π Lecture 6 β Stem-and-Leaf Display and Mode
π Overview: This lecture introduces the stem-and-leaf display, a technique invented by John Tukey (1977) that preserves individual data values while sorting and displaying data. It then transitions to measures of central tendency, focusing on the mode β the most frequently occurring value β and explains how to compute it for raw data, discrete frequency distributions, and grouped (continuous) data.
ποΈ Topics Covered
The lecture covers the construction and interpretation of stem-and-leaf displays, converting them into frequency distributions and histograms, and noting their shape similarity. It then introduces the concept of central tendency as a method for describing variable data, distinguishes between measures of central tendency and dispersion, and provides a detailed explanation of the mode, including its calculation for raw data (using a dot plot), discrete frequency distributions, and continuous grouped data using a specific formula.
π Lecture Summary
Stem-and-Leaf Display
A stem-and-leaf display is a technique for simultaneously sorting and displaying data, overcoming the disadvantage of a frequency table where the identity of individual observations is lost in the grouping process. Each number in the data set is divided into two parts: a stem, which is the leading digit(s) used for sorting, and a leaf, which is the trailing digit(s) shown in the display. A vertical line separates the leaf from the stem. For example, the number 243 can be split as "24 | 3" or "2 | 43". It is common practice to arrange the trailing digits in each row from smallest to highest, which also produces an ordered array of the data.
π Definition β Stem: The leading digit(s) of each number in a data set, used for sorting in a stem-and-leaf display. π Definition β Leaf: The trailing digit(s) of each number in a data set, shown in the display of a stem-and-leaf display.
π Example (Ages): The ages of 30 patients are: 48, 31, 54, 37, 18, 64, 61, 43, 40, 71, 51, 12, 52, 65, 53, 42, 39, 62, 74, 48, 29, 67, 30, 49, 68, 35, 57, 26, 27, 58. Using the first digit as the stem and the second as the leaf, and ordering the leaves, we get the display:
Stem (Leading Digit) | Leaf (Trailing Digit)
1 | 2 8
2 | 6 7 9
3 | 0 1 5 7 9
4 | 0 2 3 8 8 9
5 | 1 2 3 4 7 8
6 | 1 2 4 5 7 8
7 | 1 4
This display can be easily converted into a frequency distribution (e.g., class 10-19 has frequency 2, class 20-29 has frequency 3, etc.). If this frequency distribution is converted into a histogram and rotated 90 degrees, its shape is exactly like the shape of the stem-and-leaf display.
π Example (Death Rates): Construct a stem-and-leaf display for mean annual death rates per thousand: 7.5, 8.2, 7.2, 8.9, etc. Using the decimal part as the leaf and the rest of the digits as the stem, we get an ordered display:
Stem | Leaf
3 | 9
4 | 6 6
5 | 0 4 5
6 | 0 0 2 2 5 5 6 6 6 8 9
7 | 1 3 3 3 4 4 5 6 7 7 8 8 8 9
8 | 1 1 2 4 6 6 7 7 8 8 8 9 9
9 | 0 1 2 3 3 3 3 4 4 4 6 7 7 9 9
10 | 0 1 1 2 3 3 4 4 6 6 7 8 8 9 9
11 | 1 4 4 6 6 9
12 | 0 1 4 5 6 8 8
14 | 0
π‘ Why this matters: The stem-and-leaf display provides a quick visual summary of a dataset's distribution while retaining the original data values, which is a key advantage over histograms.
Description of Variable Data and Measures of Central Tendency
In any statistical enquiry, a concise numerical description of variable data is needed. Measures of central tendency (averages) enable us to measure the central tendency of variable data, indicating the location or general position of the distribution on the X-axis. Measures of dispersion enable us to measure its variability. An average is a single value representing a set of data, more or less central to the values which cluster around it. The most common types are the arithmetic mean, geometric mean, harmonic mean, median, and mode. The mathematical averages (arithmetic, geometric, harmonic) indicate the magnitude of values; the median indicates the middle position; the mode indicates the most frequent value.
The Mode
The mode is defined as that value which occurs most frequently in a set of data, indicating the most common result.
π Example (Raw Data): Marks of eight students: 2, 7, 9, 5, 8, 9, 10, 9. The most common mark is 9. Hence, Mode = 9.
Mode in Case of Raw Data (Continuous Variable)
For raw data of a continuous variable not grouped into a frequency distribution, the mode is obtained by counting the number of times each value occurs. A dot plot can be used to visualize this, where dots are placed above one another for repeating values, forming piles.
π Example (R&D Percentages): Data on percentages of revenues spent on R&D by 49 companies. The dot plot shows value 6.9 occurs 3 times, while all others occur once or twice. Hence, the modal value is 6.9. The dot plot also shows almost all percentages are between 6% and 12%, with most between 7% and 9%.
Mode in Case of a Discrete Frequency Distribution
In a discrete frequency distribution, the mode is the value with the highest frequency.
