CogAttentionCogMemoryNEUROPSYCHOLOGICAL ASSESSMENT LABPAUL JONES, ED.D.
UNIVERSITY OF NEVADA, LAS VEGAS
____________________________________________________________________
ADAPTED COGNITIVE SCALES: CogAttention and CogMemory
Assessment Lab Report 2-4
December/1998
This report includes data for two cognitive tests designed with specific intent to provide equity for use with both the sighted population and persons with visual disability. CogAttention and CogMemory have historical roots in a three-year project, funded by Title IV-C, on assessment for children and adolescents with visual disabilities (Jones, 1980). After obtaining a password from the author, each can be administered, scored, and interpreted via the Internet/WWW at no cost.
____________________________________________________________________
____________________________________________________________________
CogAttention Description and Development
CogAttention is a two part measurement of basic attending and organizing of information. Part one uses ten items to test simple pitch discrimination. In the second part, also with ten items, the person hears two series of tones and identifies which tone is different in the second series. CogAttention is based on an auditory competency scale developed as a followup to the Title IV-C project cited above. The intent of the auditory competency test was to provide cognitive screening using stimuli with neither symbolic nor semantic meaning.
CogAttention began as a 32-item auditory competency pretest, administered with a standard ability test in the spring of 1983 to a sample of 249 students enrolled in a comprehensive community college located in a Southwestern state. Prior research suggested that the student body was similar to a general population in cognitive ability, ethnicity, and socioeconomic status.
The ability test data were used in a stratified random sampling process to identify a sample of 100 subjects, selected to approximate a normal distribution. This "normal" sample was then used to choose the items using typical difficulty/discrimination criteria and to create performance norms.
To minimize overinterpretation of raw scores on a short scale, the sten scale was selected for use in normative comparisons. Sten scores of 1-3 are interpreted as low, 4-7 as average, and 8-10 as high. The high and low ranges represent approximately the top and bottom 16% of the general population. On the 20 item scale the split-half reliability estimate was .71 (SEmeas = 1.42). The KR20 reliability estimate was .70 (SEmeas = 1.44). The mean item difficulty was .78. Gender differences were not significant.
The original auditory competency scale was administered using an audio tape. Conversion to a format for administration via the Internet/WWW was completed in 1998. With Java (TM), two forms of the test were prepared. The standard form assumes that a sighted person is interacting with the display (examiner and/or subject) and uses the mouse as the primary response mode with typical graphical user interface. The enabled form is intended to facilitate use by persons with limited vision, using keyboard rather than mouse response and a larger font display. Tones matching those used in the original test were created in the AU file format.
For consistency with other tests being developed in the UNLV Neuropsychological Assessment Lab, a derivative of the sten scale was initially used to report CogAttention results. The derivative is simply the sten conversion multiplied by 10. This results in a 10 point scale with scores ranging from 10 to 100 in intervals of 10.
The CogAttention Test appears to directly tap the "attention" function in the contemporary PASS theory (Naglieri & Das, 1990) built on Luria's model of neuropsychological functions. In essence this model identifies four functions: (P)roblem solving, (A)ttention, (S)uccessive processing, and (S)imultaneous processing.
On comparable items in other tests there is significant difference in the performance of normal vs. brain impaired subjects (Golden, Purisch, & Hammeke, 1985). CogAttention part one is clearly consistent with assessment of this attention function. Part two would appear to include elements of both attention and successive processing, but the two parts of the CogAttention scale loaded on the same factor in factor analytic studies.
CogMemory Description and Development
CogMemory uses memory span for digits to measure short-term memory, attention, and successive processing. The subject listens to a series of numbers and then enters the numbers using the keyboard. Memory span for digits is a widely used task in tests of mental ability.
The same procedures and sample used in initial design of CogAttention were used in development of the CogMemory test. After pretesting, eight memory span items were selected. The number of digits in each item ranges from 5 to 9. Assigning one point for each correctly identified digit results in a 55-point scale. The split-half reliability estimate was .89 (SEmeas = 2.25). The KR20 reliability estimate was .80 (SEmeas = 3.03). The mean item difficulty was .72. Gender differences were not significant.
In 1998 this scale was also converted with Java (TM) for administration on the Internet/WWW. The standard form uses the mouse with typical graphical user interface. The enabled form uses keyboard control with a larger font display. Sound files were created to present each item in the AU file format. In both forms, the individual responds by typing the numbers. Results are reported with the sten score derivation described above for the CogAttention scale.
