Welcome to the June 1997 issue of Ergonomics Australia, I hope you find items of interest within. Highlights include Ian Gibson's presidential reflections, a summary of the ESA council's last meeting (a lengthy face-to-face affair in Melbourne), and updated conference information.
The professional development highlight of this issue is an article by Geoff Hammond (University of Western Australia) which summarises convincingly the argument for dispensing with statistical null hypothesis testing. The practice of statistical null hypothesis testing has been subject to severe, and undisputed, criticisms for at least 40 years. Yet, it persists, and is typically taught as a basic tenet of research methods. Geoff Hammond's article is a particularly well written and concise argument for replication as the only valid test of reliability, and the routine practice of reporting effect sizes and confidence intervals. The article concerns research practice in psychology, but the lesson applies equally to ergonomics.
Useful references for those wishing to peruse this matter further are also provided (Cohen 1990 and 1994 are particularly good, and entertaining, places to start). Another more recent reference is a 1997 electronic textbook by Will Hopkins titled "A new view of statistics". Apart from being an excellent comprehensive treatment of the issues involved in statistical analysis consistent with the arguments outlined in Geoff Hammond's article, the complete text is also available free! See the professional development resources section of this issue for more details.
Other items of interest in this issue include a case study concerning the redesign of a control room prepared by Elizabeth Bunker and myself. I'm hoping that, apart from being of interest to some members, the inclusion of the case study will prompt others to similarly contribute.
On more mundane matters, the deadline for submissions for the remaining 1997 issues will be progressively earlier. The deadline for material for the August issue will be August 4th, the deadline for the October issue will be September 24, and the deadline for the December issue will be November 17.
Finally, an apology is due to Phil Clark. The April issue contained extracts from a letter Phil sent to me regarding the direction of the journal. However Phil did not intend the comments for publication, and I apologise for this error.
Best wishes to all,
Robin Burgess-Limerick
Email: robin@hms.uq.edu.au
Dear fellow ergonomes
A lot of water seems to have flowed under the bridge since I last wrote some notes for the February 97 issue of EA. Which immediately reminds me of our holiday in Italy (it seems ages ago) - some interesting bridges, especially the Ponte Vecchio in Florence lined with shops (good postcard fodder for tourists) but the rivers generally seemed much dirtier than ours. Not much to add to my ergorogues gallery though there was plenty to interest the industrial hygienist, especially the noise and atmospheric pollution from the two-stroke motor scooters.
Back home it is good to be able to offer warmest congratulations to our esteemed editor and Tracey on the arrival of Benjamin on 28 May. We hope that he will bring many blessings to them as the years roll by.
The two-day Council meeting in May was a great opportunity to be able to work face-to-face on important issues facing the Society, and for councillors, the executive and the EO to get to know one another better. My public thanks to Ian Mitchell for all his efforts in getting the agenda and supporting documents together and to Jenni Miller for keeping us on track and to schedule - we actually dealt with all the items on the agenda by close of play on the Sunday. It wasn't until I was on the plane returning to Perth that I realised just how tiring a 2-day committee meeting can be! I hope that by the time you read this, councillors will have had the opportunity to discuss with you various issues arising from the meeting, so I won't attempt to report on all the outcomes. A few important issues from my standpoint are the role of SIGs, the competencies project ( a massive undertaking ably steered by Jim Carmichael, whose final report will be ready soon), and communications.
Council spent some time discussing the place of SIGs in the Society and we felt that, with the notable exception of CHISIG, the formal SIG structure was not as effective as we would like it to be. Council hopes that informal networks will expand through links between members with common interests, and this may lead to old or new SIGs appearing with renewed vigour.
Related to competencies is the complex issue of the standing of ergonomics (and ergonomists) at Government level. At the next executive meeting we shall be finalising details of how best I can approach the relevant State Ministers with a view to increasing government awareness of the role of ergonomics in the improvement of work design and the reduction of accidents. I was interested to read recently that the Australian Medical Association has instituted a "Dirty Ashtray" award for the State Government that has done LEAST to encourage people to quit smoking. The 'winner' (I suppose there had to be one) was Queensland. Perhaps in the fullness of time the ESA will be able to give awards for the best, as well as the worst, States for enlightened promotion of ergonomics in the workplace. "Dirty Ashtray" is a particularly apt term, what do you think the ergonomic equivalents should be? Let me know.
The BHP Australian Steel Awards program is a bright spot in an otherwise rather gloomy scene. Held biennially, the awards recognise outstanding examples of innovation in Australian steel, and the president of the ESA is, ex officio, a member of the judging panel. We will have selected the 1997 winners by the time you read this; being a member of the panel is one of the perks of presidency and is both a pleasure and an education. Sadly, BHP seem to be an exception: I recently saw the brief for a competition for design students in WA to design a 'commercial executive desking system'. I don't know the difference between a desking and a desk, but perhaps I'll find out. Nowhere does the design brief mention the expected user population, or suitability for its intended purpose, although the system must be 'aesthetically pleasing.' A wide range of materials are permitted, including granite! And the judging panel includes designers, a manufacturer, retailer and export consultant, but (you've guessed it) no ergonomist. The organisers will be hearing from me shortly in no uncertain terms.
Finally, communication (or rather the perceived lack of it) has been a thorny issue for some time. However, there does seem to be a new spirit abroad to attempt to improve communication between council, executive, secretariat, branches and members. In this context I would like to distinguish between communications, which covers the means whereby we communicate ( and which is becoming easier with the expansion of e-mail, etc) and communication, which is really a state of mind. Once we've got the state of mind right, then the communications can come into play, but we mustn't confuse the one with the other. I was exhorted many years ago, by a wise teacher, "Make sure your brain is in gear before engaging your mouth." In this hi-tech age perhaps the last phrase should read ".... before engaging your keyboard." but the end result should be the same - good communication.
Finally finally, don't forget the Society's conference in November.
Best regards.
Ian Gibson.
The Council held one of its two face-to-face meetings on Saturday 10 May 1997 and Sunday 11 May, (yes, even sacrificing Mothers Day, to the chagrin of many) where a good deal of work was transacted. A summary of the main decisions were:
SIGs as structural building block (of the Society): Councillors generally agreed that many SIGs as presently constituted were either not working nor effective. On the other hand Councillors recognised the value of a structured means of networking among members. The Executive Officer is to write to all SIG Convenors concerning their current activity.
CHISIG: Roger Hall the Chairperson of CHISIG was invited to address the Council and participate in the debate concerning ESA's working closer with CHISIG.
Occupational Health & Safety Trust: Christine Aickin was appointed the Society's representative on the Occupational Heath and Safety Trust.
Competencies: Jim Carmichael (Convenor of sub-committee) tabled the sub-committee's final report. As some last minute suggestions had not been incorporated into the final draft, it was resolved that the matter be referred back to the Competencies Steering Committee for further review.
The NSW Branch Chairperson had suggested that in view of Worksafe's removing ergonomists from its staff, all Branches approach the relevant Authority in their respective State / Territory and request funding over a limited period, say three years, towards the cost of ESA employing a national education officer. The President of the Society will write to relevant Ministers from each State Government requesting that he/she receive a deputation to discuss the role of ergonomics in the improvement of work design and the reduction of accidents.
Media Kit: The ACT Branch tabled a draft marketing plan for Council's consideration. Councillors are to discuss the Plan at respective Branches
Professional Enquiries & Complaints: Early this year, a company which had engaged an ergonomist (a member of the Society) wrote to the President seeking advice on the value of the ergonomist's report, the reasonableness of the fees charged and the fairness of the number of hours spent. Although the particular issue had been resolved, the Council resolved to ask the PAB to consider and to recommend terms, costs, structure and process for the establishment of an Appeal Panel to which members of the public and the Society may refer, and to recommend a set of By-Laws to Council.
PAB Rules: It was agreed that the PAB Rules be changed to bring the PAB elections into line with the Society's AGM and elections.
Electronic Communication: It was resolved that a sub-committee comprising Robin Burgess-Limerick, Rod Powell, Phil Moody, Leon Straker develop a policy on electronic communication for the Society with the terms of reference including
1. contents of the Homepage
2. how EA can be placed on line
3. accessibility to members
4. copyright issues
5. costs and legal implications
6. duty statements for those engaged in carrying the Society's responsibilities
Financial Situation: a sub-committee comprising the Treasurer, the Past President and the immediate Past Treasurer is to examine the current financial situation and propose options to the next Council meeting.
