Voicing Psychology at Big Blue
BY JAMES R. LEWIS
As an engineering psychologist at International Business Machines Corp. (IBM), I’m involved in commercializing the voice and speech technologies developed by IBM Research. These technologies include speech recognition, artificial speech production, speaker identification and translation. This work provides a tremendous opportunity to apply psychology – primarily cognitive psychology and psycholinguistics – to the design of human-computer interfaces. These opportunities include (but are not limited to): Developing design principles from the psychological and human factors literature, and applying the empirical methods of psychology to product design and to the development of product assessment instruments.I suppose most people have followed some strange path to get from where they started to where they are. My involvement in software development began in high school when I participated in New Mexico’s First Annual Programming Competition. The competition included a crash course in FORTRAN, after which I decided that whatever I did with my life, it wouldn’t be computers. That didn’t last, however, and for my master’s thesis in music at New Mexico State University, I produced NMSU’s first computer-generated musical composition. The programming language was MUSIC4BF, which was based on FORTRAN!
NOW THAT’S HARD SCIENCE
Because I had taken a number of psychology electives, I could get an undergraduate degree in psychology with the investment of one additional semester, making work in the emerging field of music therapy a possibility.
In that additional semester, I took all of the ‘hard’ psychology courses I’d been avoiding – perception, learning, experimental design, and introductory statistics. That semester changed my life: I hadn’t realized the elegance and beauty of experimental psychology-the intellectual challenge of using statistical methods to extract meaning from a collection of numbers generated by conducting an experiment. I ended up going for my master’s degree in experimental psychology, specializing in engineering psychology.
I chose engineering psychology because the job market for this type of applied psychology was growing rapidly in the late 1970s and early 1980s, and I was tired of college (and broke). This was one of the few areas in which a psychologist without a PhD could easily find work.
At the time, four major fields were hiring engineering psychologists: the military, aviation and aerospace, nuclear power plants, and computer companies. During my second year in graduate school in psychology, I had a summer internship with IBM in Boca Raton, Florida, working in their human factors department. After going back to finish my thesis (in the area of motor control), I returned to IBM in Boca Raton as a permanent employee.
RIGHT PLACE, RIGHT TIME
In 1993, IBM supported my part-time attendance at Florida Atlantic University as a doctoral candidate in experimental psychology. In 1996 (at the age of 43), I received my PhD in experimental psychology, having studied psycholinguistics under Lewis Shapiro and Betty Tuller. This was fortuitous, because it was at about this time that IBM began serious development of commercial products based on its speech technologies, and all of IBM’s human factors engineers in Boca Raton began working in this area.
Since 1980, I’ve worked on many interesting projects, including:
- IBM’s first attempt to produce a computer system for which ease-of-use and ergonomic design was a primary focus (System/23 – a precursor to the modern personal computer)
- Physical input and control devices, including IBM’s converged keyboard for personal computers, mice, tablets, touch screens and in-keyboard pointing sticks
- Assessment techniques for determining the relative usability of competitive products
- A portable tablet computer with satellite communications for long-haul truck drivers
- A cellular telephone with touch screen interface and personal organization software (Simon)
- IBM’s first continuous speech dictation system (ViaVoice)
- A formal grammar for natural-language-like control of a word processor
- User interface guidelines for Voice XML applications
THINKING ABOUT A CHANGE?
If you’re working at a university, maybe you’ve been thinking about making a change to industry. The current technology boom means there are certainly plenty of positions available.
I expect it is very different from the normal life of a university professor, though. Less vacation time, for one thing. Another is the need for approval for anything you publish. (If you hate the pressure of publish-or-perish, this might sound refreshing.) That’s not to say publishing isn’t encouraged: Over the years, I have been able to publish over 150 technical papers, including a large number of IBM technical reports (both classified and unclassified) and technical disclosures, a book chapter in the Handbook of Human-Computer Interaction, eight papers in refereed journals, 26 conference papers, and have had ten US patents issued (with about 40 more pending).
You’ll have to get used to the D-word: DEADLINE-Here’s what it sounds like:
“If the product isn’t ready in May then it won’t be in production and on store shelves in August in time for the back-to-school rush!” Or….
“If we don’t get this product out in time for Christmas, then we might as well not make it at all!”
I thought I was going to hate deadlines, but they’re what help make working in product development demanding, exciting and fulfilling. When you’ve got a deadline, you have to make things happen and you have to be as efficient as possible. You develop the flexibility to adapt when you need to throw away all of your carefully-considered research plans to tackle an immediate crisis. You must be willing to provide the best judgement you can in the absence of confident knowledge.
