As we embark on our 40th year...
RDA Capabilities: Tapping the power of the web
Making a mixed mode approach work
Prospecting for results and insights

As we embark on our 40th year . . .

More than ever, enhancing productivity and reducing costs are two of the biggest challenges on anyone's task list. Part of this responsibility is placed on your external partners and vendors. Working as a team, you ask them to help you achieve these goals.

For nearly 40 years, we have worked as partners with our clients to help them find answers to their questions in the most efficient and cost-effective ways possible. We do this by offering creative solutions and new approaches to time-tested practices, such as using a multi-mode approach for collecting information or by developing an interactive web site that does more than provide static information.

Sometimes a solution lies deep within mounds of data already collected, yet not easily identified by traditional analytical processes. There are many advanced techniques to uncover and pull together these pieces of data to help you make informed business decisions.

Examples of how our clients are successfully using or applying these ideas to their business information needs are highlighted in this newsletter. Speaking of our newsletter, you'll notice a dramatic change to its appearance and editorial content. We believe this new style more effectively shares ideas that may work for your organization's market research and informational needs.

Thank you,
Donald Pietrowski
President
RDA Group

Back to top

RDA Capabilities: Tapping the power of the web

To some, it may seem unusual to have a market research firm create a product that an IT company might normally produce.

"While many IT firms have the technological ability to program a web site, RDA Group combines this technology with the proper understanding, use and application of data analysis and business prioritization tools," notes Suzanne Battermann, Senior Vice-President, Customer Satisfaction. "This is a core area of expertise that IT firms just do not have. Given the research and data analytic knowledge that we possess, we are uniquely able to deliver a much more effective user interface in a much shorter amount of time. And after all, time is money."

Battermann speaks with first-hand knowledge as her team recently applied these advantages and skills toward developing a web site for one particular project.

""The challenge was to create a reporting web site that was very easy to navigate and uncomplicated to use," Battermann continues. "We needed to develop a reporting tool that allowed multiple users with different levels of rights to be able to access data as needed."

After being awarded the project in a competitive bid situation, Battermann's team went to work.

"We looked at the client's prior web site and talked with them to get a thorough understanding of what they wanted the web site to do." she says. "We wanted the new site to have features people were familiar with when using their computers. For example, Excel file tab-like buttons that would allow users to quickly select the information sheets of interest."

What RDA developed was a site that:

• Shares information either at one of hundreds of individual business units, a regional or a corporate level.
• Displays individual survey responses and flags those needing immediate customer follow-up.
• Tracks the number of survey returns.
• Monitors customer satisfaction progress and compares period over period scores.
• Provides the voice of the customer.
• Communicates program information or important corporate announcements, examples of program surveys and a helpful user's guide.
• Offers the ability to modify a user profile.

"Part of the project initially had us calling customers to thank them for their patronage and asking them a few satisfaction questions," explains Battermann. "Those recordings were posted on the web site and could be played by clicking on the audio button linked to the customer's survey information. They love this feature! There is nothing more powerful, and sometimes more humbling, than hearing the voice of the customer."

Another RDA Group team created a web site to allow a client to manage and measure the effectiveness of corporate sponsored events.

"We created a web site that allows our clients the ability to manage data collection for each event and then analyze the results," explains Ronald Emmert, Senior Vice-President, Strategic Accounts.

As a management tool, events are entered and posted on the client's web site. A calendar provides a quick overview by month of events, by date and by location. Clicking on a date provides additional event details including times, whether the event is being measured and if so, provides the modes of data collection planned for that event. Other web site features for event management include the ability to post examples of the questionnaires to be used at the event and the capability to blog about the event itself.

A separate reporting tool allows users to create reports about the event data collected. "Using our proprietary reporting tool, TabRight, you can create special reports, run cross tabs or view verbatim comments," notes Emmert. "It's been very well received."

Back to top

Making a mixed mode approach work

Many market research consultants will recommend against using a mixed mode approach for collecting data; suggesting that the increased variability in the data may cause erroneous results and conclusions.

However, the reality is response rates for traditional mail and telephone studies continue to decline, which actually serves to increase non-response bias over time in and of itself. Therefore, new approaches in data collection need to be considered. One such approach is using a mixed mode of mail and web-based versions of a survey to increase response rates and reduce costs.

"The 'taboo' of mixing market research methodologies is being outweighed by the business world's pressure to increase response rates while simultaneously decreasing survey costs," explains Ann Benson, Vice-President, Strategic Accounts. "More traditional options to improve response rates, such as increasing the volume of surveys mailed, boosting the number of attempts to contact a respondent by telephone, or changing the incentive, all increase costs, which is not ideal as research budgets continue to shrink."

In many situations, the blending of mail and web survey approaches can be successfully accomplished. One needs to be aware of the challenges associated with this approach and properly prepare for them. For example, when mixing modes of data collection, it is important to use the same sample source to help minimize sampling error. Also, it is important to make sure both surveys are in the field for the same amount of time and that the total population is represented.