Membership Numbers: It was resolved that Councillors report to Council's next meeting on their Branch's activities which aim to increase the number of members and member services.
ESApress: The Council had earlier considered a number of options for publishing its documents and it was resolved that the Society work in conjunction with IEApress.
Yellow Pages Advertising: Some time ago the Society convinced the Yellow Pages company to run a separate category for ergonomists; it also placed advertisements in the major capital's Yellow Pages. It was anticipated that the Society would place a "masthead" and logo advertisement under which those ergonomists who chose would add their own placement as several other societies do. The cost of the individual entries, including a proportion of the masthead, would be borne by the contributing members such that it is revenue neutral but the advantage to those members would be their business being association with the Society's logo. Councillors are to ask their respective Branches whether any members wished to participate in the scheme and to advise Secretariat by 20 June.
Communication between Branches and both Executive and Council: This items has been the subject of concern for some time. After a lengthy debate, it was agreed that -
1. Each Branch Councillor would formally advise the Branch of the decisions of the Council (using the Minutes of the meeting as a basis for reporting);
2. Each Branch would send copies of its own activities, its technical programs and its Newsletter to all other Councillors, Branch Chairpersons & Secretaries, the Secretariat and the Editor of EA and also send copies of its Branch Committee meeting Minutes to the Executive and to the Secretariat;
3. All appointments to specific projects are to be made by the Council or the Executive and a system of reporting back agreed at the time of the appointment;
4. An item on "Reports" be added to Agenda of Council and Executive Meetings in which personnel can report on current projects which they undertake. (This also requires the project leader to write regular updates);
5. When Branches or Members wish the Society to consider an item of national import, it write a submission and have it circulated through the Secretariat in time for consideration at the Executive or Council meeting.
Survey: In response to a suggestion from the Victorian Chairperson that the Society conduct a Survey of members every three or four years, Council decided that the Society would
1. work towards a climate of surveys;
2. conduct a survey of members in the next twelve months;
3. work from strategic plan to identify the type of data sought in each survey;
4. report on an analysis of the membership renewals responses at the next face-to-face meeting of Council.
Changes to Constitution: On the recommendation of the Executive, Council resolved to recommend to the membership of the Society at the next AGM the following changes to the Constitution
1. the Chairperson of the PAB be added to the Council;
2. from the Clause describing the Council's composition, remove the offices of Secretary-elect and the Treasurer-elect;
3. provision be made for the Executive to invite others to address the
Council on specific items as the need arises.
PAB: It was resolved that Council ask PAB to consider raising its own profile and to provide leadership to the Society by CPEs contributing more fully in such matters as presentation of papers and of chairing sessions at the Annual Conference, writing articles for Ergonomics Australia, etc.
Honours & Awards Committee: the immediate past-President was added to the Committee. It was also resolved to abolish the Branch medal in that it could dilute the value of the President's Medal.
Pamela Freeman's presentation, Ergonomic Reporting & Communication, on 21 May was well attended and received much praise for the clarity of ideas presented. Her workshop will follow on the 27 June and should be on the agenda for anyone serious about improving their skills in this area - Pamela is a great believer in the KISS principle for academe as well as industrial workplaces; most importantly she makes it fun as well as invaluable advice ... even for the old hands who 'have been there & done that'.
As for the rest of our state program:
June 18: Vibration & Environmental Measurement in Ergonomics - a presentation by Byron Longstaff & Bruel Kjaer.
July 23: Cognitive Ergonomics & Warnings - Austin Adams.
August 20: Work Physiology & Project Aquarious - John Brotherhood.
September 17: Designing for Useability - Roger Hall & Jonathan Talbot.
October 22: Slipping & Falling - Measurement Debate & Forensic Issues - Neil Adams & Mike Stevenson.
November 19: AGM at venue to be announced.
All our branch meetings are held at the Worksafe Auditorium, 92-94 Parramatta Road, Camperdown (opposite Sydney University)and any ESA members and friends visiting from interstate or overseas are assured of a warm welcome - just come and introduce yourselves, be stimulated by our fantastic speakers, then join us for dinner afterwards at a nearby restaurant ... chosen for monthly diversity and value (cheap!).
Congratulations to Catherine Cook who won the 1996 NSW Branch Ergonomics Prize awarded for the best performance in the core subjects of either the Master of Applied Science in Ergonomics, or the Graduate Diploma of Ergonomics, from the Department of Safety Science at the University of New South Wales. Catherine is an occupational therapist and is working at the University of Western Sydney Macarthur, in the Faculty of Health, Division of Occupational Therapy.
NSW Branch has quite a number of members in the Wollongong and Newcastle areas and the matter of regional groups has long been mooted. Earlier this year, via our state newsletter, Christine Aickin asked for expressions of interest by 6 June and has had a good response which will be further developed once she gets back from IEA in Sweden.
There is the usual annual pilgrimage to northern climes about to take place from NSW - Christine Aickin, Marcia Lusted, Maurice Oxenburgh, Barbara McPhee, Trudy Tilbury et al are off to Sweden for the IEA Conference ... are other states equally aol? Just jealous, someone has to stay behind to mind the shop!
Those still looking for a later travel opportunity may be interested in the forthcoming 15th Asian Conference on Occupational Health, Kuala Lumpar, 31 August - 3 September 1997 which is considering a request from the Technical Committee on Ergonomics of the Asian Association of Occupational Health, to include a scientific session on ergonomics. The Technical Committee has been unable to arrange a meeting since the last one at the Beijing Conference 1984 and is urgently looking for support. Some colleagues from the Phillipines and Vietnam have contacted Professor Kazutaka Kogi, Director, The Institute for Science of Labour (ISL) in Kawasaki, Japan. Anyone interested in further details could contact Barbara McPhee or Shann Gibbs - or make direct contact with Professor Kazutaka Kogi: ISL, 2-8-14 Sugao, Miyamae-ku, Kawasaki 216 Japan Telephone: 81 44 977 2121 or Facsimile: 81 44 977 7504
Shann Gibbs
PRODUCTIVITY, ERGONOMICS & SAFETY - THE TOTAL PACKAGE
Gold Coast, Queensland, 24-27 November 1997
Time is drawing closer by the month to the most exciting conference ever. The conference, held in conjunction with the Division of Workplace Health & Safety provides skills, hands-on experience and scientific knowledge about ergonomics and it's application in industry.
Preparations for a mock trial are going well, with members of the ESA (especially Justin O'Sullivan) giving valuable input to the Division of Workplace Health & Safety and our legal friends. The "production" will give life to ergonomics, health and safety in the workplace, with an emphasis on legal obligations and outcomes.
In addition to Prof. Stu McGill (see April issue for details), a second keynote speaker, Prof. Penny Sanderson, has been confirmed. Prof Sanderson is a psychologist trained at UWA and University of Toronto, who is recently returned to Australia to take up a chair in the School of Computer Science and Software Engineering at Swinburne University of Technology after previously holding joint appointments at the University of Toronto and the University of Illinios. The likely title for Prof. Sanderson's address is "Cognitive Ergonomics of Information Technology in Critical care (the green gas goes in the front hole). She will also run a workshop in the area of cognitive ergonomics.
Keynote speakers for the Occupational Health & Safety Forum include Prof. Dennis Else, Chair of the National Occupational Health & Safety Committee; and Caren Sundstrom-Frisk, a psychologist from the Swedish National Institute for Working Life who will present a paper on the perception of risk.
Further Information
For further information and general enquiries please contact:
Conference Secretariat,PO Box 177, RED HILL QLD 4059 AUSTRALIA
tel: 07 3368 2644 (international callers: + 61 7 3368 2644), Freecall: 1 800 811 510 (Australian callers from outside Brisbane only), Facsimile: 07 3369 3731 (international callers: + 61 7 3369 3731), e-mail: carillon@ozemail.com.au
Last year at the ESA Conference in Canberra Roger Hall and I discussed the need for CHISIG to become more effectively aligned with the ESA. In fact, some years ago we worked on achieving just that, had the plan ready but were not around long enough to ensure it would also be implemented. So, here we are again!
