I85PREP
⌘ K
Korea Social Survey Analyst Grade 2 study materials
Korea Social Survey Analyst Grade 2 study resources

Korea Social Survey Analyst Grade 2 practice questions

One question at a time. Build your Korea Social Survey Analyst Grade 2 understanding.

First 50 of 190 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Research design, measurement and sampling

In scientific social research methodology, which error occurs when an empirical finding observed at an aggregate group level is improperly inferred to hold true for individual members of that group?

  1. Ecological fallacy (생태학적 오류) Correct answer.
  2. Individualistic fallacy (개인주의적 오류)
  3. Reductionism (환원주의)
  4. Systematic measurement error (체계적 오차)

Explanation

An ecological fallacy (생태학적 오류) occurs when an investigator draws conclusions about individual behavior or attributes solely based on aggregate, group-level statistical observations. For example, inferring that a wealthy individual voted for a conservative party because they reside in an affluent district is an ecological fallacy.

Question 2Research design, measurement and sampling

In the process of operationalization (조작적 정의), which operational step directly converts an abstract theoretical concept into observable and measurable empirical indicators?

  1. Conceptualization (개념화)
  2. Operational definition (조작적 정의) Correct answer.
  3. Nominal scaling (명목척도화)
  4. Content validation (내용타당도 검증)

Explanation

An operational definition (조작적 정의) specifies the exact empirical operations, measurement procedures, or quantitative metrics used to observe, measure, and quantify an abstract theoretical concept in field research.

Question 3Research design, measurement and sampling

A researcher investigates the effect of job training on wage increases, while suspecting that the relationship is transmitted through enhanced labor productivity. In this research model, what role does labor productivity play?

  1. Independent variable (독립변수)
  2. Moderating variable (조절변수)
  3. Mediating variable (매개변수) Correct answer.
  4. Extraneous variable (외생변수)

Explanation

A mediating variable (매개변수) transmits the causal influence from the independent variable to the dependent variable, explaining the generative mechanism through which X influences Y (Job Training -> Labor Productivity -> Wage Increase).

Question 4Research design, measurement and sampling

Which longitudinal survey design repeatedly measures the exact same cohort of identical individuals across multiple successive time intervals?

  1. Trend study (추세조사)
  2. Cohort study (동년배조사)
  3. Cross-sectional survey (횡단조사)
  4. Panel study (패널조사) Correct answer.

Explanation

A panel study (패널조사) tracks and collects data from the exact same sample of respondents (the panel) at successive points in time, allowing direct tracking of individual-level changes over time, though it is vulnerable to panel attrition (패널탈락).

Question 5Research design, measurement and sampling

Which of the following conditions is NOT a necessary criterion for establishing causality between two variables in empirical social science research?

  1. Empirical association (공변관계)
  2. Temporal precedence (시간적 우선성)
  3. Non-spuriousness (허위관계 배제)
  4. Deterministic necessity (결정론적 필연성) Correct answer.

Explanation

Social sciences evaluate probabilistic causality rather than deterministic necessity (결정론적 필연성). The three fundamental criteria for establishing causality are: 1. Empirical correlation/covariation (공변성), 2. Temporal precedence of the cause before the effect (시간적 우선성), and 3. Non-spuriousness through the control of confounding variables (허위관계 배제).

Question 6Research design, measurement and sampling

In experimental design, which internal validity threat refers to biological, cognitive, or physiological changes occurring within research subjects simply as a function of the passage of time?

  1. History effect (역사요인)
  2. Maturation effect (성숙요인) Correct answer.
  3. Testing effect (검사요인)
  4. Instrumentation (도구요인)

Explanation

The maturation effect (성숙요인) refers to internal biological, developmental, psychological, or physiological processes within participants (e.g., aging, fatigue, intellectual development) that operate during the course of an experiment and confound treatment effects.

Question 7Research design, measurement and sampling

Which experimental design is specifically structured to assess and control for the interaction effect between pre-testing and the experimental treatment?

  1. Post-test only control group design (사후검사 통제집단설계)
  2. One-group pretest-posttest design (단일집단 사전사후설계)
  3. Solomon four-group design (솔로몬 4집단설계) Correct answer.
  4. Static-group comparison design (정적집단 비교설계)

Explanation

The Solomon four-group design (솔로몬 4집단설계) utilizes four randomly assigned groups (two pretested, two un-pretested; two receiving treatment, two control). This unique structure enables researchers to detect and isolate the pre-test testing effect and its interaction with the treatment.