"We'll mail a paper survey to a respondent that includes an option for completing the same survey online," Benson explains. "It's increasingly important to give respondents a choice. Otherwise you maybe less likely to reach all groups of your representative population because not everyone has access to the Internet; some are hesitant to complete an online survey, while others are more inclined to go online than to reply to a paper survey. Thus, this approach not only helps to increase the effective sample size, but also aids in reducing a key source for non-response bias.

"Another critical element to multi-mode success is making sure the questionnaires are identical. This requires more than simply verifying the same questions and scales are used in both versions of the survey. Ideally, the appearance of the web survey needs to be as similar as possible to the paper version, notes Benson.

"This issue is a bigger challenge with more complex paper surveys, which tend to have many pages and can be intimidating to potential respondents who may not want to complete the survey because of its perceived length," she explains. "A web survey, where full survey content is not immediately visible, gives the perception of a less lengthy, less intimidating survey. Thus, respondents are more likely to complete it."

After the fieldwork is complete, the data still requires "cleaning" prior to analysis.

"You need to create logic traps to identify outliers who, if left in the final sample, will misrepresent the data," explains Bharath Vijayendra, Senior Vice-President, Statistical Modeling & Consulting. "People, who quickly complete the survey for the incentive, often answer open-end questions with non-sense phrases or might `straight-line' a grid of questions. These are just a couple of examples of aspects you need to consider."

Adds Benson: "With a multi-mode approach, you also need to account for the demographic differences between the two groups because results will often differ. For example, in one of our quality studies, web respondents reported fewer product concerns than paper survey respondents did. By controlling for demographic differences, we found that some of the differences were in fact due to demographics, while some were due to the layout of the two surveys. We generally recommend conducting bridge studies and statistically test the data to make sure it is valid and reliable."

Back to top

Prospecting for results and insights

Your company has terabytes of data in various databases from numerous sources, but trying to understand what it all means and how it can benefit your organization can be an intimidating challenge.

If this sounds familiar, you are not alone.

Making sense of data is where data mining comes in to play. Like modern day prospectors, panning for nuggets of information, data mining allows businesses the ability to analyze large volumes of data to identify consistent patterns between variables resulting in strategic information to help make key decisions in business operations.

"Data mining evaluates a multitude of general numeric relationships using more traditional tools such as multiple regressions and cluster analysis along with heuristic algorithms such as neural networks and decision trees," explains Dr. Duane Trojniak, the RDA Group Vice-President of Statistical Modeling & Consulting. "Using these algorithms helps you look for the proverbial needle in the haystack. However, it is important to set-up the analysis to specify which needle you are looking for. To maximize your success, data mining tools and techniques are most effectively combined with advanced statistical modeling knowledge and experience. Assumptions inherent to many interface friendly packages can often lead to errant directions and conclusions."

There are three areas of data mining. These are:

• The traditional "data mining" using a statistically-based method of searching for numeric relationships.
• The search of character relationships, such as key words or phrases in a verbatim comment database.
• The review of how customers are using information on a web site by looking at the number of "hits" on the site and customer usage patterns while on the site.

"Data mining techniques are used to find relationships in data that are not normally found using the more basic statistical techniques," Trojniak continues. "They allow you to establish a multitude of predictable models. So you go through a variety of algorithms to find numeric relations which are then predicatively tested for accuracy."

In market research, these data mining techniques are often used for predicting consumer behavior. For example, grocery stores use customer loyalty cards as a way to identify purchase patterns, and predict inventory needs and potential timing for placing orders. RDA Group has helped its clients get their arms around both the technical and abstract nature of data mining.

"We have a client who offered a product with more than 1,800 different configurations," Trojniak notes. "Working with them, we were able to identify about 12 configurations that consistently had high levels of customer satisfaction. Next, we created a demographic profile of those consumers who purchased one of the 12 configurations. Finally, from this group, we were able to identify who in the market was purchasing the products with these configurations.

"From this information, our client was able to do three key things," Trojniak continues. "First, they were able to reduce the number of product configurations, which provided dramatic cost savings. Second, by offering the product in these 12 configurations, they were able to maintain a high level of customer satisfaction. Third, they were able to develop effective marketing and promotional materials targeted for this specific group."

Another example where data mining was successfully used by RDA Group, was with a client in the finance industry for business equipment, says Trojniak. The client wanted to segment its customers by purchase patterns. From this information, they were able to predict when customers would be financing a piece of equipment, identify the type of equipment they would likely purchase and the level of financing needed.

"What was challenging about this request was the client had customers all over the world," Trojniak explains. "When developing the data mining algorithms, we had to consider various factors such as local economies, currency and interest rates."

For more information about data mining concepts and techniques, visit the RDA Group Researcher's Toolbox section at www.rdagroup.com.

Back to top