In dealing with the demands of product development, you come face-to-face with the limits of psychological knowledge, which is often better-suited to the development of scientific theories than to meeting the quantitative requirements for effective human factors engineering. It is a setting that can either drive you out or drive you to high levels of productivity and discovery. I wouldn’t trade it for any other.
BY MARY MANNION-PLUNKETT
My background in psychological science has provided me with an ability to be a critical thinker and to ask appropriate questions, but to do so without being critical of others. In completing my advanced degree, I learned to synthesize a lot of information, and discern the useful and valid from the chaff. I’m not one to take the latest and greatest “fad of the month” or “most favorable consultant” at face value.I am part of the Boeing Leadership Center, an organization chartered to develop today’s and tomorrow’s Boeing leaders. My area of responsibility is Customer Relations, which includes evaluation/needs analysis and research. In the area of evaluation, we have a number of innovative “experiments” underway to measure our leadership development programs; among these are: a follow-on interview process that includes not only everyone in our organization but also a sample of our alumni; pre-post 360 competency assessments with our executives related to their participation in our two-week Executive Program; leveraging employee survey data to compare program graduates to non-graduates (aggregate and organizational level data); and other similar activities.
In the research arena, my colleagues and I are working on an exciting longitudinal study that looks at how Boeing leaders grow and develop over the life of their careers. Expanding on the work of previous researchers from the Center for Creative Leadership (a kind of a “think tank” for leadership development education and research, headquartered in Greensboro, NC), we are identifying key events that our first, middle, and executive-level leaders cite as significant turning points in their careers, and documenting the associated lessons learned from these events. In addition, we are analyzing the situational factors and personal strategies used by the leaders to navigate through those events.
Eventually, we will be defining and identifying predictors of advancement or derailment, as well as providing findings related to core values, mentoring advice, current challenges, etc. The ultimate goals are to provide our participants in the longitudinal study with a set of useful tools with which they can begin to chart their career path and/or mentor and coach others, and to provide the organization with an effective “on-the-job-development” program.
I can’t say I ever expected to be in this job; that is, when I graduated, I didn’t say ‘I want to be a Senior Manager for Customer Relations for The Boeing Leadership Center, working in The Boeing Company’s Corporate HR function.’ I arrived here in a bit of a circuitous way, although most career journeys probably are circuitous. Like many of the participants in our longitudinal study, it is only in retrospect that I can identify certain experiences and associated lessons as being significant. However, all my positions have some common elements: Appropriate challenge, ability to learn new things and interact with new people, knowledgeable colleagues, flexibility (rather than a rigid job description), opportunities to accomplish something while doing what I like to do, and supportive bosses.
I’m not as current with the field as I’d like to be. I try to keep up with journals such as the Journal of Applied Psychology and Personnel Psychology, although I’m several months behind. I recently began teaching at a local university as an adjunct professor, which has helped connect me to the “latest and greatest” of theories and research. Plus, I work with some great colleagues who will often pass along articles of interest.
I don’t get to be as “pure” in my science as I imagine my academic colleagues can be, especially when it comes to research design and statistics. Although I have several projects under way which I deem “experiments,” they are far from being true experiments. The longitudinal study is a good example: Our volunteer participants are not “subjects,” rather we consider them “partners.” As such we interact with them, and provide them information and tools along the way which we intend to have an influence on them. I also think universities – good ones, anyway – encourage people to question; sometimes our business culture doesn’t promote this as much as we could. On the other hand, I’m in a real live laboratory, and no two days are the same. Given our dynamic business environment, flexibility and agility in all that we do is a real key.
Here’s my advice to people thinking about a job in the private sector: Take a broad view when you’re thinking about the kinds of jobs that might be relevant to your background; take the approach that any job could be relevant. What’s most important is to know what your skills are, and what interests and excites you. Then, figure out how to translate this into some tangible results for the company you want to work for. In other words, sell yourself to the company of your choice by demonstrating the value you can add to the bottom line. It’s not enough to have “credentials”-you need to translate those credentials into “this is what I can do for you.”