Question 8Research design, measurement and sampling

A researcher wishes to draw a sample of n = 500 from a numbered sampling frame of N = 25,000 residents using systematic sampling (계통표본추출). What is the sampling interval (k) and how should the first element be selected?

  1. Interval k = 50; the first element must be selected at random between 1 and 50 Correct answer.
  2. Interval k = 50; the first element must always be fixed as number 1
  3. Interval k = 25; the first element must be selected at random between 1 and 25
  4. Interval k = 100; the first element must be selected at random between 1 and 100

Explanation

The sampling interval is calculated as k = N / n = 25,000 / 500 = 50. In systematic probability sampling, the initial starting element must be randomly chosen from between 1 and k (1 to 50), after which every k-th element is selected.

Question 9Research design, measurement and sampling

What primary hazard occurs when applying systematic sampling to a sampling frame that exhibits hidden cyclical or periodic patterns (주기성)?

  1. Elimination of sampling error
  2. Severe sampling bias if the sampling interval coincides with the cycle periodicity Correct answer.
  3. Automatic transformation of the sample into a stratified random sample
  4. Reduction of the required sample size by half

Explanation

Periodicity (주기성) is the primary danger in systematic sampling. If the list contains recurring cycles and the sampling interval k happens to match the cycle period, the sample will systematically overrepresent or underrepresent specific elements, creating severe bias.

Question 10Research design, measurement and sampling

In stratified random sampling (층화표본추출), how should the population be divided into strata to maximize the statistical efficiency of parameter estimates?

  1. Heterogeneous within strata, and homogeneous between strata
  2. Homogeneous within strata, and heterogeneous between strata Correct answer.
  3. Identical sample sizes allocated to all strata regardless of stratum variance
  4. Random assignment of elements into arbitrary strata without criteria

Explanation

To achieve maximum precision in stratified sampling, strata should be internally homogeneous (집단 내 동질적, minimizing within-stratum variance) and mutually heterogeneous (집단 간 이질적, maximizing between-stratum differences).

Question 11Research design, measurement and sampling

Which probability sampling method is most appropriate when an exhaustive sampling frame of individual elements is unavailable, but the population is naturally organized into identifiable geographical clusters?

  1. Simple random sampling (단순무작위표본추출)
  2. Stratified random sampling (층화표본추출)
  3. Cluster sampling (군집표본추출) Correct answer.
  4. Quota sampling (할당표본추출)

Explanation

Cluster sampling (군집표본추출 or 집락표본추출) is chosen when a complete list of individual elements cannot be compiled, but clusters of elements (e.g., schools, voting precincts, census tracts) can be listed. Clusters are sampled first, followed by census or sub-sampling within chosen clusters.

Question 12Research design, measurement and sampling

Which of the following non-probability sampling techniques relies on existing research participants recruiting subsequent participants from among their acquaintances, making it ideal for hard-to-reach or hidden populations?

  1. Convenience sampling (편의표본추출)
  2. Purposive sampling (유의표본추출)
  3. Quota sampling (할당표본추출)
  4. Snowball sampling (눈덩이표본추출) Correct answer.

Explanation

Snowball sampling (눈덩이표본추출 or 누적표본추출) is a non-probability technique where an initial set of informants who meet research criteria refer additional members of their network. It is widely used for stigmatized, illicit, or hard-to-reach populations.

Question 13Research design, measurement and sampling

Although quota sampling (할당표본추출) and stratified random sampling (층화표본추출) both divide the population into demographic sub-groups, what is the fundamental methodological difference between them?

  1. Quota sampling uses random selection within sub-groups, whereas stratified sampling relies on convenience selection
  2. Stratified sampling selects elements randomly within strata, whereas quota sampling allows interviewers to non-randomly select cases within quotas Correct answer.
  3. Quota sampling requires an exhaustive sampling frame, while stratified sampling does not
  4. Stratified sampling cannot be used when demographic proportions are known

Explanation

The crucial distinction is the selection mechanism: Stratified sampling is a probability method where elements are drawn strictly via random sampling within each stratum from an exhaustive frame. Quota sampling is a non-probability method where interviewers fill quotas using convenience or judgmental selection, introducing interviewer bias.