π Example (Airline Passengers): Number of passengers on 50 flights of a 40-seater plane:
| No. of Passengers (X) | No. of Flights (f) |
|---|---|
| 28 | 1 |
| 33 | 1 |
| 34 | 2 |
| 35 | 3 |
| 36 | 5 |
| 37 | 7 |
| 38 | 10 |
| 39 | 13 |
| 40 | 8 |
| Total | 50 |
| Highest frequency (fm) = 13, occurs against X value 39. Hence, Mode = 39. The company should be satisfied that a 40-seater is the correct size. |
Mode in Case of the Frequency Distribution of a Continuous Variable
For grouped data, the modal class is the one with the highest frequency. To find the exact mode within this class, the following formula is used:
π Formula: Mode (XΜ) = l + [ (fm - f1) / {(fm - f1) + (fm - f2)} ] Γ h
Where:
- l = lower class boundary of the modal class
- fm = frequency of the modal class
- f1 = frequency of the class preceding the modal class
- f2 = frequency of the class following the modal class
- h = length of class interval of the modal class
π Example (EPA Mileage Ratings):
| Mileage Rating | Class Boundaries | No. of Cars |
|---|---|---|
| 30.0 β 32.9 | 29.95 β 32.95 | 2 |
| 33.0 β 35.9 | 32.95 β 35.95 | 4 = f1 |
| 36.0 β 38.9 | 35.95 β 38.95 | 14 = fm |
| 39.0 β 41.9 | 38.95 β 41.95 | 8 = f2 |
| 42.0 β 44.9 | 41.95 β 44.95 | 2 |
The third class is the modal class. Using the formula: XΜ = 35.95 + [ (14 - 4) / {(14 - 4) + (14 - 8)} ] Γ 3 XΜ = 35.95 + [10 / (10 + 6)] Γ 3 XΜ = 35.95 + (10/16) Γ 3 XΜ = 35.95 + 1.875 XΜ = 37.825
Locating this value on the X-axis of the frequency curve shows the mode exists in the middle of the data values, confirming it is a measure of central tendency.
β Key Takeaways
The stem-and-leaf display is a powerful tool for exploratory data analysis because it sorts data and shows its distribution without losing individual values. Converting this display into a frequency table and histogram reveals that the shape of the distribution is preserved. The mode is a straightforward measure of central tendency representing the most frequent value, but its calculation method depends on data type: by counting for raw data, by identifying the highest frequency for discrete distributions, and via a specific formula for continuous grouped data. The formula for the mode in grouped data interpolates within the modal class using the frequencies of the classes before and after it. Understanding the mode's calculation is essential for describing the most typical or common outcome in a dataset.
π§ Quick Revision Questions
- What is the main advantage of a stem-and-leaf display over a frequency distribution?
- In the stem-and-leaf display for the age data (12 to 74), how many observations (leaves) are in the row for stem "5"?
- How is the modal class identified in a frequency distribution of a continuous variable?
- In the formula for the mode of grouped data (Mode = l + [ (fm - f1) / {(fm - f1) + (fm - f2)} ] Γ h), what does "f1" represent?
- For the EPA mileage ratings example, if the frequency of the modal class (fm) was 14, the class before it (f1) was 4, and the class after it (f2) was 8, what is the value of the term (fm - f2)?
π Lecture 7 β Measures of Central Tendency (Mode, Arithmetic Mean, Median)
π Overview: This lecture continues the discussion of measures of central tendency, focusing on the mode, its desirable properties, and when it is most useful. It then introduces the arithmetic mean as the most widely used average, covering its calculation for both ungrouped and grouped data, along with the concept of grouping error. Finally, the lecture discusses the weighted arithmetic mean and introduces the median as a measure of central tendency, particularly useful when data contains extreme values.
ποΈ Topics Covered
The lecture begins by detailing the desirable properties and appropriate uses of the mode, including the concept of a bi-modal distribution. It then formally introduces the arithmetic mean, demonstrating its calculation for ungrouped data and grouped frequency distributions using class-marks, while also explaining the concept of grouping error. The lecture covers the desirable properties and limitations of the arithmetic mean before moving to the weighted mean to handle data with unequal importance. The session concludes with the definition and calculation of the median for raw data and discrete frequency distributions, highlighting its value when extreme values distort the arithmetic mean.
π Lecture Summary
DESIRABLE PROPERTIES OF THE MODE
The mode is a valuable measure of central tendency, particularly easy to understand and ascertain for discrete frequency distributions. A key advantage is that it is not affected by a few very high or low values. The mode is most useful in practical business contexts, such as inventory management, where knowing the most common size or value is critical. The manager of a men's clothing store, for instance, needs to know the modal hat size to stock in greatest quantity.
π Definition β Mode: The value that occurs most frequently in a data set.
π‘ Why this matters: The mode is crucial for business decisions like stock control, where the most frequent item is the most important to have in large supply.
It is important to note that sometimes a frequency distribution contains two modes, in which case it is called a bi-modal distribution.
