Transformations of data sets Revision Notes

Data transformations involve changing every value in a data set by applying the same operation, such as adding or multiplying by a constant. These transformations are common in statistics to adjust data for comparison or interpretation. It is important to understand how these changes affect key summary statistics like the mean and standard deviation, which describe the central tendency and spread of the data, respectively.

• Transformations apply the same operation to every data point.
• Mean and standard deviation summarize data location and spread.
• Understanding effects on these measures aids data interpretation.

When a constant value is added to every data point, the entire data set shifts by that amount. This shift increases the mean by the same constant because the mean is the average of all data points. However, the standard deviation, which measures how spread out the data are, does not change because the relative distances between data points remain the same.

For example, consider the data set: 4, 6, 8. The mean is (4 + 6 + 8) / 3 = 6. If we add 3 to each data point, the new data set is 7, 9, 11. The new mean is (7 + 9 + 11) / 3 = 9, which is 3 more than the original mean. The spread of the data points remains unchanged, so the standard deviation is the same.

• Mean increases by the added constant.
• Standard deviation remains unchanged.
• Data spread is unaffected by uniform shifts.

Multiplying every data point by a constant scales the data set. This operation affects both the mean and the standard deviation. The mean is multiplied by the constant because each data point is scaled. The standard deviation is also multiplied by the absolute value of the constant because the spread between data points is stretched or compressed accordingly.

For example, take the data set: 2, 5, 7 with mean (2 + 5 + 7) / 3 = 4.67. Multiplying each data point by 4 gives 8, 20, 28. The new mean is (8 + 20 + 28) / 3 = 18.67, which is 4 times the original mean. The standard deviation also becomes 4 times larger, reflecting the increased spread.

• Mean is multiplied by the constant.
• Standard deviation is multiplied by the absolute value of the constant.
• Data spread scales proportionally with multiplication.

To summarise, adding a constant shifts the data set without changing its spread, so only the mean changes. Multiplying by a constant scales the data set, affecting both the mean and the spread. These properties allow statisticians to predict how summary statistics change under common transformations, which is useful for data standardisation and comparison.

It is important to note that the standard deviation is always multiplied by the absolute value of the constant because spread cannot be negative.

• Adding constant: mean changes, standard deviation unchanged.
• Multiplying constant: mean and standard deviation both scale.
• Standard deviation scales by absolute value of multiplier.

Suppose a data set has mean 10 and standard deviation 3. If each data point is increased by 5, the new mean is 10 + 5 = 15, and the standard deviation remains 3. If instead each data point is multiplied by -2, the new mean is 10 × (-2) = -20, and the new standard deviation is 3 × 2 = 6 (note the absolute value).

• Adding 5 shifts mean to 15; standard deviation stays 3.
• Multiplying by -2 changes mean to -20; standard deviation to 6.
• Absolute value used for standard deviation scaling.

Data transformations involve changing every value in a data set by applying the same operation, such as adding or multiplying by a constant. These transformations affect measures of central tendency and spread in predictable ways. Understanding these effects helps interpret transformed data correctly and compare data sets after changes.

When a constant is added to each data point, the entire data set shifts by that constant. The median, which is the middle value when data are ordered, also increases by the same constant because the position of the middle value moves up by that amount.

However, the interquartile range (IQR), which measures the spread between the first quartile (Q1) and third quartile (Q3), remains unchanged. This is because adding a constant shifts all quartiles equally, so the difference Q3 - Q1 does not change.

For example, consider the data set: 3, 5, 7, 9, 11. The median is 7, Q1 is 5, and Q3 is 9, so IQR = 9 - 5 = 4. If we add 2 to each data point, the new data set is 5, 7, 9, 11, 13. The median is now 9 (7 + 2), Q1 is 7 (5 + 2), and Q3 is 11 (9 + 2), so IQR = 11 - 7 = 4, unchanged.

• Median increases by the added constant.
• IQR remains the same because spread does not change.

Multiplying each data point by a constant scales the data set. Both the median and the IQR are multiplied by the absolute value of that constant because the positions of the quartiles and the median scale accordingly.

For example, using the same data set: 3, 5, 7, 9, 11, the median is 7 and IQR is 4. If we multiply each data point by 3, the new data set is 9, 15, 21, 27, 33. The median becomes 21 (7 × 3), and the IQR becomes 12 (4 × 3).

If the constant is negative, the data order reverses, but the absolute values of median and IQR scale by the constant's magnitude. The median's position changes accordingly, but the IQR remains positive as a measure of spread.

• Median and IQR are multiplied by the absolute value of the constant.
• Negative constants reverse data order but scale spread by magnitude.

