Descriptive Statistics and Biostatistical Notation

Foundations of Biological Statistics

  • Quantitative biological investigations result in observations composed of numerical facts known as data, where a single isolated fact is referred to as a datum.
  • Objective methods are necessary to analyze and present quantitative research data.
  • Statistics refers specifically to the organization, summarization, analysis, and interpretation of data.
  • Biostatistics refers to statistical concepts and methods applied specifically to data that are biological in nature.
  • Descriptive statistics is the branch of statistics concerned with the organization and summarization of data, typically displayed in visual formats such as tables and figures (also known as graphs or charts).
  • Inferential statistics is the branch of statistics involving the analysis and interpretation of data to formulate general conclusions and inferences.

Population Parameters vs. Sample Statistics

  • Population represents the entire collection of measurements of a variable being studied about which statistical conclusions are drawn.
    • Example: Studying the oxygen consumption of the rodent species Fattus rotundus means the population consists of all oxygen consumption measurements of all Fattus rotundus individuals.
  • Sample refers to a subset or portion of the population on which actual measurements are taken, required because measuring every individual in a population is practically impossible in almost all cases.
  • Random sampling is mandatory for deriving valid population conclusions by induction from a sample.
    • Random sampling requires that every member of a population has an equal and independent chance of being selected into the sample.
    • Careful planning and execution are required to ensure samples are drawn randomly.
  • Population Measures:
    • Measures of central tendency (or location) describe the general tendency for a majority of population measurements to lie around the middle of the range.
    • Mean (average) is the most useful measure of central tendency.
    • Median and mode are additional measures of location.
    • Measures of dispersion (or spread) characterize how dispersed measurements are around their average.
    • Standard deviation is one of the most useful measures of dispersion.
    • A complete statistical summary or description of a data set requires reporting estimates for both central tendency and dispersion.
  • Parameter vs. Statistic Distinction:
    • Parameter: Any measure of central tendency or dispersion known for an entire population. Parameters are conventionally designated using Greek letters.
    • Population mean parameter: ̢
    • Population standard deviation parameter: ̣
    • Statistic: An estimate of a population parameter calculated from a sample. Statistics are conventionally designated using Latin (Roman) letters.
    • Sample mean statistic: Xˉ\bar{X}
    • Sample standard deviation statistic: ss

Statistical Notation and Summation Rules

  • A collection of NN quantitative values in a population is denoted as:   X1,X2,…,XNX_1, X_2, \dots, X_N
    • The quantitative variable is represented by xx or XX.
    • Subscripts designate individual measurement units (X1X_1 is the first unit, X2X_2 is the second unit, up to XNX_N representing the NthN\text{th} or final unit).
  • Summation Notation Definition:   ∑i=1i=Nxi=x1+x2+⋯+xN\sum_{i=1}^{i=N} x_i = x_1 + x_2 + \dots + x_N
    • The capital Greek letter Σ\Sigma (sigma) denotes the operation of summing the quantity that follows it.
    • The subscript variable ii represents the index of summation.
    • Lower limit of summation is 11; upper limit of summation is NN
    • When summation explicitly includes all values, limits are omitted and the notation simplifies to ∑xi\sum x_i
  • Termination Rules for Summation:
    • Summation instructions terminate at a ++ or −- operator.
    • Terminated summation expression:     ∑xi−5=x1+x2+⋯+xN−5\sum x_i - 5 = x_1 + x_2 + \dots + x_N - 5
    • This indicates all xx values are summed together first, and then 55 is subtracted from the total sum.
    • Individual subtraction expression:     ∑(xi−5)\sum (x_i - 5)
    • This indicates 55 is subtracted from each individual xix_i term first, and the resulting differences are then summed.
  • Summation of Constants:
    • Summing a constant value cc across NN terms yields the constant multiplied by the number of terms:     ∑i=1i=Nc=c+c+⋯+c=Nc\sum_{i=1}^{i=N} c = c + c + \dots + c = Nc
    • Specific constant summation example:     ∑i=1i=N3=3+3+⋯+3=3N\sum_{i=1}^{i=N} 3 = 3 + 3 + \dots + 3 = 3N
    • Expanding constant subtractions:     ∑(xi−5)=∑xi−∑5=∑xi−5N\sum (x_i - 5) = \sum x_i - \sum 5 = \sum x_i - 5N
    • Calculating ∑xi−5N\sum x_i - 5N on the right side provides a computationally simpler method than performing individual subtractions on the left side.
  • Sum of Squares versus Square of Sum:
    • Sum of squares (adding squared xx values):     ∑i=1i=Nxi2=x12+x22+⋯+xN2\sum_{i=1}^{i=N} x_i^2 = x_1^2 + x_2^2 + \dots + x_N^2
    • Square of the sum (squaring the total sum of xx values):     (∑xi)2\left( \sum x_i \right)^2

