Continued article from the The Behavioral Measurement Letter, Vol. 3, No. 2 – Spring 1996
Al K. DeRoy and Fred B. Bryant
For many years psychologists and educators have used computers to facilitate measurement, first in automated scoring and analysis, and later in test administration. More recently, computers are being used as actual measurement tools. In this column we describe use of the computer to measure variables directly using the computer mouse (“the mouse paradigm”), record and store the measurements made, manipulate and analyze measurement data, and report experimental results.
This innovative computer application was developed by Vallacher, Nowak, and Kaufman (Journal of Personality and Social Psychology, 1994, 67, 20-34), who describe the use of the computer mouse to measure dynamic variables in social judgment. Vallacher et al. present a rationale for viewing social judgment phenomena as dynamic rather than static once the social judgment system is activated. Because the purpose of this column is not to discuss social judgment, that rationale is not examined here. Rather, in this column we discuss the mouse paradigm, describe its use in measuring the dynamics of social judgment, and list other potential applications.
Most prior research on social judgment has involved measurement of judgments as snapshots in time and the measurement of changes in judgment over relatively long periods of time (days, weeks, months). As discussed in Vallacher et al. (1994), this research suggests that social · judgment is typically stable over such long time periods. As indicated by cognitive theory and empirical research, however, social judgments are unstable over the shorter periods of time when they are being formed or re-examined. To test this idea, a tool was needed to measure changes in social judgment over very short time periods – intervals of seconds and fractions of seconds. Thus, Vallacher and his colleagues developed the mouse paradigm for this purpose.
The mouse paradigm is based on the idea that social judgment involves approach-avoidance, a concept put forth by Hovland, Janis, and Kelley (Communication and Persuasion, Yale University Press, 1953). This notion implies that one’s feelings about a person, object, or event at any point in time can be represented as the distance between a symbol designating the target person, object or event, and a symbol designating the judge. The nearer the judge symbol to the target symbol, the more positive the subject feels about the target person, object, or event; the farther the judge symbol from the target symbol, the more negative the subject feels about the target. If the judge symbol is neither near nor far from the target symbol, i.e., itis somewhere closer to the center than either the positive or negative pole, it may be inferred that the subject has ambivalent feelings about the target.
Given the above, Vallacher et al. (1994) innovatively employed the computer mouse to measure moment-to-moment changes in distance between the target and judge symbols to determine the judge’s continually changing feelings toward the target person, object, or event (see, also, Vallacher & Nowak, Dynamial Systems in Social Psychology, 1994, pp. 257-69). As described in these references, a small circle, representing the target, and the computer mouse pointer (an arrow symbol), representing the judge, are initially positioned in the middle of the computer screen. The circle remains stationary, while the arrow is mobile reflecting movement of the mouse. Judges are asked to think about a particular person (or object or event) with whom s/he is familiar. This target is to be someone (or something) toward whom the judge feels either positively, negatively, or ambivalently (both positively and negatively), depending upon the experimental condition being tested. The judge is then asked to move the arrow (by moving the mouse) toward or away from the circle, or to allow the arrow to remain in one place, depending upon how s/he feels about the target individual at the moment. Cartesian coordinates, representing the position of the mouse pointer, are collected ten times per second over a two-minute period, yielding a total of 1,200 data points per session.
The mouse data are then transformed through use of a FORTRAN program into four measures of social judgment: linear distance of the arrow from the target (measured in pixels), representing the judge’s overall evaluation of the target; average speed of travel of the arrow (measured in pixels per O.1 second), representing change in the judge’s evaluation of the target; acceleration/ deceleration of mouse travel (measured as change in number of pixels traveled per 0.1 second), representing the rate of change in the judge’s evaluation; and time the mouse remains at rest, representing the amount of time the judgment system is at equilibrium.
The mouse data are further manipulated using a C++ program to calculate the dimensionality of the social judgment system. Dimensionality is a measure of a dynamic system’s order (complexity) using the temporal trajectory of a single variable. The program uses the Grassberger-Procaccia algorithm (Physical Review Letters, 1983, 50, 346-350), which determines the number of variables required to specify the state of the system. This algorithm is used to distinguish between a random system (many variables, low order) and a deterministic system (few variables, high order). Because human judgment must be exercised in separating noise in the system (error variance) from dynamic functioning of the system, guidelines provided by Ben Mizarahi, Procaccia and Grassberger (Physical Review Letters, 1984, 50, 975-977) may be used to assist in making such determinations.
In their experiments, Vallacher et al. (1994) calculated dimensionality in the beginning (first 40 seconds) and ending (last 40 seconds) portions of two-minute sessions. Because they found that dimensionality was best estimated using only every other data point, dimensionality was calculated for both even-numbered and odd-numbered 0.1 second intervals. The two sets of scores were found to correlate highly (p < 0.001) across subjects, r(71) = .88 and .82 for beginning and ending period scores, respectively, indicating that the dimensionality measure is reliable.
To date the mouse paradigm has been used to explore the dynamics of social judgment phenomena and probe the structure of the social judgment system. By logical extension, the mouse paradigm could be used in the future to investigate a variety of intrapsychic phenomena, such as self-evaluation, stereotyping, causal attribution, moral judgment, pain, and emotional experience. The mouse paradigm, coupled with appropriate analytical techniques, could be employed to explicate the dynamic properties of these phenomena and the underlying structures and processes that govern them. In sum, the mouse paradigm is a creative new approach to, and conceptual advance in, psychological measurement that will find increasing application.
Read additional articles from this newsletter: