An Introduction to Market Research Tables – Request for Feedback
Most market research textbooks (for example Malhotra) do not really cover the sort of tables we use in market research. The main reason being that there is almost no academic justification for the way we market researchers use tables. We tend to use data that ought not be in tables, frequently use the wrong types of statistical tests, and our tendency to cross everything by everything is more like a fishing expedition that a scientific evaluation of hypotheses.
However, whatever the rights and wrongs of the issue, tables are probably the most widely used tool in commercial market research. A few years ago, I wrote some notes on how to use market research tables and I currently updating my notes with the intention of publishing them as a book or at least as an eBook.
I have reached the first draft stage and would really appreciate some feedback. In particular I am looking for feedback from:
- People who are unsure how to use tables – do these notes help? What else would you like to know?
- People who use tables regularly – what have I missed? What is wrong or inadequately correct?
- People who supply tables – what have I missed? What is wrong or inadequately correct?
You can download a copy of the current PDF by clicking here.
You can either supply your input directly as comments to this post, or you can send them to me (the address is in the PDF).
Many thanks
Ray
3 thoughts on “An Introduction to Market Research Tables – Request for Feedback”
Comments are closed.
Hi Ray, Nice description of tabulations and how they work. A handful of comments. Firstly, because people have different terminologies, I wondered if you need to emphasize, “here we’ll use banner” and put in it bold (or actually have a diagram with the name parts that are commonly used) – it can be confusing for say a dp house to use cross-break and the client to use banner if you’re a new researcher. For instance, I’d refer to your ‘Nets’ as ‘Summary lines’. I’ve also heard statisticians refer to tabs as contingency tables
I wondered if the idea of a stack of tables would also help. So most tabs are many pages of tables with a main banner and then break out tabs for special groups. So the idea of a main table set with one fixed banner for every question, then insert extra tables for drill downs or to check cross-relationships.
As a supplement to Nets, you might mention combination tables – eg a Yes, No question which acts as a filter, might then be combined with the yes responses to the filtered question so it gives the whole base and adds back up to 100%.
Similarly, for awareness questions I’d tend to roll the awareness questions up – so spontaneous, then a table for spontaneous+prompted as a classic for don’t just blindly tabulate every question. Or for pricing, to report the data cumulatively than price-by-price.
Having a BMRB background, TGI uses indexes all over the place. If you’re on the media side you’re likely to run into some form of indexing so it might be worth explaining how it works and what the calculation is.
On the weighting, you might also link it to oversampling (which is somewhat topical for the POTUS elections), as it can be more than just tweaking, particularly from a Rep+Boost sample design.
Mean scores are a topic in their own right. Clients not used to them can get very confused. The centre point of a scale scored 1 to 4 scale is 2.5 for instance, and in use it’s harder to push to the extremes than to the centre. You also don’t have to use a low-value scoring system for instance scoring 0 to 100 gives a mean as a much easier to read integer.
I don’t know if it’s appropriate, but the old trick of adding up all the percentages for a multiple question gives the average number of items selected might have a place.
And on numerics, the use of banding ranges (and checking for outliers) – the use of quartiles, quintiles and deciles, or 80-20 rules.
By the way, do people still really use dp or spec writers to produce tabs? I’ve been under the impression that the tools are so easy now, researchers can do it themselves faster than writing out a spec sheet.
Ray, this is some excellent work. I once had the challenge to explain to some web developers what MR tables are all about. I whish I could have such a book at hand.
One recommendation from my side: Please try to mention the meanings of SPSS-style asterisks (*** / ** / …) behind cell values (in the context of significance testing). The fact alone that asterisks mean a lot in the context of MR tables tells us quite a story about MR people. 😉
Also, manuals of Quantum (now part of the IBM SPSS family) and Vovici (now Verint) have turned out to be quite useful when trying to figure out where some of those odd table standards can be derived from.
Thank you so much for providing the detailed information and data on Market Research Tables. I agree with the statement that most of the textbook do not really cover the sort of tables we use in market research as I have personally read many of those books.
It is unavoidable fact that tables are probably the most widely used tool in commercial market research. After reading the PDF, one can understand why it is essential to understand the importance of learning insights about market research tables. I have found similar stuff on https://www.researchoptimus.com/market/data-analysis.php
You can go through this.It might add to your knowledge.