Summation Calculations and Worked Examples

  • Given an exemplar dataset with N=8N = 8 values: 2,3,5,1,4,3,2,42, 3, 5, 1, 4, 3, 2, 4
    • Individual assigned values: x1=2x_1 = 2, x2=3x_2 = 3, x3=5x_3 = 5, x4=1x_4 = 1, x5=4x_5 = 4, x6=3x_6 = 3, x7=2x_7 = 2, x8=4x_8 = 4
  • Sum of all 88 values:   ∑i=1i=8xi=2+3+5+1+4+3+2+4=24\sum_{i=1}^{i=8} x_i = 2 + 3 + 5 + 1 + 4 + 3 + 2 + 4 = 24
  • Sum of the second through fifth values (i=2i = 2 to i=5i = 5):   ∑i=2i=5xi=3+5+1+4=13\sum_{i=2}^{i=5} x_i = 3 + 5 + 1 + 4 = 13
  • Subtracting 55 from the total sum:   ∑xi−5=2+3+5+1+4+3+2+4−5=19\sum x_i - 5 = 2 + 3 + 5 + 1 + 4 + 3 + 2 + 4 - 5 = 19
  • Subtracting 55 from each individual observation prior to summation:   ∑(xi−5)=(2−5)+(3−5)+(5−5)+(1−5)+(4−5)+(3−5)+(2−5)+(4−5)=−16\sum (x_i - 5) = (2 - 5) + (3 - 5) + (5 - 5) + (1 - 5) + (4 - 5) + (3 - 5) + (2 - 5) + (4 - 5) = -16
  • Equivalent constant evaluation check:   ∑xi−5N=24−(5×8)=24−40=−16\sum x_i - 5N = 24 - (5 \times 8) = 24 - 40 = -16
  • Calculating the sum of squares (∑xi2\sum x_i^2):   ∑xi2=22+32+52+12+42+32+22+42=4+9+25+1+16+9+4+16=84\sum x_i^2 = 2^2 + 3^2 + 5^2 + 1^2 + 4^2 + 3^2 + 2^2 + 4^2 = 4 + 9 + 25 + 1 + 16 + 9 + 4 + 16 = 84
  • Calculating the square of the sum ((∑xi)2(\sum x_i)^2):   (∑xi)2=(2+3+5+1+4+3+2+4)2=242=576\left( \sum x_i \right)^2 = (2 + 3 + 5 + 1 + 4 + 3 + 2 + 4)^2 = 24^2 = 576

Algebraic Properties of Summation

  • For any constant cc that does not vary as index ii changes, three fundamental algebraic properties apply:
    1. Summation of a constant:      ∑c=Nc\sum c = Nc
    2. Multiplication by a constant:      ∑cxi=c∑xi\sum c x_i = c \sum x_i
    • Adding products of constant cc and each xix_i is equal to the product of constant cc and the total sum of xix_i
    • Division form of Property 2:        ∑(xic)=∑xic\sum \left( \frac{x_i}{c} \right) = \frac{\sum x_i}{c}
    1. Addition or subtraction of a constant:      ∑(xi±c)=∑xi±Nc\sum (x_i \pm c) = \sum x_i \pm Nc

