Statistical Officer / Government Statistical Service (GSS) Assessment Survey Methodology, Data Collection and Sampling Syllabus

Survey Methodology, Data Collection and Sampling syllabus — full chapter and topic list

Expand any chapter to see its topics and sub-topics. This is the whole examinable outline for Survey Methodology, Data Collection and Sampling in Statistical Officer / Government Statistical Service (GSS) Assessment, not a summary of it.

1. 1

   Designing Data Collection

   3 topics

   • Sources of statistical data
     • Surveys, censuses and administrative data
     • Big data and alternative data sources
     • Strengths and limitations of each source
   • Questionnaire design
     • Writing clear, unbiased questions
     • Question order and response scale effects
     • Cognitive testing and piloting
   • Modes of data collection
     • Face-to-face, telephone, postal and online surveys
     • Mixed-mode designs and mode effects
2. 2

   Sampling Methods

   3 topics

   • Probability sampling
     • Simple random, systematic and stratified sampling
     • Cluster and multi-stage sampling
     • Probability proportional to size
   • Non-probability sampling
     • Quota, convenience and snowball sampling
     • Risks of bias in non-probability designs
   • Sample size and precision
     • Determinants of required sample size
     • Trade-off between precision, cost and time
3. 3

   Errors, Bias and Weighting

   3 topics

   • Total survey error
     • Sampling error versus non-sampling error
     • Coverage, measurement and processing error
   • Non-response
     • Unit and item non-response
     • Non-response bias and follow-up strategies
   • Weighting and adjustment
     • Design weights and calibration
     • Imputation for missing data
4. 4

   UK Statistical Sources

   3 topics

   • Major ONS surveys and outputs
     • Labour Force Survey and the labour market statistics
     • Consumer Prices Index and inflation measures
     • GDP, national accounts and population estimates
   • The Census
     • Purpose, design and the move to an admin-data-based approach
   • Administrative and linked data
     • Use of admin data under the Digital Economy Act
     • Secure data access via the Integrated Data Service / SRS

Survey Methodology, Data Collection and Sampling flashcards for Statistical Officer / Government Statistical Service (GSS) Assessment

19 of 55 cards from the Survey Methodology, Data Collection and Sampling deck — real questions with worked answers.

1. What is the distinction between primary and secondary data in official statistics?

   Primary data are collected first-hand by the statistician for the specific purpose at hand (e.g. a bespoke survey). Secondary data are pre-existing data originally collected for another purpose and reused (e.g. administrative records, existing surveys, published statistics).

2. Name the four broad sources of statistical data used by the GSS/ONS.

   1) Survey/census data (collected directly from respondents); 2) Administrative data (from operating government/business systems, e.g. tax, benefits); 3) Commercial/'big' data (scanner data, web-scraped prices, mobile/sensor data); 4) Published/secondary statistics and registers.

3. What are the main advantages and disadvantages of administrative data versus survey data?

   Advantages: cheaper, large coverage, no extra respondent burden, frequent/timely. Disadvantages: collected for non-statistical purposes, so definitions/coverage may not match the statistical concept, variable quality, limited control over content, and potential coverage errors.

4. In questionnaire design, what is the difference between an open and a closed question?

   An open question lets respondents answer in their own words (rich but hard to code/analyse). A closed question offers fixed response categories (easy to code and compare, but restricts answers and can omit options).

5. List common sources of response error introduced by poor question wording.

   Leading questions, double-barrelled questions (two issues in one), ambiguous or vague terms, loaded/emotive wording, double negatives, presuming/assumptive questions, overlapping or non-exhaustive response categories, and recall/memory demands that are too high.

6. What is acquiescence bias and how can questionnaire design reduce it?

   Acquiescence bias is the tendency to agree with statements regardless of content. It can be reduced by avoiding agree/disagree formats, balancing positively and negatively worded items, and offering substantive response scales rather than yes/no.