CogMemory appears to directly tap the "successive" function in the contemporary PASS theory. Tests of memory span for digits are often used in neuropsychological assessment with differential performance typically evident between normal and brain damaged persons (Lutey, 1977). Digit span also appears to differentiate between left and right brain damage; left brain lesions are associated with markedly lower scores (Kaufman, 1990).
Method
Details about participants and other cognitive instruments used in the series two studies are available in related reports. Six protocols with evidence of random responses on the CogAttention scale were eliminated before proceeding with the studies below. CogAttention was the first scale in which participants were asked to respond to auditory stimuli and use a combination of mouse and keyboard responses. The resulting sample of 65 appears comparable to the original group on all other factors (e.g. demographics and performance on other cognitive scales). The eliminated protocols appear most likely to be an artifact of the change in response set.
As described in the development history, each of these auditory scales was originally designed for administration with a tape recorder. In the series two studies, tests were computer-administered.
Results
Reliability
Table 1 below provides descriptive data for the CogAttention and CogMemory scales. Results from the tape version are included for comparison with the series two data. Data in Table 1 show somewhat higher performance on the tape version for both scales. Reliability estimates are comparable between computer-administered and tape-administered versions. These data suggest sufficient instrument reliability on both scales for the recommended high, medium, low interpretations of performance. The somewhat lower performance on both scales may be an interaction of response mode (paper and pencil vs. computer keyboard) and the available sound quality.
____________________________________________________________________| computer Fall 1998 n=65 | tape Spring 1983 n=100 | |||||||
|---|---|---|---|---|---|---|---|---|
| mean (s.d.) | r (KR21) | SEmeas | mean (s.d.) | r (KR20) | SEmeas | |||
| CogAtt | 13.5 (3.81) | 1.96 | 15.6 (2.63) | 1.44 | ||||
| CogMem | 35.0 (7.94) | 3.43 | 39.4 (7.79) | 3.03 |
Intercorrelations
Table 2 describes the relationships among the two parts of CogAttention and the CogMemory scale. For reference, comparable data for the tape version are included in parentheses. CogAttention-Part 1 appears to have the same relationship with CogAttention-Part 2 and CogMemory in both versions. There was a stronger relationship between CogAttention-Part 2 and CogMemory in the computer version with the series two sample.
____________________________________________________________________| CogAttention-1 (tape version) | CogAttention-2 (tape version) | CogAttention-Total (tape version) | CogMemory (tape version) | ||||
|---|---|---|---|---|---|---|---|
| CogAttention-1 | 1.0 (1.0) | .38 (.34) | .87 (.82) | .07 (.05) | |||
| CogAttention-2 | .38 (.34) | 1.0 (1.0) | .79 (.82) | .32 (.16) | |||
| CogAttention-Total | .87 (.82) | .79 (.82) | 1.0 (1.0) | .22 (.13) |
Validity
Table 3 below provides correlations between CogAttention and CogMemory raw scores and the efficiency (correct responses/minute) scores on the other cognitive measures used in series two studies. Although the higher correlation coefficients were associated with scales hypothesized to be measuring the same neuropsychological construct (e.g. CogAttention and Simple 2-Choice Reaction Time as measures of attention; CogMemory and Math Processing as measures of successive processing), none of the correlations were statistically significant.
| 2-choice | react-1 | react-2 | math | spatial | match | |
|---|---|---|---|---|---|---|
| CogAttention | .23 | .05 | -.01 | -.09 | .16 | -.01 |
| CogMemory | -.18 | .00 | -.17 | .12 | .04 | .02 |
____________________________________________________________________2-choice: Simple 2-Choice Reaction Time
react-1: Procedural Reaction Time- Standard Version
react-2: Procedural Reaction Time- Degraded Stimuli
math: Mathematical Processing
spatial: Spatial Processing- Simultaneous
match: Matching To Sample
Table 4 provides additional data regarding the underlying neuropsychological structure of the CogAttention and CogMemory scales. Caution is needed in interpreting the results portrayed in Table 4 because of the small n per factor. Given this caution, the data in the exploratory factor analysis displayed in Table 4 provide tentative support for the hypothesized relationships. The Matching to Sample and Spatial tests are presumed to assess a simultaneous processing neuropsychological function. Each has highest loading on the same factor. CogMemory was hypothesized, along with the Math Processing Scale, to assess the successive processing neuropsychological function. Each has highest loading on the same factor. CogAttention, particularly part one, was assumed to tap the "attention" function, along with the three reaction time scales. In this dataset, the CogAttention subtests emerged as a distinct factor.
