Fieldworkers began training on Monday and by 9 April 2011, over 150 will have been trained on how to use Mobile Researcher to conduct house-to-house enumerations on low cost mobile phones.

Working with international social development consulting firm Ayala Consulting, data collected using Mobile Researcher on over 25,000 households will be automatically incorporated into the information system utilising the Mobile Researcher application programming interface (API).

After facing significant logistical challenges using paper to capture 25,000 households initially, the Ministry took the decision to upgrade to mobile technology to streamline data collection and allow the recording of household locations via integrated GPS. The collection of household photographs as well as an image to identify the head of the household provides invaluable information for verification and follow up purposes.

Subsequent phases are planned to enumerate a further 150,000 households making this one of the largest census-type projects utilising low cost mobile phones known.

On Tuesday, 25 January 2011 we will be deploying a major update to the Analytics section. The Analytics section provides access to raw data, reports and export functionality for all your data collected using Mobile Researcher.

A common and powerful feature used by survey designers is the ability to make a section repeat based on the value of a numeric field captured in the survey. For example, imagine your survey collects biographical information about the members of a household. Using this functionality, you can prompt a fieldworker to record the number of household members, then configure the survey to iterate through each person in turn and capture their information. The mobile application automatically repeats the relevant questions based on the number of household members entered.

As useful as this feature is, one challenge is how to analyse this data once captured (because the questions are generated dynamically). Currently, each iteration is displayed as a separate column which means that if you have a section with 5 questions which repeats 10 times, 50 columns would be generated. Although you can still analyse the data in this format, we received numerous requests for the repeats to be displayed as rows rather than as columns. Once this deployment goes live on Tuesday, repeated sections will be shown in rows which should make analysis much simpler.

Instead of the entire survey being displayed in one grid, we will now display each section separately allowing you to move from one section to the next easily.

We’ll also be releasing some upgrades to the filtering abilities, now allowing you to filter by multiple fieldworkers and languages. When generating export files, you will now be able to select from 3 options outlined below:

• Comma Separated Variables (CSV): Each section of the survey is a separate file. Files are Zipped into a single folder for download.
• Microsoft Excel 2007 (or later): One worksheet is created for each section of the survey.
• Microsoft Excel 2003: One worksheet is created for each section of the survey. To compensate for a limitation in Microsoft Excel ® 2003 and earlier, response data with more than 256 fields will be split up into separate sheets.

In addition to exporting raw data in the traditional column/row format, we’ll now also export in a format which is ready to be used for the generation of pivot tables. We’ve also reformatted the way in which we export survey metadata such as question names and option labels to make importing this information into a statistics package much simpler.

Another important change is how you edit responses (if you have permission to do so, of course). Instead of making changes in the grid which was a little confusing, we’ve refactored this process and you now edit submissions individually.

Finally, you can also now perform bulk deletes (with necessary permissions) – very useful for clearing out test or training data.

One of the encouraging aspects was the number of young people who attended the conference. PHASA provides young researchers and public health professionals with a platform to present their findings, conclusions and ideas.

Something less positive that stood out for me however me was the lack of technical attendees at the conference. This surprised me due to the fundamental, enabling role that I believe ICT can play in improving public health. Although the use of technology, and mobiles in particular is beginning to gain momentum, there need to be far more forums where health and IT professionals can meet to discuss the challenges and opportunities which exist in the sector.

I was fortunate to be seated with one of the members of the PHASA executive during the Gala dinner. We briefly brainstormed a few potential remedies and one possibility which surfaced was the creation of an mHealth Special Interest Group within the PHASA framework. Although not an ideal solution by any means, I think such an entity could be a starting point in opening the channels of communication between domain and technology experts.

Hopefully the potential direction this could take will become clearer in time and I’ll certainly share any developments as they unfold.

In support of their ongoing work in schools, they recently embarked on a baseline study to inform these efforts.

The baseline survey involves asking about a variety of topics, including sensitive questions where respondent privacy is paramount. In order to receive honest answers, respondents need to feel that no one else can observe their responses and that they will remain anonymous. WhizzKids selected Mobenzi Researcher to conduct the study and installed the mobile application on 50 low cost handsets which are handed out to learners, one class at a time. Each handset is preloaded with the baseline survey and a facilitator is present to assist each learner in completing the survey and answer questions they may have.

The use of Mobenzi Researcher in this fashion is one of the first examples of Mobile Assisted Self Interviewing (MASI), and represents a low cost, lightweight, and more personal version of the more traditional CASI (Computer Assisted Self Interviewing). MASI also leverages users’ familiarity with a mobile phone, and as WhizzKids report, “respondents find the survey more engaging”.

Read more about the use of Mobenzi Researcher for this project and their experience with the software on the WhizzKids United blog.

The API leverages Microsoft ADO .NET Data Services which currently supports exchanging entities using either JSON or ATOM (an XML-based feed format). Furthermore the API supports a rich query interface that supports filtering, sorting and query expansion. To find out more about the API and the functionality it supports, please refer to the Mobenzi Researcher API guide.

If you would like to find out more about the Mobenzi Researcher API or have any problems accessing it please feel free to send us an email at support@mobenzi.com.

