Sitonce
Country: US
Show exams for United States Hong Kong
Sign in

Market Research: Praxis Marketing Education 5561

Updated 12 min read
Key takeaway

Market research connects a marketing decision to relevant evidence.

  • For Praxis 5561, distinguish primary and secondary data, choose a collection method, identify the population and sample, and interpret descriptive statistics.
  • Check who responded and which denominator a percentage uses before turning reported interest into a claim about demand or sales.
On this page12 sections
  1. Start with a decision
  2. Turn a symptom into a research question
  3. Primary and secondary data
  4. Exploration, description and cause
  5. Choose a collection method
  6. Population, sample and sampling frame
  7. Write questions people can answer
  8. Process data without changing its meaning
  9. Mean, median, mode and spread
  10. Association does not establish causation
  11. Technology and a useful research report
  12. Original practice question

Start with a decision

Market research collects and interprets information to support a marketing decision. The useful starting point is a question the organization can act on: which customers need a service, why buyers abandon a purchase, or whether a proposed offer solves a problem. A vague instruction to collect more data leaves the population, measure and decision undefined. An impressive dashboard can still answer the wrong question. For Praxis Marketing Education 5561, connect the information need to a method, then judge whether the evidence supports the conclusion being offered.

The official outline places Market Research within Marketing-Information Management and Planning. It covers the need for data, collection methods and instruments, primary and secondary data, technology, data processing and descriptive statistics. Preparation therefore needs more than recognizing a survey. You should be able to explain what was measured, whose responses were collected and what a percentage actually describes. The examples below are original fictional teaching cases. Their numbers are practice inputs, not industry benchmarks, published research findings or a prediction of how many customers a real business will acquire.

Turn a symptom into a research question

Imagine a fictional refill shop whose monthly sales have fallen. The decline is a symptom; its explanation is still unknown. Possible causes include fewer visits, lower purchase frequency, stock shortages or customers choosing a different product. Asking whether customers like the shop cannot distinguish these explanations. A better first task is to examine transaction and stock records for the relevant period, then investigate the remaining questions. Define the product, customer group and time period before selecting a questionnaire. Otherwise, respondents may each answer a different version of the problem.

A research objective might be to identify the main reported reasons why recent customers stopped purchasing refills. That objective names a group and a behavior. It differs from estimating demand among people who have never visited the shop. Former buyers can explain their own experience, but they cannot automatically represent all potential customers. State the intended decision too: changing availability, adjusting an offer or improving communication. Research can narrow uncertainty and expose an incorrect assumption without guaranteeing that the resulting business action will succeed.

Primary and secondary data

Primary data is collected for the research project at hand. Secondary data was collected for another purpose and is reused. The distinction concerns the purpose of collection, not whether the information is numeric, recent or online. A new interview conducted for this project can be primary data. An existing company transaction file can be internal secondary data. A previously published population table can be external secondary data. Downloading information does not make it primary, and collecting a new set of answers does not automatically make it accurate or relevant.

Use available secondary evidence to refine the question before commissioning new collection. Check the definition, date, geographic coverage and population. A report about all households may not describe buyers of a specialized service, and an annual figure may conceal a seasonal change. Primary collection can be tailored to a missing question, but that advantage comes with time, cost and design responsibilities. Neither category is inherently superior. The strongest choice supplies relevant evidence at an appropriate level of quality for the decision, while making the limitations visible.

Original situationData classificationMain limit
Interview former buyers for the current investigationPrimaryTheir accounts may not represent nonbuyers
Reuse last quarter’s transaction recordsInternal secondaryTransactions do not explain every motive
Use an existing local population reportExternal secondaryIts population and date may not match the target market

Exploration, description and cause

Exploratory research helps clarify an unfamiliar problem, discover possible explanations and develop better questions. Descriptive research characterizes a population or situation, such as reported preferences or purchasing patterns. A causal investigation asks whether changing one factor changes an outcome. These purposes are related but different. An interview may suggest a reason for declining purchases; a survey may estimate how often respondents report it; a properly designed experiment may test an intervention. Do not convert the first suggestion directly into a measured market share or a proven cause.

Qualitative evidence includes descriptions, explanations and accounts of experience. Quantitative evidence includes numerical measurements or counts. The purpose and the data form are separate choices: an exploratory project can use existing numerical records, and a descriptive survey can contain open responses alongside closed questions. Counting comments can be useful when categories are defined consistently, but the count does not remove the meaning or limitations of the original answers. Avoid deciding a research design solely from whether its report contains words, numbers or a chart.

Choose a collection method

Interviews allow follow-up questions and can reveal how a respondent understands a problem. They take time, and the interviewer can unintentionally influence answers. A focus group enables discussion and interaction, but a confident participant may dominate or discourage disagreement. These methods are useful for generating explanations and examining language. A small discussion group is not automatically a representative estimate of the wider market. An original case with enthusiastic comments from selected members establishes what those participants said, rather than the proportion of all residents who would buy.

A survey uses a questionnaire to collect responses with a degree of consistency. It can reach many people, although access, wording and nonresponse affect the result. Observation records behavior, such as where shoppers stop or whether a displayed item is handled. It can show what happened without establishing the participant’s motivation. An experiment compares outcomes under different conditions; its ability to support a causal conclusion depends on assignment, controls and measurement. Match these strengths to the question. Asking people to remember a behavior and directly observing it are different forms of evidence.

Population, sample and sampling frame

The population is the group the research aims to describe. A sample is the subset from which information is collected. A sampling frame is the practical list or access mechanism used to reach that group. In an original case, the population is all eligible local households, but invitations go only to a shop’s email subscribers. That frame may omit households who have never bought anything, do not subscribe or cannot use email. A large number of subscriber responses does not by itself fix the mismatch between the population and the frame.

Probability sampling uses a selection procedure with known selection probabilities; simple random sampling is one example. Convenience sampling relies on people who are easy to reach, and voluntary responses come from people who choose to participate. These approaches can be useful for a limited purpose, but their limitations affect generalization. Larger samples can improve precision under an appropriate design without curing a biased selection process. Do not infer representativeness merely because a sample is large, respondents seem diverse or the results appear in a professional-looking report.

Stratified sampling separates a population into defined groups and samples within them. Cluster sampling selects groups or units such as locations, then studies people within the selected units according to the design. A sample is still a sample regardless of which approach selects it. For a scenario question, identify what is selected at each stage. Also distinguish selection from weighting: adjusting responses after collection can address specified differences under assumptions, but it cannot automatically recover the views of every group that was missed. Avoid treating any method name as a guarantee of unbiased evidence.

Write questions people can answer

A questionnaire should use a defined reference period and language respondents understand. “How many times did you buy a refill in the past four weeks?” is more precise than “Do you shop regularly?” The latter leaves regularly undefined. A question that combines price and service quality may produce an answer whose meaning cannot be separated. Ask about each concept independently when the decision requires that distinction. Leading wording also matters: describing an offer as convenient or excellent before asking for an opinion can push respondents toward agreement.

For closed questions, consider whether response options overlap, omit plausible experiences or require knowledge the respondent lacks. Someone who has never used a service should not have to rate its quality. A not-applicable route can preserve that distinction. Open questions allow an unexpected explanation but require a consistent approach to interpreting answers. Pilot the instrument with people similar to the intended respondents. If they interpret a phrase differently or cannot recall the requested information, revise the question before treating the answers as a reliable measure of the target concept.

Process data without changing its meaning

Data processing includes organizing records, checking duplicates, defining codes, examining missing values and preparing summaries. Keep a record of decisions so that another reader can understand how the reported result was produced. If several responses come from the same person, do not silently count them as several independent customers. If an answer is missing, distinguish it from a zero or a negative response. These values mean different things. Cleaning should correct identified problems under stated rules rather than delete inconvenient answers until a preferred conclusion appears.

An original survey invites 500 eligible contacts and receives 125 completed responses. Under this simplified definition, completion among invited contacts is 25%. Of the 125 respondents, 75 report interest in a proposed service, which is 60% of respondents. The 75 interested respondents also equal 15% of all invitations, but that does not establish that only 15% of the population is interested. The other 375 invitations lack completed answers. Write the numerator and denominator beside every percentage. Do not describe interest as a confirmed purchase or assume nonrespondents share respondents’ preferences.

Mean, median, mode and spread

Descriptive statistics summarize the observed data. The mean adds values and divides by their count. The median is the middle value after ordering them, or the average of the two middle values when the count is even. The mode is a most frequent value and need not be unique. These measures answer different questions about a distribution. A measure of center does not describe its full spread. A report should also show enough information to reveal unusual values, variation or groups whose experiences differ substantially from the overall summary.

