Although each is best considered as experimental, sufficient data are now available for the five adapted scales to be available for use outside the lab. Users are encouraged to treat results from these scales as tentative with additional confirmation needed before an impairment could appropriately be determined or ruled out. These results should not be used alone for diagnostic identification and are not intended to substitute for face-to-face consultation with a qualified provider.
The test instruments require a Java enabled browser, for example Netscape or Microsoft Internet Explorer, running in Windows 95/98/NT/ME/2000 or Macintosh equivalent. Tests are loaded in an applet window. Test items and the keyboard response modality have been selected to facilitate equitable use by persons with and without a visual disability. However, if the user has no functional vision, some assistance in navigation to and between tests may be required.
The four cognitive scales available for online use were designed to be consistent with the Luria's identification of four primary neuropsychological functions as reflected in the P(lanning), A(ttention), S(imultaneous), S(uccessive) model (Naglieri & Das, 1997). The related online tests from this lab are CogPlanning, CogAttention, CogListening, and CogMemory, respectively.
CogStyle provides an assessment of four core personality traits. It is included in the context of neuropsychological appraisal, because there is sufficient evidence of a relationship between personality traits and cognitive performance to warrant attention to this affective area in a comprehensive screening.
There is no charge for use of the tests, but an access code for entry into the private consult area is required along with a password for the test menu. To obtain an access code and password, send an e-mail to Dr. Jones jones@nevada.edu including brief explanation of qualifications and intended use.
Links for information about the scales now available for use on the Internet/WWW are below:
CogAttention is comprised of ten pairs of tones with the task to identify whether the first is higher in pitch, the second is higher in pitch, or if they are the same pitch. This is intended as a test of "attention", not musical aptitude; fine pitch discriminations are not required. Sound files were prepared in the au format and are loaded prior to the instructions for the test.
The user hears a sample prior to the actual test items. All responses are on the computer keypad. The alpha reliability coefficient for this ten-item scale based on responses from Series Four (n=65, m = 7.67, s.d. = 2.23) participants was .73 with a standard error of measurement = 1.16. Evidence related to criterion and construct validity of the CogAttention scale is available in the Series Two studies and the Series Four studies.
Precise scaling of the results of the CogAttention is not warranted, given the brief length of the instrument and the unpredictable impact of the quality of the local sound system and isp connection. Considering the intent to serve as a screening tool, normative comparisions are reported only as follows:
The 15th percentile has been used in prior normative studies in the lab as the cutting score for clinical significance, based on Gastaldo, Reeves, Levinson, and Wenger (1997) and Kay (1995). Normative data for CogAttention using Series Four (n=65) participants are in Table 1 below:
| raw score | |||||
|---|---|---|---|---|---|
| above the norm | 9-10 | ||||
| normal | 8 | ||||
| below norm but within normal range | 6-7 | ||||
| possible impairment | 0-5 |
Technical Information For Other ScalesCogMemory is similar to a standard digit-span test with the exception that one point is assigned for each correctly identified digit. As with the CogAttention scale, sound files were prepared in the au format and are loaded prior to the instructions for the test. The user hears a series of numbers with the series length ranging from 5 to 9. The only numerals used in this scale are 1, 2, 3, 4, or 5.
The user is instructed to listen to the number series, press Ø on the keypad when the series ends, and then use the keypad to enter the numbers which were heard. The test begins with a sample series and one practice opportunity in number entry. There are eight number series files, for a maximum total of 55 on this scale. The digit-span item type has been identified as a measure of sequential (succcessive) processing in other instruments, for example the Kaufman Short Neuropsychological Assessment Procedure (Kaufman & Kaufman, 1994).
The reliability coefficient (KR-21) for this 55-item scale based on combining data from Series Four and a subset of Series Six participants (n=80, m = 38.76, s.d. = 11.14) was .92 with a standard error of measurement = 3.15. Evidence related to criterion and construct validity of the CogMemory scale is available in the Series Two studies and the Series Four studies.
Normative scores reported for CogMemory are intentionally limited using the same rationale cited above for the CogAttention scale. A "below norm" (between 15th and 25th percentile) level is added given the larger spread of scores in CogMemory.