Question 14Research design, measurement and sampling

In sample size determination for estimating a population proportion with a 95% confidence level (z = 1.96) and a margin of error d = 0.05, when no prior estimate of the population proportion (p) is available, what conservative value should be assigned to p to guarantee maximum sample size?

  1. p = 0.10
  2. p = 0.25
  3. p = 0.50 Correct answer.
  4. p = 0.95

Explanation

In the sample size formula n = z^2 * p * (1-p) / d^2, the variance component p*(1-p) reaches its mathematical maximum when p = 0.50 (0.50 * 0.50 = 0.25). Setting p = 0.50 provides the most conservative estimate, ensuring sufficient statistical power regardless of the true population parameter.

Question 15Research design, measurement and sampling

If a researcher wants to reduce the margin of sampling error by half (from 4% to 2%) at the same confidence level, by what factor must the sample size be increased?

  1. 2 times
  2. 4 times Correct answer.
  3. 8 times
  4. 16 times

Explanation

Because the margin of error d is inversely proportional to the square root of sample size n (d = z * sigma / sqrt(n)), sample size is inversely proportional to d squared (n = z^2 * sigma^2 / d^2). Halving the error (d/2) means n must be divided by (1/2)^2 = 1/4, which requires multiplying the sample size by 4.

Question 16Research design, measurement and sampling

Consider the survey question: 'Do you agree that the government should reduce corporate taxes and expand public healthcare spending?' What fundamental questionnaire construction error does this item violate?

  1. Leading question bias (유도질문)
  2. Double-barreled question (이중질문 / 복합질문) Correct answer.
  3. Prestige bias (위신편향)
  4. Social desirability bias (사회적 바람직성 편향)

Explanation

A double-barreled question (이중질문 or 복합질문) combines two separate, distinct issues into a single prompt requiring one answer. A respondent might favor corporate tax cuts but oppose healthcare spending expansion (or vice versa), making it impossible to interpret their response.

Question 17Research design, measurement and sampling

When designing a closed-ended survey question with multiple response choices, which two logical principles must the response categories strictly fulfill?

  1. Subjectivity (주관성) and Ambiguity (모호성)
  2. Mutual exclusivity (상호배타성) and Exhaustiveness (포괄성) Correct answer.
  3. Linearity (선형성) and Normalcy (정규성)
  4. Discreteness (불연속성) and Parity (등가성)

Explanation

Response options in closed-ended survey questions must be mutually exclusive (상호배타성: no respondent can logically fit into more than one category) and collectively exhaustive (포괄성: every eligible respondent must find at least one category that applies to them, often achieved by adding 'Other').

Question 18Research design, measurement and sampling

In questionnaire flow and layout, which sequencing principle is generally recommended to optimize respondent cooperation and minimize early dropouts?

  1. Placing highly sensitive and intrusive personal demographic questions at the very beginning
  2. Starting with easy, interesting, and non-threatening introductory questions before moving to core substantive topics Correct answer.
  3. Placing open-ended essay questions in the opening section and simple multiple-choice questions at the end
  4. Randomizing the order of all questions across the survey without thematic grouping

Explanation

Standard questionnaire sequencing dictates starting with engaging, straightforward, and non-threatening questions (warm-up questions) to establish rapport. Sensitive, complex, and demographic questions (e.g., income, marital status) are placed toward the end.

Question 19Research design, measurement and sampling

What is the primary role of a pre-test (사전조사 or 파일럿 테스트) conducted prior to launching full-scale survey fieldwork?

  1. Generating the final empirical findings to test formal research hypotheses
  2. Identifying confusing item phrasing, inappropriate question flow, and formatting errors in the draft questionnaire Correct answer.
  3. Selecting the final sample of respondents for the primary study
  4. Calculating definitive population parameters

Explanation

A pre-test (사전조사) is administered to a small sub-sample (typically 20–50 subjects similar to the target population) to detect ambiguous wording, response category omissions, excessive length, interviewer difficulties, and logical flow defects before finalized fieldwork.

Question 20Research design, measurement and sampling

In qualitative research, what is the primary structural characteristic of a Focus Group Interview (FGI / 표적집단면접법)?

  1. Administering a strictly closed-ended questionnaire to a large representative probability sample
  2. Facilitating an interactive, guided group discussion among 6 to 10 relatively homogeneous participants led by a trained moderator Correct answer.
  3. Conducting secret observational field research without the subjects' knowledge or consent
  4. Analyzing historical documents and archival statistical registers using automated text miners

Explanation

A Focus Group Interview (FGI) brings together a small group (typically 6 to 10 participants sharing relevant characteristics) to engage in an open, interactive discussion guided by a skilled moderator, capitalizing on group dynamics to generate rich qualitative insights.

Question 21Research design, measurement and sampling

In a Focus Group Interview (FGI), what is the primary risk associated with the phenomenon of 'groupthink' (집단사고) or dominant respondent bias?

  1. Interviewer bias, where the moderator's own wording steers participants toward a preferred answer
  2. Recall bias, where participants misremember the timing of events they are asked to describe
  3. Social desirability bias, where participants overstate behaviour they believe the researcher approves of
  4. Suppression of dissenting minority opinions and an artificial consensus driven by dominant or high-status participants Correct answer.

Explanation

A major vulnerability in FGI sessions is groupthink or dominant personality influence, where vocal or high-status participants dominate the conversation, causing more reserved participants to conform and withhold divergent experiences or opinions.

Question 22Research design, measurement and sampling

Which qualitative interviewing format utilizes an outline of core thematic topics but gives the interviewer freedom to modify question wording, reorder prompts, and probe deeper into emergent responses?

  1. Standardized structured interview (표준화/구조화 면접)
  2. Semi-structured interview (반구조화 면접) Correct answer.
  3. Completely unstructured projective test (완전 비구조화 투사검사)
  4. Self-administered postal questionnaire (자기기입식 우편설문)

Explanation

A semi-structured interview (반구조화 면접) uses an interview guide (면접지침) covering predetermined key inquiry domains while granting the interviewer latitude to probe, ask follow-up questions, and adjust phrasing according to participant responses.

Question 23Research design, measurement and sampling

In planning a statistical survey (통계조사계획), which sequence correctly orders the planning decisions that precede fieldwork?

  1. Establish the survey purpose, then decide the survey content, then determine the survey method, then draft the execution plan Correct answer.
  2. Determine the survey method, then establish the survey purpose, then decide the survey content, then draft the execution plan
  3. Decide the survey content, then determine the survey method, then establish the survey purpose, then draft the execution plan
  4. Draft the execution plan, then establish the survey purpose, then determine the survey method, then decide the survey content

Explanation

The official planning sequence runs from purpose to operations. The researcher first fixes what the survey is for (통계조사목적 수립), then decides which items of information are needed to serve that purpose (조사내용 결정), then selects a data collection method capable of obtaining those items within budget and schedule (조사방법 결정), and only then writes the detailed execution plan covering sample, fieldwork, timetable and cost.

Question 24Research design, measurement and sampling

Under which condition is a complete enumeration (전수조사) generally preferred over a sample survey (표본조사)?

  1. When the population is very large and geographically dispersed
  2. When the measurement process destroys or consumes the unit being tested
  3. When the population is small and results must be reported for very fine subgroups Correct answer.
  4. When the budget and the reporting deadline are both tight

Explanation

A complete enumeration is justified when the population is small enough to cover exhaustively and when estimates are required for very small domains, such as individual districts or rare industries, where a sample would leave too few cases per cell to report. Korea's Population and Housing Census is the standard example. In most other situations a well-designed sample gives usable precision faster and more cheaply, and can even be more accurate, because a smaller field operation is easier to control.

Question 25Research design, measurement and sampling

What is the decisive methodological advantage of probability sampling (확률표본추출) over non-probability sampling (비확률표본추출)?

  1. It always produces a larger sample for the same budget
  2. It removes the need to construct a sampling frame
  3. It guarantees that the sample mean equals the population mean
  4. Every unit has a known, non-zero probability of selection, so sampling error can be estimated and a confidence interval computed Correct answer.

Explanation

In probability sampling each unit's selection probability is known and greater than zero. That known probability is what makes it possible to derive the sampling distribution of the estimator, quantify sampling error and attach a margin of error to a result. Non-probability designs such as convenience, purposive, quota and snowball sampling remain useful for exploratory work or hard-to-reach populations, but no defensible margin of error can be computed from them.

Question 26Research design, measurement and sampling

In stratified random sampling with a fixed total sample size n, how does Neyman (optimum) allocation (네이만 배분) differ from proportional allocation (비례배분)?

