d2 • d2 Test of Attention

---

R. Brickenkamp, T. Merten, K.-D. Hänsgen
Hogrefe Ltd. The Test People, Oxford

Overview

The Test

With the ever increasing application of computers in psychological assessment and rehabilitation, and particularly in assessment and diagnosis of attention (see, e.g., Säring, 1988; Prosiegel, 1991; Sturm, Hartje, Orgaß, & Willmes, 1994), the widely used d2 test has now been adapted for use on computers. The adaptation makes use of the special advantages that computers can offer, in particular the impact on subject motivation of attractive onscreen design. In addition, a computer-based version makes the test more widely available for use with normal, healthy subjects and in the clinical field (Merten & Hänsgen, 1994; Merten, 1995).

Description of the scale and characteristics

Test Construction: Ten different symbols are used, which are made up of combinations of the letters "d" and "p" with one to four dashes arranged either above or below the letter. Nine items are presented per screen (Figure d2-C.1). The task of the subject is to mark all the items that consist of "d" with two dashes (by pressing the appropriate numerical key). Corrections are permitted (by pressing an alternative numerical key). The subject should, as instructed, always work from left to right. After completing the first screen of 9 items, the subject moves to the next screen by pressing the space bar.

Twelve series of items are shown in total. Each series begins with a new screen (nine items); within a session, multiple screens of a series are processed. The processing time per series is set at 30 seconds. The timing begins when the display comes up on the screen and ends when the first key is pressed after the 30 seconds processing time is over. Depending on the computer being used, it takes approximately three seconds for the computer to change screens. The screen changing time is additional to the processing time. Completion time, including instructions and practice, is approximately 10 minutes.

Construction of the series: Among the nine items in a screen there are two to five target items (i.e. the letter d with two dashes). The number of available target items in each series was determined randomly with the following constraints:

a) the first to fourth screens of every series include three to five target characters, with a total of 16 target items appearing (out of total 4 x 9  =  36 items);
b) the fifth and sixth screens each include two and four target items, the seventh screen contains five; the total number of target items is 11 (out of a total 3 x 9  =  27 characters);
c) for the eighth to eleventh and the twelfth to fourteenth screens the same above construction principles were also applied.

The order in which the items (targets and distractors) are presented on screen is randomized subject to the above constraints.

Instruction and practice are carried out interactively on-screen. The practice series includes target items in a random order. Three consecutive screens in the practice series must be correctly answered before the test proper begins. It is particularly recommended that the administrator reads the instructions out loud to the test takers, to ensure that instructions have been fully understood.

Comments: The administration of the test generally should only be carried out in the presence of a test administrator. The role of the administrator is to check that the test instructions have been understood, to provide support if necessary, and to ensure the test taker is motivated to give of his or her best. Any relevant observations of the participants behavior should be incorporated when interpreting results.

All the data presented here were collected during individual test sessions under the supervision of a test administrator. A standard computer screen (14-inch color monitor) and keyboard were used; the standardization data are only useable under these conditions. The data of healthy subjects were obtained by 12 different researchers who were instructed to maximize the motivation of subjects to complete the test. The administration of the d2-C followed an initial phase of at least 30 minutes in all groups during which other tests were carried out. Even subjects without any experience with computers therefore gained enough experience with the input keys by the beginning of the test.

The guidelines in the Validity section should be consulted when interpreting low performance levels in organically brain-damaged subjects.

	Population-based norms; associated with items
	Population-based norms; calculated
	Derived/calculated score
(x)	Dealing with missing values: (0) ignore them (1) use regression to estimate an answer (2) use the middle scale point as answer (3) use a defined score as answer

Characteristics	Mouse	Keyboard
TN Total number of symbols processed
TN-E Symbols recognized correctly
E% Error percentage
FL Fluctuation
E Total number of errors (E1+E2)
E1 Omitted
E2 Confused
E2L Wrong letter
E2D Wrong number of dots

As in the paper-and-pencil version, the following test scores are considered (see Brickenkamp, 1994, pp. 20-22); the parameters enclosed in parentheses appear on the score print-out.

• TN Total Number of Items Processed (1) Total number of the 12 series processed, i.e., the measure of performance of all items that were processed, both relevant and irrelevant ones. Norm values (C norms) are given for two age groups of healthy subjects and three age groups for the patient sample.
• E Raw Score of Errors (2) is the sum of all mistakes (two forms): E1 are errors of omission and E2 the less common (also with the computer test) errors of commission. For errors of commission, it is displayed whether the error was due to a wrongly identified character, a letter permutation (p instead of d) with two correctly identified dashes, or due to wrongly identified dash numbers (d with more or fewer dashes than two).
• E% Percentage of Errors (3) measures the proportion of errors made (E) within the area of all items processed (TN) and is a more exact measure than the Raw Score of Errors (E). This score is lower for the computer version than the paper-and-pencil version; it is given to two decimal places; the corresponding norm values (percent ranks) are given.
• TN-E Total Number of Items Minus Error Scores (4). This score is standardized like the Total Number of Items Processed and Percentage of Errors scores (C norms).
• FR The Fluctuation Rate (5) is the maximum difference between the Total Number of Items Processed (TN) scores on the individual series.

Due to the recent controversy over error-corrected performance scores (see Brickenkamp, 1991, 1993; Oehlschlägel & Moosbrugger, 1991, 1993), the two following test scores were taken into account, even if they do not appear in the analyses print-out of the scores:

• TN-2E As opposed to the TN-E, double the error score is used to correct for the total performance.
• CP (Concentration Performance) is derived from the number of correctly crossed out relevant items (d2) minus the errors of commission (E2).