Insuring Psychological Science
BY ROGERS TAYLOR
State Farm has been actively involved in applied psychological research since the 1950s; I’ve been here at the corporate headquarters of the State Farm Insurance Companies since 1967. Between 1967 and 1999, the research department grew from about six employees to close to 100, and the research function had enlarged its scope of activities considerably beyond the area of human resources research, to include units concerned with auto collision technology, building technology, market research, consumer research and business/economic/competitive research.I supervise a group of applied researchers in two areas of activity: one is reasonably well described by the term “human resources research” and the other by the term, “usability.” In the area of human resources research, we use the methods of psychological science, measurement and experimental statistics to develop, collect, analyze and summarize empirical data in a form that can be understood and used by executives. The process of developing such information has afforded those of us who make up the Human Resources Research unit many opportunities to contribute to psychological science over the years, but that is a secondary purpose of the group; our primary purpose is to assist executives with business decision-making.
My staff and I have presented papers at professional meetings, participated in workshops, panels and symposia at annual meetings of The Society for Industrial and Organizational Psychology, have published articles in professional journals, and have contributed a chapter to a book. At State Farm, as is the case with most business organizations I have been acquainted with, professional activity is encouraged to the extent that it does not interfere with work activities seen as more directly related to the purposes of the organization.
HAVING AN IMPACT
Working as psychological scientists in an organization like State Farm presents both opportunities and challenges. Among the opportunities, working internally allows a person to have an important impact on an organization. Early in my career I was able to complete one of the first studies in the industry dealing with validity generalization. Later, I suggested to an insurance trade association that they sponsor a consortium validation study of employment tests for the entire insurance industry, to pursue the topic of validity generalization on a larger scale. This research ultimately became one of the first industry-wide studies completed in the United States.
The Human Resources Research unit spent several years helping an association of private hospitals complete a longitudinal study of treatment effectiveness for chemical dependency. More recently, the unit became heavily involved in the use of computers for administering employment tests, and has published in that area. As part of that research, members of the unit developed, validated and implemented a computer adaptive test for use in the selection of computer programmers.
On numerous occasions, research performed by the unit to evaluate the effectiveness of programs has resulted in the programs being dropped, saving the company substantial amounts of money that would otherwise have gone to those ineffective programs. Alternatively, much of our program evaluation research has suggested ways to improve already-effective programs or to guide decisions about which among competing programs should receive additional funding.
The Human Resources Research unit also is responsible for developing and evaluating the company’s employee opinion survey program, an important method of employee temperature-taking used by the organization. Questionnaires assessing work related attitudes and opinions are administered to all employees approximately every three years. New norms for this survey are developed annually using sample data. Working with an outside vendor, State Farm will soon have an almost fully automated survey capability utilizing the company intranet for questionnaire administration. A final area of work activity in which members of the unit staff are engaged involves consulting with other areas of the company, such as Auditing, on matters relating to sampling, experimental statistics and mathematical modeling.
I should note parenthetically that only a fraction of the survey work performed by the Research Division is conducted by members of my unit. In addition to the attitude survey program managed by my unit, the Research Division includes a different unit devoted to qualitative research and to surveying. This group works closely with the company business units of the company to conduct focus groups and surveys both within and outside the State Farm organizations, e.g. with claimants and with policyholders. This group also hires psychological scientists, with many members holding bachelors or masters degrees in psychology.
As I noted above, the other major area for which I am responsible rubric is Usability. Usability is a sub-area of Ergonomics, is closely related to Human Factors, and deals with “the making of devices and programs convenient and practical for a user doing work” (Bennett, 1979, p. 5). Bennett suggested the term “usability” as a replacement for the previously used expression “user-friendly.” The Usability area of the unit operates two labs at State Farm corporate headquarters, consists of 6 employees at the present time, one of whom is a PhD, and is actively seeking to hire another PhD-level Ergonomics, Human Factors or Cognitive Psychologist. The Usability function is looked upon increasingly by the organization as a valuable way of ensuring that products and materials work the way they were intended to work once they get in the hands of end users.
VARIETY OF PROJECTS
I mentioned previously that working as a psychological scientist within an organization like State Farm is ripe with opportunities and with challenges. The variety of the projects we have become involved with is a major opportunity. Over the course of my years with the company, I have taken on projects covering most of the topics included in the table of contents of textbooks in the area of industrial and organizational psychology. Another advantage, I believe, is being able to work on projects that are important to the organization. Yet another advantage to working in a large company is access to sample sizes that are generally quite adequate for our purposes and do not consist of undergraduate students.