For an original set of five transaction amounts, $10, $12, $12, $16 and $100, the total is $150 and the mean is $30. The median and mode are both $12. The range is $90, found by subtracting the smallest from the largest. The large transaction pulls the mean above most observed purchases. That does not make the mean mathematically wrong; it changes what the summary communicates. If the question concerns a typical transaction, explain why the median may provide a useful additional view rather than hiding the unusually large purchase.

Compare distributions as well as totals. In an original fictional week, one location records 100 visits and 20 purchases, while another records 40 visits and 12 purchases. Purchase conversion under the stated visit-based definition is 20% and 30%, respectively. The first location has more purchases but a lower conversion percentage. Neither observation alone identifies the more profitable location because prices and costs are absent. Visits also need not mean unique visitors. A good interpretation names the unit counted and answers the measure actually requested before recommending an action.

Association does not establish causation

Suppose sales rise in the week after a fictional shop changes its display. The timing is consistent with the display helping, but a holiday, stock improvement or different traffic could also explain the result. A before-and-after comparison without a suitable design does not isolate the display’s effect. An A/B label is not enough either: showing version A on one busy day and version B on a quiet day changes more than the display. Look for comparable conditions and a defensible assignment procedure before accepting a causal conclusion.

In a simplified original experiment, comparable visitors are randomly assigned to two messages, and purchase outcomes are measured consistently. That design provides stronger evidence about the message than a voluntary survey of people who noticed it. The interpretation still depends on implementation, sample size, uncertainty and the population studied. A reported difference is not necessarily a practically meaningful or statistically convincing effect. For an exam scenario, identify the design feature that improves the comparison and the limitation that remains. Do not claim that research removes uncertainty or guarantees future customer behavior.

Technology and a useful research report

Technology can distribute questionnaires, record transactions, organize responses and create summaries. It can also make a flawed process run faster. A dashboard with duplicate events, inconsistent definitions or a mismatched date range remains misleading. Distinguish event counts from unique people and reported interest from completed transactions. When a tool groups open comments, check the category definitions and some original answers rather than assuming the automated label is correct. For this conceptual article, avoid unsupported claims about privacy compliance or a platform’s current tracking capabilities; those require separate current verification.

A useful report states the decision question, population, collection method, dates, sample and limitations before presenting results and recommendations. Separate what was observed from what is inferred. In the refill-shop example, the finding might be that surveyed former buyers often reported unavailable stock. The recommendation could be to investigate replenishment and test an improvement. That recommendation is supported as a next step, not a promise that sales will recover. A decision maker should be able to see how the evidence connects to the action and which questions remain unanswered.

Original practice question

Example question

A fictional firm surveys 80 current subscribers about a proposed new service. Sixty say they are interested. Which conclusion is best supported?

  1. 75% of all local residents will buy the service
  2. 75% of the surveyed subscribers report interest
  3. The service will cause sales to increase by 75%
  4. Nonrespondents have no interest in the service
Answer: B. Sixty divided by 80 equals 75%, describing the surveyed subscribers’ reported interest. The case does not establish representativeness of all residents, purchases, a causal sales effect or nonrespondents’ views.

To practise interpretation, rewrite each rejected option as a separate question requiring additional evidence. Estimating local demand needs an appropriate population and sample. Forecasting purchases needs more than an interest statement. Establishing a causal sales effect needs a suitable comparison. Describing nonrespondents’ views requires information that the survey did not provide. This exercise makes the evidence gap concrete. On the official test, read each item’s selection instructions carefully; selected-response questions can require more than one choice. The original question here asks for one best-supported conclusion only.

Common questions

Is an online survey always primary research?

No. A new survey collected for the current project is primary data. Reusing answers collected for an earlier purpose makes them secondary data for this project.

Does a large sample guarantee representativeness?

No. Sample size does not automatically correct a frame that misses part of the target population or a selection process dominated by volunteers.

Are descriptive and quantitative research identical?

No. Descriptive refers to a research purpose; quantitative refers to numerical data. A descriptive investigation may include numerical summaries and open responses.

Does reported interest measure purchases?

No. Interest is an expressed response. Actual purchasing is a different outcome and needs its own evidence.

Should every report use the mean?

No. Choose summaries that fit the question and distribution. The mean, median, mode and spread can reveal different aspects of the same observations.