THE ARITHMETIC MEAN
The arithmetic mean is the statisticianβs term for the average and is the most widely used average. It is the value that is numerically MOST representative of the whole series. Its formal definition is: βThe arithmetic mean or simply the mean is a value obtained by dividing the sum of all the observations by their number.β
For ungrouped data, the formula is:
π Formula: XΜ = Ξ£X / n β The mean (XΜ) equals the sum of all observations (Ξ£X) divided by the number of observations (n).
π Example: A news agent's receipts for a week are: Monday Β£9.90, Tuesday Β£7.75, Wednesday Β£19.50, Thursday Β£32.75, Friday Β£63.75, Saturday Β£75.50, Sunday Β£50.70. The total is Β£259.85. The mean sales per day are Β£259.85 / 7 = Β£37.12.
For grouped data in a frequency distribution, the observations in each class are assumed to be identical with the class-mark (midpoint, X). The formula becomes:
π Formula: XΜ = Ξ£fX / Ξ£f or XΜ = Ξ£fX / n β The mean equals the sum of (frequency * midpoint) divided by the total frequency.
π Example: For the EPA mileage ratings of 30 cars, the midpoints (X) are computed (e.g., 31.45, 34.45), multiplied by their frequencies (f), and summed (Ξ£fX = 1135.5). The mean is 1135.5 / 30 = 37.85 miles per gallon.
GROUPING ERROR
Grouping error refers to the error introduced by the assumption that all values in a class are equal to the class midpoint. In reality, this is rarely true, so the mean from a frequency distribution is an approximation. However, this error is usually small for the arithmetic mean. In the EPA example, the true mean from raw data was 37.82, while the grouped data gave 37.85, a very slight difference.
DESIRABLE PROPERTIES OF THE ARITHMETIC MEAN:
- Best understood average in statistics.
- Relatively easy to calculate.
- Takes into account every value in the series.
LIMITATION: A few very high or very low values can drag the arithmetic mean towards them, making it unrepresentative.
π Example: The number of floors in a cityβs buildings: 5, 4, 3, 4, 5, 4, 3, 4, 5, 20, 5, 6, 32, 8, 27. The mean is 9, even though 12 out of 15 buildings have 6 floors or less. The three skyscrapers disproportionately affect the mean.
WEIGHTED MEAN
The weighted mean is used when data values cannot be regarded as having equal weightage. Each value (X_i) is assigned a weight (W_i) according to a suitable criterion.
π Formula: XΜ_w = Ξ£W_i X_i / Ξ£W_i β The weighted mean is the sum of (weight * value) divided by the sum of the weights.
π Example: In a high school with 100 Freshmen (15% absent), 80 Sophomores (5% absent), 70 Juniors (10% absent), and 50 Seniors (2% absent), the simple average is incorrectly 8%. Using the number of students as weights gives the correct answer: (100*15 + 80*5 + 70*10 + 50*2) / (100+80+70+50) = 2700/300 = 9%.
MEDIAN
The median is the middle value of a series when the variable values are placed in order of magnitude. It is defined as a βvalue which divides a set of data into two halves, one half comprising of observations greater than and the other half smaller than it.β The median is a better average than the mean when data contains extreme values.
For ungrouped data with an odd number of observations, it is the middle value. π Example (Odd n): For the building floors data sorted as 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 6, 8, 20, 27, 32, the median is 5. This is much more representative than the mean of 9.
For ungrouped data with an even number of observations, the median is the arithmetic mean of the two middle values. π Example (Even n): For passengers sorted as 5, 14, 18, 23, 34, 47, the median is (18+23)/2 = 20.5.
For a discrete frequency distribution, the median is found by locating the value corresponding to the (n+1)/2 th item in the cumulative frequency distribution. π Example: In a school with 45 classes, the median class-size is the 23rd class. The cumulative frequency column shows 20 classes have up to 28 pupils and 28 classes have up to 29 pupils. Therefore, the 23rd class is the one with 29 pupils per class.
β Key Takeaways
The mode is the most frequent value, useful for inventory but can be absent or not unique (bi-modal). The arithmetic mean is the most common average, calculated as the sum divided by the number of values, but it is sensitive to extreme outliers. The weighted mean is essential when data points have different levels of importance. The median, as the middle value, is a more robust measure for skewed data or when outliers are present. For grouped data, these measures are approximations due to grouping error, though this error is usually small for the mean.
π§ Quick Revision Questions
- What is the formal definition of the arithmetic mean, and what is the symbol used to represent it?
- Explain the concept of "grouping error" and why the mean calculated from a frequency distribution is an approximation.
- In what type of situation is the median a more appropriate measure of central tendency than the arithmetic mean? Provide a brief example.
- Describe the difference between calculating the median for a raw data set with an odd number of observations versus one with an even number of observations.
- When is it necessary to use the weighted arithmetic mean instead of the simple arithmetic mean?
π Lecture 8 β Median, Empirical Relation, and Quantiles
π Overview: This lecture explains how to compute the median for a continuous frequency distribution and open-ended distributions. It introduces the empirical relation between the mean, median, and mode, and then extends the concept of partitioning data into quartiles, deciles, and percentiles. Understanding these measures is essential for describing data distribution and relative standings.
