• Subscribe

  • Back

    How to design Cost-Effective, Powerful Market Research – Data Analysis (Part 3 of 3)

    Stephen Walker

    Now that you understand samples and surveys, it’s time to discuss what you do with all that data you’ve collected.

    What is the overall response rate? Response rates can run the gamut, so how do you know if you’ve received enough responses to make decisions based on your data?

    Consider your study objective.

    Do you merely want an overview of the issue? If so, a lower response rate might be sufficient to give you directional data. However, if you want to segment the data, users versus non-users for example, your response rate will need to be high enough to provide reliable numbers in your segments.

    It is helpful to consider response rate in advance of data delivery, especially if you already know you’re surveying a hard-to-capture or low incidence sample. In general, a sample size less than 50 is “qualitative.” The results are useful from a directional standpoint but lack the rigor of a larger quantitative sample size.

    Providing context is critical. Numbers are just numbers until you put them into a relevant context. For example, a survey question asks which brands of crackers respondents have purchased in the past month. Your brand comes up with a purchase rate of 25%. Is this good? The only way to determine if this number is one to cheer about is to look at it in context with the other brands mentioned. This may seem obvious, but data can be misleading and incorrect assumptions can be made if context isn’t part of the equation.

    Statistics is a thing. We are a market research firm, so statistics are important! There are two main types of statistics to consider in survey research. These are descriptive and inferential.

    • Descriptive Statistics are the “who” of the data set. These illustrate, or describe, the data. Examples of descriptive statistics include overall sample size, gender percentages, average age, etc. Descriptive statistics are the bones that inferential statistics hang on.
    • Inferential Statistics allow researchers to presume, or infer, something about the study respondents. Further, if the sampling is done correctly, assumptions can be made of the population at large that the smaller sample represents. Statistical significance is a result of analysing data using inferential statistics.

    Data analysis has come a long way. It used to be necessary to have a variety of data analysis tools available to run a wide assortment of statistical tests. Toluna sought to demystify statistics and make data analysis as easy-to-execute as possible by creating TolunaAnalytics, a one-stop research resource software solution.

    It’s a visual platform that refreshes as survey data is collected, so changes are in real-time. The detailed results presented by TolunaAnalytics are automatic with an insights presentation deck just a print command away. The program also allows data from other sources to be included creating a holistic data analysis. Deliverables beyond the presentation include data tables, verbatims from open ends/other specify questions and access to raw data. You can learn more about TolunaAnalytics at http://www.toluna-group.com/en-gb/products/toluna-analytics.

    Now you have a complete picture from sample selection to data analysis. If you missed the earlier posts in this series, you can find Sampling here and Survey design here.