A Quantitative Comparison of SSA Interviewing Activities with and without Computer Assistance

Abstract

The purpose of the reported experiments was to learn the effects of different forms of computer assistance on the performance of field personnel of the Social Security Administration as they interview potential beneficiaries. Twenty-four subjects, twelve qualified in each of two interviewer positions, learned to operate different interfaces (labelled Systems W, P, and T) that simulated interaction with computer data-handling systems. They then conducted simulated interviews with each other, during which predefined categories of interviewer activity were sampled at 15-second intervals. Results showed that for the interviews simulated and regardless of the systems tested, computer-assisted interviews lasted more than 4 minutes longer on the average than did conventional paper-process interviews. This was a 97 per cent increase for two short interviews that averaged 4.7 minutes in the paper-process condition, and a 27 per cent increase for a moderately long interview that averaged 20.2 minutes in that condition. The additional time required for computer-assisted interviews was primarily due to two factors: (1) time spent waiting for a response from the terminal, and (2) more time being required for data entry (i.e., keyboarding) in the computer-assisted process than for its analog (i.e., writing plus handling) in the paper-forms process. These data point in some initial directions for optimizing person/machine system performance in this context, for example by controlling system response delays at different levels of interaction. Further, these data, in combination with other data developed by SSA, provide a means for attaching dollar values to particular system and interface configurations.