Norms (n=80), with data from Series Four and a subset of Series Six studies are provided in Table 2 below:
| raw score | |||||
|---|---|---|---|---|---|
| above the norm | 39-55 | ||||
| normal | 38 | ||||
| below norm but within normal range | 30-37 | ||||
| below norm | 28-29 | ||||
| possible impairment | 0-27 |
Technical Information For Other ScalesThe CogListening scale is comprised of six declarative sentences presented in sound files with increasing levels of compression. After each sentence, the user, in multiple-choice format identifies the actor (man, woman, boy, or girl), the target (package, paper, pencil, or plate), and the setting (kitchen, garage, office, or storeroom) presented in the sentence. Sound files, prepared in au format, were compressed using a standard sound editor. All sound files are loaded prior to the instructions for the test. The user has one practice question from a sound file without compression and one example of a compressed sound. The user is instructed to listen to the sentence, press Ø on the keypad when the sentence ends, and then use the keypad to respond to the multiple-choice questions.
This scale is intended to assess the simultaneous processing in the Luria model of neuropsychological function. The guiding principle in development of the scale was the premise that the perceptual process in responding to "incomplete sounds" would be comparable to that in the incomplete pictures format often used in visual assessment of the simultaneous processing function. The test was introduced in the Series Four studies. Additional information about the compression process is available with that link.
Results from the Series Four studies suggested that the higher rates of compression could be introduced earlier in the test. A subset of the participants in the Series Six studies completed the refined version of the CogListening scale with fewer items than the original scale and a steeper compression slope.
Table 3 below presents the compression rates and mean scores on the three-item sets for each of the six sentences in the refined version. These data combine the Series Six subset of participants with a re-analysis of responses of the Series Four participants on the items which were maintained (n=80, m = 14.76, s.d. = 2.14).
| compression rate | mean score (3 items) |
||||
|---|---|---|---|---|---|
| sentence 1 | |||||
| sentence 2 | |||||
| sentence 3 | |||||
| sentence 4 | |||||
| sentence 5 | |||||
| sentence 6 |
The alpha reliability coefficient for the refined scale using the combined sample was .61 with a standard error of measurement = 1.34. While somewhat lower than the comparable reliability coefficient (.71) with the larger number of items in the Series Four version, there was a corresponding decrease in the standard error of measurement (1.48 in the longer version). The concurrent criterion validity cofficient between the original and refined version using data from the Series Four participants (n=65) was .94. It thus appears reasonable to predict that the other validity evidence obtained in the Series Four study would generalize to the refined scale.
Normative comparsions for the CogListening scale use the same categories and rationale described above for the CogMemory scale. Conversions with data from Series Four and a subset of Series Six studies (n=80) are provided in Table 4 below:
| raw score | |||||
|---|---|---|---|---|---|
| above the norm | 16-18 | ||||
| normal | 15 | ||||
| below norm but within normal range | 14 | ||||
| possible impairment | 0-13 |
Technical Information For Other ScalesThe CogPlanning scale was introduced in the Series Three studies and adapted for administration via the Internet/WWW in the Series Four studies. Detail about the history and rationale in development of the instrument is available through those links and in the Series Six test descriptions. This scale is intended to assess the "planning" function in the Luria model.
In essence, the CogPlanning scale provides an assessment of the user's "catching on" to the optimal response pattern in a series of 30 items. The task in each item is to evaluate two statements in regard to whether one or both is consistent with a symbol series.
The symbol series is comprised of three elements: * @ # and is presented at the top of the screen. In consideration of possible use by persons with visual disability, these three symbols were chosen as likely to be reported accurately by a screen reader and as one syllable words (e.g. star, at, pound).
Below the symbol series are two statements describing the relationship between two of the symbols, for example @ after *. The user is instructed that at least one of the statements will correctly describe the order and responds on the keypad by pressing 1 if only one of the statements is correct or 2 if both correctly describe the order. Three sample items with feedback after the response are provided before the actual test begins. The instructions are to respond accurately and rapidly.