As our analyses for the computer version show, however, the error-corrected measures TN-E, TN-2E and CP correlate so highly with one another that a separate standardization can be dispensed with for the latter two measures. Rarely occurring individual cases of behavior contrary to instructions can easily be identified by the test administrator and stand out due to an excessively high E% score.

In addition to the named parameters, the score print-out displays a numerical and graphic distribution of the analyses of the TN-E, TN-E1 and TN-E2 scores across the 12 series.

Norms

Norm	Mouse	Keyboard
Apprentice selection Germany	Age groups Not gender specific Mouse Keyboard Touch screen from 17 to 26;11 yrs. N = 1187 - 1188
Apprentice selection Germany, High school students	Age groups Not gender specific Mouse Keyboard Touch screen from 18 to 26;11 yrs. N = 553
Apprentice selection Germany, non-High school students	Age groups Not gender specific Mouse Keyboard Touch screen from 17 to 26;11 yrs. N = 635
Apprentice selection Switzerland	Age groups Not gender specific Mouse Keyboard Touch screen from 15 to 30;11 yrs. N = 905
Person undergoing rehabilitation, total	Age groups Not gender specific Mouse Keyboard Touch screen from 21 to 58;11 yrs. N = 261
Person undergoing rehabilitation, age specific	Age groups Not gender specific Mouse Keyboard Touch screen from 21 to 35;11 yrs. N = 145 from 36 to 58;11 yrs. N = 116
Healthy persons, age specific		Age groups Not gender specific Mouse Keyboard Touch screen from 18 to 40;11 yrs. N = 264 from 41 to 60;11 yrs. N = 59
Patients, age specific		Age groups Not gender specific Mouse Keyboard Touch screen from 15 to 40;11 yrs. N = 83 from 41 to 60;11 yrs. N = 94 from 61 to 81;11 yrs. N = 39

Standardization of the Keyboard Version

The test can be completed using either the computer keys or the mouse. It was important to standardise the two methods separately since it was hypothesised that the key version may result in more errors due to the respondent having to switch between looking at the keys when typing and looking at the screen.

The standardization studies carried out so far on the key version are based on four samples, that is, three normal groups and one patient group.
The normal subjects were enrolled during three diploma research projects:

(1) Sample I (Samp I) consists of 108 persons with an average age of 36.7 years (SD  =  12.4; range 18 to 75 years), who took part in a study by Belter (1995). The following (German) tests from the Hogrefe Testsystem (H_T_S) were carried out: "Visual Analogy Scale" (Visuelle Analogieskala: VAS), "General Well-Being Scale" (Befindlichkeits-Skala: Bf-S), "Speed-Learning Test" (Tempo-Lern-Test: TLT), subtest three of the "Performance Test System" (Leistungsprüfsystem: LPS-3), "Symbol Digit Modalities Test" (Symbol-Zahlen-Lerntest: SZT), "d2 Test of Attention" (Aufmerksamkeits-Belastungs-Test: d2-C), subtests one and two of the "Peformance Test Sytem" (Leistungsprüfsystem: LPS-1/2), "Attentional Shift Test" (Umstell-Interferenz-Test: JZT), "Scale of Cerebral Insufficiency" (Skala Cerebrale Insuffizienz: c.I.). On one of the two follow-up days, the test was repeated by 106 persons, who completed first the computer version of the d2 and then the paper-and-pencil version.

(2) Sample II (Samp II) was recruited during the diploma research project of Schwettmann (1995), and consists of 91 subjects with an average age of 29.2 years (SD  =  8.3; range 18 to 58 years), who completed the German versions of the following tests: VAS, Bf-S, TLT, LPS-3, LPS-4, SZT, d2-C, JZT, paper-and-pencil version of d2 (d2-pp), and the "Multiple Choice Vocabulary Test" (Mehrfachwahl-Wortschatztest: MWT-A).

After completing the paper-and-pencil form of the d2, the subjects of Samples I and II completed an answer-preference questionnaire (eight items).

(3) Sample III (Samp III) consists of 140 subjects with an average age of 34.4 years (SD  =  13.1; range 18 to 83 years), who completed the German versions of the following tests in the study of Siebert (1995): VAS, Bf-S, revised version of the Eysenck Personality Questionnaire EPQ-R (Ruch, in preparation; see Ruch & Hehl 1989), Carroll Rating Scale for Depression (Deutsche Carroll Rating-Skala für Depression), d2-C, TLT, LPS-3, SZT, JZT, re-administration of the d2-C, and MWT-A.

The 339 subjects of Samples I to III consist of 166 men and 173 women in total.

A more precise description of the tests, which, unless otherwise indicated, were all administered under the H_T_S and for which no exact bibliographic references are provided here, can be found in the printed H_T_S manuals.

(4) The patient sample consists of 216 patients (134 men and 82 women) with an average age of 46.3 years (SD  =  15.5; range 15 to 81), who were administered the d2-C within the framework of a neuropsychological diagnosis. Within this group of patients, 64 (29.6 %) had a craniocerebral injury, 68 (31.5 %) had a vascular illness, 33 (15.3 %) were diagnosed with a neoplasm, 14 (6.5 %) presented dermatitis, 11 (5.1 %) had brain atrophy, 9 (4.2 %) had a diagnosis of (predominantly alcohol-induced) toxic encephalopathy (including Korsakow syndrome), 6 (2.8 %) presented with so-called idiopathic, normal pressure hydrocephalus, 8 (3.7 %) belonged to other diagnostic groups (Parkinson's disease, B12 vitamin deficiency, etc), and 3 (1.4 %) had no diagnosis.