  1. Neyman allocation assigns each stratum a share of n proportional only to that stratum's share of the population
  2. Neyman allocation assigns an equal number of units to every stratum regardless of stratum size
  3. Neyman allocation abandons random selection inside each stratum in favour of purposive selection
  4. Neyman allocation assigns a larger share of n to strata that are both larger and more internally variable Correct answer.

Explanation

Proportional allocation sets the stratum sample size n_h in proportion to the stratum population size N_h alone. Neyman allocation sets n_h in proportion to N_h multiplied by the stratum standard deviation S_h, so an internally heterogeneous stratum receives more of the sample than its size alone would justify. This minimises the variance of the overall estimator for a fixed n. Where every stratum has the same internal variability, Neyman allocation reduces to proportional allocation.

Question 27Research design, measurement and sampling

In questionnaire construction, what is the function of a filter question (여과질문) placed before a contingency question (수반질문)?

  1. It confirms the respondent's identity before the interview begins
  2. It records how long the respondent took to answer the previous item
  3. It establishes whether the respondent is eligible for the follow-up item, so that ineligible respondents skip it Correct answer.
  4. It repeats an earlier item in reversed wording to detect careless responding

Explanation

A filter question screens for the condition that makes the follow-up item meaningful, for example asking whether the respondent is currently employed before a block of questions about working hours. Respondents who do not meet the condition are routed past the contingency questions. Filters stop respondents being asked items that do not apply to them, which would otherwise produce meaningless answers and higher break-off rates, so clear skip instructions are essential in self-administered questionnaires.

Question 28Research design, measurement and sampling

When is an open-ended question (개방형 질문) the better choice over a closed-ended question (폐쇄형 질문)?

  1. When the plausible range of answers is not yet known, as in exploratory research or a pilot study Correct answer.
  2. When the questionnaire is long and the field budget is tight
  3. When results must be compared across many survey waves using identical coding
  4. When respondents are known to have limited literacy or very little time

Explanation

An open-ended item lets the respondent answer in their own words, which is what is needed when the researcher cannot yet enumerate the realistic response categories: early exploratory work, a pilot used to build the answer list for the main questionnaire, or a topic where an imposed list would bias answers. The cost is that the answers must be post-coded, which is slow, expensive and requires intercoder reliability checks.

Question 29Research design, measurement and sampling

A researcher must explore how individual employees experienced being dismissed in a redundancy programme, including their financial anxieties. Why is an in-depth interview (심층인터뷰) preferable to a focus group interview (FGI) here?

  1. In-depth interviews produce results that can be generalised to the population
  2. In-depth interviews cost less per participant than focus groups
  3. Focus groups cannot be audio-recorded or transcribed
  4. The topic is sensitive and personal, so a one-to-one setting avoids the social pressure and self-censorship of a group Correct answer.

Explanation

In-depth interviewing suits sensitive, stigmatised or highly individual experiences. With no other participants present the respondent is less likely to censor themselves or to be swayed by dominant group members, and the interviewer can follow one person's account in depth. Focus groups are stronger where the researcher wants to observe how opinions form and are contested through interaction, as in concept testing on non-sensitive topics.

Question 30Fieldwork quality and data preparation

Which data collection method generally achieves the highest response rates, allows for complex skip patterns, and facilitates the use of visual aids, but entails the highest financial cost per completed interview?

  1. Face-to-face personal interview (대인면접조사) Correct answer.
  2. Telephone survey (전화조사)
  3. Mail questionnaire survey (우편조사)
  4. Online web survey (온라인/웹조사)

Explanation

Face-to-face personal interviewing (대인면접조사) yields the highest response rates and best rapport, handles long and complex surveys, and permits physical visual aids, but incurs heavy labor, travel, and supervision expenses.

Question 31Fieldwork quality and data preparation

What is the primary methodological weakness of traditional Computer-Assisted Telephone Interviewing (CATI / 전화조사) in contemporary public opinion research?

  1. Inability to obtain rapid survey turnaround within 24–48 hours
  2. Exclusion of cell-phone-only populations if relying solely on landline RDD, and declining contact/cooperation rates Correct answer.
  3. High cost compared to in-person household interviews
  4. Total prohibition of Computer-Assisted Interviewing software under telecommunications law

Explanation

Telephone surveys face severe coverage error if sampling frames rely on landline directories due to the proliferation of mobile-only households, compounded by caller-ID screening, spam filters, and plummeting response cooperation rates.

