Paper tables - not automatable

Before Christmas I spent some time working with tables on a seven-country study, and going Grrr. It seems to me that the main purpose of tables is to provide the aggregated data that populates a visual report. I’d be interested to hear from anyone who disagrees, but in my experience the research results are really only analysed once charted, making the tables a passing phase on the reporting process flow and a repository for looking things up rather than the tool which is used for interpreting the data. So, why not design them so that they are easier to use for reporting? Here are some ideas for doing just that.

First: use the text you want to see in the report, not the text from the questionnaire. Text in tables should be abbreviated and converted into PowerPoint chart-friendly statements, not whole sentences lifted from the script which will never appear in the report itself. Here’s a version from real life of a table title:

“S10. S10. Now, looking at this list, please tell me which of these products/solutions do you know, even if only having heard about them? * QUESTYPE. Questionnaire Type + SEC + S3. S3. In which city do you permanently stay + S2. gender + AGE + Awareness_type + Usership_Type + Usership_Criteria_2 + Usership_Criteria_1 + Stated_Condition”

In the report, you would probably want “Prompted Awareness of Products” with possibly the question identifier. That’s all. That should be the question text in the table as well, so that you can use the text that’s there either by copying and pasting or lifting it via a simple automation program.

Second: apply the same reasoning to column headers. If you want it to say “All respondents” – specify that in the tables, plus the relevant, abbreviated wording for the subgroups you’ll be using. I still see columns where words have been separated oddly with hyphens so that the header text fits nicely over the column. We really need to stop doing that. It’s much more important that the text matches what’s required in the report than looks neat on the page. I am assuming here that people are looking at tables in Excel or a similar copy and pastable-format – not a pdf or a printed page. Does anyone print tables any more?

Third: – and related to above – don’t separate a long banner into shorter banners and have them one below each other on the same sheet. Have all the columns you need in one long banner that stretches off to the right. That way each question is only on the sheet once, which makes it much easier to work with. We don’t have to worry any more about whether it fits onto a page. We can just freeze the first column and scroll across these days, can we not.

Fourth: Keep the number formats consistent. Most research reports contain column percents plus a few mean scores and unweighted base sizes. Remove everything else, particularly absolutes. If you’re working with statistical significance indicators, have them output into a different sheet so you don’t have to remove them manually if you don’t need them. You want to be able to highlight a block, including text and figures, and copy and paste it all as one movement, not constantly backwards and forwards because there are blank rows or absolutes mixed in with percentages. Speaking of which, decide whether you want percentage signs or not, thoughout the end report, and have them in the tables, or not, accordingly. I’m a little fed of up pasting into a blank Excel sheet and doing (=B1*100) 15 times in a day…

Fifth: Give each question a title – a single title, not two parts. That has two advantages – it means that all your tables will exactly match in terms of layout and the column headers will always be in the same position relative to the title text etc. And it means the title will be more useful in the report. In other words, don’t have this:

“Q2X1. Q2. Please tell me how much you agree or disagree with each of the following statements about your attitudes towards life and personal care in general.

I buy environmentally friendly products”

Have this:

“Agree/Disagree: I buy environmentally friendly products”

Sixth: Order scale questions how you will need them in the report. This will need a bit of forethought in terms of the PowerPoint charts you are using. For example you may want to work negative to positive left to right, with DK/NA at the end. Or the other way round. Whichever it is, get the tables set up to match so that you are not constantly repeating the same copy/paste operation.

Seventh: Don’t rank or zero suppress, particularly if you will be working with more than one set of tables across a project. the more similar they are, the better. Besides, you will probably want to rank on different things depending on what you are reporting, not just the total column all the time, so better to do this in the chart datasheets or via automation.

Eighth: If you have an index page at the front with tables on different worksheets, this can help navigation enormously. Make sure the names of the tabs relate to the question and will always match that question even if new questions are added. ie, not just Table1, Table2 etc, with the danger that adding a new table will mean that they no longer match the content of the sheet.

Usable and automatable tables will be a lot easier to work with manually, and also very easy to use in an automation package without a lot of tweaking and moving stuff about, thus allowing more time to actually look at what the results mean rather than spending tedious hours just getting to a stage of having charts.