Ambitions to bring the ESA and OZCHI conferences at least back-to-back have been aired, and it was hoped that this could happen for the first time in Melbourne in 1998. Alas, CHISIG Victoria is responsible for arranging the technical programme for INTERACT'97 which takes place in Sydney July 14-18. The technical programme includes papers, panels, doctorial consortia, workshops, tutorials, new technologies, interactive experiences, and others, so most Melbourne people have been up to their eye balls in conference preparations since the beginning of 1996. None of us have the puff to host another conference here in 1998, so the tete-a-tete won't happen next year. We are working on making it happen later though (I can't pin myself to a date, but trust me!), so tune in to this column again.
Even if you are not interested in interactive computing, the tutorial programme at INTERACT'97 offers a number of tutorials of interest to ergonomes, including:
Johnson, P., Wilson, S. & Johnson, H.Designing user interfaces from analyses of user tasks... This is all about structured task analysis.
Bevan, N. Planning and implementing user-centred design... This is about design and evaluation techniques evolved under the European ESPRIT programs.
Wilding, C. Practical GUI screen design: Making it usable... For folks who are interested in visual design principles that include and go beyond GUIs.
Wright, P. Fields, A. Dearden, A. & Harrison, M. Techniques for the design of human error tolerant systems.... Presents simple techniques for tackling human error issues in [product, including computer]design.
Hewett, T. Cognitive factors in design: Basic phenomena in human memory and problem solving .... An introduction to the various phenomena relating to memory and problem solving via hands-on exercises.
In addition, there are tutorials focusing on methods for measuring usability, cost-justifying usability, introduction to activity theory, and many others. For further information, please call Dr. Gitte Lindgaard (03 9846 3656, e-mail: glindgaard@acslink.net.au).
SBESIG: David Caple, the Convenor of SBESIG, has taken on the task of Convenor for the 1998 Conference in Melbourne. As a result he seeks volunteers to relieve him as SBESIG Convenor. He mentions that although there has been little activity of late in the SIG, he has received a number of expressions of interest in reviving it and hopes that someone can take up the cudgels. Any offers please? Let the Secretariat know and we will fill you in with the background.
The Sydney based OHS Auditor Certification Panel, with ESA representatives Christine Aickin, Mike Stephenson and Maurice Oxenburgh as alternate members have been trained and on 6 May 1997 registered the first six OHS auditors. Monthly meetings are planned for this year. Ros Kushinsky and Derek Viner have been nominated for the Melbourne panel but it will not meet for some considerable time. Malcolm Dispain attended the Auditor Certification Review Board on 22 May 1997. There is a strong interest in OHS Auditor Certification. If you require information, call, write or fax
Register of Certified Auditors, 69 Christie Street, St Leonards NSW 2065
tel 02 9901 9944, fax 02 9901 4677
ESA must now look to providing training for persons wishing to be OHS / Ergonomics registered auditors
Malcolm G. Dispain
WorkSite Services and Systematic Marketing, PO Box 25, North Adelaide, SA, 5006, ph: 08 8363 1583. fx: 08 8362 2596. mobile: 041 981 2130, dispain@senet.com.au
From John Skaczkowski
"After nearly 5 years within the Rehabilitation and Claims Management industry in South Australia I have set up my own company. Not surprisingly I provide a service relating to the afore mentioned tasks along with OH&S consultancy and Training (development and presentation). I can be contacted on 08 8248 5377, (mobile) 0411 612 379 and fax 08 8248 4471."
The Objections to Null Hypothesis Testing as a Means of Analysing Psychological Data.
Geoff Hammond
Department of Psychology, The University of Western Australia, Nedlands, WA 6907 e-mail: geoff@psy.uwa.edu.au
This article first appeared in the Australian Journal of Psychology (1996), 48, 104-106. Copyright 1996 The Australian Psychological Society Ltd. Reproduced with permission.
Abstract
The objections which have been made to testing statistical null hypotheses as the means of analysing the results of psychological research are assembled. None of these criticisms is by itself new and none is controversial. Together, however, these criticisms demand a fundamental change in the way we analyse the results of psychological research and the way in which we teach students to analyse data. In order to cumulate a reliable body of knowledge psychologists should dispense with statistical hypothesis testing and use instead empirical replication as the test of reliability and report effect size measures together with an indication of their likely error.
Introduction
Despite periodic criticisms (e.g., Bakan, 1966; Meehl, 1978; Carver, 1978; Guttman, 1977) the practice of null hypothesis testing has flourished in psychology. More recently, however, the criticisms of this practice have become more frequent, more insistent, and consequently more difficult to ignore (Cohen, 1990, 1994; Gigerenzer, 1993; Guttman, 1985; Loftus, 1993; Meehl, 1990; Rosenthal, 1991, 1993). The criticisms themselves are not in dispute and come from psychologists adept at research and the quantitative analysis of psychological data. The criticisms of hypothesis testing are now a ground swell that can no longer be ignored. The purpose of this paper is not to make an original contribution to the case against hypothesis testing in psychology but to assemble the objections to hypothesis testing as the means of analysing data from psychological experiments and to point to methods more appropriate to developing a cumulative body of psychological knowledge. By raising the issues in this journal I hope to bring them to the attention of Australian research psychologists, and so to encourage a change in the way we analyse research findings.
The objections to statistical hypothesis testing as the means of analysing psychological data will be described in two sections:- first, what the faults are in our current practices, and second, why these faults aren't worth correcting.
What is wrong with our current practices?
Two important faults have been identified in our current practices. The first fault is that the result of a statistical significance test is often misunderstood and misinterpreted. Significance tests are treated as if they convey some information about the probability that the null hypothesis is true. This is not so - the significance test says nothing about the likely truth or falseness of the null hypothesis. The significance test begins with the assumption that the null hypothesis is true and gives the probability of obtaining the data given that assumption. This is not the same as the probability that the null hypothesis is true given the data; this probability is indeterminate (Guttman, 1985; Cohen, 1990, 1994). The sense from reading research reports is that this misinterpretation of statistical significance is in part a convenient fiction which we share in order to get on with the business of finding things out. However, the appearance of this misinterpretation of significance tests in some standard statistical texts shows it also to be in part a deeply-seated misconception (Cohen, 1994; Gigerenzer, 1993). As Cohen (1994) has made clear by argument and example, the deductive logic of hypothesis testing is invalid. Statistical significance is a conditional probability, that the obtained data were unlikely given the truth of the null hypothesis, and does not imply that the null hypothesis is itself certainly or even probably false. In turn, despite the commonly-used phrase `statistical reliability', statistical significance carries no implication that the result is repeatable, nor does the level of significance indicate the likelihood that the result will be repeatable. Gigerenzer (1993) has called this belief the replication fallacy - the mistaken belief that the greater the level of significance, the greater the probability the result will be repeated in future replications of the experiment.
The second important fault in our current practice is that significance tests are often made without regard for statistical power. The existence of a real treatment effect does not guarantee a statistically significant result. Whether a treatment effect appears as statistical significance depends on the power of the test. Statistical power is the probability of rejecting a false null hypothesis so is the complement of , the probability of making a Type II error, that of failing to reject a false null hypothesis. Despite repeated exhortations for psychologists to consider power when designing experiments by Cohen (1962, 1988, 1992), the average power of published experiments apparently remains around .4 to .5 (Cohen, 1992). The high cost of achieving high power makes low-powered experiments endemic. For example, Cohen's (1992) tables show that in order to achieve what he regards as an adequate power level (.8) for a test of a medium-sized effect between the means of two independent samples (defined as a difference of 0.5 standard deviation units) with an level of .05, 64 subjects are needed in each group. The same power for a test of a medium-sized product-moment correlation (defined as r = .30) requires a sample size of 85. Cohen's tables make for sobering reflections on how we usually design research. Experiments with power levels around .5 have only a 50-50 chance of producing statistical significance in the presence of a real treatment effect. It is not the case that low power is self-protective, in the sense that large experimental effects will be statistically significant and small experimental effects will fail to reach significance. Effect size is one of the variables that determines power, and is therefore factored in to its calculation; experiments with low power will therefore often fail to reject statistical hypotheses no matter how large an effect is actually present. Furthermore, if (as is usually the case) statistical significance is taken as the criterion of successful replication, there is only a one-in-four chance of experiments with a power of .5 both obtaining significance even when a real effect is investigated (Rosenthal, 1993).1
Power analyses have two salutary effects for those who test statistical hypotheses: first, they force the recognition that sample size is crucial to the outcome of a statistical test and is not a matter of convenience, and second, they draw attention to the size of the effect produced by the experimental conditions.