Resources:

Mobenzi Researcher API Guide – Download

However, where data to be collected by fieldworkers are primarily quantitative in nature, here are my top 10 reasons (in no particular order) to use a good mobile data collection service (I’d like to include Mobile Researcher in that category but will let others decide that):

1. Capture once: With Mobile Researcher, you only capture the data once – in the field during survey conduction – it is then available for export as an Excel file (for example) once it has uploaded from the handset.
2. Validate in the field: Mobile Researcher allows you to design logic and validation rules which prevent many common erroneous entries (the clichéd “pregnant male”, accidentally skipped questions, etc) in the field. As there’s no separate data capture involved, there are also fewer points of failure.
3. Reduce latency: Data collected via Mobile Researcher is available for analysis as soon as it is uploaded – typically within a few minutes. This allows for faster analysis, near real-time data cleaning (automated or manual) and the opportunity to pick up anomalies while the study is in progress rather than weeks or months afterwards.
4. Conserve the environment: Obviously not using paper saves money and trees. We’ve had thousands of survey responses, each equivalent to 50 printed pages or more, captured using Mobile Researcher.
5. Reduce cost: In many cases, using Mobile Researcher is actually cheaper than paper (apart from the other benefits).  This is due to the removal of the exhausting print, distribute, collect and capture processes.
6. Increase flexibility: Surveys can be modified very easily – even with fieldworkers in the field. Survey design errors can be corrected, missed questions added, confusing questions rephrased, etc.  That’s very expensive and logistically taxing with paper.
7. Monitor activity: Using Mobile Researcher you can see what time surveys were started and completed. If the handset is GPS-enabled, we will also stamp the location of where the survey took place.
8. Improve data integrity: Once data has been captured using Mobile Researcher a full audit trail is maintained recording who modified data (if they even have permission to do so), when it was modified and what it was modified from and to.  Ensuring data is able to be cleaned without being incorrectly altered is more challenging on paper.
9. Preserve confidentiality: Once data has been captured on Mobile Researcher, it is securely transmitted to our servers.  Only authorised personnel you grant access to can view the data (there are other very granular permissions as well).  Our systems utilise the same level of encryption used by online banks.
10. Simplify survey conduction: Fieldworkers don’t need to understand branching or skip logic – this is built into the design of the survey.  Based on their response to each question, they are automatically directed to the next appropriate question.  This simplifies training substantially – particularly where prerequisites are complex.

I’ve covered some of the challenges involved in paper-based data collection in several previous posts but here I wanted to summarise the key factors I feel are important.

Perhaps there are other factors which are more important to you? Or are their areas where you disagree? Feel free to post your thoughts.

Breadcrumb

A Breadcrumb is a common navigational element used on many websites to allow users to orientate themselves and quickly move from one place to another with relatively few clicks. The term originates from the age-old fairy tale of Hansel and Gretel who leave breadcrumbs behind them in the hope of finding their way home.

In the research console, the breadcrumb has been implemented to indicate the current:

1. Organisation / account
2. Selected study
3. Selected survey. If a survey is selected, the survey’s status will also be displayed.

Account: Clyral; Study: Training; Survey: Training Demo

Clicking anywhere on the Breadcrumb will open the Quick-switch panel which has been reworked to make navigation substantially easier.

Quick-switch

Although the Quick-switch is not altogether new (you were always able to open it by clicking on the arrow at the top left of the screen) we have redesigned this feature to provide a much more intuitive and comprehensive set of navigation options.

The Quick-switch provides instant access to every study and survey which you have permission to access. Simply select the study and then choose the survey you’d like to access. You will be redirected to the same tab you’re currently viewing but with the chosen survey in context. Alternatively, if you’d like to see an overview of activity for a study, select it from the Quick-switch then click the “Dashboard” link at the top right.

The Quick-switch panel provides access to all studies and surveys.

Depending on the permissions you have been granted, you may also be able to create new surveys and studies from the Quick-switch panel.

Ribbon

If you’ve used Microsoft Office 2007, you’ll be familiar with the concept of a ribbon. When you select one of the main function groups (such as Home, Insert, View, etc), a ribbon with additional actions related to the group appears. The use of a ribbon in the research console provides access to common functionality based on the main tab selected.

Actions available from the dashboard ribbon

By enabling this service for a fieldworker’s SIM card, you are able to automatically specify recharges consisting of airtime, data, SMS bundles or a combination. Configured recharges take place without your intervention and only if a fieldworker’s credit drops below a predefined threshold.

By recharging with data only, misuse of provided airtime can be dramatically reduced along with the associated logistics of monitoring, purchasing and loading of traditional airtime.

Currently this service is available for all prepaid Vodacom and MTN SIMs in South Africa.

Please contact us should you wish to enable this service and for pricing information.

With contributions from individuals such as Erik Hersman (of White African and Ushahidi acclaim), Steve Vosloo (a fellow at the Shuttleworth Foundation), Vincent Maher (internet services portfolio manager at Vodacom) and a range of others including representatives from the Kenyan government and industry players, the conference promises to be a melting pot of ideas and a platform to share experiences.

Our own Mark Fowles will be in attendance and will no doubt be tweeting about the event. He’ll also share some of the new Mobile Researcher features due to be launched officially in February and our latest initiative – Mobenzi – which seeks to create sustainable jobs by croudsourcing solutions to business problems from low income mobile phone users.

If you’re attending the conference, please be sure to get in touch.

The article is available for download here but I’d like to highlight some key findings and comments:

Overall, our findings demonstrate that mobile phone based data collection is feasible at scale.

The software application and web-based interface enhanced real-time supervision of data collectors.

Real-time supervision of CHW performance was a significant advance over previous implementation work.

The automated graphs and real-time information allowed supervisors to focus their time on other aspects of quality control and solving logistical difficulties in the field.

Our experience with a large scale baseline survey suggests that the real-time quality control and data collector supervision enabled by the use of a mobile phone based survey system make this an attractive management option and preferable to a paper based approach.

One clarification I’d like to make on a point mentioned in the publication is that although the source code for the custom management system as described in the article may be released under an open source license agreement (at the election of the software sponsor), there is no intention (at present) for the core Mobile Researcher source code to be released under such an agreement.