For analysis, the 30 items are categorized as Condition A (first statement was incorrect), Condition B (second statement was incorrect), or Condition C (both statements were correct). A cognitive efficiency score is calculated for each of the three conditions. The concept of the cognitive efficiency score was introduced in the first series of studies in this lab and is a ratio of accuracy and response time. The nature of the task is such that without consideration of response time, extremely high accuracy rates would be anticipated. Levinson and Reeves (1997) suggest that the cognitive efficiency scale provides a more useful measure of the individual's level of CNS function.
Split-half reliability estimates for the CogPlanning scale were calculated using data obtained from the Series Three participants. The CogPlanning (identified as Grammatical Reasoning in Series Three) responses were re-analyzed as two 15-item parallel forms. Reliability coefficients (with Spearman-Brown correction) for the Condition A, Condition B, and Condition C efficiency scores were .83, .81, and .81, respectively. The corresponding standard errors of measurement were 2.49, 1.97, and 1.70. Criterion and construct validity information is available in the Series Three studies and Series Four studies.
When adapting this instrument for administration via the Internet/WWW, it was evident that use of the absolute Condition A efficiency score would be problematic for normative comparisons. The cognitive efficiency score is heavily influenced by speed of response, and the measurement of response speed was obviously a function of both user response and quality of isp connection. To compensate, both studies in both Series Three and Series Four included an additional score, the absolute difference in cognitive efficiency scores for Condition A and Condition B. A higher score with Condition A is associated with higher "planning" performance; the assumption was that online connection factors which could impair response speed would typically be equally evident in Condition A and Condition scores.
Normative comparisons for the CogPlanning scale thus are based on the A minus B efficiency score difference. To avoid negative numbers in reporting the outcome, a linear T-score (capped at 20 and 80) was calculated, combining the Series Four and Series Six participants (n=100, m = 2.29, s.d. = 2.86). With rationale and categories as described for other scales, conversions are provided in Table 5 below:
| T-score | |||||
|---|---|---|---|---|---|
| above the norm | 51-80 | ||||
| normal | 50 | ||||
| below norm but within normal range | 43-49 | ||||
| below norm | 41-42 | ||||
| possible impairment | 20-40 |
Technical Information For Other Scales
The CogStyle scale, based on an artificial intelligence model postulated by Lowen (1982), measures four core personality traits:
The four traits measured in CogStyle appear comparable to the dimensions in the Myers-Briggs Type Indicator (Myers & McCaulley, 1985). They are also, with the exception of omission of an anxiety scale, consistent with traits suggested by the so-called "big five" (Costa & McCrae, 1992) model of personality.
Assessment of two or more of these core personality traits was a part of Series One, Series Two, Series Four, and Series Five studies in this lab. Rationale and development history are detailed in lab report 2-5. Series Six introduces a refined version of the scale with the number of items reduced from 35 to 30.
CogStyle begins with the user expressing prior beliefs about preferences in each of the four core areas. This is followed by paired comparisons of six adjectives, one each for the Holland RIASEC dimensions, and then paired comparisons of six action verbs, each also corresponding to one of the Holland dimensions. Adjectives are practical, curious, flexible, sympathetic, ambitious, and efficient. The actions are fixing things, investigating things, designing things, assisting others, persuading others, and organizing things.
The initial scoring step is calculation and ranking of the RIASEC dimensions based on user responses to the 30 paired comparisons. The three highest ranked dimensions are used with formulae provided by Prediger, Swaney, and Mau (1993) to obtain scores for the approach and focus traits. The formula for the approach trait results in scores ranging from -10 to +10 with the negative numbers associated with an "ideas" preference and positive numbers associated with "data" preference. The formula for the focus trait results in scores ranging from -11 to +11 with negative numbers associated with "people" preference and positive numbers associated with "things" preference. To avoid use of negative numbers in user reports, and particularly to avoid inadvertent valuing of one trait over another, the score is multiplied by 5 and added to 50 with arbitrary caps of 1 and 99. Thus, a zero outcome from the formula becomes a score of 50.
