For two main reasons:
- To reduce the risk of reading the survey results in a wrong way and making wrong decisions
- To avoid missing important information due to a superficial survey report
Marketing research as a discipline applies the scientific approach to investigate situations we may be interested in. This scientific approach has rules, mainly from statistics. When applied in a rigorous manner these rules ensure that we collect reliable survey data and read it appropriately.
Unfortunately, surveys are often conducted by improvisation, or the information is misused, or both. An infamous example is that of the Brexit survey YouGov conducted to predict the outcome of the consultation. The survey wasn’t (completely) wrong; The problem was that the information channels broadcasted incomplete information because they did not read the survey results correctly. Surveys are an extremely useful tool. But as with any other tool, we must make sure we use it correctly.
LogRatio is the only, fully automated solution able to create survey reports like marketing research agencies do. We apply the scientific approach to surveys in a rigorous and error-free manner.
- Numerical survey report. An Excel file containing all the numbers you need, and
- Verbal survey report. A PDF file with the written interpretation in plain English of the numerical survey report.
You can collect survey data yourself, for instance using a form at your website; or you can use an online survey data gathering provider like Google forms, SurveyMonkey, and others. There are also many datasets online; take a look at Dataset Search.
To test LogRatio, we make two datasets available. Go to Try LogRatio… and select Test File 1 or 2 from the listbox “Select a survey file”.
A few rules must be followed in order to arrange the survey data for LogRatio. For all details, read chapter “Survey data: How to format your input file” from the LogRatio User’s Guide.
Your input data to LogRatio can be in any language. However, the PDF written report will be in English while the Excel file will be in the language of the input file. Special characters, like ö, ä, ç, ñ, ǽ, and others, could be misinterpreted and reported in a wrong way.
LogRatio is free of charge for unlimited use. We promote the professional use of surveys to contribute to educating a generation of technically savvy marketers and business analysts.
Like other free services, we at LogRatio operate on a tight budget. To support our development efforts, you can help us with a donation.
It is for a good purpose.
Yes. All you do is fill in a few simple fields and select the dataset you want analyzed.
LogRatio does the rest, including a professional interpretation of survey results written in plain English.
Yes. LogRatio applies deep learning algorithms to transform the input data in analysis-ready format, it uses natural language processing techniques to treat the answers to open-ended questions, as well as other deep learning algorithms to qualify variables, sort answer options, and more.
We are developing algorithms to make surveys easier, deeper, and faster.
Anywhere from a few seconds to a few minutes.
LogRatio applies efficient and appropriate algorithms to:
- Make the raw survey data analysis-ready. Raw survey data must be coded and edited in a way that allows conducting the necessary survey analyses to extract all information contained in the data.
- Analyze the survey data in a professional manner. Cross tables are the bread and butter of marketing research agencies. LogRatio goes even deeper, also presenting a Sample Size Analysis, Descriptive Statistics for each variable, Correlation analysis, and two kinds of Cluster analysis.
- Interpret survey results. Together with the numerical analysis, LogRatio provides a verbal report written in plain English that summarizes the results of the survey.
LogRatio survey reports are made of two parts:
- The written report. This is a PDF file written in plain English that interprets the results of LogRatio survey analyses.
- The numerical report. This is an Excel file containing the figures of the five standard survey analyses LogRatio conducts with the raw data of a survey:
- Sample size analysis. This sub-report computes the sampling error of a survey and shows a matrix of sample sizes using different levels of confidence and sampling error.
- Correlation matrix. This analysis shows how two closed-ended questions vary together. It is useful for predictive purposes and to identify and validate any models hidden in the data.
- Descriptive statistics. This sub-report supplies descriptive information about the variables (questions) of the survey. It is useful for interpreting the survey results as well as when building models such as simulation models and others.
- Cluster analysis. This is a statistical technique used for segmenting items or people in homogeneous groups. See also What is a cluster analysis? on this FAQ page. LogRatio uses it to identify homogeneous groups of respondents in your survey data.
- CrossTabs. Contingency tables, also called CrossTabs or cross tables, summarize the joint counts by answer class of two variables. They are the core survey analysis technique used by professional marketing research agencies. LogRatio tables include significance test, error levels, and much more information about your survey.
LogRatio’s professional survey report can teach a lot about surveys and how to use them correctly. This is why we say that LogRatio makes every user a survey expert.
From an operating standpoint, a survey must include at least 5 respondents for LogRatio to be able to generate a report. However, to produce reliable survey results, more interviews than 5 are required. As for open-ended questions, at least 30 answers to a single question should be available in order for a survey to deliver meaningful results.
We encourage LogRatio users to think in terms of “sampling error” rather than “number of interviews”. First determine what level of accuracy the survey should deliver in order to support the decisions you want to make. The sample size depends on the accuracy you want, and not the other way around.
Yes. LogRatio uses natural language processing techniques to treat the answers to open-ended questions. It codes the data in analysis-ready format, creates cross tables, performs other analyses, interprets it all, and summarizes findings in a survey report written in plain English.
However, as of January 2022 LogRatio’s development team is not happy with the quality of our open-ended coding, so we decide to restrict data coding to closed-ended questions only.
The automated coding and analysis of open-ended questions will be reactivated soon.
LogRatio applies the proprietary Entropy Auto-Recovery (EAR) algorithm to produce survey reports with no human intervention. EAR uses advanced methods from statistics, deep learning, and natural language programming to analyze the answers to open-ended questions. It begins by identifying the kind of questions in a survey, for instance closed-ended single answer, closed-ended multiple answers, open-ended, and so on. It then assigns a whole open answer, or only part of it, to one or more of the classes it creates according to all the answers respondents gave to one open question.
The automated analysis of open-ended questions is a complex process. We are aware that LogRatio does not yet deliver perfect results, although they are continually improving.
Rest assured we are working hard to make LogRatio the fastest, most accurate, and most convenient way for you to analyze text answers.
Yes, of two kinds:
- CrossTabs between all survey variables.
- CrossTabs between survey variables and respondent groups as found with the Cluster Analysis conducted on all closed-ended questions of your survey
For each pair of questions, LogRatio makes two CrossTabs: One with column totals and one with row totals. Although the counts do not change between tables, the percentages (proportions) and the results of significance tests do change. For a professional survey report both views are necessary.
True, this is a lot of CrossTabs, but as the saying goes: The devil is in the details! And you want to have them all because it is easier to get rid of useless tables than to produce a missing one.
Yes. LogRatio runs two separate cluster analyses: with the variables of the survey and with the respondents:
- The former cluster analysis helps to understand any models hidden in the survey data, for instance, such as the causal relationship between some survey variables and a derived construct, say, “Satisfaction”.
- The latter kind of cluster analysis finds groups of respondents that, according to the questions of the survey, may share similar profiles and therefore may reply in a similar way to the same stimulus, such as a promotional offer, for instance.
Cluster analysis is a multivariate statistical technique used for segmenting items or people in homogeneous groups.
Segmentation needs arise frequently in strategic marketing linked to differentiation matters, which relate to the strategic positioning statement and require behavioral and attitudinal data, for instance from surveys. But it is also often used in the analysis of performance of affiliate companies, points of sale, distributors, and so on.
We are working on it. For now, LogRatio delivers a survey report made of two files: (1) an Excel file holding the numbers and (2) a PDF file with the interpretation of the numbers written in plain English.
Yes. And if you have your own unique needs and you think LogRatio does not fit, Contact us. Chances are we can help you.