Using PDA consistency checks to increase the precision of profits and sales measurement in panels

Personalized Digital Assistants (PDAs) and other forms of hardware needed to collect survey data electronically have become more affordable and powerful in recent years, leading to their use in a number of surveys in developing countries.  Simple use of these devices can offer the prospect of more timely data entry and greater accuracy in guiding respondents through skip patterns.  Further benefits are possible through the use of more complex consistency checks.  We use PDAs to measure sales and profits for microenterprises, which are notoriously noisy.  Consistency checks in the cross-section compare sales and profits, while those in the panel query responses which result in large changes from one period to the next.  Cross-sectional checks also served as a second prompt in the case of missing profits.  These checks do succeed in reducing the standard deviation and in increasing the correlation of the observations for which corrections are made.  However, we find that the vast majority of large changes in enterprise sales and profits are confirmed by firm owners as genuine, highlighting the volatility of income in this sector.  As a result, the overall impact of these consistency checks on the full sample is rather limited, suggesting that while such checks are useful if computerized forms of data collection are being used, the consistency checks per se are not a strong reason for computerized data collection in collecting firm profits and sales.