Path: utzoo!mnetor!uunet!lll-winken!lll-tis!ames!mailrus!tut.cis.ohio-state.edu!bloom-beacon!mit-eddie!bbn!rochester!PT.CS.CMU.EDU!cadre!pitt!mullins From: mullins@pitt.UUCP (Paul M. Mullins) Newsgroups: comp.cog-eng Subject: stress measurement (need comments & suggestions) Message-ID: <3182@pitt.UUCP> Date: 6 Apr 88 16:44:04 GMT Organization: Computer Science Dept., Univ. of Pittsburgh Lines: 68 Keywords: stress, human-computer interaction, experimental method This is a duplicate of earlier requests to sci.research, sci.misc, and sci.psychology (na distribution). Results -- a couple of requests for information, but no offers. I would really be interested in just opinions at this point. ---- A few years ago, voice stress analyzers (PSEs) were fairly popular with prospective employers, et cetera. I am aware of the debate about the usefulness of "lie detectors", but I have seen little about the stress analyzers, and proponents continue to claim accuracies of 80-93% (i.e. an experienced operator will correctly identify the 'liar' over 90% of the time - I know, not good if you are one of the errors :-<). Can anyone point me to any study indicating a fair amount of correlation between actual stress levels and those indicated by such a device? A lack of correlation? Most of the information I have found in this regard is for polygraphs. Note that I am interested in an indicator of general stress level, not a lie detector. I am considering the use of such a device in determining how comfortable users are with software interfaces. This would be used in conjunction with voice/video protocols. The idea is that such an objective measure would aid the trained observer in analyzing the protocols. The advantage to voice stress analyzers is the relatively low cost and ease of use (a high quality recording can be obtained unobtrusively). Some implicit assumptions are that people have a hard time identifying stress and its causes, that the physiological responses measured by these devices (FM modulations in the voice, BP, HR, respiration, and galvanic response) are actually indicative of stress, and that the indicated stress can be attributed to a specific environmental cause (lots of background noise to be considered). Specific questions: Are you aware of an previous attempts similar to this? (I do recall an experiment where at least two of these variables were measured for computer users, but I cannot locate the source - results were inconclusive though.) It seems obvious that stress causes measurable physiological responses (sympathetic and parasympathetic). Can it be assumed that parasympathetic ('play-dead') responses are the result of extreme stress and are therefore unlikely to interfere with the measurement of sympathetic responses normally associated with polygraphs. Some current research seems to indicate that relationship between relaxation (I read that as low stress/anxiety) and physiological responses. Has anyone shown anything significant here? Other suggestions would also be appreciated. Please e-mail responses. If there is a significant response I will summarize for the net. Thanks in advance. -------------------------------------------------------------------------- Paul Mullins UUCP: psuvax1!pitt!mullins or {allegra,cadre}!... Univ. of Pittsburgh mullins@pitt.uucp Computer Science Dept. Internet: mullins%pitt@relay.cs.net 313 Alumni Hall BITNET: MULLINS@PITTVMS Pittsburgh, PA 15260 CSNET: mullins@pitt Disclaimer: I'm not a psychologist, and I need help ...