Geographical Data: Sources, Compilation and Processing

1. Data, information and why geographers need numbers

Read the detailed note →

• Data are numbers that stand for measurements of the real world. Examples: 20 cm of rain in Barmer, or the 1,385 km Delhi–Mumbai rail distance via Kota and Vadodara (1,542 km via Itarsi and Manmad). A datum is one single measurement.
• Information is a meaningful answer to a question, or a meaningful clue that leads to more questions. It comes from data through logical, step-by-step (algorithmic) or statistical working.
• A variable is any feature that changes from place to place or from time to time.
• Quantitative variable: the differences can be measured (rainfall).

• Qualitative variable or attribute: it has only categories (sex).

• Why geographers need data: many variables act on each other, and numbers describe them best.

• Cropping pattern: cropped area, yield, irrigated area, rainfall, fertiliser and pesticide use.

• Growth of a city: population, density, migrants, jobs, wages, industry and transport.

• Statistical fallacy (a wrong conclusion drawn from correct numbers): a man measured river depths of 0.6, 0.8, 0.9 and 1.5 m. The average was 0.95 m, so his 1 m tall child should have been safe, but the child drowned. "Lekha Jokha Thahe, to Bachha Dooba Kahe?" An average can hide the extremes.

• Because of this, geography has moved from describing things in words to measuring them. Careful methods are used at every step, from collecting the data to drawing the conclusion.

2. Primary sources of data

Read the detailed note →

• Primary data are collected first-hand by a person, a group or an institution for a set purpose (Class 12, Data – Its Source and Compilation, Fig. 1.1).
• Personal observation: a field survey records relief, drainage, soils, vegetation, population structure, sex ratio, literacy, transport and settlements. The observer needs knowledge of the subject and must judge without bias.
• Interview: information is gathered by talking to the respondent. NCERT lists eight precautions:
• prepare an exact list of items to ask about, and be clear about the aim of the survey;
• win the respondent's trust and promise secrecy; keep the mood friendly;
• use simple, polite language; never ask questions that hurt self-respect or religious feelings;

• at the end, ask for any extra information, then thank the respondent.

• Questionnaire vs schedule (a common trap):

	Questionnaire	Schedule
Filled in by	the respondent	a trained enumerator who asks the questions
Who can answer	only literate people	literate and illiterate people
Strength	cheap; can be posted to faraway places; suits large areas	wider coverage, as in the Census

• Other methods: soil-testing kits and water-quality kits used in the field, and transducers (sensors) that measure crop health (Fig. 1.2).
• Beyond NCERT: sampling (random or stratified) makes it possible to survey a huge population by studying only a part of it, as the NSS does.

3. Secondary sources: published and unpublished

Read the detailed note →

• Secondary data have already been collected by someone else. The user takes them from published or unpublished records.
• Published sources:
• Government: Census of India (Office of the Registrar General), NSS reports, IMD weather reports, state Statistical Abstracts, District Bulletins and commission reports (Fig. 1.3).
• Semi- or quasi-government: urban development authorities, municipal corporations and zila parishads.
• International: UN bodies (UNESCO, UNDP, WHO, FAO) and their Demographic Yearbook, Statistical Yearbook and Human Development Report (Fig. 1.4).
• Private: yearbooks, surveys and reports published by newspapers and private organisations.
• Newspapers and magazines.

• Electronic media, especially the internet.

• Unpublished sources of data:

• Government documents, such as the village revenue (land) records kept by the patwari.
• Quasi-government records, such as development plans and reports of municipal corporations and district councils.

• Private documents, such as records of companies, trade unions, political and non-political organisations and residents' welfare associations.

• Exam angle: the same Census table is primary data for the Census office and secondary data for a student who uses it.

4. India's statistical system and data sources today

Read the detailed note →

• Census of India (beyond NCERT): the first count in 1872 was not held everywhere at the same time. It has been held across the country on the same dates every ten years since 1881. It runs under the Census Act 1948, led by the Registrar General and Census Commissioner (Ministry of Home Affairs).

