A Research Challenge: Extracting Data from Handwritten Surveys
May 15, 2013 by Jennifer Cobb
When it comes to data capture, survey researchers often find themselves stuck between a rock and hard place. In many settings, handwritten data on paper surveys remains the gold-standard for ease of use and higher response rates. Paper is simple, reliable and familiar, needs minimal training and rarely fails in the field. At the same time, data collected via hand-filled paper forms is hard to use. And in today’s research environment, the demands to extract data quickly and with a high degree of accuracy is more critical than ever.
We get it. Captricity offers a full-service data entry/capture platform that improves data extraction and integration for customers worldwide. The demand for accuracy, speed and privacy have all been baked into our solution from the earliest days. Our technology was developed in health clinics in Uganda, so we have experience addressing even the toughest challenges offered by handwritten survey information. Here are some recent examples:
Catapult Health: Catapult Health improves employee health and lowers costs by offering on-site (e.g. at-work) preventive health exams, risk assessments, and personalized treatment plans. Though most of Catapult’s information flows electronically into its health management information system, the company routinely collects paper-based patient satisfaction surveys at all of its locations. With paper, Catapult achieves response rates of 90%. These surveys are scanned on site directly to a Box.com folder. From there, they are uploaded to Captricity and pushed to Catapult’s Salesforce.com information system.
- Captricity streamlines Catapult’s business process through integrations with Box.com and Salesforce.com.
- Catapult’s COO estimates that the current volume of 3500 forms a month, Captricity saves the company 400 employee hours per year while providing higher customer satisfaction and more accurate and efficient data analysis.
New York University’s Center for Technology and Economic Development (CTED): A team of researchers from CTED conducted a randomized control trial of 1000 farmers in Ghana who were using a novel text messaging system (Esoko) to send farmers up-to-date crop pricing data. In order to map Esoko’s impact, the CTED team administered a baseline paper survey and then monthly paper follow-ups. Paper was the clear choice for coordinating and conducting the research. With paper, the team knew what to expect and could entrust thousands of surveys to a dozen enumerators without worrying about additional training or technology failures.
- Captricity worked with CTED to digitize all the paper surveys much more quickly than manual entry. Each form would have taken 20 minutes to enter manually (333 hours total!) The 1000 form baseline would have taken 2 weeks; Captricity completed the job in 24 hours.
- Manual entry was error prone. In such a large spreadsheet, the correct information would often be entered on the wrong line, throwing off the whole data set in ways that were difficult to correct. Captricity returned double-entry quality data with no manual error issues.
Political Science Research in Mali: Political scientist Jessica Gottlieb sought to understand whether providing citizens with information about local government and the basics of democracy led to increased demand for good governance and accountability among elected officials. Her study looked at the effect of educating citizens in Mali, a country that had its first democratic elections in 1992. Gottlieb designed a three-month paper survey of citizens across 570 villages in 95 rural towns in Mali. A total of 5,700 seven-page paper surveys were completed for a total of 36,819 pages, with an average of 162 fields per survey — approximately 1 million distinct data fields.
- The project’s original estimate indicated it would have taken two data clerks 32 weeks to finish the job at single-entry quality (based on a rate of three surveys/hour). After receiving the images, Captricity digitized the one million values in less than a week, at near-double-entry quality.
- The results were returned to Gottlieb as a .csv file which she then imported into STATA where she performed her analysis. “Captricity was a simpler and more reliable way to getting to a better result,” said Gottlieb.
Psychology Department at a Major US University: A multi-year survey conducted in middle schools sought to develop insight and tools for combatting school bullying. In each school, all children completed a paper-based survey at the beginning and end of the school year. The research produced hundreds of thousands of pages of handwritten, paper-based data during two survey rounds. A survey of this size would take many months and a large budget to process using conventional manual data entry. In addition, there were accuracy and privacy concerns with manual data entry. They required that all the data be housed in a secure location and only be accessible to the researchers.
- The project team worked with Captricity to create a template based on the survey and sourced a company to scan the paper surveys into image files. The data was returned as a .csv file in hours with a high degree of accuracy.
- Using manual entry, this project would have taken close to 800 hours for single-entry quality. To achieve double entry quality would have taken roughly 2,000 hours. Instead, Captricity achieved double-entry accuracy in days.
- All data was processed with privacy protections, and results were stored securely.
Can Captricity Help You?
You many know that paper is best for data collection but dread the data entry phase. If you are designing or running a paper-based survey, we can help you get your data faster and more accurately. Drop us a line. We’d love to learn more about your use case and help get you started. If you’re a non-profit research institute or university, be sure to ask about our special non-profit discount!