Adding a constant shifts the data set, increasing the median by that constant but leaving the IQR unchanged.

Multiplying by a constant scales the data set, multiplying both the median and the IQR by the absolute value of that constant.

These effects allow quick calculation of median and IQR after transformations without recalculating from raw data.

Understanding these effects is useful when data are transformed for analysis, such as converting units or adjusting scores.

For example, if test scores are increased by 5 points for all students, the median score increases by 5, but the spread of scores (IQR) remains the same, indicating no change in variability.

If scores are doubled, both the median and IQR double, indicating that the spread of scores has increased proportionally.

• Use transformations to interpret changes in data summaries.
• Recognise when spread changes or remains constant after transformations.

The median is a measure of central tendency that identifies the middle value in a data set when the values are arranged in ascending order. If the number of data points is odd, the median is the middle value. If the number of data points is even, the median is the average of the two middle values.

For example, consider the data set: 3, 5, 7, 9, 11. There are 5 values (odd number), so the median is the 3rd value, which is 7.

If the data set is 3, 5, 7, 9, 11, 13 (6 values, even number), the median is the average of the 3rd and 4th values: (7 + 9) / 2 = 8.

• Median is the middle value in an ordered data set.
• For odd number of values, median is the middle value.
• For even number of values, median is the average of the two middle values.

When new values are added to a data set, the median may increase, decrease, or remain unchanged depending on the values added and their position relative to the current median.

If the added values are all greater than the current median, the median tends to increase or remain the same, because the middle position shifts towards higher values.

If the added values are all less than the current median, the median tends to decrease or remain the same, as the middle position shifts towards lower values.

• Adding values greater than the median can increase the median.
• Adding values less than the median can decrease the median.
• Adding values equal to the median may leave the median unchanged.

Removing values from a data set also affects the median depending on which values are removed.

If values above the median are removed, the median may decrease or remain the same because the data set loses higher values, shifting the middle towards lower values.

If values below the median are removed, the median may increase or remain the same because the data set loses lower values, shifting the middle towards higher values.

• Removing values above the median can decrease the median.
• Removing values below the median can increase the median.
• Removing values equal to the median can change the median depending on the data set size and distribution.

Consider the data set: 4, 6, 8, 10, 12. The median is 8 (middle value of 5 values).

Add two values: 14 and 16 (both greater than 8). The new data set is 4, 6, 8, 10, 12, 14, 16 (7 values). The median is now the 4th value: 10.

The median increased from 8 to 10 because the added values were greater than the original median.

• Original median: 8
• Added values: 14, 16 (both > 8)
• New median: 10 (increased)

Consider the data set: 3, 5, 7, 9, 11, 13. The median is the average of 7 and 9, which is 8.

Remove the values 11 and 13 (both above the median). The new data set is 3, 5, 7, 9 (4 values). The median is the average of 5 and 7, which is 6.

The median decreased from 8 to 6 because the removed values were above the original median.

• Original median: 8
• Removed values: 11, 13 (both > 8)
• New median: 6 (decreased)

The median depends on the position of values in an ordered data set. Adding or removing values changes the size and composition of the data set, which can shift the middle position.

To predict how the median changes, consider whether the added or removed values are above, below, or equal to the current median.

This reasoning helps in understanding data transformations and their impact on statistical measures, which is essential for data analysis and interpretation.

• Median is sensitive to the position of added or removed values.
• Changes depend on whether values are above or below the median.
• Understanding these effects aids in interpreting data set modifications.

The mean is a measure of central tendency that represents the average value of a data set. To calculate the mean, add all the data values together and then divide by the total number of values.

The formula for the mean \( \bar{x} \) is: \[ \bar{x} = \frac{\sum x}{n} \] where \( \sum x \) is the sum of all data points and \( n \) is the number of data points.

For example, consider the data set: 4, 7, 9, 10, 5. The sum is \(4 + 7 + 9 + 10 + 5 = 35\), and there are 5 values. Thus, the mean is \(35 \div 5 = 7\).

• Mean is the sum of all data values divided by the number of values.
• It provides a central value representing the data set.
• Always include units if data values have units (e.g., cm, kg).

The standard deviation measures how spread out the data values are around the mean. A small standard deviation means the data points are close to the mean, while a large standard deviation indicates more spread.

For a sample data set, the standard deviation \( s \) is calculated using: \[ s = \sqrt{\frac{\sum (x_i - \bar{x})^2}{n - 1}} \] where \( x_i \) are the individual data points, \( \bar{x} \) is the mean, and \( n \) is the number of data points.