The standardization has, until now, been based upon an aggregated healthy sample (N  =  339), which consists of two age groups: 18 to 40 year olds (N  =  264; 138 male, 126 female; average age 28.4 years, SD  =  5.3) and 41 to 60 year olds (N  =  59; 20 male, 39 female; average age 48.5 years, SD  =  5.4). As there is little data available for older healthy subjects (61 to 83 years; N  =  16), they will not be taken into account here. It is of importance to note that, as with other attention tests, there is a significant slowing down with increasing age, so that the standardization of the younger age groups is not useable with older persons. For the group of over 60 year olds in the healthy sample (average age: 68.2 years, SD  =  6.7), the TN has an average of only 245 characters (SD  =  54.3), TN-E 239 characters (SD  =  54.7), and E% 2.59 (SD  =  1.90).
The first age group consists of 2 schoolchildren, 13 trainee/apprentices, 49 students, 138 persons with a completed vocational training, 52 with a completed university level education, and 10 with no completed vocational training. The second age group consists of 16 university graduates, 41 persons with completed vocational training and 2 without.

A segregated standardization was carried out for the patient sample encompassing the following three age groups: 15 to 40 years (N  =  83; 53 male, 30 female; average age 30.0 years, SD  =  67), 41 to 60 years (N  =  94; 60 male, 34 female; average age 51.5 years, SD  =  5.2), and 60 to 81 years (N  =  39; 21 male, 18 female; average age 68.7 years, SD 4.9).

Standardization of the Mouse Version

Norms are available for the mouse version for a vocational trainee sample in Switzerland, a vocational trainee sample in Germany, and return-to-work samples (occupational training institute):

1. Vocational Training Students (Switzerland)
Data source: Tassera (2002), Novartis AG Basel

Description of the sample:

Norm population		Gender		Age in years
	Total	m	f	Mean	Min	Max
Vocational training students (Switzerland)	N  =  905	N  =  660 (72.9%)	N  =  245 (27.1%)	17.88	15	30

Positions: Biology laboratory assistants, Chemistry laboratory assistants, business occupations, information technology and electronics

Test score distribution for vocational training student sample Switzerland:

	Minimum	Maximum	Mean	Standard deviation
TN Total performance score	99	752	472.09	72.27
TN-E Performance score of correct reactions	9946	747	464.30	73.05
FR Fluctuation range	4	36	13.10	4.40
E% Error percent	0.00	53.54	1.76	3.27

2. Vocational Training Students (Germany)
Data source: Heese (2000), Merck KGaA Darmstadt

Description of the sample:

Norm population		Gender		Age in Years
	Total	m	w	Mean	Min	Max
Vocational training students (Germany) Total sample	N  =  1188	N = 608 (51.2%)	N = 580 (48.8%)	19.6	17	26
Vocational training students  (Germany) Education specific
Grammar/high school	N  =  553	N = 206 (37.3%)	N = 347 (62.7%)	21.2	18	26
Non-grammar/high school	N  =  635	N = 402 (63.3%)	N = 233 (36.7%)	18.3	17	26

Test score distribution apprentice sample Germany (G = grammar, N-G = Non-Grammar):

Scale	Min			Max			Mean			Standard deviation
	Total	G	N-G	Total	G	N-G	Total	G	N-G	Total	G	N-G
TN Total number of items processed	275	361	275	813	813	742	532.41	564.14	504.78	77.94	75.60	68.96
TN-E Total number of items minus error scores	269	348	269	805	805	732	523.96	557.15	495.06	78.31	75.23	68.94
FR Fluctuation rate	4	4	4	45	45	31	13.29	13.17	13.40	4.37	4.38	4.37
E% Percentage of errors	0.00	0.00	0.00	9.47	7.50	9.47	1.63	1.25	1.95	1.37	1.10	1.51

3. Occupational Rehabilitants (Leipzig)
Data source: H. Rabenberg (2001); occupational training institute (OTI) Leipzig

HTS-specific norms were calculated from the data of adults returning to work in the framework of job seeking and job trials. The most commonly occurring disability of this sample are illnesses of the muscle and skeletal systems (approx. 40%), injuries, poisoning, and other conditions with external causes (approx. 15%), mental and behavioral disorders (approx. 8%), skin diseases (approx. 8%), circulatory disorders (approx. 6%), and a range of other illnesses.

Description of the sample:

Norm population		Gender		Age in years
	Total	m	f	Mean	Min	Max
Occupational rehabilitants total sample	N = 261	N = 191 (73.2%)	N = 70 (26.8%)	33.7	21	58
Occupational rehabilitants age specific
21 to 35 years	N = 145	N = 102 (70.3%)	N = 43 (29.7%)	28.9	21	35
36 to 58 years	N = 116	N = 89 (76.7%)	N = 27 (23.3%)	39.8	36	58

Test score distribution of occupational rehabilitants OTI Leipzig:

Scale	Min				Max				Mean				Standard deviation
	Total	Years			Total	Years			Total	Years			Total	Years
		21-35	36-58			21-35		36-58		21-35		36-58		21-35	36-58
TN Total number of items processed	187	210		187	758	641	758		430.70	441.97	416.62		88.56	74.65		101.96
TN-E Total number of items minus error scores	156	207		156	755	630	755		424.16	436.07	409.28		88.69	74.61		102.04
FR Fluctuation rate	6	7		6	31	28	31		13.89	13.42	14.47		4.37	4.12		4.61
E% Percentage of errors	0.00	0.00		0.00	16.58	8.24	16.58		1.60	1.37	1.89		1.77	1.20		2.26

4. Occupational Rehabilitants (OTI) Birkenfeld
Data source: occupational training institute (OTI) Birkenfeld, 2002-2003

HTS-specific norms were calculated on the basis of data from adult occupational rehabilitants in the framework of job seeking and job trials. The subjects were disabled (physically and mentally) or under threat of being disabled. It was explained beforehand that they were being considered for an occupational rehabilitation and that they would be tested for psychological aptitude with a battery of tests.

