Revision summary
Simple random and systematic sampling require a usable frame. Stratification ensures analytically important minorities are represented. Cluster and multistage designs suit dispersed regional surveys but need design-effect adjustment. Purposive sampling selects information-rich cases; theoretical sampling develops concepts. Snowball sampling reaches hidden groups but carries network bias. Anthropological sampling should combine transparent inference, ethnographic depth, non-response analysis and ethics.
Model answer
Introduction
Sampling selects cases from a defined population so that fieldwork is feasible. Its quality depends not on one universally best technique but on fit among research question, sampling frame, mobility, social boundaries and the inference claimed.
Body
Probability designs
- Simple random sampling gives every listed unit a known equal chance; suitable for a complete village household register, but costly over dispersed terrain.
- Systematic sampling takes every kth unit after a random start; efficient in settlement lists, unless periodic ordering creates bias.
- Stratified sampling samples within relevant strata such as sex, age, caste or ecological zone. It ensures small subgroups are represented and can improve precision.
- Cluster and multistage sampling first select villages or census blocks, then households and persons. This is practical for regional nutrition or morbidity surveys, though intra-cluster similarity enlarges sampling error.
Probability designs support estimable sampling error and population inference, provided non-response and frame omissions are addressed.
Non-probability and adaptive designs
Purposive sampling selects information-rich ritual specialists, healers or fossil collections; quota sampling fills categories without random selection; convenience sampling is fast but weak for generalisation. Snowball sampling, associated with referral chains, reaches stigmatised or hidden groups such as migrants, sex workers or drug users, but over-represents network hubs. Respondent-driven sampling attempts to model referral probabilities under strong assumptions. Theoretical sampling in grounded theory selects new cases to develop emerging concepts until saturation.
Classic intensive ethnography—Malinowski in the Trobriands or M. N. Srinivas in Rampura—often chose a site purposively, then used a census, genealogy and case studies within it. Statistical representativeness and ethnographic depth answer different questions.
Gatekeeper access, seasonal absence, household definitions, refusal and digital exclusion can bias any design. Tribal or vulnerable populations require community consent and protection from re-identification; oversampling a small group must not become extractive surveillance.
Flow diagram
flowchart TD
Q[Research question] --> P{Population frame exists?}
P -->|yes| PR[Probability sample]
PR --> R[Random stratified or multistage]
P -->|no or hidden| NP[Non-probability or adaptive]
NP --> N[Purposive snowball theoretical]
R --> I[Population inference]
N --> D[Depth or network access]
Conclusion
Use random, stratified or multistage designs when estimating population patterns; use purposive, theoretical or network designs when mechanisms, rare expertise or hidden populations are the object. Triangulation and honest limits matter more than calling every sample representative.
Quick related
Students also ask
-
Examine critically the concept of social stratification as a basis for sustaining social inequality.
Next question in the 2024 paper (Q8). View answer →
-
Is a large convenience sample representative?
No. Size reduces random error but cannot repair systematic selection bias.
-
Can one study mix sampling techniques?
Yes. A multistage household survey can be combined with purposive life histories, provided each supports a distinct claim.
PYQ trend
When UPSC asked this
Related PYQs from other years, newest first. Open a question to read it.
-
2023 · Q1(d) · Anthropology GS 1 · 10 marks
Customary laws and Environmental conservation. -
2015 · Q1(c) · Anthropology GS 1 · 10 marks
Incest Taboos.
More from this paper
Q1(a) · UPSC Mains 2024 · Anthropology GS 1 · 10 marks
Attributes of culture
The Nature of Culture
Culture is a learned and symbolically mediated design for living. Its core attributes are sharing, symbolism, patterning, adaptation, transmission and change. Tylor supplies the complex whole; Geertz supplies public meaning; Bourdieu links culture to embodied practice. Kula shows integration of exchange, rank, ritual and inter-island relations. Shared culture is never identical experience: gender, class, age and power matter. Treat integration as a question, not as automatic harmony.
Q1(b) · UPSC Mains 2024 · Anthropology GS 1 · 10 marks
Harappan maritime trade
Meaning, Scope and development of Anthropology.
Harappan maritime trade joined Indus rivers and coastal sites to Oman, Bahrain and Mesopotamia. Lothal, Dholavira, Kuntasi, Balakot and Sutkagen-dor were important nodes. Meluhha texts, Indus seals, Gulf materials and etched carnelian beads provide converging evidence. Exports included crafted beads, shell, ivory, timber and probably textiles; metals moved inward. Lothal’s basin may be a dock or reservoir, so do not make the whole case depend on it. Trade supported craft specialisation and urban integration but was not the sole cause of Harappan civilisation.
Q1(c) · UPSC Mains 2024 · Anthropology GS 1 · 10 marks
Critical perspective on avoidance and joking relationship
The Nature of Culture
Avoidance prescribes reserve; joking licenses teasing between named kin categories. Radcliffe-Brown read both as complementary ways of regulating conjunction and separation. Mother-in-law/son-in-law avoidance and cross-cousin or junior-affine joking are common patterns. Symmetrical joking permits reciprocity; asymmetrical joking protects one-sided licence. Gluckman highlights ritualised expression of contradiction. Critical analysis asks whether joking reproduces gendered power and whether rules match actual conduct.
Toppers' copies
Toppers' copies for this question will be uploaded soon.