Question 32Fieldwork quality and data preparation

Which of the following data collection modes is most vulnerable to self-selection bias (자발적 참여 편향) and coverage error regarding older or digitally marginalized citizens?

  1. In-person home visit survey (방문면접조사)
  2. Central location intercept survey (출구조사)
  3. Online web/opt-in survey (온라인/인터넷조사) Correct answer.
  4. Standardized workplace audit (사업체 방문조사)

Explanation

Online surveys (especially opt-in web panels) suffer from coverage error (the digital divide excluding older, lower-income, and rural citizens lacking internet access) and volunteer self-selection bias, where highly motivated individuals choose to participate.

Question 33Fieldwork quality and data preparation

In survey fieldwork quality control, what procedure is conducted by field supervisors to detect interviewer fabrication (interviewer cheating / 면접원 부정실사)?

  1. Double data entry (이중 자료입력)
  2. Verification re-contact / Back-checking (실사 검증/사후 재확인) Correct answer.
  3. Imputation of missing values (결측치 대체)
  4. Factor extraction (요인추출)

Explanation

Back-checking or verification re-contact (실사 검증) involves supervisors contacting a random sample of completed respondents (typically 10–20%) via telephone or re-visit to verify that the interview actually took place and key demographic items were recorded truthfully.

Question 34Fieldwork quality and data preparation

Which category of survey error includes errors arising from non-response, flawed sampling frames, interviewer bias, and respondent comprehension errors, and can occur in both sample surveys and complete censuses?

  1. Sampling error (표본오차)
  2. Standard error of the mean (평균의 표준오차)
  3. Non-sampling error (비표본오차) Correct answer.
  4. Random sampling fluctuation (무작위 표본추출 오차)

Explanation

Non-sampling errors (비표본오차) encompass all errors that arise during research design, frame compilation, data collection, interviewing, non-response, data coding, and analysis. Unlike sampling error, non-sampling error exists in full population censuses as well as sample surveys.

Question 35Fieldwork quality and data preparation

In content analysis (내용분석), which analytical unit represents the specific linguistic element (such as a word, theme, character, or sentence) that is individually categorized and tallied?

  1. Context unit (맥락단위)
  2. Sampling unit (표집단위)
  3. Recording/Coding unit (기록단위 / 코딩단위) Correct answer.
  4. Systematic unit (체계단위)

Explanation

In content analysis, the recording unit or coding unit (기록단위 or 코딩단위) is the smallest specific textual segment (e.g., a single word, sentence, paragraph, or thematic clause) that is evaluated and classified into coding categories.

Question 36Fieldwork quality and data preparation

Which of the following statements correctly characterizes secondary data analysis (2차 자료분석)?

  1. It requires conducting original field interviews and primary questionnaire administration
  2. It utilizes pre-existing data collected by other researchers, government bureaus (e.g., Statistics Korea KOSIS), or institutions for a different primary purpose Correct answer.
  3. It can only be performed on qualitative interview transcripts and never on quantitative statistical databases
  4. It completely eliminates the possibility of measurement mismatch between research concepts and available variables

Explanation

Secondary data analysis (2차 자료분석) re-analyzes existing datasets that were previously gathered by primary investigators, administrative agencies (e.g., KOSIS microdata), or research institutes, saving time and money.

Question 37Fieldwork quality and data preparation

Which level of measurement classifies objects into mutually exclusive categories that possess rank order, but lack equal intervals between values and have no absolute true zero point?

  1. Nominal scale (명목척도)
  2. Ordinal scale (서열척도) Correct answer.
  3. Interval scale (등간척도)
  4. Ratio scale (비율척도)

Explanation

An ordinal scale (서열척도) possesses the properties of categorization and ranking (greater than, less than), but intervals between ranks are unequal or unknown, and there is no true zero point (e.g., class rank, military rank, customer satisfaction tiers).

Question 38Fieldwork quality and data preparation

Which of the following variables represents a ratio scale (비율척도) measurement?

  1. Body temperature measured in degrees Celsius (섭씨온도)
  2. Intelligence Quotient (IQ 점수)
  3. Monthly household earned income in Korean Won (월소득) Correct answer.
  4. Academic letter grades: A, B, C, D, F (학점 등급)

Explanation

Monthly household income in Won is a ratio scale (비율척도) because it possesses an absolute true zero (0 Won means no income) and meaningful arithmetic ratios (e.g., earning 4 million Won is exactly twice as much as earning 2 million Won).