A total norm as well as age-specific norms are included. It was possible, after homogeneity was tested for, to aggregate the data with that of OTI Leipzig. In relation to this, however, further data from other occupational training institutes are needed to develop the total OTI norm.

Description of the sample:

Norm population		Gender		Age in years
	Total	m	f	Mean	Min	Max
Occupational rehabilitants total sample	N = 549	N = 441 (80.3%)	N = 108 (19.7%)	34.0	17	53
Occupational rehabilitants age-specific
17 to 30 years	N = 169	N = 136 (80.5%)	N = 33 (19.5%)	25.2	17	30
31 to 35 years	N = 127	N = 102 (80.3%)	N = 25 (19.7%)	33.1	31	35
36 to 40 years	N = 146	N = 119 (81.5%)	N = 27 (18.5%)	38.2	36	40
41 to 53 years	N = 107	N = 84 (78.5%)	N = 23 (21.5%)	43.4	41	53

Test score distribution of occupational rehabilitants OTI Birkenfeld:

Scale	Min	Max	Mean	Standard deviation
TN Total number of items processed
Total sample	89	659	435.6	82.7
17 to 30 years	223	659	453.4	72.7
31 to 35 years	208	620	440.0	78.9
36 to 40 years	138	600	433.9	84.3
41 to 53 years	89	608	404.3	91.2
TN-E Total number of items minus error scores
Total sample	33	653	428.4	84.3
17 to 30 years	215	653	446.9	73.7
31 to 35 years	206	609	432.9	79.2
36 to 40 years	89	588	426.6	87.5
41 to 53 years	33	602	396.5	92.9
FR Fluctuation rate
Total sample	4	52	13.3	4.8
17 to 30 years	4	52	13.1	5.2
31 to 35 years	4	30	13.4	4.5
36 to 40 years	4	38	13.1	4.6
41 to 53 years	4	30	13.7	4.7
E% percentage of error
Total sample	0	116	7.1	8.8
17 to 30 years	0	56	6.5	6.1
31 to 35 years	0	59	7.2	8.0
36 to 40 years	0	116	7.3	11.8
41 to 53 years	0	56	7.8	8.6

Test Score Distribution

In Table d2-C.3, the means, standard deviations, minimums and maximums of the total sample of healthy and patient groups are presented. Only the total number of processed items (TN) as well as the error-corrected performance parameters TN-E, TN-2E, and CP demonstrate an approximate normal distribution in the graphic analyses. Scores based on skewed distributions should be interpreted with care.

In the Validity section (Table d2-C.13), the results for the two sub-samples are presented parallel to one another according to age, gender, and education. A good congruence can be seen in the scatter of the four named test values.

The error level, as already seen in the standardization data, is low on the whole. On average, healthy subjects commit less than 2% errors. As this value is lower than that for the corresponding paper-and-pencil test, this is most certainly due to the significantly larger presentation of the characters on the computer screen. This is, however, aimed at not disadvantaging neurological patients as well as psychiatric patients taking neuroleptic medication, who frequently have visual impairments, from the outset.

Table. d2-C.3: Distribution characteristics of the total sample of healthy and patient subjects.

		Mean	SD	Minimum	Maximum	Normal distribution
	TN	372.60	84.21	153	628	yes
Normal	TN-E	365.67	83.96	151	625	yes
sample	TN-2E	358.75	84.05	149	622	yes
(N  =  339)	CP	151.80	36.28	61	263	yes
	E	6.93	5.29	0	42	no
	E%	1.94	1.52	0	8.67	no
	FR	12.81	4.29	5	39	no
	TN	236.79	84.07	77	466	yes
Patient	TN-E	229.06	86.41	61	464	yes
sample	TN-2E	221.32	89.33	43	462	yes
(N  =  216)	CP	95.03	41.12	16	201	yes
	E	7.73	7.57	0	47	no
	E%	4.32	5.42	0	24.48	no
	FR	9.63	3.63	3	31	no

Table. d2-C.4: Intercorrelation matrix of individually calculated test parameters. Upper diagonal: complete healthy subsample (N  =  339). Lower diagonal: patient sample (N  =  216). Product-moment correlations for the variables TN, TN-E, TN-2E and CP; otherwise Spearman's rank correlations.

	TN	TN-E	TN-2E	CP	E	E%	FR
TN	-	0.998*	0.992*	0.985*	0.06	-0.20*	0.20*
TN-E	0.996*	-	0.998*	0.992*	0.01	-0.25*	0.19*
TN-2E	0.987*	0.997*	-	0.996*	-0.04	-0.30*	0.18*
CP	0.982*	0.994*	0.998*	-	-0.06	-0.32*	0.16*
E	-0.28*	-0.34*	-0.39*	-0.40*	-	0.95*	0.09
E%	-0.57*	-0.61*	-0.66*	-0.66*	0.93*	-	0.04
FR	0.22*	0.20*	0.17*	0.17*	0.22*	0.11	-

*p < 0.05

Intercorrelations Between Parameters

The intercorrelations between parameters for the complete healthy sample are given in Table d2-C.4. As can be seen, TN, TN-E, TN-2E, and CP correlated with one another so strongly that there is hardly any additional differential information that can be gained in the normal range through the use of these values. Individual cases where many errors are made can be easily identified through the calculation of E and E%.