Calculation of Descriptive Statistics

  • Population Mean (μ\mu):
    • Formal notation [3.1]:     μ=∑i=1NXiN\mu = \frac{\sum_{i=1}^N X_i}{N}
    • Simplified shorthand notation [3.2]:     μ=∑XN\mu = \frac{\sum X}{N}
  • Sample Mean (Xˉ\bar{X}):
    • Serves as the best estimate of population mean μ\mu
    • Sample size is denoted as lowercase nn
    • Formal notation [3.3]:     Xˉ=∑i=1nXin\bar{X} = \frac{\sum_{i=1}^n X_i}{n}
  • Population Variance (σ2\sigma^2) and Standard Deviation (σ\sigma):
    • Population variance σ2\sigma^2 measures dispersion and is termed the "mean squared deviation":     σ2=∑i=1N(Xi−μ)2N\sigma^2 = \frac{\sum_{i=1}^N (X_i - \mu)^2}{N}
    • The term (Xi−μ)(X_i - \mu) represents the deviation from the mean.
    • Summing all individual deviations ∑(Xi−μ)\sum (X_i - \mu) without squaring equals zero.
    • Population standard deviation σ\sigma is termed the "mean deviation" and equals the square root of population variance [3.5]:     σ=∑i=1N(Xi−μ)2N\sigma = \sqrt{\frac{\sum_{i=1}^N (X_i - \mu)^2}{N}}
  • Population Corrected Sum of Squares (SSSS):
    • The expression ∑(X−μ)2\sum (X - \mu)^2 is called the corrected sum of squares (SSSS).
    • Computational formula for SSSS [3.6]:     SS=∑i=1NXi2−(∑i=1NXi)2NSS = \sum_{i=1}^N X_i^2 - \frac{\left( \sum_{i=1}^N X_i \right)^2}{N}
    • Difference between terms:
    • Uncorrected sum of squares ∑X2\sum X^2: Each XX value is squared first, then squared values are summed.
    • Square of sum (∑X)2\left( \sum X \right)^2: All XX values are summed first, then the total is squared.
    • Efficient computational formula for population variance [3.7]:     σ2=∑i=1NXi2−(∑i=1NXi)2NN\sigma^2 = \frac{\sum_{i=1}^N X_i^2 - \frac{\left( \sum_{i=1}^N X_i \right)^2}{N}}{N}
  • Sample Variance (s2s^2) and Sample Standard Deviation (ss):
    • Sample variance s2s^2 and standard deviation ss provide the best unbiased estimates of population variance σ2\sigma^2 and standard deviation σ\sigma
    • Sample size nn replaces population size NN
    • Denominator replaces NN with n−1n - 1, defined as degrees of freedom (dfdf
    • Sample variance formula [3.8]:     s2=∑i=1nXi2−(∑i=1nXi)2nn−1s^2 = \frac{\sum_{i=1}^n X_i^2 - \frac{\left( \sum_{i=1}^n X_i \right)^2}{n}}{n - 1}
    • Sample corrected sum of squares numerator in shortcut notation:     SS=∑X2−(∑X)2nSS = \sum X^2 - \frac{\left( \sum X \right)^2}{n}
    • Sample variance expressed via mean square [3.9]:     s2=SSn−1s^2 = \frac{SS}{n - 1}
    • Variance is defined as a mean square (MSMS); mean squares are always computed by dividing a sum of squares (SSSS) by its degrees of freedom (dfdf
    • Sample standard deviation formula [3.10]:     s=∑i=1nXi2−(∑i=1nXi)2nn−1s = \sqrt{\frac{\sum_{i=1}^n X_i^2 - \frac{\left( \sum_{i=1}^n X_i \right)^2}{n}}{n - 1}}