Improving data accuracy: Selecting the best data checking technique

•Data checking is used to correct errors that occur during manual data entry.•Double entry is substantially more effective than visual checking and read aloud.•VC and RA fail to correct many existing errors and also introduce many errors.•After checking, most errors are within the allowable range for the variables.•Therefore, histograms are not an adequate substitute for entry-by-entry checking.

Data entry errors can have catastrophic effects on the results of a statistical analysis. Therefore, researchers often spend considerable effort checking their data. This paper compared the effectiveness of three data checking methods – double entry, read aloud, and visual checking – using the types of data and data entry personnel that are typically used in psychological research. To compare these techniques, we created 20 data sheets and entered them into the computer. Next, we deliberately introduced errors into this data set. Participants’ job was to locate and correct these errors.A total of 340 undergraduates participated in this study. Of these, 80 had previous data entry experience and 260 did not. Double entry was far superior to read aloud and visual checking, both among people with previous data entry experience and among people without previous experience. Among people with no previous experience, read aloud and visual checking had more than 20 times as many errors as double entry. In addition, double entry was preferred over visual checking. Thus, although double entry takes slightly longer, it is clearly worth the extra effort.