Why our current practices aren't worth fixing.
A former teacher, Jim Lumsden, used to say that if a job wasn't worth doing it wasn't worth doing properly. This seems a fitting maxim for the practice of null hypothesis testing. The null hypothesis, phrased as a statement of no treatment effect, is seldom if ever true in psychological research.2 It is unlikely that, in psychology, the difference between two population means will be exactly zero, or that a population correlation will be exactly zero. This is so because, as Meehl (1990) has pointed out, most variables of interest in psychology are intercorrelated to some degree. These deviations from a point null hypothesis, however trivial, will become statistically significant with sufficient power. These are not chance effects that are being detected by sufficiently powerful tests, but real effects, although trivial and probably uninterpretable. They are the inevitable result of what Meehl has called the "complex and obscure causal influences" that affect psychological variables (1990, p 207). Loftus (1993) has taken this point up, arguing that statistical significance is nothing more than a measure of the statistical power of the experiment. Putting aside for a moment that the logic of rejecting the null hypothesis is flawed, what is the purpose of rejecting a hypothesis known on general grounds to be almost certainly false before any data are collected? The probability of obtaining the data given the truth of a highly improbable null hypothesis is a value of little if any interest.
Null hypothesis testing is working to retard development of psychological science, not advance it. Reliance on it as the means of identifying real phenomena has inevitably distorted psychology's cumulative knowledge, the body of published research. Important effects are missed leading to apparent failures to replicate, and unimportant - and even chance - effects are reported as if they were important. It is disturbing that we can't trust the published research literature. Furthermore, the strategy of testing statistical hypotheses leads us to think of a world in which effects are either present or absent, and to reduce our analysis of quantitative data to a binary decision. Its use deflects attention from the most important quantitative aspect of data, the size of the effect of the independent variable. The p value becomes the result.
What should we do instead?
The emerging consensus is that statistical estimation is a more suitable data analytic procedure than hypothesis testing for psychology (to pick a few: Becker, 1991; Berry, 1986; Cohen, 1994; Gardner & Altman, 1986; Gigerenzer, 1993; Rosenthal, 1993). As Rosenthal (1993) has put it, there are two important aspects to an experimental result: an estimate of effect size and an indication of the error associated with the effect size estimate. When the dependent variable is measured in units that are widely accepted and understood, these two aspects of the results can be conveyed as the mean (or median) of the parameter under investigation together with specified confidence limits (such as a 95% confidence interval) to show the precision of the estimate.3 The potential for widespread use of this simple and informative means of describing data should encourage psychologists working in particular areas to develop agreed-upon units of measurement. There is also a variety of effect size measures available which are free from the particular measurement units of an experiment which therefore allow ready comparison of effect magnitudes obtained in different experiments. McGraw and Wong (1992) have given a useful discussion and comparison of some of these different effect size statistics. It is worth mentioning some of them here to encourage their use; as they become more commonly used so it will become easier to integrate the results of different experiments.
Cohen's d is calculated as the difference between sample means divided by the unbiased estimate of the population standard deviation; it expresses the difference between the means in standard deviation units. It is a versatile statistic, but it has the disadvantages that it assumes equal within-group variance and, perhaps more importantly, that it does not give an intuitive grasp of effect size to the non-statistician (McGraw and Wong, 1992). The correlation coefficient (r) and the coefficient of determination (r2), the proportion of variance in Y accounted for by its linear relationship with X, are probably the most commonly used measures of the strength of a relationship. Rosenthal and Rubin (1982) noted that psychologists tend to underestimate the practical importance of effects with small r2 values, and showed that treatments which account for only a small proportion of variance may nevertheless have substantial practical importance. They proposed the use of a `binomial effect size display' to convey the practical importance of a treatment. Eta squared and omega squared are more general proportion-of-variance measures which give the proportion of variance in Y accounted for by its linear and non-linear relationships with X; eta squared is the effect size in the sample data whereas omega squared is an estimate of the population effect size. Both are easily computed from terms in analysis of variance summary tables.4 Rosenthal and Rosnow (1991) recommend the unsquared eta, the correlation ratio, as a measure of the strength of relationship between two variables, again on the grounds that the unsquared value conveys the practical importance of the relationship more readily. McGraw and Wong (1992) have suggested a new statistic, the Common Language effect size statistic (CL), as an easily interpretable measure of effect size for discrete groups. For the two-sample case, CL is the probability that a score drawn randomly from one of two distributions of scores will be greater than a companion score drawn randomly from the second; for more than two groups, CL is the probability that a score drawn randomly from one group will be larger than scores drawn randomly from the remaining groups. More recently, Dunlap (1994) has described a generalisation of CL to the correlation of two continuous variables. In this application, the CL is the probability that a randomly-selected individual who has a higher score on the first variable than a second randomly-selected individual will also have a higher score than the second individual on the second variable. Dunlap's paper includes a table showing the bivariate CL values which correspond to different values of r.
Approaching data analysis in this way loses the surety which seems to accompany hypothesis testing, and provokes two questions: Is an observed effect reliable? and Is an observed effect important? These two questions will be discussed in turn.
How do you know if the effect seen in any experiment is reliable? There is no objective method by which the reliability of a single observation (such as a difference between two sample means) can be determined, no matter how much we might wish for one. The only test for reliability is the empirical test of replication. Reliable effects are repeatable in independent observations (whether of groups or individuals), whereas unreliable effects - those that appeared due to chance error fluctuation - are not. Replication addresses the reliability question and at the same time gives more precise estimates of the size of reliable effects. In contrast to hypothesis testing, where larger sample sizes lead to progressively more trivial (although real) effects being declared `significant', hence distorting the cumulative body of research findings, statistical estimation gets better the larger the sample. We would do well to follow Lykken's (1991) recommendation and design our research to replicate and extend previously reported findings. It is only in this way that we will establish reliable relationships and accumulate progressively more precise estimates of the relationships between variables of interest.
How large does an effect have to be to be important? As is the case for reliability of an empirical effect, there is no purely objective method that can be applied automatically to determine the importance of an effect. Effect size measures are not an end in themselves. Judgment of how large an effect must be in order to be important for psychological theory or practice is inescapably part of the researcher's or clinician's task. Prentice and Miller (1992), after listing the advantages of statistical effect size measures, noted that small effects can be important. This is much the same point made by Rosenthal and Rubin (1982), and underlines again the need for interpretation of any statistical measure within its scientific or clinical context. Prentice and Miller (1992) suggested two practical methodological tests as supplements to statistical effect size measures to indicate the importance of an independent variable. The first of these is to determine how small a manipulation of the independent variable will affect the dependent variable (important or powerful independent variables will exert an effect at a low level) and the second is to determine how obdurate a dependent variable will show an effect of the independent variable (important or powerful independent variables will exert an effect on dependent variables that are usually resistant to influence).
Although the criticisms made of hypothesis testing are not in dispute they have to date been ignored. The problem is that hypothesis testing has become entrenched in our research practice and in our undergraduate `design and analysis' and `statistics' courses. The version of hypothesis testing which appears in psychological texts has become institutionalised in psychology (John, 1992) and its use has gone unquestioned. We must now stop teaching hypothesis testing routinely as if it were a set of objective techniques whose application leads to sure knowledge and teach instead methods more suited to the progressive accumulation of reliable quantitative knowledge. It is our responsibility to teach our students, both by example by classroom instruction, that data analysis is a thoughtful process which demands professional judgment and which is always accompanied by the possibility of error. We are wrong to lead students to believe that there are simple and objective procedures which enable us to tell what is real from what is not.
References
Bakan, D. (1966). The test of significance in psychological research. Psychological Bulletin, 66, 423-437.
Becker, G. (1991). Alternative methods of reporting research results. American Psychologist, 46, 654-655.
Berry, G. (1986). Statistical significance and confidence intervals. The Medical Journal of Australia, 144, 618-619.
Carver, R.P. (1978). The case against statistical significance testing. Harvard Educational Review, 48, 378-399.
Cohen, J. (1962). The statistical power of abnormal-social psychological research: A review. Journal of Abnormal and Social Psychology, 65, 145-153.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.), Hillsdale, NJ: Lawrence Erlbaum Associates.
Cohen, J. (1990). Things I have learned (so far). American Psychologist, 45,1304-1312.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159.
Cohen, J. (1994). The earth is round (p<.05). American Psychologist, 49, 997-1003.
Dunlap, W.P. (1994). Generalizing the common language effect size indicator to bivariate normal correlations. Psychological Bulletin, 116, 509-511.
Gardner, M.J., & Altman, D.G. (1986). Confidence intervals rather than P values: estimation rather than hypothesis testing. British Medical Journal, 292, 746-750.
Gigerenzer, G. (1993). The superego, the ego, and the id in statistical reasoning. In G. Keren & C. Lewis (Eds.), A handbook for data analysis in the behavioral sciences: Methodological issues (pp. 311-339). Hillsdale, NJ: Lawrence Erlbaum Associates.
Guttman, L. (1977). What is not what in statistics. The Statistician, 26, 81-107.
Guttman, L. (1985). The illogic of statistical inference for cumulative science. Applied Stochastic Models and Data Analysis, 1, 3-10.
Hays, W.L. (1963). Statistics for psychologists. New York: Holt, Rinehart, and Winston.
John, I.D. (1992). Statistics as rhetoric in psychology. Australian Psychologist, 27, 144-149.
Lentner, C. (Ed.). (1981). Geigy scientific tables, Vol. 2. Basel: Ciba-Geigy.
Loftus, G.R. (1993). A picture is worth a thousand p values: On the irrelevance of hypothesis testing in the microcomputer age. Behavior Research Methods, Instruments, & Computers, 25, 250-256.
Loftus, G.R., & Masson, M.E.J. (1994). Using confidence intervals in within-subject designs. Psychonomic Bulletin & Review, 1, 476-490.
Lykken, D.T. (1991). What's wrong with psychology anyway? In D. Cicchetti & W.M. Grove (Eds.), Thinking clearly about psychology, Vol. 1: Matters of public interest (pp. 3-39). Minneapolis: University of Minnesota Press.
Meehl, P.E. (1978). Theoretical risks and tabular asterisks: Sir Karl, Sir Ronald, and the slow progress of soft psychology. Journal of Consulting and Clinical Psychology, 46, 806-834.
Meehl, P.E. (1990). Why summaries of research on psychological theories are often uninterpretable. Psychological Reports, 66, 195-244 (Monograph Supplement 1-V66).
McGraw, K.O., & Wong, S.P. (1992). A common language effect size statistic. Psychological Bulletin, 111, 361-365.
Prentice, D.A., & Miller, D.T. (1992). When small effects are impressive. Psychological Bulletin, 112, 160-164.
Rosenthal, R. (1991). Cumulating psychology: An appreciation of Donald T. Campbell. Psychological Science, 2, 213-221.
Rosenthal, R. (1993). Cumulating evidence. In G. Keren & C. Lewis (Eds.), A handbook for data analysis in the behavioral sciences: Methodological issues (pp 519-559). Hillsdale, NJ: Lawrence Erlbaum Associates.
Rosenthal, R., & Rosnow, R.L. (1991). Essentials of behavioral research (2nd ed.). New York: McGraw-Hill.
Rosenthal, R., & Rubin, D.B. (1982). A simple, general purpose display of magnitude of experimental effect. Journal of Educational Psychology, 74, 166-169.
Sedlmeier, P., & Gigerenzer, G. (1989). Do studies of statistical power have an effect on the power of studies? Psychological Bulletin, 105, 309-316.
Winer, B.J., Brown, D.R., & Michels, K.M. (1991). Statistical principles in experimental design. New York: McGraw-Hill.
Footnotes
1. The consequences of low statistical power can be even worse. Sedlmeier and Gigerenzer (1989) noted that 11% of the studies published in the Journal of Abnormal Psychology in 1984 framed the research hypothesis as the null statistical hypothesis and treated a non-significant result as confirmation of the research hypothesis. The median power of these experiments was .25.
2. Cohen (1994) has drawn attention to the fact that the null hypothesis of the Fisherian system was the hypothesis to be nullified, not necessarily a hypothesis of no effect. He has characterised the point hypothesis of no effect usually tested in psychology as the `nil' hypothesis.
3. A simple but little-used method for calculating a confidence interval for a median (which is often the most suitable measure of central tendency) is described in the Geigy scientific tables (Lenter, 1981). Loftus and Masson (1994) have recently described a method for calculating confidence intervals for parameters estimated from within-subject and mixed designs when the interest is in change of the parameter across the different experimental conditions rather than in estimating the absolute values of the population parameter.
4. The computational procedures for between-subject designs have been described by Hays (1963) and those for within-subject designs by Winer, Brown, & Michels (1991).
[Editors note: Lest anyone think that this problem of fixation upon p values is restricted to Psychology, a quick flick through the first 4 issues of the 1997 volume of the journal Ergonomics reveals that of the 22 articles which report experimental results, only 6 explicitly report effect size (the coefficient of determination in all cases). A further 8 report sufficient information (mean and standard deviation, or standard error) to enable the interested reader to calculate effect size, while the remaining 8 report results without any indication of variability].
Robin Burgess-Limerick (The University of Queensland) and
Elizabeth Bunker (Elizabeth Bunker & Associates).
Introduction
This case study concerns the re-arrangement of a control room for a minerals refining process. The plant runs continuously and the control room is staffed continuously (12 hour shifts) by a Power Controller, and a Gas Controller. The Power Controller has the primary task of monitoring and regulating electricity consumption. The Power Controller is also responsible for issuing permits to visitors and sub-contractors. In addition to computer control systems used by the controllers, the control room contains other control panels, space for clerical tasks, and a separate workstation for engineering development.
The re-arrangement of the control room is being prompted by the commissioning of new computer system to monitor and control electricity and gas consumption. Within the next two years the facility will be expanded and the specifications for the new control room include provision for this expansion.
Specifications
Some aspects of the existing control room are to be retained.
A curved mimic board occupies one wall of the control room. Although monitoring and control of electricity consumption will primarily occur via the new computer system and the mimic board is being phased out, the power controller requires clear view of the board and it may be used in emergencies.
Two other large control panels (the Rectiformer desks) will also be retained in their current locations. The expectation is that all other furniture will be replaced.
Schematic plan of existing control room.
A photo of the existing control room
Operational requirements
The redesigned control room requires the following features:
* Power control desk - space for 2 x 21 inch monitors (about 60 cm cube), keyboards, mice; space to contain additional two monitors and keyboards within the next two years.
* Power controller requires sufficient desk space to perform calculations and make notes on paper.
* Gas control desk - identical requirements to power control desk i.e., two keyboard/monitors initially, space for two more within two years.
* General purpose workstation with desktop PC for performing calculations, writing reports, filing space.
* Engineering workstation with desktop PC.
* Physical separation of visitors/subcontractors from sensitive power control functions.
* Telephone on each desk.
Your Job
You have the job of providing management with a report to assist in making decisions about the control room redesign.
This report should include:
* A conceptual design for the workstations required.
* A proposed plan of the control room layout.
* Specifications for furniture to be purchased.
Your report should include the logical and/or empirical basis for your recommendations.
You will gain maximum benefit from this case study if you now spend some time thinking about, and researching, the issues involved, before looking at the sample design provided
Prompts
In your report it might be helpful to consider the following issues:
Clearly there are many potential design solutions and this sample design is not intended to be definitive, but rather is intended to stimulate students thinking about their own responses to the case study.
One possible design solution to the Control Room case study, and the rationale for this design, follows.
The conceptual design includes the following operational requirements as specified:
Power control system
A workstation with facilities for four desktop PC's and 21 inch monitors (two for existing lines, and two for future expansion). The workstation includes freestanding drawer unit which can be located under the desk initially, then alongside when the two additional monitors are in use. A moveable bookshelf could also be located on this drawer unit.
Gas control system.