| Factor 1 | Factor 2 | Factor 3 | Factor 4 | ||
|---|---|---|---|---|---|
| 2-choice | .43* | .33 | .17 | -.45 | |
| react-1 | .87* | .05 | .04 | .07 | |
| react-2 | .89* | -.07 | .08 | -.06 | |
| math | .16 | -.20 | .07 | .68* | |
| spatial | .40 | .16 | .72* | .12 | |
| match | -.08 | -.05 | .91* | -.04 | |
| cogat-1 | .01 | .75* | -.11 | -.13 | |
| cogat-2 | .03 | .83* | .17 | .08 | |
| cogmem | -.15 | .42 | .02 | .71* | |
| percent of variance | 22% | 18% | 16% | 14% |
____________________________________________________________________* highest factor loading for the scale
2-choice: Simple 2-Choice Reaction Time Efficiency Score
react-1: Procedural Reaction Time Efficiency Score- Standard Version
react-2: Procedural Reaction Time Efficiency Score- Degraded Stimuli
math: Mathematical Processing Efficiency Score
spatial: Spatial Processing Efficiency Score- Simultaneous
match: Matching To Sample Efficiency Score
cogat-1: CogAttention Part One Raw Score
cogat-2: CogAttention Part Two Raw Score
cogmem: CogMemory Raw Score
Additional data regarding the hypothesized relationships was obtained using the impairment levels described in the general cognitive efficiency norms. On the Mathematical Processing scale, five scores were at the impaired level. The CogMemory mean for those five subjects was 30.8. The mean CogMemory score for the nonclinical group was 35.4.
On the Simple 2-Choice Reaction Time scale, seven scores were at the impaired level. The CogAttention mean score for those subjects was 11.3; the corresponding score for the other subjects was 13.8. Four subjects were at the impaired level on the Standard Version of the Procedural Reaction Time scale. The CogAttention mean score for those subjects was 11.8; the mean score for the nonclinical group was 13.6. Differences in CogAttention were not as evident when the impaired (n=5) and nonclinical group means on the Degraded Stimuli version of that scale were compared. The obtained means were 13.2 and 13.5, respectively.
New Norms
With the data from the series two studies, the adapted sten scale previously used with CogAttention and CogMemory is replaced by a conversion procedure more consistent with clinical applications. Using the traditional stanine scale as a base, raw scores are converted as low, medium, or high (corresponding to stanines 1-3, 4-6, and 7-9, respectively. The new conversion scales are in Tables 5 and 6 below.
____________________________________________________________________| raw score | |||||
|---|---|---|---|---|---|
| low | 0-9 | ||||
| medium | 10-17 | ||||
| high | 18-20 |
| raw score | |||||
|---|---|---|---|---|---|
| low | 0-29 | ||||
| medium | 30-40 | ||||
| high | 41-55 |
Summary and Discussion
The data in this study support the hypothesized relationship between the ANAM/TWB math processing scale and the adapted CogMemory scale. Each appears to be assessing the "successive processing" function in the PASS model. The relationship between the CogAttention scale and other measures of attention is equivocal. Additional study is needed to explore whether a single underlying attention function is being tapped or whether the auditory application provides a confounding stimulus.
The CogAttention and CogMemory scales appear to have sufficient precision for clinical application, but there are many factors which may influence measurement in the Internet/WWW format. The low-medium-high scaling reduces the likelihood of an overly precise interpretation of performance, but the user will want to be aware that even with such scale there are points in which a single item makes a dramatic difference. The revised program for these tests identifies raw scores on the boundary points (e.g. a raw score of 29 on CogMemory is identified as being at the upper boundary of the low range.
References
Golden, C.J., Purisch, A.D., &; Hammeke, T.A. (1985). Manual for the Luria-Nebraska Neuropsychological Battery: Forms I and II. Los Angeles, CA: Western Psychological Services.
Jones, W.P. (1980). Assessing basic competencies: visually impaired. International Journal of Rehabilitation Research, 3, 84-87.
Kaufman, A.S. (1990). Assessing adolescent and adult intelligence. Boston: Allyn &; Bacon.
Lutey, C.L. (1977). Individual intelligence testing: A manual and sourcebook (2nd Edition). Greeley, CO: Carol L.Lutey Publishing.
Naglieri, J.A., &; Das, J.P. (1990). Planning, attention, simultaneous, and successive cognitive processes as a model for intelligence. Journal of Psychoeducational Assessment, 8, 303-337.
________________________________________________________________________________________________________________________________________