (IQR)
IQR = Q3 − Q1 and measures the spread of the middle 50% of data. It is less sensitive to extreme values than range or standard deviation and often used to detect outliers. Values lying outside Q1 − 1.5·IQR and Q3 + 1.5·IQR are often flagged as potential outliers.

Quartiles for grouped data
For grouped data use cumulative frequency to find classes corresponding to N/4 and 3N/4. Then use linear interpolation within those classes: Q1 = l + [(N/4 − cf)/f]·h and Q3 = l + [(3N/4 − cf)/f]·h, where l is lower limit of respective quartile class, cf is cumulative frequency before that class, f is its frequency, and h is class width. This treats frequency as uniformly distributed within the class to estimate quartile values.

Using IQR for comparison and outlier detection
IQR is useful to compare spreads across samples because it ignores extreme tails. In boxplots, the box spans Q1 to Q3 with median marked inside; whiskers extend to non-outlier minima and maxima. When comparing two datasets, the one with larger IQR has greater mid-range spread even if standard deviations differ.

Practical tips
When computing quartiles, be consistent with the method (the (N+1)/4 rule is standard for many exam problems). Always list ordered data, compute positions, and show interpolation steps clearly. Use IQR along with median and boxplot to explain shape and presence of outliers in your answer.

What is an ogive?
An ogive, or cumulative frequency curve, is a graph of cumulative frequency against class boundaries. It provides a visual method to estimate medians, quartiles and percentiles for grouped data. There are two types: less-than ogive (cumulative frequency up to an upper class boundary) and more-than ogive (cumulative frequency from lower boundaries). Usually the less-than ogive is used to find medians and percentiles.

Constructing an ogive step by step
1. Prepare a grouped frequency table and compute cumulative frequencies up to each upper class boundary. 2. Use upper class boundaries as x-coordinates and cumulative frequencies as y-coordinates. 3. Plot points at each upper boundary with corresponding cumulative frequency. 4. Join the points with a smooth curve or a polygonal line. For the less-than ogive start at the lower boundary with cumulative frequency 0, and for the more-than ogive start at the upper limit with frequency 0, depending on convention.

Reading median and percentiles from ogive
To find median using ogive: on the y-axis mark N/2 and draw a horizontal line to meet the ogive; from the point of intersection draw a vertical down to x-axis to read the median value. For p-th percentile, mark p% of N on y-axis (i.e. p·N/100) and follow the same steps. This graphical method uses interpolation within classes and matches grouped median formula when drawn accurately.

Advantages and limitations
Ogives give clear visual representation of distribution and cumulative behaviour; they are particularly helpful to estimate medians and quartiles when exact calculation is cumbersome. However, precision depends on correct plotting and smooth interpolation; exact values are better obtained using formulas when possible. Ogives also display how rapidly cumulative counts increase, showing concentration of data in certain ranges.

Practical tips for drawing
Label axes clearly and mark class boundaries precisely; use uniform scales on axes to avoid distortion. Ensure cumulative frequencies increase monotonically to form a rising curve. When estimating values read horizontally and vertically with a ruler to reduce parallax error. Compare ogive results with histogram or frequency polygon to interpret the overall shape and spread.

Exam use
Questions often ask to draw an ogive from a grouped table and read the median or a percentile. Show cumulative frequency table, plot points at class boundaries, join smoothly and clearly mark how you read the required value from the graph.

Histograms
A histogram is a bar graph representing frequency distribution of continuous data. Classes are on the x-axis and frequencies on the y-axis. Bars are drawn adjacent to each other with heights equal to class frequencies; their widths represent class widths. Histograms visually show the shape of distribution — whether symmetric, skewed, or uniform — and indicate modes and concentration of data.

Constructing histograms correctly
1. Use class intervals on x-axis with equal widths if possible and mark class boundaries clearly. 2. For each class draw a rectangle with base equal to class width and height equal to frequency density if class widths vary, or simply frequency if widths are equal. 3. Ensure bars touch each other, indicating continuity of the variable. Label axes with units and include a scale on the y-axis that shows frequencies or frequency density clearly.

Frequency polygons
A frequency polygon is formed by joining mid-points of the tops of histogram bars with straight lines. To close the polygon, add a point at each end with zero frequency at a boundary before the first class and after the last class. Frequency polygons are useful for comparing two distributions on the same axes and for smoothing visual appearance of the data shape.

Unequal class widths and frequency density
If class widths are unequal, use frequency density (frequency divided by class width) as the height of a histogram bar so that the area of each bar is proportional to the frequency. This preserves the correct visual proportion between classes. Students must calculate density = frequency / width and plot bars with widths equal to class widths and heights equal to densities; area = width × density = frequency.

Interpreting histograms and polygons
Histograms reveal modes (peaks), symmetry, and skewness. A bell-shaped histogram suggests normal-like distribution; a long tail to the right indicates right-skewness and left tail indicates left-skewness. Multiple peaks suggest bimodal or multimodal distributions. Frequency polygons help compare multiple datasets by overlaying them on the same axes.

Practical exam advice
Draw neat axes, use equal scales and clearly mark class boundaries. When asked to compare two histograms or polygons, comment on central tendency, spread and modality. When classes are unequal, always use frequency density to draw correct histograms and mention density calculations in your workings.

