`Home     Our Services     Your Resources     News & blog     About us     Contact

Jambo reviewed 

Is your research hampered by silo based software? Tim Macer says "hello!" to a multi-modal solution.

 For those who despair of the rigid, silo based methods of MR, and the resulting difficulties in linking surveys together to get the bigger picture, here is a multi-modal survey package that takes the harvest straight to the market without bothering with the silo.

Jambo is Swahili for ‘hello’ - an apt name for this product and one that goes well beyond its welcoming user-friendly interfaces. Try shouting “hello!” and you start to get the picture: this is wake-up software that turns cherished assumptions on their heads.

MR software has come late to the relational database scene, and few packages capitalise on the power of the relational approach. Surveys remain marooned as separate islands, often with the same questions and even some of the same respondents popping up repeatedly but inaccessibly beyond the shores of each survey. Jambo, by contrast, views all your research as essentially one life-long voyage, crossing between these islands and turning unfathomable depths into navigable passages.

A life long voyage of discovery

It does this by putting all of your research into one large database of questions, respondents and answers, with the potential for millions of each. The purpose of the survey becomes an operational one: a filter to select the respondents, a set of channels to direct respondents to interviewers or self-completion interviews and a script for the interview to follow.

This is not at the expense of powerful survey capabilities. Jambo is impressive for its multi-modal support. It offers full-feature web interviewing, pretty decent CATI and astonishingly seamless CAPI using always-on GPRS communications to allow real-time data collection from out in the field. Switching between modes, even street to CATI seat, is effortless, with an on-screen matrix of options to manage the handover efficiently.

When constructing a survey, you still write questions and a script, and may find it convenient to open a previous survey to copy questions across, but you can just as easily perform a search on the database, using texts that appear in questions or in answer lists. For researchers, it is a dream come true to have all of your firm’s previous research laid out on a plate, both the questions and the results from those questions.

Questions can be used unmodified, or adapted for that survey, whereupon a new question is added to the database.

My only concern was that the tools to manage and categorise a library of questions were rudimentary, and that over time it could become unmanageable. The analysis tools built into Jambo are on the simplistic side. Regrettably there is no automated topline reporting, but the assumption is you will export your data to other tools, and SPSS and triple-s links are provided for this. You extract data through a ‘matrix’, a clever, two-dimensional query on the database where you apply the survey as a filter, and extract only the variables and respondents from that survey to populate your matrix. Widen the filter and you can pick up other questions answered by the same respondents in other surveys, or answers to the same questions from other respondents, go hunting for sample or make benchmark comparisons against your back data.

Assumptions start to tumble

This architecture is responsible seeing many hallowed assumptions tumble: questions shared across surveys can easily be used to calibrate other surveys; respondents can be shared or moved easily from one job to another; panels can be constructed easily, and anyone that has consented at the end of a survey to participate in further research could become part of your ‘panel’. At this point, though, you would need to develop your own incentive management tools for panels, as these are not provided in Jambo.

Web enabled

Being a fully web-enabled program also means that you can check your results from home through a web browser. A modular web-based architecture also allows for firms to link their Jambo servers together, and for agencies to share work and resources in an efficient but secure way, based on permissions and passwords.

De Vos and Jansen Market Research, a Dutch full-service agency, were attracted to Jambo because it appears uncomplicated, despite its sophistication. Joint MD, Cor de Vos explains: “Our philosophy is to keep it simple because a complex solution is no solution at all.”

By including a question on every survey asking respondents for their permission to include them in further research, alongside other recruitment, DV&J has built up a respondent database of half a million records that supports several panels. De Vos states: “You can easily re-use all the information you have been given in previous surveys and have a very reactive relationship with your panel membership.”

The firm has invested in a stock of GPRS-enabled handhelds. De Vos, notes: “Clients cannot believe it when they log in and see live data coming in from interviewers out in the field.”

De Vos estimates that combining CATI and web, and allowing respondents to switch, improves response by around 25%. But is it valid to combine results from different interviewing channels? “We have found the results are very similar,” he states. “We always check - sometimes there is a slight difference, but usually the figures are identical.”

Published in Research, the magazine of the Market Research Society, June 2004, Issue 457.

© Copyright Tim Macer/Market Research Society 2004. All rights reserved. Reproduced with permission.

Mixed-mode interviewing suite which uses a common database to link together survey work, including questions, answers and respondents.