Sampling Plans and Data Preparation in Market Research

Posted on Feb 3, 2025 in Mathematics

Topic 5: Sampling Plan and Data Collection

Sample or Census

Differences: A census is a complete enumeration of all individuals and asks them what their opinion is. The problem is that it is impossible to survey the entire population. Sampling involves choosing people to learn something about the population.

Sampling Design Process

  1. Determine the Target Population
  2. Determine the Sampling Frame (how to identify them)
  3. Select a Sampling Technique

Sampling Techniques

Non-Probability Sampling Techniques

  • Convenience Sampling
  • Judgmental Sampling
  • Quota Sampling: This technique tries to use some framework to define the sample.
  • Snowball Sampling: Used when the subject is very complex.

Probability Sampling Techniques

  • Simple Random Sampling: All elements have the same probability of selection to participate in the research.
  • Systematic Sampling: The first element of the sample is selected using the random sample approach.
  • Stratified Sampling: A two-step process to partition the population into subsequent populations (or strata). The strata are supposed to be homogeneous within them and heterogeneous between them.
  • Cluster Sampling: A two-step process to create clusters. The groups should be mutually exclusive and collectively exhaustive. The groups are homogeneous between them but heterogeneous within. The idea is to have several subgroups that can be compared.
  • Other Sampling Techniques
    • Sequential Sampling: A small group is chosen, and more are sampled if necessary to achieve certain statistical criteria.
    • Double Sampling: A sample is chosen to conduct the research, and then a subsample of the sample is selected.

Determine the Sample Size

Sample Size: The number of elements to be included in the research. Qualitative and quantitative factors should be included in their calculation. The qualitative factors are the importance of the decision, the nature of the research, the number of variables, the nature of the analysis, sample sizes used in similar studies, incidence rates (number of people who will answer the questionnaire), completion rates, and resource constraints.

Execute the Sampling Process

  • Decide how the decisions about the population, sampling unit, sampling frame, sampling technique, and sample size are going to be implemented.
  • This is necessary because the persons that design the research and the person that collect the data are not the same.
  • In the majority of research designs, more than one person is involved.

Validate the Sample

  • The sampling frame error needs to be addressed (validate the population).
  • The usual process is to use filter questions, like screening participants during data collection.
  • The screening can be applied with filter questions about:
    • Demographic characteristics
    • Knowledge of the product
    • Usage of the product

Topic 6: Data Preparation

The Data Preparation Process

Preliminary Plan of Data Analysis

What data do we want to analyze? This was previously done when we defined our Research Approach:

  • What is the Marketing Decision Problem?
  • What is the information needed in our Marketing Research Problem?
  • What are our objectives and hypotheses about the Research Process?

Based on these, we have an idea of the type of information that we need, the nature of the information (Exploratory, Descriptive, or Conclusive), and what answers we should obtain.

Editing

This is done to correct in order to go deeper. It is the review of the questionnaire for increasing accuracy and precision. It includes reviewing answers that are illegible, incomplete, inconsistent, or ambiguous.

The idea is to try to make some type of arrangement to recover the data. Unsatisfactory Responses:

  • Return to the field: When respondents are easily identifiable and the sample size is small.
  • Assigning Missing Values: When the proportion of these questions is small, the number of questionnaires is small, and they do not affect key questions.

Coding

Take the questionnaire and decide which ones will be coded. Many questionnaires and data entry software nowadays are automatic. In any case, a good coding method for the questions is suggested to make interpretation easier. The basics imply assigning a number to each possible question. The open-ended questions should be included in another column (for example, the option”othe”).

The idea is to use simple numerical codes. This comes from the “ASCII” file tradition, which is a type of coding used for computer format that requires low memory to store.

Codebook

An encyclopedia of our research, to interpret the data analysis that explains the names of each question and the possible numbers that answer each question. Then, we have our codes.

Transcribing

Keying the coded data into computers. If we use CATI or CAPI, this step is already done automatically when we are recording the interviews.

Cleaning the Data

Thorough and extensive checks for consistency and treatment of missing responses.

Statistically Adjusting the Data

  • Weighting: Accounting for non-response, giving different weights to the responses based on the response rates.
  • Variable Respecification: The transformation of variables or creation of new ones, so they are more consistent with the purpose of our study.
  • Scale Transformation: A manipulation of scale values to ensure compatibility with other scales or otherwise make the data suitable for analysis (e.g., standardization).

Select Data Analysis Strategy