ποΈ Topics Covered
This lecture covers the median formula for continuous frequency distributions and its interpretation, handling open-ended frequency distributions with median, the empirical relation between mean, median, and mode for skewed distributions, the concept of quantiles including quartiles, deciles, and percentiles with their formulas, and the graphic location of quantiles using an ogive.
π Lecture Summary
Median in case of a frequency distribution of a continuous variable
The median for a grouped frequency distribution is found using the formula: $$\widetilde{X} = l + \frac{h}{f}\left(\frac{n}{2} - c\right)$$ where l = lower class boundary of the median class, h = class interval size, f = frequency of median class, n = total frequency, and c = cumulative frequency of class preceding median class. The median class is the first class with cumulative frequency exceeding n/2.
π Formula: $$\widetilde{X} = l + \frac{h}{f}\left(\frac{n}{2} - c\right)$$ β The median equals the lower boundary plus a proportional part of the class interval based on how far into the class the middle observation lies.
π Example: For EPA mileage ratings with n=30, n/2=15. Median class is 36.0-38.9 (boundaries 35.95-38.95). Here l=35.95, h=3, f=14, c=6. Calculating: $$\widetilde{X} = 35.95 + \frac{3}{14}(15-6) = 35.95 + 1.93 = 37.88$$. This means half of cars have mileage less than or equal to 37.88 mpg, half above.
Median in case of an open-ended frequency distribution
An open-ended frequency distribution has classes like "less than 2000" or "5000 and above" without defined boundaries. The median is especially valuable here because unless the median falls in an open-ended class, we don't need to estimate boundaries. If the median falls in an intermediate class, the first or last open-ended class is not involved in computation.
Empirical relation between the mean, median and the mode
The word 'empirical' means 'based on observation.' For a unimodal curve of moderate skewness, the median lies between the mean and the mode. The relationship is: Median - Mode = 2(Mean - Median) and equivalently Mean - Mode = 3(Mean - Median).
π Formula: Mode = 3 Median β 2 Mean β In moderately skewed distributions, if you know any two measures, you can estimate the third. π‘ Why this matters: This helps verify consistency of calculated measures or estimate missing ones.
This relation does NOT hold for J-shaped or extremely skewed distributions.
QUARTILES
Quartiles divide the total area under the frequency curve into four equal parts. The formulas follow the same pattern as median but with different numerators:
π First Quartile (Qβ): $$Q_1 = l + \frac{h}{f}\left(\frac{n}{4} - c\right)$$ β Divides lower 25% of data π Second Quartile (Qβ): $$Q_2 = l + \frac{h}{f}\left(\frac{2n}{4} - c\right) = l + \frac{h}{f}\left(\frac{n}{2} - c\right)$$ β Same as median π Third Quartile (Qβ): $$Q_3 = l + \frac{h}{f}\left(\frac{3n}{4} - c\right)$$ β Divides upper 25% of data
DECILES & PERCENTILES
Deciles divide the distribution into 10 equal parts, and percentiles divide into 100 equal parts.
π Formula for first decile: $$D_1 = l + \frac{h}{f}\left(\frac{n}{10} - c\right)$$ π Formula for first percentile: $$P_1 = l + \frac{h}{f}\left(\frac{n}{100} - c\right)$$
The 5th decile (Dβ ) equals the median, the 25th percentile (Pββ ) equals Qβ, the 75th percentile (Pββ ) equals Qβ, and the 40th percentile (Pββ) equals the 4th decile (Dβ).
All these measures β median, quartiles, deciles, and percentiles β are collectively called quantiles (or fractiles).
π Example: If Company A's yearly sales are at the 90th percentile, this means 90% of all companies have sales less than Company A's, and only 10% have sales exceeding Company A's. Percentile rankings are of practical value only for large data sets.
Graphic location of quantiles
The ogive (cumulative frequency polygon) allows convenient graphical determination of quantiles. To find the median graphically:
- Compute n/2 (here 30/2 = 15)
- Locate 15 on the y-axis
- Draw a horizontal line from y=15 to the ogive curve
- From the intersection point, drop a vertical line to the x-axis
- The x-value where the vertical line touches is the median
Similarly, for Qβ draw horizontal line at n/4, and for Qβ draw at 3n/4. For deciles, draw at n/10, 2n/10, 3n/10, etc. For percentiles, draw at n/100, 2n/100, 3n/100, etc.
β Key Takeaways
The median formula for grouped data is the foundation for computing all quantiles, with the only change being the numerator (n/2 for median, n/4 for Qβ, etc.). The empirical relation Mode = 3Median β 2Mean applies only to moderately skewed unimodal distributions, not to symmetrical or extremely skewed ones. The second quartile, 5th decile, and 50th percentile are all equivalent to the median. Percentile ranking is particularly useful for describing relative standing in large datasets. The ogive provides a quick graphical method to locate any quantile by drawing horizontal lines at appropriate cumulative frequency values.
π§ Quick Revision Questions
- What are the symbols l, h, f, n, and c in the median formula for grouped data, and what does each represent?
- Why is the median preferred over the mean for open-ended frequency distributions?
- If the mean of a moderately skewed distribution is 50 and the median is 48, what is the approximate mode?
- Which decile is equivalent to the median, and which percentile is equivalent to the first quartile?
- How would you find the 60th percentile graphically using an ogive?
π Lecture 9 β Geometric Mean, Harmonic Mean, Relation Between the Arithmetic, Geometric and Harmonic Means, Some Other Measures of Central Tendency
π Overview: This lecture introduces two additional measures of central tendency beyond the arithmetic mean: the geometric mean and the harmonic mean. It explains their formulas for both raw and grouped data, provides detailed examples of their application, and clarifies the specific situations where each measure is most appropriate. The lecture also establishes the mathematical relationship between the arithmetic, geometric, and harmonic means and briefly discusses other measures like the mid-range and mid-quartile range.
ποΈ Topics Covered
This lecture covers the geometric mean, its definition, formula, and computation using logarithms for both raw data and grouped frequency distributions, with detailed examples including the EPA mileage data and a firm's turnover growth. It then covers the harmonic mean, its definition and formula for raw and grouped data, illustrated by a car speed example to show its correct application for averaging rates. The lecture ends with the relation between the arithmetic, geometric, and harmonic means, and introduces other measures of central tendency: the mid-range and mid-quartile range.
π Lecture Summary
Geometric Mean
The geometric mean, G, of a set of n positive values Xβ, Xβ,..., Xβ is defined as the positive nth root of their product. The formula is G = βΏβ(Xβ Γ Xβ Γ ... Γ Xβ) where Xα΅’ > 0. When n is large, computing the geometric mean becomes laborious because you have to extract the nth root of the product of all values. The calculation is simplified by using logarithms. Taking logarithms to base 10, the formula becomes log G = (1/n)[log Xβ + log Xβ + ... + log Xβ] = Ξ£log X / n. Therefore, G = antilog [Ξ£log X / n].
π Definition β Geometric Mean (G): The positive nth root of the product of n positive values. It is used when averaging relative changes or rates of growth. π Formula: G = antilog [Ξ£log X / n] (for raw data) β This means the geometric mean is the antilogarithm of the average of the logarithms of the data values. π Example: Find the geometric mean of the numbers: 45, 32, 37, 46, 39, 36, 41, 48, 36. Solution: The product is βΉβ(45Γ32Γ37Γ46Γ39Γ36Γ41Γ48Γ36), which is cumbersome to compute directly. Using logarithms, the sum of log X is 14.3870. Then, log G = 14.3870 / 9 = 1.5986. Hence, G = antilog 1.5986 = 39.68.
Geometric Mean for Grouped Data
In case of a frequency distribution having k classes with midpoints Xβ, Xβ,..., Xβ and corresponding frequencies fβ, fβ,..., fβ (such that Ξ£fα΅’ = n), the geometric mean is given by G = βΏβ(Xβ^fβ Γ Xβ^fβ Γ ... Γ Xβ^fβ). Each value of X must be multiplied by itself f times, making this quite difficult. In terms of logarithms, the formula becomes log G = (1/n)[fβlog Xβ + fβlog Xβ + ... + fβlog Xβ] = Ξ£f log X / n. Hence, G = antilog [Ξ£f log X / n]. π Example: For the EPA mileage ratings data:
| Mileage Rating | No. of Cars (f) | Class-mark (X) | log X | f log X |
|---|---|---|---|---|
| 30.0 - 32.9 | 2 | 31.45 | 1.4976 | 2.9952 |
| 33.0 - 35.9 | 4 | 34.45 | 1.5372 | 6.1488 |
| 36.0 - 38.9 | 14 | 37.45 | 1.5735 | 22.0290 |
| 39.0 - 41.9 | 8 | 40.45 | 1.6069 | 12.8552 |
| 42.0 - 44.9 | 2 | 43.45 | 1.6380 | 3.2760 |
| Total | 30 | 47.3042 | ||
| G = antilog (47.3042 / 30) = antilog 1.5768 = 37.74 miles per gallon. |
π‘ Why this matters: The geometric mean is used when relative changes in a variable are to be averaged, such as rates of growth over time. Using the arithmetic mean can exaggerate the average rate. π Example: A firmβs turnover increased over four years, with yearly percentages of 125%, 200%, 150%, and 140% compared to the previous year. The arithmetic mean of these percentages is (125+200+150+140)/4 = 153.75%. Applying this rate to the initial Β£2,000 gives an incorrect final turnover of Β£11,176, whereas the actual was Β£10,500. The geometric mean is β΄β(125Γ200Γ150Γ140) = 151.37%. Applying this rate gives the correct final turnover of Β£10,500. This shows the average annual increase was 51.37%, not 53.75%.
Harmonic Mean
The harmonic mean is defined as the reciprocal of the arithmetic mean of the reciprocals of the values. In case of raw data: H.M. = n / Ξ£(1/X). In case of grouped data: H.M. = n / [Ξ£f(1/X)], where X represents the midpoints of the classes.
π Definition β Harmonic Mean (H.M.): The reciprocal of the arithmetic mean of the reciprocals of the data values. It is the appropriate average when values are given as x per y, where x is constant and y is variable. π Formula: H.M. = n / Ξ£(1/X) (for raw data) β This means the harmonic mean is the total number of observations divided by the sum of the reciprocals of the observations. π Example: A car travels 100 miles in 10 intervals of 10 miles each, at speeds of 30, 35, 40, 40, 45, 40, 50, 55, 55, and 30 mph. The arithmetic mean of the speeds is 42 mph, which is incorrect. The correct average speed is total distance (100 miles) divided by total time (2.4881 hours), which equals 40.2 mph. The harmonic mean correctly computes this: n = 10, Ξ£(1/X) = 0.2488, so H.M. = 10 / 0.2488 = 40.2 mph.
Rules for Using Different Means
- When values are given as x per y where x is constant and y is variable, the Harmonic Mean is the appropriate average to use.
- When values are given as x per y where y is constant and x is variable, the Arithmetic Mean is the appropriate average to use.
- When relative changes in some variable quantity are to be averaged, the Geometric Mean is the appropriate average to use.
π Example: If 10 students have marks out of 20 (13, 11, 9, 9, 6, 5, 19, 17, 12, 9), the average marks by the arithmetic mean is 11. This is correct because all marks are "marks per 20", where the denominator (y=20) is constant, and the numerator (marks, x) is variable.
Relation Between Arithmetic, Geometric and Harmonic Means
For any set of positive data, the following inequality holds: Arithmetic Mean β₯ Geometric Mean β₯ Harmonic Mean.
Some Other Measures of Central Tendency
- Mid-Range: For n observations with xβ and xβ as the smallest and largest observations, the mid-range is (xβ + xβ) / 2.
- Mid-Quartile Range: For n observations with Qβ and Qβ as the first and third quartiles, the mid-quartile range is (Qβ + Qβ) / 2. This is also known as the mid-hinge.
A measure of central tendency is a single number that represents a whole set of data, often called an "average". The choice of which average to use depends on the nature of the data and the specific question being asked.
β Key Takeaways
The geometric mean (antilog of the average of logs) is essential for averaging rates of growth or relative changes, as using the arithmetic mean in such cases leads to an overestimation of the average rate. The harmonic mean (reciprocal of the average of reciprocals) is the correct average for data expressed as a ratio where the numerator is constant and the denominator varies, such as calculating average speed when distances are equal. A fundamental inequality states that for any set of positive data, the arithmetic mean is always greater than or equal to the geometric mean, which is always greater than or equal to the harmonic mean. The mid-range (average of the minimum and maximum values) and the mid-quartile range (average of the first and third quartiles) are additional, though less commonly used, measures of central tendency. The key skill is to correctly identify which measureβarithmetic, geometric, or harmonic meanβis appropriate for a given problem based on the nature of the data.
π§ Quick Revision Questions
- What is the formula for the geometric mean of raw data using logarithms?
- Why does the arithmetic mean give an incorrect "average" rate of increase for a firm's turnover over several years, and which measure should be used instead?
- In the car speed example, why did the arithmetic mean of the speeds (42 mph) give an incorrect average speed, and how did the harmonic mean provide the correct answer?
- State the mathematical relationship between the arithmetic mean, geometric mean, and harmonic mean for a set of positive values.
- Define the mid-range and the mid-quartile range.
π Lecture 10 β Dispersion: Absolute and Relative Measures, Range, Coefficient of Dispersion, Quartile Deviation, Coefficient of Quartile Deviation
π Overview: This lecture introduces the concept of dispersion as a measure of the variability or spread in a dataset. It explains why knowledge of the average alone is insufficient, and differentiates between absolute and relative measures. The first two specific measuresβRange and Quartile Deviationβare defined, computed, and compared, along with their associated coefficients for relative comparison.
ποΈ Topics Covered
The lecture begins by establishing the need for measures of dispersion using examples of datasets with the same mean but different spreads. It then classifies measures into absolute and relative types. The Range and its associated Coefficient of Dispersion are explained, followed by the Quartile Deviation (Semi-Interquartile Range) and its relative measure, the Coefficient of Quartile Deviation. Finally, the lecture introduces the rationale for the upcoming measures of Mean Deviation and Standard Deviation.
π Lecture Summary
Concept of Dispersion
Just as variable series differ in their location (average), they also differ in the amount of variability or scatter they exhibit. Two datasets can have the same arithmetic mean but be entirely different in how their values are distributed. Therefore, a measure of dispersion is needed to accompany the relevant measure of central tendency.
π Definition β Dispersion: The amount of variability, scatter, or spread exhibited by the values in a dataset. It distinguishes datasets that share the same average but have different distributions. π Example: A first-year class (ages 17, 18, 19) shows consistent ages, while an evening class (ages 18 to 58) shows great variability. Both groups have ages, but their dispersion is very different. π‘ Why this matters: Knowing the average alone can be misleading. Dispersion provides a fuller description of the data.