• The 2021 round was postponed. Census 2027 has a reference date of 1 March 2027 (1 October 2026 for snow-bound areas). It will be the first digital census and will include caste enumeration (verify current). Census results are covered in india-population.

• National Sample Survey (NSS): nationwide sample surveys since 1950 (P.C. Mahalanobis), covering consumption, employment, health and other social and economic topics.

• National Statistics Office (NSO): formed in 2019 under the Ministry of Statistics and Programme Implementation (MoSPI) by merging the CSO and the NSSO. It runs the Periodic Labour Force Survey (from 2017-18) and the Household Consumption Expenditure Survey (2022-23, 2023-24), and compiles national accounts (see economic-data-statistics).
• Agriculture Census: counts operational holdings (their number, size, tenure and use) every five years since 1970-71. The 11th (2021-22) was the first to use smartphones and digitised land records (verify current).
• Livestock Census: held since 1919-20. The 20th (2019) counted about 536 million livestock. The 21st began in October 2024 (verify current).
• Other sources: NFHS (Ministry of Health with IIPS Mumbai; NFHS-5, 2019-21) on health, fertility and nutrition; IMD for weather; data.gov.in under the National Data Sharing and Accessibility Policy (NDSAP) 2012.
• Issues: delays, comparing results across survey rounds, debates over sample design, and trust in official numbers.

5. Tabulation and compilation: absolute data, percentages and index numbers

Read the detailed note →

• Raw data are data as first collected. They are a jumble of figures that must be sorted before they make sense.
• Tabulation arranges data in the rows and columns of a statistical table, with a title, headings, footnotes and source. It makes data simpler to read, easier to compare and quick to look up, and it saves space. Footnotes explain limits; for example, the J&K figure in Census 2011 excludes PoK.
• Data can be compiled in three forms:
• Absolute data: the original whole numbers. India's population in Census 2011 was 1,21,05,69,573 (62,31,21,843 males and 58,74,47,730 females). Uttar Pradesh had 19,98,12,341.
• Percentage data: values out of a hundred of a common base. This allows a fair comparison between places of different size. Literacy rate = literates ÷ total population × 100. It rose from 18.33% (1951) to 73.0% (2011; male 80.9, female 64.6) (Census 2011).

• Index number: shows change compared with a base value set at 100. Simple aggregate method: Σq₁ ÷ Σq₀ × 100. Iron-ore output was 32.5 million tonnes in 1970-71 (= 100) and 67.4 million tonnes in 2000-01, giving an index of 207 (India: Economic Year Book 2005).

• Index numbers can compare times, places, industries or countries. Beyond NCERT: WPI, CPI and IIP use weighted formulas with base years that are revised from time to time (see economic-data-statistics).

6. Grouping data: classes, frequencies and ogives

Read the detailed note →

• Grouping of data sorts raw data into classes to cut down its bulk. The number of classes and the class interval (upper limit minus lower limit of a class, for example 10 in 20–30) depend on the range of the data. NCERT's 60 geography scores range from 02 to 96, so they go into ten classes of 10.
• Tally marks ("four and cross" method): draw one stroke for each item. Every fifth stroke crosses the four before it, so you can count in fives.
• A frequency distribution shows how many items (f) fall in each class. Σf = N = 60.
• Cumulative frequency (Cf) is the running total. In NCERT's example, 27 students scored under 50 and 45 scored under 70. The last Cf always equals N.
• Exclusive method (20–30, 30–40): the upper limit of one class is the lower limit of the next. A boundary value goes to the class where it is the lower limit, so 30 goes into 30–40 ("20 and under 30").
• Inclusive method (20–29, 30–39): both limits belong to the class, so the next class starts one unit higher. Each class still covers ten values.
• Frequency polygon: plot each class frequency, usually at the class midpoint, and join the points. It is useful for comparing two distributions (Fig. 1.5).
• Ogive: a curve of cumulative frequencies.
• The less-than ogive uses upper limits and rises.
• The more-than ogive uses lower limits and falls.