The number of errors itself shows no correlation with the total item performance; however a high performance level is statistically related to a low error rate and is also related to a greater fluctuation rate, which is not surprising.

Table. d2-C.5: Values of the internal consistency for the d2-C, calculated for each first test administration of the patient, healthy and the three subsamples as well as for the retest administration.

		TN	TN-E	TN-2E	CP	F
Cronbach's	Healthy total	0.977	0.979	0.976	0.975	0.681
Alpha	Samp I	0.977	0.979	0.977	0.975	0.687
	Samp II	0.963	0.964	0.960	0.953	0.717
	Samp III	0.980	0.980	0.979	0.978	0.603
	Retest Samp I	0.982	0.983	0.982	0.980	0.578
	Retest Samp III	0.985	0.985	0.984	0.982	0.628
	Patient	0.986	0.987	0.987	0.986	0.822
Split half	Healthy total	0.954	0.956	0.954	0.953	0.638
upper vs.	Samp I	0.952	0.954	0.952	0.949	0.694
lower Test half1	Samp II	0.933	0.934	0.931	0.923	0.681
	Samp III	0.959	0.960	0.959	0.960	0.479
	Retest Samp I	0.975	0.977	0.977	0.976	0.488
	Retest Samp III	0.974	0.974	0.974	0.974	0.678
	Patient	0.973	0.954	0.977	0.977	0.802
Split half	Healthy total	0.983	0.984	0.982	0.980	0.828
even vs.	Samp I	0.988	0.990	0.989	0.985	0.845
odd	Samp II	0.973	0.974	0.973	0.967	0.810
series number1	Samp III	0.981	0.981	0.979	0.978	0.815
	Retest Samp I	0.987	0.986	0.984	0.980	0.771
	Retest Samp II	0.986	0.986	0.984	0.980	0.781
	Patient	0.990	0.991	0.991	0.991	0.924

¹ With equal-length adjustment.

Psychometric properties

Reliability

To check the reliability of the different parameters of the d2-C, the internal consistency (Cronbach's alpha and split-half reliabilities) as well as test-retest correlations were calculated. Table d2-C.5 contains the values of the consistency evaluation from Series 1 to 12 for the total sample of 339 healthy subjects and 216 patients, separated by the three sub-samples of the healthy subjects, as well as the retest administrations of Samples I und II. The split-half reliabilities are calculated separately according to upper vs. lower test halves as well as even vs. odd series number.

Despite the low number of items in a series, (compared with the paper version of test), the values of the internal consistency, the TN and the error-corrected performance values are surprisingly high and can be regarded as being very satisfactory. In contrast, the pattern of the internal consistency for the error rates is different. It must be noted here, however, that the error rate per series of the healthy subjects is, on the whole, low, and, in addition, certain processing dynamics within the test have an effect, e.g., higher error rates at the beginning of the test or to the contrary, loss of motivation with increasing inaccuracy during the course of the test. As a result, the split-half reliabilities of even vs. odd series numbers are higher than those for the upper vs. lower test halves and reach an acceptable level.

Table. d2-C.6: Intercorrelation matrices between the parameters in pre- and posttest. Product-moment correlations for the variables TN, TN-E, TN-2E, and CP; otherwise Spearman's rank correlation coefficients.

		R E T E S T
Samp I	N = 106	TN	TN-E	TN-2E	CP	E	E%	FR
	TN	0.91*	0.92*	0.91*	0.91*	-0.19	-0.41*	0.18
	TN-E	0.92*	0.92*	0.92*	0.92*	-0.22*	-0.44*	0.16
PRE-	TN-2E	0.92*	0.92*	0.92*	0.92*	-0.24*	-0.46*	0.15
TEST	CP	0.92*	0.92*	0.93*	0.92*	-0.25*	-0.47*	0.13
	E	-0.02	-0.03	-0.04	-0.03	0.37*	0.35*	0.10
	E%	-0.27*	-0.28*	-0.29*	-0.29*	0.44*	0.48*	0.06
	FR	0.25*	0.24*	0.23*	0.22*	0.15	0.09	0.06
Samp III	N = 138	TN	TN-E	TN-2E	CP	E	E%	FR
	TN	0.88*	0.88*	0.88*	0.87*	-0.10	-0.30*	0.20
	TN-E	0.88*	0.88*	0.88*	0.88*	-0.13	-0.32*	0.20
PRE-	TN-2E	0.88*	0.88*	0.88*	0.88*	-0.15	-0.34*	0.19
TEST	CP	0.88*	0.88*	0.88*	0.88*	-0.16	-0.35*	0.19
	E	-0.05	-0.07	-0.09	-0.09	0.51*	0.50*	0.04
	E%	-0.28*	-0.30*	-0.32*	-0.32*	0.51*	0.56*	-0.05
	FR	0.24*	0.24*	0.24*	0.24*	0.02	-0.04	0.14

For Samp I (retest-interval 1 to 2 days) and Samp III (retest-interval approx. 30 minutes), the intercorrelation matrices of the individually calculated parameters between pre- and post-tests are presented in Table d2-C.6. (the bold highlighted coefficients specify the retest reliabilities). In turn, the value of TN, as well as the error-corrected performance parameters can be regarded as being very satisfactory, however, not for E and E%. The FR values don't reach levels of significance for the test-retest correlation.

Table d2-C.7 displays the test-retest correlations of the individual measurements of patients for three different retest intervals. The individual groups demonstrate no systematic differences regarding age and initial performance on the d2. Because the existing sample sizes were small for the different retest intervals, in particular those of the 4 to 18 day interval, we cannot yet comment on retest reliabilities with increasing time intervals.