Another advantage is that we rarely have to actively sell our services. We usually are sought out by managers or executives who have questions that need empirical answers. While being sought out to provide a service is an advantage when it comes to filling one’s work day with meaningful work, it is a disadvantage when it comes to conducting programmatic research. Because of the wide variety of topics we research, there tends to be little time to acquire expert knowledge in any single area or research field. I have always found it personally satisfying, however, to be able to concentrate on performing research in my field without having to fit that research around other activities such as course preparation and teaching.
VEXING, YET GRATIFYING
The challenges in working in a business organization can sometimes be as vexing as the opportunities for doing meaningful work can be gratifying. Once I had to wait 10 years after completion of one project before being allowed to publish the research. The importance of sharing of information with colleagues, on which science depends, is foreign to many business people. It is more common to find executives in business organizations who treat research findings the way an inebriate treats a lamp post, more for support than enlightenment!
Research in business organizations also tends to be used for purposes of advocacy, with pressures to not find “truth,” but to obtain results that strengthen one particular point of view. In such instances, research findings sometimes become clubs which one executive uses to figuratively beat another executive as they espouse different sides. Being in the middle of feuding executives is a very uncomfortable position. On more than one occasion I’ve been asked to “bet my job,” that research conclusions were correct (which is especially disquieting when reporting results having a .05 alpha level!).
It is not unusual to develop and validate test batteries and interview protocols, implement them and, a year later, see their use halted because the job market has become tight and openings more difficult to fill. I have also come to believe at times, however erroneously, that all I needed to do to put a halt to hiring at State Farm was to initiate an in-use study of the predictive validity of one of our selection tests!
Please don’t think that the above challenges and frustrations are unique to State Farm or that State Farm is worse than any other business organization in its treatment of research-based information (or researchers). In fact, State Farm, because it is a mutual company and not subject to the short-term demands and expectations of stock owners, has always impressed me as one of the saner companies a researcher could work for.
I would like to make another point by way of advice-giving, and that is to encourage individuals anticipating either a move into industry or a career as an industrial-organizational psychologist to develop strengths in the areas of psychological measurement, experimental design, quasi-experimental design, and multivariate statistical analysis. Not only will such skills be invaluable as you are presented with a wide array of topics to research; they will serve to differentiate you from people trained in other disciplines against whom you may be competing, either for a job or once actually on the job. Business schools are offering business and industry individuals trained in organizational behavior as alternatives to persons trained as psychological scientists.
In my opinion, psychology departments offer better training in the design of experiments and in methods of data analysis than is available to most business school students. I would urge you to take full advantage of the depth and breadth of training in research methods you can receive through psychology departments.
PREOCCUPATION WITH QUALITY
My next comment may come across as drum beating and I apologize if it does, but my group and I have endeavored where possible to do work of sufficiently high quality that we would not be embarrassed to have it reviewed by a jury of our peers. We do this for two reasons. First, because it has made it much easier to get up in the morning and to come to the office over the years knowing that we are attempting to do good work. Ideally, I would like our work to be replicable, to have ties where possible to theory, to involve reliable measures of constructs, and to possess internal validity. I also urge the use of statistical analyses that have been thought through in a advance as contrasted with having been applied because there is a computer program that will run them. “Quick and dirty” research can be done by almost anyone. The value we add to the research process as psychological scientists is to provide the organization with research of the highest quality possible.
A second reason for my preoccupation with quality is that we have in fact found our work challenged by others after the fact, when no challenge was anticipated before the research was completed. This tends to occur when our findings were not those desired by managers or executives espousing a different viewpoint in an organizational disagreement.
As my time as a graduate student at Purdue came to a close, I shopped around for employment and found opportunities readily available. I decided to accept State Farm’s offer because I sensed a chance to get in on the ground floor of an applied research endeavor which had the potential both for career growth and for having an impact on a work organization. I have now been with State Farm almost 33 years and the opportunities for professional growth and for having an impact have exceeded even my optimistic expectations. I have also been able to have a family and, with my wife, to raise two fine sons and to live comfortably on my income. While no work career spanning more than 30 years is entirely a bed of roses, I feel that my choice of vocation and of employer were good ones and were ones that I would recommend to any scientific psychologist having similar interests and ambitions.
(A list of selected papers and publications of staff members from the State Farm Human Resources Research unit is available from the author at rogers.taylor.av02@StateFarm.com.)
Bennett J.L. (1979). The Commercial Impact of Usability in Interactive Systems. In: Shackel B Man/Computer Communication: Infotech State of the Art Report,2, 1-17. Maidenhead, Infotech International.