Absolute Versus Relative Measures of Dispersion
There are two types of dispersion measurements: absolute and relative.
An absolute measure of dispersion measures dispersion in the same units (or the square of units) as the data.
A relative measure of dispersion is expressed as a ratio, coefficient, or percentage and is independent of units. It is useful for comparing datasets of different natures (e.g., comparing dispersion of heights in meters with weights in kilograms).
This lecture discusses four measures: the Range, the Quartile Deviation, the Mean Deviation, and the Standard Deviation.
Range
The range is the difference between the two extreme values of a dataset.
π Formula: R = Xm β X0, where Xm = highest value and X0 = lowest value.
It is a simple measure of mental arithmetic but gives no idea of the distribution of observations between the ends. For grouped data, the range is the difference between the upper boundary of the highest class and the lower boundary of the lowest class. It is appropriately used in quality control charts, daily temperatures, and stock prices. Disadvantages:
- It ignores all information from intermediate observations.
- It can give a misleading picture of the spread.
Viewpoint of Range: The range can be seen as twice the arithmetic mean of the deviations of the smallest and largest values from the mid-range. As such, it is associated with the mid-range as the measure of central tendency.
Coefficient of Dispersion
The range is an absolute measure; its relative measure is the Coefficient of Dispersion (C.D.).
π Formula: Coefficient of Dispersion = (Range) / (Mid-Range) / 2 = (Xm β X0) / (Xm + X0)
This is a dimensionless (pure) number used for comparing dispersion across different datasets. π Example: If C.D. for one dataset is 0.6 and for another is 0.4, the first dataset has greater dispersion.
Quartile Deviation
The Quartile Deviation (Q.D.) is defined as half the difference between the third quartile (Q3) and the first quartile (Q1). It is also known as the Semi-Interquartile Range.
π Formula: Q.D. = (Q3 β Q1) / 2
It is not an extremely satisfactory measure as it only uses two values. An attractive feature is that the range Median Β± Q.D. contains approximately 50% of the data.
Viewpoint of Quartile Deviation: The Q.D. can be seen as the arithmetic mean of the deviations of the first and third quartiles from the Median. As such, it is associated with the median and should be used when the median is the chosen average. π Example: For Company X: Q1=60, Q3=270 β Q.D. = (270-60)/2 = 105 shares (high scatter). For Company Y: Q1=165, Q3=210 β Q.D. = (210-165)/2 = 22.5 shares (concentrated around median). π‘ Why this matters: A larger Q.D. indicates greater scatter. Q.D. is superior to the Range as it is not affected by extreme values.
Coefficient of Quartile Deviation
The quartile deviation is an absolute measure; its relative measure is the Coefficient of Quartile Deviation.
π Formula: Coefficient of Quartile Deviation = Q.D. / Mid-Quartile Range = (Q3 β Q1) / (Q3 + Q1)
This is a pure number used for comparing variation in different datasets.
Introduction to Mean Deviation and Standard Deviation
Unlike the Range and Quartile Deviation, the Mean Deviation and Standard Deviation are based on each and every data value.
To measure dispersion around the arithmetic mean, the natural idea is to compute distances from the mean. However, the sum of deviations from the mean is always zero.
The solution is to ignore the signs of these deviations. Summing the absolute deviations gives a non-zero total. Averaging these absolute differences yields the Mean Deviation.
π Definition β Mean Deviation (M.D.): The arithmetic mean of the absolute deviations of observations from their mean. Its full name is Mean Absolute Deviation. π Formula: M.D. = Ξ£ |d| / n, where |d| = absolute deviation from the mean.
This concept will be discussed in detail in the next lecture, followed by the Standard Deviation.
β Key Takeaways
The most critical concepts from this lecture are: (1) Dispersion measures the spread of data and is essential alongside central tendency for a complete description. (2) Absolute measures are in the original units; relative measures (coefficients) are unitless and allow comparison between different datasets. (3) The Range is the simplest but most limited measure, ignoring all intermediate data; it is associated with the mid-range. (4) The Quartile Deviation (Semi-Interquartile Range) is a more robust measure that focuses on the middle 50% of data and is associated with the median. (5) Both the Range and Quartile Deviation have relative coefficients (Coefficient of Dispersion and Coefficient of Quartile Deviation, respectively) for comparative analysis.
π§ Quick Revision Questions
- Why is it necessary to use a measure of dispersion alongside a measure of central tendency? Give a specific example from the lecture.
- What is the fundamental difference between an absolute measure and a relative measure of dispersion?
- Write the formula for the Range. What are its two main disadvantages?
- Calculate the Quartile Deviation for a dataset where Q1 = 25 and Q3 = 75. What does this value represent?
- Why can't the simple sum of deviations from the mean (Ξ£ (X β XΜ)) be used as a measure of dispersion? What solution does the Mean Deviation propose?