This workstation will also initially consist of two keyboard/monitor configurations, with space for two additional monitors.
Miscellaneous.
The design incorporates the pre-existing rectiformer desks and mimic board. There are two additional desks, an engineering workstation with desktop PC, and a desk with PC and printer for general use.
The conceptual design is illustrated in the following drawings. The major features of this design are:
Power control station
* Power control from a custom designed desk with bank of four monitors (two immediately, two when expansion occurs), keyboards and pointing devices (trackball or joystick) allowing vision and operation of each by the Power Controller with minimal movement.
* Desk allows with sufficient area for keyboard and writing space.
* Telephone located on each desk.
* Monitors located on separate custom designed tables behind working surface for ease of relocation.
* Monitors located low, and at an angle to increase postural variation and allow clear vision of mimic board.
* Keyboards which incorporate a pointing device (trackball or joystick) should be used rather than separate mouse and mouse pad.
* Additional filing cabinets, draws, and printers located on the wall right of the mimic board.
Gas control system
* As for the power system operator's workstation, initially one operator workstation incorporating two desktop PC's and monitors expandable to include two further monitors.
* Other features as for the power supply control system operator's workstation.
Reception desk
* For issuing permits, and clearances. Includes a screen which, with the reception desk, provides a physical barrier to operational workspace.
* A sit/stand workstation to provide power controller with healthy postural variation and ability to easily move to the operational screens.
Furniture Specifications
* Dimensions of the custom designed desks and tables are provided in the following figures.
* The design concept for power control station, and gas control system, involves a four sided desk front to accommodate the four computer workstations.
* Separate tables to hold video display units at an (adjustable) angle should be attached to the rear of the desks.
* All other desks should be at least 90 cm deep, and 150 cm wide. Height adjustable (motor powered) desks may be desirable to accommodate multiple users, however adjustable seating may be sufficient.
* Chair specifications - seat width at least 45 cm, adjustable seat height (at least 42 - 52 cm), adjustable seat tilt (10 degrees forward, 5 degrees backward), adjustable height backrest (height of lumbar support 17 to 25 cm above seat), backrest angle adjustable (90 - 115 deg from horizontal), adjustments easily made in sitting position, five castors, rounded front edge of seat.
Schematic Plan of Re-designed Control Room.
Conceptual Design for Power Controller Desks
Ergonomic principles
Ergonomic principles which require consideration in the re-arrangement of the physical layout of the power supply control room include the following:
* Postural variability
The importance of allowing and promoting postural variability can not be overstated and, apart from operational requirements, should be the overriding ergonomic consideration. No single posture is beneficial if maintained for long periods (12 hour shifts make this especially important). The best workstation will allow a large range of comfortable postures to be adopted. One way of preventing restricted postural variation is for users to move between workstations. This is one reason for favouring a re-arrangement in which the power controller moves to a reception desk to issue approvals and permits. (The reception desk, and screens also have the function of separating visitors from the sensitive work areas.) Moving to remove printout similarly provides beneficial postural variation. The distances to be travelled should be minimised, however, by reasonably close proximity of the destinations.
* Workstation design
The detrimental effects of postural fixity are exacerbated by postures or movements which involve extremes of joint ranges of motion. Consequently the posture required by workstation should involve middle of range of movement, especially wrist and neck.
Use of mice is a common cause of excessive wrist deviation, and hence the design concept includes a recommendation that pointing devices such as a trackball or joystick should be incorporated within the keyboard. This has the added advantage of reducing the desk space required by the input devices, and increases the ease with which the input devices can be moved to clear space for desk work.
The human visual system is such that a relatively steep downward gaze is preferred (about 40 degrees below the ear/eye line). If the monitor is placed at eye level the user has fewer comfortable postures that can be adopted. For this reason video display monitors should be located relatively low and substantially inclined backwards. Monitor inclinations of this magnitude may require glare control measures such as diffusers or changes in lighting above workstation screens (reducing the number of fluorescent tubes directly over the workstation may be sufficient). Task lighting may be provided by shaded strip lighting below each monitor.
Adjustability of workstation dimensions is very important in multi-user workstations. Adjusting chairs is not a habitual activity for most people, and training in how to adjust the chairs and regular refresher courses are necessary to gain maximum benefit from the design.
Decisions about the re-arrangement should be taken in consultation with users. The next stage should be mock ups or simulations to ensure user acceptability prior to final implementation.
Dennis Ankrum
Director of Human Factors Research, Nova Solutions, USA. email: ankrum@aol.com
This paper first appeared in: Workplace Ergonomics (1996) 2(5), Sept/Oct. p 10-13.
A data entry clerk is comfortable with the monitor about 75cm from his eyes. The company's ergonomic guidelines, however, say viewing distance should be from 45 to 60 cm. A member of his company's ergonomics committee says that the viewing distance should conform to the guidelines.
Who is right? 45 to 60 cm? 50to 70 cm? An arm's length? Recommendations that place a maximum limit on viewing distance to reduce eyestrain all have one thing in common: they have no scientific basis. To understand why, let's look at how viewing close objects can contribute to eyestrain. When we look at any close object, our eyes do two things: they accommodate and converge. Both of these can contribute to eyestrain (Collins 1975 and Fischer 1977).
So what's accommodation?
Accommodation is when the lens capsule in your eye changes shape to focus on a close object. The eyes have a default accommodation distance, called the resting point of accommodation (RPA). That is the distance at which the eyes focus when there is nothing to focus on. In total darkness our eyes are set to focus at a particular distance, so that if the lights were turned on, an object at that distance would be in clear focus. The RPA averages 80cm. for younger people and gets farther away with age. In the mid 1980s it was thought that people would have less eyestrain if the monitor were placed at the distance that coincided with a person's resting point of accommodation. More recent research has shown that the RPA is not the only consideration.
Convergence
Convergence is when the eyes turn inward toward the nose when we view close objects. Convergence allows the image of the object to the projected to the same relative place on each retina. Without accurate convergence, we see double images. The closer the object, the greater the strain on the muscles that converge the eyes.
The visual system also has a resting point of vergence (RPV). It is similar to the resting point of accommodation, but it's the distance at which the eyes are set to converge when there is no object to converge on. It's also known as dark vergence. The RPV averages about 115cm when looking straight ahead and comes in to about 89cm with a 30-degree downward gaze angle. Recent studies by Jaschinski-Kruza (1988) and Owens and Wolf-Kelly (1987) have shown the stress of convergence contributes more to visual discomfort than the stress of accommodation.
Jaschinski-Kruza (1988) divided subjects into two groups, near and far resting points of accommodation. The first (near) group had RPAs of around 50cm. The second (far) group's RPAs averaged 100cm. Both groups worked on computers at viewing distances of 50cm and 100cm. As expected, the near group had less eyestrain working at 50 cm than the far group. But both the near and far groups had less eyestrain at the 100-cm distance. Both groups judged the 50cm monitor distance as "too near", and both groups accepted the 100cm distance. Although their resting points of accommodation were different, both groups had far resting points of vergence. Jaschinski-Kruza concluded that "the stress on the convergence system may be the crucial factor for visual strain."
When Jaschinski-Kruza measured performance, he found that both groups performed better at the 100cm distance than they did at the 50cm distance.
Research by Owens and Wolf-Kelly (1987) found that after one hour of near work, the resting points of both accommodation and vergence shifted to a distance closer to the eyes. The size of the shifts depended on the resting points before the near work: subjects who began the session with far resting points had the greatest inward shifts.
They found that the greater the inward shift in the resting of accommodation, the greater the reduction in visual acuity, or keenness, when viewing a distant target. Changes in the resting point of accommodation did not correlate with subjective eye fatigue. On the other hand, greater inward shifts in the resting point of vergence were associated with greater eye fatigue, but not with changes in visual acuity.
Then you work at close distances, the visual system adapts by bringing the resting point of vergence closer. That inward movement could be the visual system's reaction to fatigue. While continually viewing objects closer than the resting point of vergence has been found to contribute to discomfort, no studies have shown greater fatigue with viewing distances farther than the resting point of vergence.
What does this mean in practical terms?