Example: Using the previous data set (4, 7, 9, 10, 5) with mean 7, calculate each squared deviation: \((4-7)^2=9\), \((7-7)^2=0\), \((9-7)^2=4\), \((10-7)^2=9\), \((5-7)^2=4\). Sum these: \(9+0+4+9+4=26\). Divide by \(n-1=4\): \(26/4=6.5\). Finally, take the square root: \(\sqrt{6.5} \approx 2.55\). The standard deviation is approximately 2.55 units.

• Standard deviation quantifies data spread around the mean.
• Use \( n-1 \) in the denominator for sample standard deviation.
• Calculate squared deviations, sum, divide by \( n-1 \), then square root.

When two data sets are combined, each with known means and sizes, the mean of the combined data set is the weighted average of the two means.

If data set A has mean \( \bar{x}_A \) and size \( n_A \), and data set B has mean \( \bar{x}_B \) and size \( n_B \), then the combined mean \( \bar{x}_C \) is: \[ \bar{x}_C = \frac{n_A \bar{x}_A + n_B \bar{x}_B}{n_A + n_B} \]

Example: Data set A has 8 values with mean 12, data set B has 12 values with mean 15. The combined mean is \( \frac{8 \times 12 + 12 \times 15}{8 + 12} = \frac{96 + 180}{20} = \frac{276}{20} = 13.8 \).

• Combined mean is a weighted average based on sizes of data sets.
• Multiply each mean by its data set size, sum, then divide by total size.
• This method assumes data sets are combined without overlap or omission.

Calculating the exact standard deviation of a combined data set requires more information than just the means and standard deviations of the individual sets, such as the raw data or sums of squares.

However, understanding the combined mean is essential before attempting standard deviation calculations.

In some cases, if sums of squares or variances and means are known, the combined variance can be calculated using the formula: \[ s_C^2 = \frac{(n_A - 1)s_A^2 + (n_B - 1)s_B^2 + n_A(\bar{x}_A - \bar{x}_C)^2 + n_B(\bar{x}_B - \bar{x}_C)^2}{n_A + n_B - 1} \] where \( s_A \) and \( s_B \) are the standard deviations of the two sets, and \( \bar{x}_C \) is the combined mean. The combined standard deviation \( s_C \) is the square root of \( s_C^2 \).

• Exact combined standard deviation calculation requires means, standard deviations, sizes, and differences of means.
• Formula accounts for within-group variation and between-group variation.
• This is more advanced and often beyond basic syllabus requirements but useful for deeper understanding.

Always check the number of data points when calculating means and standard deviations.

When combining data sets, do not simply average the means without weighting by the number of data points.

Remember to use \( n-1 \) in the denominator for sample standard deviation, not \( n \). This is important for unbiased estimation of population standard deviation from a sample data set.

• Use weighted averages for combined means.
• Do not confuse population and sample standard deviation formulas.
• Double-check calculations of sums and squared deviations to avoid errors.

When working with data from different sources or measured in different units, direct comparison can be misleading. For example, comparing test scores from two different subjects where one is out of 50 and the other out of 100 requires adjustment to make a fair comparison.

Standardizing data transforms values so that they share a common scale, specifically a mean of 0 and a standard deviation of 1. This process removes the original units and scale differences, allowing us to compare data points from different distributions directly.

• Standardization enables comparison across datasets with different units or scales.
• It helps identify how unusual or typical a data point is within its distribution.
• It is essential in many statistical analyses, such as hypothesis testing and regression.

The key to standardizing data is calculating the z-score for each data point. The z-score measures how many standard deviations a value is from the mean of its dataset.

The formula for the z-score is: \( z = \frac{x - \mu}{\sigma} \), where:

\( x \) is the original data value, \( \mu \) is the mean of the dataset, \( \sigma \) is the standard deviation of the dataset.

• Subtract the mean from the data value to find the deviation from the mean.
• Divide this deviation by the standard deviation to scale it.
• The result, z, is dimensionless and indicates relative position.

A z-score tells us how far and in what direction a data point lies from the mean, measured in standard deviations.

A z-score of 0 means the data point equals the mean.

A positive z-score indicates the data point is above the mean; for example, z = 2 means 2 standard deviations above the mean, which is relatively high compared to the dataset's average value. A negative z-score indicates the data point is below the mean; for example, z = -1.5 means 1.5 standard deviations below the mean, which is relatively low.

• Z-scores allow identification of outliers or unusual values.
• They provide a standard scale to compare data points from different datasets.

Suppose a student scored 78 marks in a maths exam where the mean score was 70 and the standard deviation was 8. To find the student's z-score:

\[ z = \frac{78 - 70}{8} = \frac{8}{8} = 1 \] This means the student's score is 1 standard deviation above the mean.

