Organisation and Representation of Data

Statistical data collected during an economic survey or census investigation are known as Raw Data (or crude data). Raw data are an unstructured, chaotic collection of observations recorded in the order of their enumeration. In this unorganized form, they possess three major deficiencies: they fail to convey any immediate economic meaning, occupy immense space, and cannot be utilized for comparative analysis or algebraic computations. Organisation of data refers to the systematic arrangement and condensation of raw data into orderly classes, groups, and tables to bring out their underlying relationships and salient features.

Classification is the primary operational step in data organisation. It is defined as:

"Classification is the process of arranging data into sequences and groups according to their common characteristics or separating them into different but related parts." — Horace Secrist

The principal objectives of classification in economic statistics include:

1. Simplification and Condensation: Reducing immense, unwieldy mass data into compact, homogeneous groups that human intelligence can easily absorb.
2. Highlighting Similarities and Contrasts: Segregating data points possessing common attributes from those with distinct traits (e.g., separating literate agricultural labourers from illiterate ones).
3. Facilitating Meaningful Comparison: Enabling comparative evaluations between demographic groups, time periods, or geographic regions.
4. Preparing Data for Tabulation: Serving as the indispensable preparatory bridge between raw collection and formal statistical tabulation.

• Exhaustiveness / Comprehensiveness: Every single item in the raw dataset must find a place in one of the classes without any omission.
• Mutual Exclusivity: Classes must be non-overlapping; no single observation should qualify for more than one class.
• Stability: The basis of classification must remain uniform throughout the entire investigation to prevent bias.
• Flexibility: The classification scheme should be adaptable to changing circumstances without destroying comparative validity.
• Homogeneity: All units placed within a given class must share fundamentally similar characteristics.

In quantitative classification, a measurable phenomenon that varies in magnitude across observation units is termed a variable:

• Discrete Variable: A variable that increases by distinct, finite, disconnected jumps and assumes only isolated exact integers. Fractional values are physically impossible (e.g., number of children per household, number of road accidents per month, number of printing errors per page).
• Continuous Variable: A variable capable of assuming any real numerical value (including infinite fractions and decimals) within a continuous numerical interval (e.g., worker wages, body weight, temperature, monthly electricity consumption, agricultural yield per acre).

A frequency distribution is a statistical table that presents values of a variable along with their corresponding frequencies (the number of times each value or class of values recurs in the dataset). Constructing a frequency distribution condenses raw data into a structured mathematical form using the standard Tally Mark method (the four-and-cross technique: $ ext{||||}\mkern-10mu/$ denoting bundles of 5).

1. Class Interval: A designated numerical band or bracket within which individual data points are grouped (e.g., $10 - 20$).
2. Class Limits: The extreme numerical boundaries of a class interval:
   • Lower Class Limit ($L_1$): The lowest value that can belong to the class.
   • Upper Class Limit ($L_2$): The highest value that can belong to the class.
3. Class Width / Interval Length ($c$): The numerical difference between the upper and lower boundaries of a class: $c = L_2 - L_1$.
4. Class Midpoint / Class Mark ($m$): The central value of a class interval, defined mathematically as: $$m = rac{L_1 + L_2}{2}$$
5. Frequency Density: The frequency of a class divided by its class width. Crucial when class intervals are unequal: $$ ext{Frequency Density} = rac{ ext{Class Frequency } (f)}{ ext{Class Width } (c)}$$
6. Relative Frequency: The proportion of the total frequency belonging to a particular class: $$ ext{Relative Frequency} = rac{f}{\sum f} = rac{f}{N}, \quad ext{Percentage Frequency} = \left(rac{f}{N} ight) imes 100$$

To convert an inclusive discontinuous series into an exclusive continuous series, calculate the correction factor:

Cumulative frequency expresses how many observations fall below or above a given class boundary:

• "Less-Than" Cumulative Series: Frequencies are progressively cumulated downwards starting from the lowest class. Plotted against the upper true class boundaries. Answers questions like: "How many workers earn less than Rs. 20,000?"
• "More-Than" Cumulative Series: Frequencies are cumulated upwards starting from the highest class, or subtracted progressively from the total $N$. Plotted against the lower true class boundaries. Answers questions like: "How many workers earn more than Rs. 10,000?"

Tabulation is the systematic and logical presentation of classified numerical data in horizontal rows and vertical columns. While classification is the preliminary grouping of observations according to common traits, tabulation is the final architectural presentation of that grouped data for analytical dissemination.

Core objectives of tabulation:

1. Economizing Space: Compresses thousands of narrative observations into a compact, easily readable matrix.
2. Facilitating Visual Comparison: Placing interrelated figures in adjacent columns or rows highlights ratios, percentages, and trends instantaneously.
3. Eliminating Repetition: Headings, stubs, and measurement units are stated once in the header, obviating redundant explanatory text.
4. Providing Reference Foundation: Serves as the official archive and source data for statistical testing and computational software.

According to standard statistical conventions, every formal table must incorporate eight structural parts:

• According to Purpose:
  • General Purpose Table (Reference Table): Comprehensive repository storing extensive raw census or survey aggregates for archival use (e.g., Census of India tables).
  • Special Purpose Table (Summary / Analytical Table): Concise, targeted table constructed to illuminate a specific economic relationship, hypothesis, or policy variable (e.g., comparing inflation rates across five years).
• According to Construction / Complexity:
  • Simple (One-way) Table: Classifies data on the basis of a single characteristic (e.g., number of students classified solely by faculty stream).
  • Complex Table: Classifies data simultaneously on two or more interrelated characteristics:
    • Two-way (Double) Table: Data classified across two characteristics (e.g., faculty stream $ imes$ gender).
    • Three-way (Treble) Table: Data classified across three characteristics (e.g., faculty stream $ imes$ gender $ imes$ residential domicile).
    • Manifold Table: Data cross-classified across four or more simultaneous attributes.

Diagrammatic presentation translates classified and tabulated numerical data into visual geometric forms such as bars, rectangles, and circles. While statistical tables provide exact mathematical precision, visual diagrams possess unmatched power to convey quantitative relationships instantly to laymen, policymakers, and corporate executives without requiring mathematical training.

A Bar Diagram consists of equidistant rectangular pillars (bars) erected vertically or horizontally on a common baseline. The width of every bar is strictly equal and arbitrary; the height (or length) of each bar is directly proportional to the numerical magnitude it represents. The major variants include:

A Pie Chart is a circular statistical diagram partitioned into sectors, where the entire circle of $360^\circ$ represents the aggregate total ($100\%$), and each individual sector represents the proportional share of a specific component category.

• Ideal Applications: Visualizing sectoral composition of Gross State Domestic Product (Primary, Secondary, Tertiary sectors), sources of government tax revenue (GST, Income Tax, Corporate Tax), and household budget expenditure shares.
• Limitations of Pie Charts: Ineffective if the number of components exceeds 7 or 8 (creates cluttered, unreadable slivers), cannot display negative values (negative angles do not exist geometrically), and humans struggle to visually evaluate angles compared to linear bar heights.

A Histogram is a two-dimensional graphical representation of a continuous grouped frequency distribution consisting of contiguous vertical rectangles erected over class intervals:

• Class boundaries are plotted on the horizontal $X$-axis; frequencies are plotted on the vertical $Y$-axis.
• The Area Principle: Unlike bar charts where only height matters, in a histogram the area of each rectangle is directly proportional to the frequency of that class.
• No Spaces: Because the variable is continuous, rectangles stand shoulder-to-shoulder with zero gaps between them.
• Unequal Class Intervals Adjustment: If class intervals are unequal, plotting raw frequencies distorts the visual area. In such cases, the height of each rectangle must represent Frequency Density: $$ ext{Adjusted Height} = rac{ ext{Class Frequency}}{ ext{Class Width}} imes ext{Standard Base Width}$$
• Graphical Determination of Mode ($Z$): The mode is located by identifying the highest modal rectangle, drawing intersecting diagonal lines from its top corners to the adjacent rectangles, and dropping a perpendicular to the $X$-axis.

• Frequency Polygon: Constructed by joining the midpoints of the upper horizontal tops of the histogram rectangles with straight line segments. To close the polygon, both extremities are extended to touch the horizontal baseline at the midpoints of hypothetical adjacent classes with zero frequency on either side.
• Smoothed Frequency Curve: Drawn through the midpoints of the histogram columns as a smooth, continuous freehand curve, eliminating the sharp angular vertices of the polygon to reflect the theoretical underlying population distribution.

An Ogive (pronounced oh-jive) is a cumulative frequency graph plotted on coordinate axes:

Ogives also graphically determine all partition values: the First Quartile ($Q_1$ at $rac{N}{4}$), Third Quartile ($Q_3$ at $rac{3N}{4}$), Deciles, and Percentiles.

When statistical data are recorded over successive chronological intervals (years, months, days), the resulting graphical plot is termed a Time Series Graph or a Historigram (distinct from a continuous 'histogram'):

• Time (independent variable) is invariably plotted on the horizontal $X$-axis.
• The economic variable (output, inflation, exports, share prices) is plotted on the vertical $Y$-axis.
• Individual coordinate points are plotted and connected using straight line segments.
• Multiple series can be plotted on the same graph using distinct line styles (solid, dashed, dotted) to examine macroeconomic co-movements.

Normally, the vertical $Y$-axis must begin strictly at zero ($0$). However, if the observed economic values are very large and fluctuate within a narrow, elevated band (e.g., a company's share price fluctuating between Rs. 950 and Rs. 980), starting the axis at zero creates two severe flaws:

1. The lower 90% of the graph paper remains completely blank and wasted.
2. The actual economic fluctuations appear as an imperceptible, flattened line, concealing significant market volatility.