Farther is better
If we just consider viewing distance, farther is better. So where do recommendations for maximum viewing distance come from? The arm's-length limitation most likely came from recommendations on monitor placement in cockpits. NASA Standard 3000 (1995) limits the displays that have associated controls. That is based on reach distance. While that is only for displays located close to their associated controls, the notion of reach distance has been used in other guidelines. Some try to justify a limit to how far away the monitor can be placed with the argument that if the screen is beyond a certain distance, you might not be able to read the letters.
It's clear that if you can't read the characters, the viewing distance is too great. Or is it? Instead of moving the monitor closer, why not make the characters larger? In fact, guidelines recommending close viewing distances can only encourage the computer industry to maintain relatively small characters. That in turn forces close viewing distances and can perpetuate eyestrain.
How close is too close?
It is difficult to set an exact limit for a minimum viewing distance. If sustained viewing closer than the resting point of vergence contributes to eyestrain, perhaps we should say that eye-screen distance should not be closer than the resting point of vergence. (On average, about 115cm away at horizontal eye level and 89cm away with a 30-degree downward gaze angle.)
But there are no cliffs in ergonomics (one inch closer and you fall): If your RPV at a 30-degree downward gaze angle is 89cminches, your eyes are not going to fall out with a viewing distance of 75cm.
Somewhere between your resting point of vergence and 6 inches in front of your nose, you are going to experience discomfort. That distance is a combination of gaze angle, how long you've been working at the computer, your individual visual system's capabilities, and a number of other factors.
Does this help to put an absolute number on how close is too close? I recommend at least 65cm. But some people are not bothered by closer viewing distances.
How far is too far?
The reality is that there is no limit, based on visual fatigue considerations, to maximum viewing distance at computer workstations. From what we know about visual strain, farther viewing distances are better, at least up to the RPV. For example, if the RPV is 89cm, an eye--to-screen distance of 65cm is preferred to 50cm, 89cm is better than 25 inches. Viewing distances beyond 89cm (the RPV in this case) should neither increase not decrease eyestrain.
To allow for greater eye-to-screen distances, we need software programs and monitors that allow font sizes to be increased easily. We need guidelines that don't force people to sit closer to their monitors than the distances at which they are comfortable.
References
Collins, C., O'Meara, D., and Scott, A.B. (1975). Muscle strain during unrestrained human eye movements. Journal of Physiology, London, 245, 351-369.
Fisher, R.F. (1977). The force of contraction of the human ciliary muscle during accommodation. Journal of Physiology, London, 270, 51-74.
Jaschinski-Kruza, W. (1988). Visual strain during VDU work: the effect of viewing distance and dark focus. Ergonomics, 31, 10, 1449-1465.
NASA, (1995). NASA-STD-3000, Man Systems Integration Standards, National Aeronautics and Space Administration: Houston.
Owens, D.A., Wolf-Kelly, K. (1987). Near work, visual fatigue, and variations of oculomotor tonus. Investigative Ophthalmology and Visual Science, 28, 743-749.
HEALTH & SAFETY TRUST
1998 GRANTS
APPLICATIONS NOW INVITED
The Australian Occupational Health & Safety Trust is established as a non profit making organisation the purpose of which is to provide grants to occupational health & safety practitioners.
These grants will be awarded for projects involving the development of occupational health & safety education/training material or for personal professional education.
The ultimate goal of any grant must be to improve the health & safety of people at work in Australia and therefore preference may be given to projects relating to the development of educational material due to its wider application in this area.
Written applications should state the objectives of the project, its methodology and how this will benefit people at work in Australia. Applications must conform to the Guidelines for Grant Applications, available from the Trust at the address below.
Applications will be judged on merit. The decision of the Trust will be final in all matters associated with the awarding of grants. Trustees are ineligible.
Upon receipt all applications will be acknowledged in writing.
The Australian Occupational Health & Safety Trust
PO Box R804
SYDNEY NSW 2000
Telephone: (02) 9220-6374
Closing date for applications - 30th September 1997
A selection of electronic resources.
A New View of Statistics
Will Hopkins
http://www.sportsci.org/resource/stats/index.html
[Editors note: This is an electronic text which presents the use of statistics in an exceptionally clear and comprehensible fashion. The complete text is available free, although for a fee subscribers receive additional benefits (spreadsheets etc) and local hosting (faster access). Although the examples are drawn from sport science, the principles hold for ergonomics (indeed the distinction is rather artificial to my mind).
I recommend it highly, and would be interested in comments (a comprehensive review perhaps?) from others.
The text is part of the SportScience site (http://www.sportsci.org/) which is also well worth a look as an example of what a dedicated bunch of people can achieve via electronic publishing.]
From the introduction to the text:
"The title has a double meaning. If you're new to stats, most of what you read here will be a new view. But even if you have done stats--or if it has done you--there's plenty here that's new. For example, I've discarded all details of computation, in the hope that you will get a better understanding of the concepts. Let's leave the computations to the computers! You'll also find a new unified treatment of effect statistics and their magnitudes, a new emphasis on validity and reliability, new valid methods to calculate reliability, a new exalted position for confidence intervals, a new attack on statistical significance and hypothesis testing (hard climbing, and the view is obscured by cloud), a new way to understand all statistical models, a new simple treatment of non-parametric analyses, a new method of doing repeated measures with missing values (yes, it's true!), new simple ways to estimate sample sizes, and best of all, a highly ethical new way to reduce sample sizes. And as you may have noticed, I am blazing a trail with the use of plain language for a text of this sort."
NIOSH Exploratory Study on Keyboard Design Finds No Major Differences in User Comfort, Fatigue
http://www.cdc.gov/niosh/keyboard.html
From the web page:
Results of an exploratory study by the National Institute for Occupational Safety and Health (NIOSH) suggest that computer keyboard design is not a significant factor for user comfort.
Over a two-day period, test subjects reported no significant differences between the levels of discomfort and fatigue they experienced when using conventional video display terminal keyboards and three alternative keyboard designs, the study found. In all cases, reported levels of discomfort and fatigue were low.
Additional research is needed to determine whether alternative keyboard designs have benefits for users over longer periods, or for workers who previously have reported symptoms of discomfort or fatigue, NIOSH said.
"The findings of this study provide a vital ingredient for the further research needed to protect keyboard users effectively from upper-extremity musculoskeletal disorders," said NIOSH Director Linda Rosenstock, M.D., M.P.H.
NIOSH has conducted pioneering studies on VDT safety issues for several years, including a landmark 1991 study that resolved long-standing questions about potential effects from VDTs on reproductive health. The institute also has led research to find practical, effective ways to protect workers from job-related musculoskeletal risks.
Some scientists have theorized that changes in keyboard configuration can protect users from musculoskeletal disorders of the wrist, arm, and back. Within the past five years, as public concern over musculoskeletal disorders has grown, manufacturers have begun to market keyboards based on alternative designs. However, few studies have examined actual performance on alternative keyboards to evaluate whether they are more beneficial than standard designs in preventing discomfort, fatigue, and strain.
The NIOSH study involved 50 female clerical workers, each of whom typed on either a conventional keyboard or one of three alternative keyboards. The alternative units all featured a split design -- keys for the left hand and the right hand were on separate panels -- but differed in other ways. The study will be published soon in the peer-reviewed technical journal Applied Ergonomics.
Results of the study, conducted by NIOSH researchers Naomi G. Swanson, Traci L. Galinsky, Libby L. Cole, Christopher S. Pan, and Steven L. Sauter, are available from NIOSH by calling toll-free 1-800-35-NIOSH. For further information on other NIOSH research on VDTs and musculoskeletal issues, contact the toll-free number or visit the NIOSH home page on the World Wide Web, http://www.cdc.gov/niosh/homepage.html.
Ergonomics Program Management Guidelines for Meatpacking Plants (1993)
U.S. Department of Labor
Occupational Safety and Health Administration
The complete document in PDF format (40 pages, approx 4 Mb) is available at http://www.osha-slc.gov/Publications/Osha3123.pdf
Correction to Vibration Data Base URL
Ian Gibson notes: "in February's issue there was a reference to a vibration database (on p. 29) from Ronnie Lundstrom. The URL was incorrect, it should be http://umetech.niwl.se, not umtech. It's in Swedish at present, but Lundstrom expects to have an English translation in a few months time"
A Comparison between Anir&tm; Mouse & a traditional mouse.