Table. d2-C.7: Test-retest reliabilities coefficients for neurological patients with different retest intervals. Product-moment correlations for the variables TN, TN-E, TN-2E, and CP; otherwise Spearman's rank correlation coefficients.

	Interval 1 day	Interval 2-3 days	Interval 4-18 days
	(N  =  36)	(N  =  34)	(N  =  36)
TN	0.95*	0.96*	0.92*
TN-E	0.95*	0.96*	0.94*
TN-2E	0.95*	0.97*	0.94*
CP	0.95*	0.96*	0.93*
E	0.57*	0.36*	0.78*
E%	0.66*	0.60*	0.81*
FR	0.20	0.19	0.08

*p < 0.05

Altogether, the trend is towards a somewhat high reliability coefficient - relating to the internal consistency as well as stability values - for the patient group.

Impact on the Test Scores of Test Repetition

While test-retest reliabilities provide an indication of the correlation between scores when a test is repeated, in order to judge whether an improvement has been attained (e.g., resulting from remission or successful therapy), the improvements in performance that arise simply through practice and being more familiar with the test requirements must also be considered. Westhoff and Dewald (1990) have proved that "Performance in concentration tests can be greatly improved through sheer repetition" (p. 44). On the other hand, practice effects do not transfer from one concentration test to another.

Fay and Meyer (1993) found marked performance increments for the Bonner concentration test and a decrease in the percent of error scores between the first and second testing.

For the d2-C, such performance increases between the pre- and post-test have also been established, as can be seen in Table d2-C.8. The parameters TN, TN-E, TN-2E, and CP show significant increases for all groups and retest intervals (verified with t tests for related data). The changes in the other parameters, that is, low absolute error rates, lower percentage error rates, and lower fluctuation rates, are significant for both of the healthy subject groups (verified with Wilcoxon signed-rank tests), while the differences between the patient groups with low sample sizes are only significant for E% in the retest interval of 1 day, and 4 to 18 days and E for 4 to 8 days.

As can be seen, performance increases are, on the whole, significantly lower for the patient groups. Systemic investigations are necessary for a differentiated picture of the differences between healthy subjects and those with organic brain illnesses, as well as at different retest intervals.

For the results to date, it can be determined, however, that there is no statistically verifiable dependence of the performance increases in the parameters on the initial level of these variables.

Table. d2-C.8: Changes in the analyzed parameters of the d2-C between first and second testing for different samples and retest intervals.

Group	Healthy				Patient
Retest interval	approx. 30 min (N  =  106)		1 to 2 day(s) (N  =  138)		1 day (N  =  36)		2 to 3 days (N  =  34)		4 to 18 days (N  =  36)
	m	(SD)	m	(SD)	m	(SD)	m	(SD)	m	(SD)
TN	50.48	(34.19)	71.70	(50.17)	36.28	(31.22)	37.85	(28.13)	31.89	(26.93)
TN-E	52.87	(33.12)	72.72	(49.92)	38.17	(31.35)	39.03	(27.67)	34.67	(24.61)
TN-2E	55.25	(32.84)	73.75	(49.96)	40.06	(32.68)	40.21	(28.43)	37.44	(24.34)
CP	24.09	(13.77)	31.11	(21.13)	18.64	(14.72)	18.00	(14.07)	17.58	(11.87)
E	-2.39	(5.20)	-1.02	(3.82)	-1.89	(6.17)	-1.18	(5.87)	-2.78	(7.29)
E%	-0.87	(1.41)	-0.51	(1.04)	-1.37	(3.62)	-0.93	(3.19)	-2.49	(4.26)
FR	-1.14	(4.99)	-1.35	(5.86)	-0.53	(3.61)	0.24	(4.51)	-1.03	(5.17)

Relationships Between Computer- and Paper-and-Pencil Forms

It can be seen from the description of the construction of the d2-C that the paper-and-pencil version has not been directly transferred to the computer version, which would in fact be highly unsatisfactory. There is therefore no direct 100% equivalence between the two tests, but there is a clear similarity at the level of the construct "selective attention"; the data presented can be seen as supporting the construct validity.

Table d2-C.9 includes the correlations between the parameters of the computer version and the paper-and-pencil version of the d2 for Samples I and II. In each case, the d2-C was carried out first, with sample I twice, before the administration of the paper-and-pencil version.

The correlations between the error corrected performance parameters are completely satisfactory and are higher after an immediately consecutive administration of the d2-C (retest) and the d2-PP. On the other hand, the values for the other parameters are far from satisfactory, particularly for the fluctuation rate.

A similar correlation matrix between the computer test and the paper-and-pencil version of the d2, incidentally, can be found for the Bonner Concentration Test in the work of Fay and Meyer (1993, p. 38, Table 5) .

Acceptability of the Computer Version

In order to assess the acceptability of the computerised version of d2, an 8-item answer-preference questionnaire was completed by respondents. The questionnaire was given to Samples I and II following the administration of the two versions. In particular, the subjects were asked to judge: (1) on which of the two versions they could concentrate better; (2) which version was more enjoyable; (3) which form they would prefer, if they were to take more similar items; (4) which variant felt easier; (5) which version they would chose in a job application situation; (6) which was more fun; (7) with which they could better grasp the tasks; and finally (8) with which version the time went by faster. For each question, a decision had to be made in favor of the paper-and-pencil version or the computer version or indifference (no difference); these alternatives were scored with 1, 3, or 2 points, respectively.

