Even accomplished strategic thinkers can sometimes be slow to appreciate innovation. Take Napoleon’s reaction upon being told of Fulton’s steamship. “You would make a ship sail against the wind and currents by lighting a bonfire under her decks?" he asked. “I pray you excuse me. I have no time to listen to such nonsense.”
Consultants and pollsters ought to seriously consider recent innovations in polling analytics. While toplines, crosstabs, and static PowerPoints have served us well for more than a quarter of a century, turning a blind eye to new tactics will spell disaster for our industry. As Napoleon demonstrated, hey, what can possibly go wrong with the old way of doing things?
Interactive analytics (IA) lets consultants ask strategic questions of their polling data, and does so in real time. The premise is that the broader the functionality of the analytics, the better your strategic analysis will be.
For example, just two mouse clicks duplicate every function performed by toplines and a stack of crosstabs, plus a lot more. The first mouse click tabulates the survey results for any given survey question. A second click commands the software to instantaneously pass through the entire data file, crosstabulating the selected question with every independent variable.
IA then calculates the statistical significance of each crosstabulation, calculates interpretive indices, sorts the variables in descending order of importance, and displays the entire list. With a third mouse click you can display any crosstabulation from the list as both a table and a graphic, along with the exact wording of the question plus statistics pertaining to that crosstabulation.
The big benefits of IA come as it answers strategic questions that simply cannot be answered with traditional toplines and crosstabs. And interactivity means that your questions are answered easily and at the moment the question arises. A few examples:
1. Voter Profiles
Just who are the voters who were undecided on the first ballot question? Who are the respondents who found a particular argument convincing?
Using IA you simply click on the label of any response within any question in the survey. IA then builds a profile of the respondents who gave that particular response. The profiling variables can include demographics, psychographics, and behavioral variables.
2. Voter Migration
Which voters moved in our direction from the first ballot test to the second ballot test? Who did we lose?
If you want to know which subgroups within the study moved from one question to another, and how much they moved, IA will not only tell you, but will score and sort the groups from top to bottom or bottom to top.
3. Message Strength
Which messages proved most persuasive among voters who moved in our direction from a first ballot test to a second ballot? Which messages did intractable voters find appealing?
A common survey research design asks a first (uniformed) ballot question, then tests a block of messages, then repeats the ballot test again a second or even third time. IA has a large set of tools designed especially for the analysis of blocks of questions. IA, for example, will isolate the respondents who moved in one direction or the other, or remained unchanged. Consultants can then examine message blocks within those isolated groups of respondents.
Exactly how did issues tested in the survey array themselves across multiple dimensions such as issue priority and issue performance?
Questionnaire designs often repeat measurements on a list of topics across multiple dimensions. In a market research study we might measure a set of product attributes across dimensions such as importance of the attributes and performance of the product on those attributes. In politics we might ask how important a series of issues are to a voter, and then ask voters how well the party in power has performed on those issues. IA will score the attributes on the multiple dimensions then visually superimpose one dimension on top of another, thus visually simplifying the analysis.
How can a list of matchups (candidates, ballot designations, descriptive phrases, etc.) be scored then considered across key voter segments?
To do this three pieces of information are desirable: (1) a matrix comparing each item with every other item, (2) an overall preference score that reveals each item’s aggregate score across comparisons, and (3) the capacity to re-measure the preference scores within subgroups of the sample. IA doesn’t even break a sweat.
What will the ballot numbers look like if voter turnout is lighter than expected? Or higher than expected for one party but not the other?
IA allows for the application of both static (raked) and dynamic weights to create simulations of a variety of scenarios.
The headwinds and currents of political campaigns can be formidable. With interactive analytics consultants can “sail against the winds and currents.” Or to paraphrase drug companies, ask your pollster if interactive analytics might be right for you. Side effects may include winning campaigns, happy candidates, and a slight bulge in your bottom line.
Val Smith, Ph.D. and Dan Garvin are Principals at Data Analysis and Display, LLC, makers of Porpoise Survey Analytics and Orca Data Editor. They can be reached at (916) 932-2374 or PorpoiseAnalytics.com