If another student scored 65 in a science exam with a mean of 60 and a standard deviation of 5, their z-score is: \[ z = \frac{65 - 60}{5} = \frac{5}{5} = 1 \] Both students have the same z-score, so relative to their respective exams, they performed equally well, despite different raw scores and exam scales.

• Calculate mean and standard deviation first.
• Apply the formula carefully with correct substitution.
• Interpret the z-score in context.

Standardization simplifies comparison and interpretation across datasets with different scales or units. It is widely used in statistics, including in normal distribution analysis and z-tests.

However, standardization assumes the mean and standard deviation are meaningful summaries of the data, which may not hold for highly skewed or non-numeric data.

Also, standardizing does not change the shape of the distribution; it only rescales it. Outliers can still affect the mean and standard deviation, thus influencing z-scores.

• Useful for comparing and interpreting data across different scales.
• Depends on meaningful mean and standard deviation values.
• Does not eliminate outliers or skewness.
• Always interpret z-scores in the context of the original data.

When analysing data, it is often necessary to compare two or more distributions to understand their characteristics. Key aspects to consider are the typical values around which data cluster and how spread out the data are. Measures of central tendency and dispersion provide numerical summaries that help in this comparison.

Central tendency measures, such as the mean, median, and mode, describe the centre or typical value of a data set. Dispersion measures, including range, interquartile range (IQR), and standard deviation, describe the variability or spread of the data.

By examining these measures together, we can determine which distribution tends to have higher values on average and which is more consistent or variable. This is essential in fields like quality control, education, and research where understanding differences between groups or conditions is important.

• Central tendency summarises the centre of data.
• Dispersion summarises the spread or variability.
• Comparing both helps understand differences between distributions.

The mean is the arithmetic average of all data points and is calculated by summing all values and dividing by the number of values.

Formula: \( \bar{x} = \frac{\sum x_i}{n} \), where \(x_i\) are data values and \(n\) is the number of values.

The mean is sensitive to extreme values (outliers) but provides a useful overall average for comparison purposes.

• Calculate mean by adding all values and dividing by count.
• Higher mean indicates higher average values in the distribution.
• Example: Distribution A has values 4, 5, 6, 7, 8; mean = (4+5+6+7+8)/5 = 6.
• Distribution B has values 3, 4, 5, 6, 7; mean = (3+4+5+6+7)/5 = 5.
• Distribution A has a higher mean, so its values tend to be higher on average.

Standard deviation measures how spread out the data values are around the mean. A low standard deviation means data points are close to the mean, indicating consistency. A high standard deviation means data are more spread out, indicating variability.

Formula for standard deviation (population): \[ \sigma = \sqrt{\frac{\sum (x_i - \mu)^2}{N}} \]

For a sample: \[ s = \sqrt{\frac{\sum (x_i - \bar{x})^2}{n-1}} \] where \( \bar{x} \) is the sample mean and \( n \) is the sample size.

• Calculate deviations of each value from the mean.
• Square these deviations and find their average (adjusted for sample or population).
• Take the square root to get standard deviation.
• Lower standard deviation means more consistent data.
• Example: Distribution A values: 4, 5, 6, 7, 8; mean = 6.
• Calculate squared deviations: (4-6)²=4, (5-6)²=1, (6-6)²=0, (7-6)²=1, (8-6)²=4.
• Sum = 10; variance = 10/5 = 2; standard deviation = √2 ≈ 1.41.
• Distribution B values: 3, 4, 5, 6, 7; mean = 5; squared deviations sum = 10; standard deviation also ≈ 1.41, so both have similar spread.

To compare two distributions, first look at their means to see which has higher average values. Then examine their standard deviations to assess consistency.

A distribution with a higher mean but similar or lower standard deviation indicates generally higher and more consistent values.

If one distribution has a higher mean but also a much higher standard deviation, it means values are higher on average but more spread out, so less consistent.

• Compare means to identify which distribution has higher values.
• Compare standard deviations to identify which distribution is more consistent.
• Example: Distribution X: mean = 50, standard deviation = 5.
• Distribution Y: mean = 45, standard deviation = 3.
• Distribution X has higher average values but is less consistent than Y.
• Distribution Y has lower average values but is more consistent.

Suppose two classes take the same test. Class A has a mean score of 75 with a standard deviation of 10, while Class B has a mean score of 70 with a standard deviation of 5.

Class A’s higher mean suggests better average performance, but the higher standard deviation indicates more variability in scores.

Class B’s lower mean suggests lower average performance, but the lower standard deviation shows more consistent scores among students.

• Higher mean = better average performance.
• Lower standard deviation = more consistent results.
• Interpret both measures together for a full comparison.