7. Why is question order important in a questionnaire, and what is a context (order) effect?

   Earlier questions can prime or frame answers to later ones (a context/order effect). Good practice: start with easy, non-threatening questions, group topics logically, place sensitive/demographic questions later, and randomise item order where context effects are a concern.

8. What is the purpose of cognitive testing and piloting a questionnaire?

   Cognitive testing (e.g. think-aloud, probing) checks how respondents comprehend, retrieve, judge and respond to questions, exposing misunderstanding. A pilot tests the whole instrument and field procedures at small scale before main fieldwork to catch routing, timing and wording problems.

9. Compare the four main modes of survey data collection on cost and interviewer presence.

   Face-to-face (CAPI): high cost, interviewer present, high response, good for complex surveys. Telephone (CATI): medium cost, interviewer present. Postal/paper: low cost, no interviewer, slower, lower response. Web/online (CAWI): lowest cost, no interviewer, fast, but coverage/digital-exclusion issues.

10. What does 'mixed-mode' data collection mean and why is it used?

    Using two or more modes (e.g. web-first then telephone or face-to-face follow-up) within one survey. It is used to cut costs while maintaining coverage and response rates, but introduces a risk of mode effects, where the mode itself influences answers.

11. Define interviewer-administered versus self-completion modes and a key trade-off.

    Interviewer-administered (face-to-face, telephone): interviewer asks questions, improving response, clarification and complex routing, but can introduce interviewer effects and social-desirability bias. Self-completion (postal, web): more honest answers to sensitive items and cheaper, but higher item non-response and no clarification.

12. What is a probability (random) sample, and why is it the gold standard for inference?

    A sample in which every unit in the population has a known, non-zero probability of selection. This allows design-based estimation of sampling error, unbiased estimators, and valid statistical inference (confidence intervals) to the whole population.

13. Describe simple random sampling (SRS) without replacement.

    Every possible sample of size $n$ from a population of $N$ has an equal chance of selection, so each unit has selection probability $\pi_i = \frac{n}{N}$. It requires a complete sampling frame and gives unbiased estimates but can be inefficient and operationally costly.

14. What is systematic sampling and how is the sampling interval defined?

    After a random start, every $k$-th unit is selected from an ordered frame, where the interval is $k = \frac{N}{n}$. It is simple to implement and spreads the sample evenly, but bias arises if the list has a periodic pattern matching $k$.

15. Explain stratified random sampling and one key benefit.

    The population is divided into mutually exclusive, exhaustive strata (e.g. region, age) and an independent random sample is drawn within each. Benefit: guarantees representation of each stratum and reduces sampling variance when strata are internally homogeneous, improving precision over SRS.

16. State the formula for proportional allocation in stratified sampling.

    Under proportional allocation each stratum's sample is proportional to its size: $n_h = n \cdot \frac{N_h}{N}$, where $N_h$ is the population size of stratum $h$ and $N = \sum_h N_h$.

17. State the formula for Neyman (optimum) allocation in stratified sampling.

    Neyman allocation minimises variance for fixed $n$ by allocating more to larger and more variable strata: $$n_h = n \cdot \frac{N_h S_h}{\sum_{k} N_k S_k},$$ where $S_h$ is the standard deviation within stratum $h$.

18. What is cluster sampling and why does it usually increase sampling variance?

    The population is divided into clusters (e.g. postcodes, schools); a random sample of whole clusters is selected and units within are surveyed. It cuts travel/frame costs but units within a cluster are correlated (intra-cluster homogeneity), so it is less efficient than SRS for the same sample size.

19. What is multi-stage sampling?

    Sampling carried out in successive stages, e.g. first select primary sampling units (areas), then households within them, then individuals. It avoids needing a full national frame of individuals and concentrates fieldwork geographically to reduce cost.

See more Survey Methodology, Data Collection and Sampling flashcards →

Planning Survey Methodology, Data Collection and Sampling for Statistical Officer / Government Statistical Service (GSS) Assessment

Survey Methodology, Data Collection and Sampling (Statistical Officer / Government Statistical Service (GSS) Assessment) FAQ