Need for relative measure of dispersion
Standard deviation measures spread but depends on the unit and scale of data. To compare variability between two datasets with different units or widely different means, use a relative measure: coefficient of variation (CV). CV expresses spread as a percentage of the mean so it is unitless and directly comparable across datasets.

Definition and interpretation
Coefficient of variation CV = (Standard deviation / Mean) × 100%. It is a dimensionless percentage that expresses SD relative to mean. A higher CV indicates greater relative variability. CV is useful in finance (comparing investment risks), biology (variability among species measurements), and exam results for comparing consistency across classes. It helps answer questions such as which dataset is more consistent relative to its average.

Steps to compare distributions using CV
1. Compute mean and SD for both datasets using the appropriate formulas. 2. Calculate CV for each dataset. 3. Compare CVs: the dataset with larger CV is relatively more variable. When CVs are very close, consider other measures like IQR or visual plots to decide. Also, check if means are large enough to make CV meaningful.

Limitations and cautions
CV is meaningful only for data measured on ratio scales where zero has a real meaning and means are positive. When mean is zero or near zero, CV becomes unstable or meaningless. CV also does not indicate direction of skewness or shape of distribution; use CV along with other measures (skewness, boxplots) for fuller comparison. Be cautious when comparing CVs across datasets with different data quality or measurement error.

Practical application examples
Compare two machines producing screws: if machine A has mean length 10 mm and SD 0.2 mm (CV=2%), machine B has mean 20 mm and SD 0.4 mm (CV=2%), both have equal relative variability despite different absolute spreads. If CV differs, choose the machine with lower CV for consistent production.

Exam tips
Round CV to required decimal places and state which dataset is more variable. Mention if CV is not appropriate due to near-zero means. When used in a solution, show calculation of mean and SD clearly and then compute CV as a final comparative step.

What are outliers?
Outliers are observations that lie far away from the rest of the data. They may result from measurement error, data entry mistakes, or they may be genuine extreme values. Detecting outliers is important because they can distort numerical summaries like mean and standard deviation and may affect conclusions drawn from data.

Detecting outliers using IQR
A common method is to use quartiles and IQR. Compute Q1 and Q3 and IQR = Q3 − Q1. Observations below Q1 − 1.5·IQR or above Q3 + 1.5·IQR are considered potential outliers. This rule is simple and robust because it relies on quartiles which are not much affected by extremes and provides a standard procedure for identification.

Effect on mean and SD
Outliers can significantly change the mean and increase standard deviation due to squaring of large deviations when computing variance. The mean moves toward the outlier, while median remains much less changed because it depends on position. Standard deviation increases as outliers inflate average squared deviations, making datasets appear more variable than most of their observations suggest.

Handling outliers
Investigate outliers before deciding how to handle them: check for entry errors or instrument faults. If an outlier is due to mistake, correct or remove it. If the outlier is genuine, consider reporting results both with and without the outlier, or use robust measures like median and IQR that are less influenced by extreme values. Transformations (e.g., logarithm) can sometimes reduce the effect of outliers by compressing large values.

Exam-style tasks and interpretation
In exam questions students may be asked to compute mean with and without a given outlier and comment on the effect, or to identify outliers using IQR. Clearly state any assumption made when removing outliers and explain how conclusions change. Show calculations for both scenarios to support your analysis.

Visualization and communication
Use boxplots to show outliers graphically; points beyond whiskers are typically flagged. When reporting results, mention potential outliers and their influence so readers understand limitations of the summary statistics and can interpret conclusions appropriately.

Understanding the question
Begin by reading the question carefully to know what type of data is given (grouped or ungrouped), what is asked (mean, median, mode, SD, etc.), and whether any special instructions (assumed mean, combine groups) are provided. Identify totals and check units. Underline numbers and note whether the data represent a full population or a sample if the question specifies.

Choosing methods
For ungrouped data calculate mean directly and use computational formula for SD where convenient. For grouped data prefer assumed mean method for mean and variance to reduce arithmetic. For median and quartiles in grouped data, use cumulative frequency and interpolation formulas. When asked to draw graphs, label axes and use correct class boundaries. Decide early whether to use direct or shortcut formulas to save time.

Use of calculators and presentation
Use a calculator for sums and squares to reduce mistakes. Show essential steps in exams: frequency table, class marks, Σfi·xi and Σfi·xi^2 or Σfi·di and Σfi·di^2, then apply formulas. Keep working neat in columns so examiners can follow. Write final answers rounded as instructed and include units where relevant.

Checking answers
Verify sums of frequencies equal given totals, check that computed mean lies within data range, and confirm that SD is non-negative and plausible. If combining groups, cross-check combined mean formula with weighted sums. When possible do a quick reasonableness check: e.g., if most data are between 40 and 60, an SD of 100 is unlikely.

Time management and shortcuts
Practice faster computation methods: assumed mean, computational variance formula, and neat tabular layout to save time. For multi-part problems, compute shared quantities like Σfi·xi once and reuse where needed. For graphical parts, draw axes first and mark scales to avoid re-drawing. Allocate time so you can check arithmetic at the end.

Interpreting results
Beyond calculations, be ready to interpret: comment on skewness using mean and median, compare variability using SD or CV, and discuss whether outliers affect results. In longer answers include a short sentence explaining what the computed measures suggest about the data (e.g., 'data are tightly clustered', 'distribution is right-skewed').