Question 39Fieldwork quality and data preparation

What type of scale asks respondents to evaluate an attitude object along a continuum anchored at opposite ends by polar opposite adjectives (e.g., 'Clean ... Dirty', 'Efficient ... Inefficient')?

  1. Likert scale (리커트 척도)
  2. Semantic differential scale (의미분화척도 / 어의차이척도) Correct answer.
  3. Guttman scale (거트만 척도)
  4. Thurstone scale (서스톤 척도)

Explanation

The semantic differential scale (의미분화척도 or 어의차이척도), developed by Charles Osgood, measures the psychological meaning of concepts using 5-point or 7-point rating intervals bounded by bipolar adjective pairs.

Question 40Fieldwork quality and data preparation

Which cumulative scaling technique is designed such that items are ordered hierarchically by difficulty or intensity, so that an individual who agrees with item 4 is mathematically presumed to have agreed with items 1, 2, and 3?

  1. Guttman scale (거트만 척도 / 누적척도) Correct answer.
  2. Borgadus social distance scale (보가더스 사회적 거리척도)
  3. Likert summated rating scale (리커트 총화척도)
  4. Semantic differential scale (의미분화척도)

Explanation

A Guttman scale (거트만 척도), also known as a scalogram or cumulative scale, organizes items unidimensionally such that acceptance of any higher-threshold statement logically guarantees endorsement of all lower-threshold items.

Question 41Fieldwork quality and data preparation

In Guttman scalogram analysis, what minimum threshold for the Coefficient of Reproducibility (재생계수) is traditionally required to conclude that an item set constitutes a valid cumulative unidimensional scale?

  1. 0.50
  2. 0.70
  3. 0.90 Correct answer.
  4. 1.00

Explanation

The Coefficient of Reproducibility (CR = 1 - Total Errors / (Total Responses)) measures how accurately individual item responses can be reproduced from the total score. A CR of 0.90 or higher is the universally accepted benchmark for Guttman scalability.

Question 42Fieldwork quality and data preparation

Which type of measurement validity assesses whether the operational indicators of a construct correlate strongly with indicators of theoretically related constructs (수렴타당도) and correlate weakly with indicators of theoretically distinct constructs (판별타당도)?

  1. Content validity (내용타당도)
  2. Criterion-related predictive validity (예측타당도)
  3. Construct validity (개념타당도 / 구성타당도) Correct answer.
  4. Face validity (안면타당도)

Explanation

Construct validity (개념타당도 or 구성타당도) verifies that an instrument actually measures the underlying theoretical construct it purports to measure. It is demonstrated through convergent validity (수렴타당도) and discriminant validity (판별타당도).

Question 43Fieldwork quality and data preparation

An entrance exam is evaluated by comparing candidates' scores with their actual first-semester university grade point averages six months later. What form of measurement validity is being assessed?

  1. Concurrent validity (동시타당도)
  2. Predictive validity (예측타당도) Correct answer.
  3. Discriminant validity (판별타당도)
  4. Content validity (내용타당도)

Explanation

Predictive validity (예측타당도) is a subtype of criterion validity where test scores are correlated with an external performance criterion that is measured at a future point in time.

Question 44Fieldwork quality and data preparation

Which method of estimating reliability evaluates internal consistency by dividing a multi-item test into two equal halves (e.g., odd vs even items) and calculating the correlation between the halves, corrected by the Spearman-Brown formula?

  1. Test-retest method (재검사법)
  2. Parallel-forms method (동형검사법)
  3. Split-half method (반분법) Correct answer.
  4. Inter-rater reliability (검사자간 신뢰도)

Explanation

The split-half method (반분법) administers a test once, divides the items into two matched halves (such as odd and even numbered items), correlates scores between the halves, and applies the Spearman-Brown prophecy formula to estimate full-test reliability.

Question 45Fieldwork quality and data preparation

What is the relationship between measurement reliability (신뢰도) and validity (타당도)?