Table d2-C.10 displays subjects' answers to the individual questions, separated according to sample. Without taking indifference decisions into account, the differences in choice between the two versions were tested using a binomial test. In the table, significance levels are presented. A total value was calculated for all eight items. The indifference point lies at 16; that is, for the binominal test, all values under 16 are contrasted with those over 16 points.

Table. d2-C.9: Intercorrelations between the test parameters from the d2-C and the d2-PP. Sample I: N  =  108 for the pre- and N  =  106 for the posttesting; Sample II: N  =  91. Product-moment correlations for the variables TN, TN-E, TN-2E und CP; otherwise Spearman's rank correlation coefficients.

		d2-PP
d2-C		TN	TN-E	TN-2E	CP1	F	F%	FR
Samp I	TN	0.63*	0.70*	0.74*	0.72*	-0.20*	-0.32*	-0.14
PRE-	TN-E	0.63*	0.71*	0.75*	0.73*	-0.22*	-0.34*	-0.16
TESTING	TN-2E	0.63*	0.71*	0.76*	0.74*	-0.25*	-0.36	-0.17
	CP	0.63*	0.71*	0.76*	0.75*	-0.25*	-0.37*	-0.16
	E	0.06	-0.02	-0.09	-0.13	0.34*	0.32*	0.13
	E%	-0.13	-0.23*	-0.32*	0.35*	0.40*	0.42*	0.17
	FR	0.23*	0.19*	0.15	0.10	0.03	0.02	-0.18
		d2-PP
d2-C		TN	TN-E	TN-2E	CP1	F	F%	FR
Samp I	TN	0.69*	0.75*	0.78*	0.75*	-0.17	-0.30*	-0.13
POST-	TN-E	0.69*	0.75*	0.78*	0.76*	-0.19*	-0.32*	-0.13
TESTING	TN-2E	0.68*	0.75*	0.79*	0.76*	-0.21*	-0.34*	-0.14
	CP	0.67*	0.75*	0.78*	0.76*	-0.22*	-0.35*	-0.14
	E	-0.08	-0.17	-0.28*	-0.31*	0.51*	0.53*	0.21*
	E%	-0.27*	-0.36*	-0.47*	-0.49*	0.52*	0.57*	0.23*
	FR	0.02	-0.01	-0.02	0.02	0.11	0.10	0.06
Samp 2	TN	0.62*	0.65*	0.65*	0.72*	-0.05	-0.17	-0.22*
	TN-E	0.62*	0.65*	0.66*	0.71*	-0.08	-0.20	-0.23*
	TN-2E	0.61*	0.64*	0.66*	0.69*	-0.10	-0.23*	-0.23*
	CP	0.60*	0.64*	0.66*	0.67*	-0.12	-0.25*	-0.23*
	E	0.16	0.10	0.04	0.27*	0.30*	0.28*	-0.07
	E%	0.03	-0.03	-0.09	0.05	0.30*	0.31*	-0.01
	FR	0.26*	0.26*	0.23*	0.27*	0.03	-0.02	0.15

¹ For Sample II, only the CP values of  N = 31 subjects of the PP version were calculated. * p < 0.05.

Table. d2-C.10: Answers of Samples I (N  =  106) and II (N  =  91) regarding the preference of a version of the d2, significance of the binomial tests and Kendall rank correlations between the item and self evaluation with regards familiarity with computers. Upper section: Sample I; Lower section: Sample II.

Item	Average	Preference			Binomial	correlation with
		d2-PP	Indifference	d2-C	test	familiarity
(1) Better concentration	2.65 2.25	13 27	11 14	82 50	0.000 0.012	-0.01 0.06
(2) More enjoyable	2.68 2.32	12 28	10 6	84 57	0.000 0.002	0.10 0.15
(3) Continuation preference	2.70 2.42	11 23	10 7	85 61	0.000 0.000	0.00 0.10
(4) Easier	2.59 2.24	16 25	11 19	79 47	0.000 0.013	0.07 0.11
(5) Preference in job application situation	2.54 2.22	21 31	7 9	78 51	0.000 0.036	0.01 0.12
(6) More fun	2.71 2.52	6 15	19 14	81 62	0.000 0.000	0.01 0.09
(7) Easier to grasp requirements	2.48 2.10	11 23	33 36	62 32	0.000 0.281	-0.03 0.10
(8) Time seemed to pass more quickly.	1.89 1.65	50 53	18 17	38 21	0.241 0.000	-0.20* 0.08
Total points1	20.24 17.71	18 28	3 3	85 60	0.000 0.001	-0.07 0.12

¹ The indifference point is set at 16 points for these values (8 x 2); see Text. * p < 0.05

The Kendall rank correlations with self-evaluation of familiarity with computers are displayed (6-point rating). As the results show, preference for one of the two versions is not related to personal experience with computers.

For both groups, there is a distinct preference for the computer version in the subjective evaluation regarding most of the items, with the difference for Sample I being more pronounced. Whether the two administrations of the d2-C are responsible for this difference can not be answered in the framework of the experimental design.

In contrast to the results of a study of the adaptation of a questionnaire (EPQ-R), it was shown that the time seemed to pass quicker with the paper-and-pencil version of the d2 (but, significance was only reached for Sample II). This reflects, however, only the difference in the actual time taken (PP: 14 series at 20 s  =  4:40 min; Computer: 12 series at 30 s =  6:00 min).

Validity

AGE:
As was already made clear in the presentation of the standardization data, performance on the d2-C is dependent on age. For the total healthy sample, there is a correlation between age and TN and/or TN-E of r = -0.49 in each case (Patient: r = -0.43 or -0.44). The error rate correlates only with age for the patient group (rS  =  0.25); for both groups there is a significant correlation between age and E% (Spearman rank correlation: rS = 0.14; Patient: rS = 0.37).