Arne Aarås,
Alcatel STK
http://www.animax.no/reportamouse.html
[Editors note: Complete text of research report comparing electromyographic data from subjects using a conventional and alternate mouse. The report includes a good discussion of the issues involved with mouse use, and how electromyography might be utilised to assess benefits of alternate devices.]
May 1997
Ten goals for IEA revisited
MARTIN HELANDER
President IEA
In 1994, at the beginning of my term as President, I suggested 10 priorities for work within the IEA. Now I take the opportunity to report on the progress.
1. IEA is the only global organization in ergonomics and must take a global responsibility. Partnerships with other international or regional organizations (United Nations, the European Union, ASEAN) are important but few of them employ ergonomists. Their awareness should increase. During 1996-97, our contacts with ILO and WHO improved. This is partly due to new officials at ILO and WHO but also the joint publication of Ergonomics Checklists with ILO. Since international organizations have frequent reorganizations and shifting personnel and funding, IEA must take the initiative in contacts rather than wait for them to contact IEA.
2. IEA must continue to support ergonomics in countries where it has been neglected. The first reason is we want to help. The second is that IEA wishes to recruit more members. During the period 1994-97, IEA training activities increased. Courses were held in Columbia (2 courses), Hong Kong, Indonesia, Lithuania, Latvia, Malaysia (3), South Africa (5), Thailand (2) and Vietnam. Some courses were in cooperation with ILO and WHO. The IEA Committee for IDCs (Kamil Vanwonterghem, chair) was responsible for these activities. Through the aid of the Brazilian and South African Ergonomics Societies, IEA has been able to stimulate ergonomics interests elsewhere in South America and Africa. The book donation program has flourished. There are now 10 library depositories in IDCs. The student chapter of the HFES at SUNY Buffalo, USA (email: CALEVINE@ACSU.BUFFALO.EDU) has been very active in this effort. However, we need more donations of current textbooks and journals.
3. IEA must remain at the forefront of technological and scientific development. The Science and Technology Committee (Holger Luczak, chair) has the main responsibility; it has 16 Technical Groups, which can arrange conferences, symposia and meetings of experts. During 1994-97, IEA sponsored 17 conferences; many were held in IDCs. The Technical Groups also helped in soliciting half of the papers presented at IEA 97 in Tampere.
4. IEA must disseminate information about ergonomics on a global scale. The newsletter Ergonomics International (Stephan Konz, editor), published quarterly, has been distributed four ways: 1) by email to the newsletter editor of each of the national ergonomic societies, 2) by the IEA www page, 3) by being published in the journal Ergonomics, and 4) as an offprint mailed to about 250 people around the world. The Publications and Promotions Committee (Waldemar Karwowski, chair) has established the IEA Press. Its main objective is to make conference proceedings available at low cost. The IEA Journal of Ergonomics Research and Applications will appear as a refereed Web journal; its goal is to make information available at low cost, especially to IDCs.
5. IEA must facilitate global communication among professionals. IEA was well-prepared for the global communications explosion through the Internet. Communications among Council members and the IEA Executive are faster and there is no postage for email. The IEA has a web page as do some of our Federated Societies. By linking these pages, communication will be enhanced. The first Web conference in ergonomics, the CybErg conference, organized by Curtin Univ., Australia, was a great success.
6. IEA must continue its leadership role in developing guidelines for educational accreditation and professional certification. Draft certification guidelines were developed by the Education and Training Committee (Margaret Bullock, chair). The final guidelines, intended as recommendations for adoption by Federated Societies, should be available in 1998.
7. IEA must increase its budget and seek new funding sources. Since IEA relies on volunteers, it is possible to accomplish much at little cost. But lack of funding restricts IEA activities so IEA needs to seek additional sources of revenue.
8. IEA Council members must be given opportunities to help with IEA business, for example by doing special projects. Although some Council members are active on committees, there are many opportunities for special projects of limited duration. I would like to encourage Council members to suggest activities in which they would like to become involved. For example, see the following item.
9. IEA must actively recruit new members. In addition to ergonomic societies from more countries, the IEA needs to increase the number of Sustaining (company) members. This activity rests with the Policy and Development Committee (O. Brown, chair).
10. IEA must promote an international debate on the goals of ergonomics. Ergonomics is practiced differently in different counties. Ergonomics must be explained so that the message is easy to understand for non-professionals. With a clear message, we can promote ergonomics more effectively. One possibility is "The purpose of ergonomics is to design systems that enhance productivity, safety and user satisfaction."
HCI
The HCI Technical Committee of IEA was reorganized in January 1997. The goal is to promote international exchange of scientific and technical information concerning Human-Computer Interaction. For information, comments or questions concerning membership of the Technical Committee, contact Prof. Susuma Saito at saitos@niih.go.jp
Psychosocial/organizational factors
An electronic network is being set up for people interested in psychosocial and organizational factors in occupational and environmental health. The plan is to have an electronic newssheet and maintain a www page. The web page would include concise articles. Contact mca@easynet.co.uk
Maslow
Abraham Maslow is famous for his hierarchy of needs (nourishment, safety, social, esteem and self-actualization). He studied further on how to implement self-actualization to the workplace. He decided teams were important. His thoughts, long out of print, are now available on the web at http://www.scarlet- fire.com/maslow
Cognitive systems
Human Factors is soliciting papers for a special issue on "Cognitive Systems in Context". Manuscripts should illustrate or sample the diversity of Cognitive Task Analysis or Cognitive Field Research methods in the context of particular content areas, projects or results. Papers are due 31 August 1997. For information, contact Prof. David Woods, ISE Dept., Ohio State Univ., Columbus OH 43210, USA. Email: woods@csel.eng.ohio- state.edu
Anthropometry
Civilian and European Surface Anthropometry Resource (CAESAR) is a cooperative research effort which will use a whole-body laser scanner to generate 3-D data. The plan is to measure 8000 subjects in the USA, 3000 in the Netherlands, and 3000 in Italy. Potential applications include 3-D computer-aided design in areas such as automated manufacturing, rapid prototyping, crash dummy modeling, virtual design and evaluation, and computer animation. They wish to have other organizations join the study. Contact Gretchen Stokes, SAE, 400 Commonwealth Drive, Warrendale, PA 15096-0001, USA. Email: stokes@sae.org
Product design
Kansei is a Japanese word which implies a customer's psychological feeling and image regarding a new product. Nagamachi has developed a detailed design procedure to turn consumer's preferences into product designs. For example, on the Miyata automobile by Mazda, the basic concept was "human-machine unity". This was divided into four 1st level concepts "Tight feeling", "Direct feeling", "Speedy feeling" and "Communication". One of the subconcepts of Direct Feeling led to semantic-differential testing of various "My own way of controlling" which led to a shift lever length of 9.5 cm. Speedy Feeling led to the concept of "pushing forward" which led to the sound from the exhaust pipe which led to the selection of an exhaust pipe with the most satisfying power sound. The first Japan-
Korea Symposium on Kansei Engineering was held 17 February 1997 in Tokyo. Among the examples were:
* Shapes and Impressions of Vehicle Front-End Design
* Spectacle Design
* Cross Cultural Preferences for Colors and Forms
* Application of Kansai Engineering to Kitchen Design
* Seating Discomfort
The 140 page Proceedings (in English) is available through Prof. Nagamachi, President, Kure National Institute of Technology, 2-2-11, Aga-minami, Kure, Hiroshima 737, JAPAN. Email: mituo@kure-nct.ac.jp
1997
November 25-27. ESA National Conference Gold Coast (see elsewhere in this issue).
1998
The preferable form of submission is via email, either in the body of a message, or as an attachment. Files may also be mailed on floppy, (or Zip disc if very large). Virtually any format of files can be accommodated. Otherwise contributions should be printed in a large (14 pt preferred) non-serif font (such as Helvetica) and faxed to +61 07 33793545. Printed pages of similar specification may also be sent by post. Handwritten submissions will only be accepted in exceptional circumstances.
Intending contributors are invited to contact the editor to discuss potential submissions.
All enquiries or feedback should be addressed to the editor, Robin Burgess-Limerick PhD.
Email: robin@hms.uq.edu.au
These pages are hosted by the Department of Human Movement Studies, The University of Queensland, AUSTRALIA