A Bayesian scaling procedure (Jones, 1989) is used to calculate scores for the attitude and process traits. The user's initial perceptions of status on these scales from the initial questions on the test are used as the prior probabilities for the Bayesian scaling. To illustrate, the first question on the CogStyle scale is: "Do you think you are usually more: 1) extraverted and outgoing; 2) mid-range or uncertain; 3) introverted and thoughtful" The initial probability for an extraverted preference on the attitude scale is either .70, .50, or .30, contingent on the response to that question. Using data from a prior large-scale research project (Jones, 1996), a Bayesian approach is then used with the rankings on each of the RIASEC dimensions to obtain the posterior probability for "belonging" to either extroverted or introverted groups on the attitude scale. A comparable process is used to obtain the posterior probability for "detail" or "context" on the process trait. The obtained posterior probabilities are then multiplied by 100 and rounded to whole numbers for the user report.
Although two different procedures are used to obtain the trait scores, the outcome is comparable with 50 indicating a mid-level preference and scores of 1 or 99 indicating the clearest preference for either pole on the bipolar dimensions. For user reports, scores greater than 70 or less than 30 are arbitrarily identified as strong preferences.
The design of the paired comparisons enabled also the calculation of a "consistency" score. To illustrate, when "curious" is chosen over "efficient" and "practical" is chosen over "curious", then the "practical" should be selected when comparing to "efficient". Four triads for consistency are scored and reported to the user.
Split-half reliability estimates for the refined CogStyle scale were calculated using data obtained from the Series Six participants. Their CogStyle responses were re-analyzed as two 15-item parallel forms. RIASEC rankings were determined for each form; scores were then calculated using the procedures described above. The reliability coefficients (after Spearman-Brown correction) for the attitude, approach, focus, and process traits were .93, .43, .77, and .78, respectively. The relatively low coefficient for the approach trait may have been an artifact of this sample. When the split-half correlation on this trait was calculated using only participants who had expressed a strong preference on one of the forms, the Spearman-Brown corrected coefficient for the trait was .89. Criterion and construct validity information related to the CogStyle scale is available in the Series One, Series Two, Series Four, and Series Five studies in this lab.
____________________________________________________________________
Technical Information For Other ScalesReferences
Costa, P.T., Jr., &; McCrae, R.R. (1992). NEO PI-R: Professional manual. Odessa, FL: Psychological Assessment Resources.
Gastaldo, E., Reeves, D., Levinson, D., & Wenger, C.B. (1997). ANAM normative data: USMC-1995 hyponatremia outbreak studies. San Diego, CA: National Cognitive Recovery Foundation.
Jones, W.P. (1989). A proposal for the use of Bayesian probabilities in neuropsychological assessment. Neuropsychology, 3, 17-22.
Jones, W.P. (1996). Career assessment for patients with visual disability: Consideration of instrument distortion. Journal of Vision Rehabilitation, 10, 2-7.
Kaufman, A.S., & Kaufman, N.L. (1994). Manual: Kaufman Short Neuropsychological Assessment Procedure . Circle Pines, MN: American Guidance Service, Inc.
Kay, G. (1995). Cogscreen: Aeromedical edition: professional manual. Odessa, FL: Psychological Assessment Resources, Inc.
Levinson, D.M., & Reeves, D.L. (1997). Monitoring recovery from traumatic brain injury using Automated Neuropsychological Assessment Metrics. Archives of Clinical Neuropsychology, 12, 155-166.
Lowen, W. (1982). Dichotomies of the mind: A systems science model of the mind and personality. New York: John Wiley.
Myers, I.B., & McCaulley, M.H. (1985). Manual: A guide to the development and use of the Myers-Briggs Type Indicator. Palo Alto, CA: Consulting Psychologists Press.
Naglieri, J.A., & Das, J.P. (1997). Cognitive Assessment System. Itasca, ILL: Riverside Publishing
Prediger, D., Swaney, K., & Mau, W. (1993). Extending Holland's hexagon: Procedures, counseling applications, and research. Journal of Counseling & Development, 71, 422-428.
___________________________________________________________________
Guide to Reports