• The point where the two curves cross gives the median.

• NCERT error: the more-than values in Table 1.9 are wrong. From "more than 30" onward they should be 46, 39, 33, 23 and 15, not 44, 38, 28, 20 and 14.

7. Measures of central tendency: mean, median and mode

Read the detailed note →

• Measures of central tendency (statistical averages) give one value that stands for the centre of the data, the point around which the values gather.
• Mean = sum of all values ÷ number of values.
• Direct method: the normal rainfall of seven Malwa plateau districts (Mandsaur 825 mm to Dewas 1,083 mm) gives 6,484 ÷ 7 = 926.29 mm.
• Indirect method of mean (assumed mean, or "coding"): subtract an assumed mean A from each value, then use X̄ = A + Σd/N. Here 800 + 884/7 = 926.29 mm.

• Grouped data use class midpoints: X̄ = Σfx/N. Daily wages of 99 workers: 10,160 ÷ 99 = Rs 102.6. Indirect method: 100 + 260/99 = Rs 102.6. The indirect method works with equal or unequal class intervals.

• Median: a positional average (found by position, not by adding values) that splits an ordered series into two equal halves.

• Ungrouped data: the value of the (N+1)/2th item. Seven Himalayan peaks (7,817–8,848 m) give the 4th item, 8,172 m (NCERT misprints this as "8.172 m").

• Grouped data: M = l + (i/f)(N/2 − c). Here 80 + (10/16)(25 − 21) = 82.5. The median class is 80–90 (NCERT misprints it as "40–44").

• Mode (Z or M₀): the value that occurs most often. A series can be unimodal, bimodal, trimodal, multimodal or without mode.

• Properties:
• The mean uses every value but is pulled by extreme values (the river-depth fallacy).
• The median is not affected by extreme values.
• The mode always sits at the hump (peak) of a distribution and suits category data, such as the most common crop.

8. Shape of a distribution: normal curve and skewness

Read the detailed note →

• Normal distribution: a symmetrical, bell-shaped curve. Most values gather in the middle and thin out evenly towards rare extremes on both sides.
• Mean, median and mode are equal (100 in Fig. 2.1).
• Half the values lie above the middle and half below.

• Traits such as intelligence and achievement scores come close to it. A normal curve can have high or low spread.

• Skewness means the curve is lopsided, so the three averages separate (Fig. 2.2, 2.3).

	Tail	Order
Positive skew	long tail to the right	mean > median > mode
Negative skew	long tail to the left	mean < median < mode

• Positive-skew examples: incomes, or rainfall in a dry region with a few extreme storms.
• Beyond NCERT: for moderately skewed data, mode ≈ 3 median − 2 mean (Karl Pearson). The median always lies between the mean and the mode.
• Exam use: use the median for skewed data such as incomes or landholdings. Use the mean for symmetrical data.

9. Dispersion and relationship: spread and association

Read the detailed note →

• Class 12, Data Processing names these two groups of measures but does not calculate them. The glossary defines several of them.
• Measures of dispersion show how spread out the values are, usually around a central value.
• Range = maximum − minimum. It is the simplest measure and is also used to set choropleth map classes (range ÷ 5).
• Mean deviation: the average distance of values from the mean or median, ignoring plus and minus signs.
• Standard deviation: the positive square root of the average of the squared deviations from the mean. It is the most widely used measure; a bigger value means more spread.

• Coefficient of variation (beyond NCERT) = SD ÷ mean × 100. It compares variability between data sets. India's rainfall CV is above 50% in western Rajasthan and Kutch, and below 25% on the west coast and in the north-east. So the driest areas also have the least dependable rain (see indian-climate-monsoon).

• Measures of relationship show how closely two variables are linked, such as rainfall and floods, or fertiliser and yield.

• Correlation coefficient: a number from −1 to +1.
  • Positive: both variables rise together (rainfall and paddy area).
  • Negative: one falls as the other rises (altitude and temperature).
  • Near zero: little or no link.
• Beyond NCERT: Karl Pearson's r is used for measured values and Spearman's rank correlation for ranks. A correlation does not prove that one thing causes the other.

10. Spatial statistics used by geographers

Read the detailed note →

• Beyond NCERT: geographers adjust statistics so that they take location into account.
• Location quotient (LQ) = a region's share of an activity ÷ the nation's share.
• Example: a district's share of textile workers ÷ India's share of textile workers.

• LQ above 1 means the region specialises in that activity. It is used to find industrial clusters and a region's economic base (see manufacturing-industries-india).

• Nearest neighbour analysis (Clark and Evans, 1954) works in three steps:

• measure the average distance from each settlement to its nearest neighbour;
• compare it with the distance expected if settlements were scattered at random;

• read the index Rn: near 0 = clustered; 1 = random; up to about 2.15 = perfectly regular (hexagonal) pattern.

• It is used to classify how rural settlements are spread out (see human-settlements).

• Today these measures run inside GIS (geographic information systems), which joins statistics with map-based spatial analysis (see remote-sensing-gis).

---

Exam angles

Prelims — high-yield facts and traps

• Questionnaire: filled in by the respondent, so literate people only, but it can be posted. Schedule: filled in by an enumerator, works for illiterate people, used in the Census.
• Patwari land records = unpublished government source. Census and NSS reports = published government sources. Human Development Report = international (UNDP) publication.
• Exclusive classes leave out the upper limit (30 goes to 30–40). Inclusive classes include both limits (20–29).
• Ogive = plot of cumulative frequency. The less-than ogive rises and the more-than ogive falls; where they cross is the median. "An ogive plots simple frequency" is FALSE.
• Median is not affected by extreme values. Mode marks the hump. In a normal curve, mean = median = mode. Positive skew: mean > median > mode. "The mean is unaffected by extremes" is FALSE.
• Index number = Σq₁ ÷ Σq₀ × 100, with the base year = 100.
• Census since 1872, held at the same time nationwide from 1881, under the Census Act 1948. NSS since 1950. NSO formed in 2019 by merging CSO and NSSO. Census 2027 reference date 1 March 2027 (verify current).
• Agriculture Census every five years since 1970-71. Livestock Census since 1919-20 (20th in 2019: about 536 million).
• Correlation coefficient runs from −1 to +1. CV = SD ÷ mean × 100. Nearest-neighbour Rn: 0 clustered, 1 random, about 2.15 regular. LQ above 1 = specialisation.

Mains — GS-I themes

1. "Averages can mislead." Discuss when to use the mean, median or mode in geographical analysis, using the river-depth fallacy, skewed incomes and rainfall.
2. Why official statistics must be timely and trusted for planning. What the delayed census means for welfare targeting, delimitation and urban planning.
3. Compare primary and secondary data for a field study of a village or city. Discuss sources of bias and the precautions to take.
4. Rainfall variability, measured by the coefficient of variation, as a basis for drought-proofing and crop planning in India.
5. How spatial statistics (location quotient, nearest neighbour analysis) and GIS make geography more precise.

Current-affairs hooks

• Progress of Census 2027: digital enumeration, caste data, houselisting phase (verify current).
• New survey releases: monthly and quarterly PLFS, HCES, NFHS-6, Agriculture Census and Livestock Census results (verify current).
• Revision of base years for GDP, CPI, WPI and IIP (verify current).
• Debates on the credibility and independence of official statistics, and the role of the National Statistical Commission.
• data.gov.in, NDSAP and MoSPI's data-sharing initiatives (verify current).