Does ad hoc mean the survey happens only once forever?

It means no repetition is planned in the stated definition. An organization can later decide to repeat or redesign it, but that is a subsequent decision.

Can an ad hoc survey use random sampling?

Yes. The label concerns repetition, not the method of selecting respondents.

Is an ad hoc survey the same as a cross-sectional survey?

Not exactly. Cross-sectional describes observation at a point or period, while ad hoc describes the lack of planned repetition. The terms refer to different aspects of design.

AI Glossary / Statistics and probability / Beginner

Ad hoc survey

Definition

An ad hoc survey is a survey undertaken without a plan to repeat it. It usually addresses a particular information need arising at a particular time, rather than forming part of a scheduled statistical series. The label describes the absence of planned repetition; it does not mean that the survey is informal, improvised or methodologically weak.

What it means

A business might commission a single survey to understand demand for a proposed service or to investigate an unexpected problem. The survey can be carefully designed, use a probability sample and produce documented estimates even though no second wave is planned.

Ad hoc is therefore different from a sampling method. It does not specify whether respondents are randomly selected, whether data are collected online or by interview, or whether the study is descriptive or analytical. Those are separate design choices that determine what conclusions the results can support.

How it works

Begin with the decision or research question and check whether existing information can answer it. If a new survey is needed, define the target population, reference period, variables and intended outputs. Choose a sampling and collection approach consistent with those objectives and available resources.

Develop and test the questionnaire, plan contact and follow-up procedures, and specify how responses will be processed and analyzed. A one-time survey still needs clear eligibility rules, quality checks and treatment of missing responses.

Because repetition is not planned, the survey may not provide a time trend or support before-and-after comparisons. Preserve its questionnaire, sampling description, fieldwork dates and analysis decisions so users can interpret the findings and any later study can assess comparability. A later decision to repeat the survey does not retroactively create an original panel design.

Three practical examples

1. A company conducts a one-time survey of eligible customers to assess interest in a proposed training service before deciding whether to develop it.

2. A local organization surveys members after an unusual service disruption to understand the problems experienced during that specific event.

3. A research team commissions a targeted survey to fill a documented evidence gap that an existing recurring survey does not cover.

Think of it this way

Think of a photograph taken to answer a specific question, rather than a camera programmed to capture the same scene every month. The photograph can be carefully composed and accurately labeled. What it does not provide by itself is a sequence showing how the scene changed over time.

Worked example

Consider an original hypothetical ad hoc survey of 2,000 customers on a complete list. The team draws a simple random sample of 200 without replacement, and all 200 respond. Of these respondents, 120 say they would consider the proposed service, giving an estimated proportion of 120/200 = 0.60.

Using the usual approximate standard error for a proportion with a finite-population correction gives sqrt[(0.60 x 0.40 / 200) x (2000 - 200)/(2000 - 1)], approximately 0.0329. A rough 95% normal interval is 0.60 plus or minus 1.96 x 0.0329, approximately 53.6% to 66.4%.

This interval reflects sampling uncertainty under the stated assumptions. It does not cover misleading wording, incomplete customer lists or the gap between stated interest and actual purchase. The one-time design also supplies no measured trend in interest.

Illustrative case study

Fictional case study: Harbour Learning commissions a survey about a proposed weekend workshop. There is no scheduled repeat. Its team first plans to share an open link on social media, then realizes that the intended decision concerns existing subscribers rather than whoever chooses to click the link.

The team revises the target population and sampling approach, tests the questions and records the collection dates. Its report clearly distinguishes respondents' stated interest from confirmed registrations and avoids comparing the result with an older survey that used a different audience and question.

This fictional example illustrates design alignment. It does not establish demand or predict attendance; those decisions would require the actual evidence and the limitations of the implemented survey.

Common misconceptions and mistakes

An ad hoc survey is not necessarily a convenience sample, a pilot or a poorly planned questionnaire. Conversely, giving a survey a specific purpose does not guarantee representativeness.

A one-time observation cannot by itself demonstrate change over time or a causal effect. Asking respondents to recall an earlier period does not turn the study into a prospective longitudinal survey. Another mistake is attaching a conventional sampling margin of error to an unrestricted volunteer sample without a justified inference method and clearly stated assumptions.

Business use

Ad hoc surveys can help businesses answer focused questions quickly enough to inform a decision. Their flexibility is useful when the topic is new or absent from routine reporting, but the study should still be designed around the required inference.

Specify who the results describe and what decision they can reasonably inform. Plan subgroup comparisons before choosing sample sizes, because an adequate overall sample may leave small segments poorly measured. Include the exact questions and methodological notes with findings. If ongoing monitoring becomes necessary, design a repeatable program deliberately rather than assuming later one-off surveys will automatically be comparable.

Check your understanding

A carefully sampled survey is commissioned to investigate a single event, with no future waves planned. Does its rigorous design prevent it from being ad hoc?

Answer: No. It is ad hoc because repetition is not planned. Methodological quality and repetition schedule are separate characteristics.

Related terms

cross-sectional survey
repeated cross-sectional survey
panel survey
target population
sampling frame
probability sample
nonresponse and questionnaire design

Learn next

write a short survey specification containing the decision, target population, reference period and intended estimates. Identify which conclusions require repeated observations and which can be supported by a single well-designed survey. Then assess existing data before collecting new responses.

Sources

Primary sources: OECD Glossary of Statistical Terms, Ad hoc survey, citing the United Nations survey-reporting definition: https://www.oecd.org/content/dam/oecd/en/publications/reports/2008/09/oecd-glossary-of-statistical-terms_g1gh9ad7/9789264055087-en.pdf

Statistics Canada, Planning a survey: https://www150.statcan.gc.ca/n1/edu/power-pouvoir/ch14/spd_0001-eng.htm

Statistics Canada, Questions to ask about surveys: https://www.statcan.gc.ca/en/customized-products-services/custom-surveys/questions-to-ask

Previous: Ad hoc retrieval Next: ADALINE