π Lecture 11 β Mean Deviation, Standard Deviation and Variance, Coefficient of Variation
π Overview: This lecture introduces two important measures of dispersion that involve all data values: the mean deviation and the standard deviation. It explains how to compute these for raw data and frequency distributions, and introduces the coefficient of variation as a relative measure for comparing variability across different datasets.
ποΈ Topics Covered
The lecture covers the mean deviation as an absolute measure of dispersion based on absolute deviations from the mean, followed by the standard deviation and variance which square the deviations to overcome mathematical limitations. It then discusses the coefficient of variation as a relative measure useful for comparing variability between different datasets or variables.
π Lecture Summary
Mean Deviation
The mean deviation is a measure of dispersion that involves each data value. Since the sum of deviations from the arithmetic mean is always zero, we take the absolute values of these deviations. The mean deviation for raw data is the average of these absolute deviations from the mean. For grouped data (frequency distribution), the formula accounts for frequencies.
π Definition β Mean Deviation: The average of the absolute deviations of observations from their mean. π Formula: M.D. = Ξ£|dα΅’| / n (for raw data); M.D. = Ξ£fα΅’|xα΅’ - xΜ| / n (for grouped data) π Example: For fatalities data (4,6,2,0,3,5,8), mean = 4. Absolute deviations: 0,2,2,4,1,1,4. Sum = 14. M.D. = 14/7 = 2 fatalities. π‘ Why this matters: The mean deviation provides a quick and simple measure using all data, but ignoring signs creates mathematical limitations for further analysis.
The coefficient of mean deviation is the relative form obtained by dividing the mean deviation by the average used (mean or median). The median is preferred for datasets with extreme values. π Formula: Coefficient of M.D. = M.D./Mean or M.D./Median
Standard Deviation and Variance
The variance overcomes the sign problem by squaring the deviations from the mean. The standard deviation is the positive square root of the variance, bringing the measure back to the original units. A short cut formula simplifies computation using Ξ£x and Ξ£xΒ².
π Definition β Variance: The average of the squared deviations of values from their mean. π Definition β Standard Deviation: The positive square root of the variance. π Formula: Variance = Ξ£(x - xΜ)Β²/n; Standard Deviation S = β[Ξ£(x - xΜ)Β²/n] π Short Cut Formula: S = β[(Ξ£xΒ²/n) - (Ξ£x/n)Β²] π Example: For fatalities data, Ξ£(x-xΜ)Β² = 42. Variance = 42/7 = 6 squared fatalities. Using short cut: Ξ£x=28, Ξ£xΒ²=154. S = β[(154/7) - (28/7)Β²] = β(22-16) = β6 = 2.45 fatalities.
For grouped data, the standard deviation formula includes frequencies: π Formula: S = β[Ξ£f(x - xΜ)Β²/n] = β[(Ξ£fxΒ²/n) - (Ξ£fx/n)Β²] π Example: For light bulb data (life in hundreds of hours): Ξ£fx=2437.5, Ξ£fxΒ²=78781.25, n=100. Mean=24.375. S = β[(78781.25/100) - (2437.5/100)Β²] = β(787.8125-594.14) = β193.67 = 13.9 hundred hours = 1390 hours.
Coefficient of Variation
The coefficient of variation (CV) is the most important relative measure of dispersion, expressed as a percentage. It is used to compare variability between different variables or datasets with very different means.
π Definition β Coefficient of Variation: A relative measure of dispersion that expresses standard deviation as a percentage of the mean. π Formula: C.V. = (S/xΜ) Γ 100 π Example 1 β Comparing earnings variability: Country 1: mean=$19.50, S=$4, CV=20.5%. Country 2: mean=Rs.75, S=Rs.28, CV=37.3%. Country 2 has greater variability. π Example 2 β Comparing crop yields: Untreated land: mean=35 bushels, S=10, CV=28.57%. Treated land: mean=58 bushels, S=10, CV=17.24%. The treated land actually has lower relative variability despite the same absolute standard deviation.
β Key Takeaways
The mean deviation is the average absolute distance from the mean, but ignoring signs makes it unsuitable for further mathematical work. The standard deviation and variance overcome this by squaring deviations, with the variance in squared units and the standard deviation in original units. The short cut formulas using Ξ£x and Ξ£xΒ² simplify computation significantly. The coefficient of variation is essential when comparing variability between different datasets or variables, as it normalizes standard deviation by the mean to show relative dispersion. For light bulb data with open-ended classes, midpoint approximation is necessary and standard deviation was 1390 hours.
π§ Quick Revision Questions
- Why must we use absolute deviations for mean deviation instead of actual deviations from the mean?
- What is the mathematical advantage of squaring deviations in standard deviation over taking absolute values in mean deviation?
- Calculate the standard deviation (using short cut formula) for the dataset: 10, 12, 8, 15, 10.
- Why does the coefficient of variation show that variability has "decreased" for treated wheat land even though standard deviation remains 10 bushels?
- In which type of frequency distribution situation would we prefer mean deviation from median over mean deviation from mean?