  1. Reliability is a necessary but not sufficient condition for validity Correct answer.
  2. Validity is a necessary but not sufficient condition for reliability
  3. A measure that is highly reliable is guaranteed to be valid
  4. Reliability and validity are completely independent and mutually exclusive concepts

Explanation

Reliability (consistency) is a necessary condition for validity (accuracy), but it is not sufficient. An instrument can produce perfectly reliable and consistent measurements of the wrong concept (high reliability, zero validity), but a measure cannot be valid without first being reliable.

Question 46Fieldwork quality and data preparation

Which coefficient of internal consistency is most widely used in survey research for multi-point items (e.g., 5-point Likert scales) to assess the degree to which all items measure the same underlying construct?

  1. Kuder-Richardson 20 (KR-20)
  2. Cronbach's alpha (크론바흐 알파) Correct answer.
  3. Guttman's lambda (거트만 람다)
  4. Kendall's tau (켄달 타우)

Explanation

Cronbach's alpha (크론바흐 알파 계수) mathematically represents the mean of all possible split-half correlation coefficients. It is the gold standard metric for internal consistency reliability across multi-category, continuous, and Likert-scaled items (commonly benchmarked at alpha >= 0.70).

Question 47Fieldwork quality and data preparation

In survey data processing, what is a codebook (코드북)?

  1. The interviewer's manual describing how to introduce the survey to respondents
  2. The list of sampled addresses issued to each interviewer
  3. The register of revisions made to the questionnaire during the pilot
  4. The document recording, for every variable, its name and position, the code assigned to each response category, and the codes reserved for missing data Correct answer.

Explanation

A codebook is the map between the questionnaire and the data file. For each variable it records the variable name and location, the numeric code standing for each response category, and the reserved codes for don't know, refused and not applicable. Without it the data file cannot be interpreted, re-analysed or archived, and it is the reference against which coders and data-entry staff are checked.

Question 48Fieldwork quality and data preparation

A survey asked 'What was the main reason you changed jobs?' as an open-ended item, and 500 verbatim answers must now be converted to numeric codes. What is the correct procedure, and what quality check applies?

  1. Assign the codes before fieldwork begins, because categories must always be fixed in advance
  2. Let each coder devise categories independently so that every nuance in the responses is captured
  3. Read a sample of the answers to build a mutually exclusive and exhaustive category scheme, then have two coders code overlapping cases and measure intercoder reliability Correct answer.
  4. Discard answers that do not fit an existing category so that the distribution stays clean

Explanation

Open-ended answers are post-coded (사후부호화). The analyst reads a subset of the verbatim responses, derives a category scheme that is mutually exclusive and exhaustive with a documented residual category, records it in the codebook and then applies it. Because assigning a verbatim answer to a category is a judgement, part of the material is double-coded and agreement between coders is measured; low agreement means the scheme or the coder training must be revised before the full file is coded.

Question 49Fieldwork quality and data preparation

In preparing a survey data file, why are distinct reserved codes normally assigned to 'don't know', 'refused' and 'not applicable' instead of leaving those fields blank or coding them all as zero?

  1. Because each represents a different reason for the value being absent, which affects how the case is treated in analysis and in non-response assessment Correct answer.
  2. Because statistical software cannot process blank fields at all
  3. Because reserved codes reduce the physical size of the data file
  4. Because respondents have a legal right to have refusals stored separately

Explanation

'Not applicable' means the item was correctly skipped and no information was lost. 'Refused' and 'don't know' are genuine item non-response, and separating them shows whether an item was sensitive or simply unknown. Coding all three as blank discards that distinction, and coding them as zero is worse, because a zero enters means and totals as if it were a real measured value. Reserved out-of-range codes such as 8 and 9, or 98 and 99, keep them visible and excludable.

Question 50Fieldwork quality and data preparation

During data editing, one record shows age 17 together with 25 years of work experience, and another record shows age 250. Which editing checks detect these two problems?

  1. A consistency check detects both errors
  2. A range check detects both errors
  3. A consistency check detects the first and a range check detects the second Correct answer.
  4. Neither is detectable before analysis; both appear only in the frequency distribution

Explanation

A range check (범위검토) tests whether each value falls inside the permissible values for that variable, so age 250 is caught by the range check alone. A consistency check (일관성검토) tests whether values of different variables can logically coexist: age 17 with 25 years of work experience is within range on each variable separately but impossible in combination. Double entry, in which the same questionnaires are keyed twice and the two files compared, catches keying errors that are both in range and internally consistent.