EDUCATIONAL LEVEL:
The correlation with educational level (number of educational years) is likewise significant, although has less influence than age (healthy and patient groups, respectively - TN: each r  =  0.27; TN-E: r  =  0.28  and  0.27; F: rS  =  -0.07 ns and -0.17; E%: rS  =  -0.15 and -0.26).

EXPERIENCE WITH COMPUTERS:
For the healthy groups, it was also established that subjects who had more experience with computers obtained a higher score on the test (correlation with TN and TN-E: rS = 0.29), although these correlations are relatively low. There is no relationship between E and E%.

GENDER:
In checking the gender differences, a significant but relatively low correlation (t test: p < 0.05) with TN and TN-E was found for the healthy sample, with male subjects performing slightly better than females. Since the difference was less than a quarter of a standard deviation, no gender-specific norms are provided. The male subjects revealed a somewhat lower E% value. No such differences were found for the patient sample.

CRITERION-RELATED VALIDITY: Healthy Subjects
Spearman rank correlation coefficients between d2 and other tests for the three healthy sub-samples are listed in Table d2-C.11. Satisfyingly high correlations are to be found, as expected, between the parameters TN, TN-E, TN-2E, CP, and construct-similar test values, i.e., the total performance in the SZT and the reaction time in the JZT. There is evidence of a correlation with psychomotor reaction speed (TLT), which is, however, very low. The relationship to verbal and nonverbal intelligence measures (LPS-1 to -4) lies in the average ranges. It is worth noting here that the these tests are time-limited and the corresponding results - as with the d2 - are clearly time-dependant. The correlation with the point values on the MWT-A are clearly lower suggesting that a person's general well-being doesn't have an influence on their performance on the d2-C. Self-indicated performance and psychovegetative problems on the c.I. Scale are associated with a low performance, with which age is also observed to be a confounding factor. For the other three variables, E, E%, and FR, the data do not reach significant statistical levels. In all cases, accuracy performance (E%) demonstrates comparatively low correlations with intelligence parameters, with psychomotor reaction times, and with accuracy performance on the SZT (rS  =  0.16).

Here it must be emphasized again that the most important interpretable value is the error-corrected performance rate, whereas the accuracy performance can only provide valid additional information when there are large deviations from the norms.

Table. d2-C.11: Rank correlation coefficients between the test values of the d2-C and other tests for the healthy sample. The differing sample sizes arise form the different experimental designs of Subsamples I to III.

Variable	N	TN	TN-E	TN-2E	CP	E	E%	FR
MWT-A, correct	230	0.20*	0.21*	0.22*	0.22*	-0.18*	-0.23*	-0.01
LPS-1/2. correct	108	0.41*	0.41*	0.41*	0.41*	-0.07	-0.19	0.04
LPS-3, correct in 1st trial	277	0.56*	0.57*	0.58*	0.58*	-0.24*	-0.39*	0.14*
LPS-4, correct in 1st trial	91	0.37*	0.38*	0.40*	0.40*	-0.09	-0.16	0.19
TLT, average reaction time	339	-0.27*	-0.28*	-0.28*	-0.28*	0.11*	-0.20*	-0.17*
TLT, SD of the reaction time	339	-0.20*	-0.21*	-0.22*	-0.22*	0.17*	0.23*	-0.07
SZT, performance score 1st-3rd series	339	0.71*	0.71*	0.72*	0.72*	-0.07	-0.26*	0.17*
SZT, error rate 1st-3rd series	339	0.19*	0.18*	0.17*	0.16*	0.14*	0.09	0.16*
SZT, error percentage	339	0.03	0.02	0.01	0.00	0.18*	0.16*	0.12*
JZT, average reaction time	339	-0.65*	-0.66*	-0.66*	-0.66*	0.05	0.22*	-0.14*
VAS: General well-being	339	0.04	0.04	0.05	0.05	0.03	0.01	-0.01
Bf-S: Well-being disorders	337	0.08	0.09	0.08	0.07	0.03	0.01	0.07
Performance difficulties (c.I.-Scale)	108	-0.31*	-0.30*	-0.29*	-0.29*	-0.11	-0.04	-0.09
Depressive symptoms (CRS-D)	200	-0.07	-0.08	-0.08	-0.09	0.12	0.14*	-0.00

* p < 0.05

CRITERION-RELATED VALIDITY: Patient Sample
For the patient sample (see Table d2-C.12), correlations with selected conventionally administered tests, (i.e. in paper form) are presented as well as computer administered tests from the Hogrefe Test System, (H_T_S), Descriptions of these are given in the handbook of the H_T_S. The following conventionally administered tests have been selected for this presentation: the two broad tests Mini Mental Status Test (MMST: Folstein, Folstein, & McHugh 1990; higher scores signify lower fewer problems) and the "Short Syndrome Test" (der Syndrom-Kurztest: SKT: Erzigkeit 1989; higher error points signify larger noticeable problems), the Benton or Visual Retention Test (Benton 1990; Form C, immediate reproduction), the subtests Block Design and Digit Span of the WAIS-III (Hamburg-Wechsler Intelligence Test for Adults HAWIE; Wechsler, 1964), the Standard Progressive Matrizes of Raven (1956), the Trail Making Test (TMT: Reitan, 1958), as well as the Hooper Visual Organization Test (VOT: Western Psychological Services, 1983).

Table. d2-C.12: Rank correlation coefficients between test scores of the d2-C and other tests for the patient group. The conventionally administered tests are listed first.

Variable	N	TN	TN-E	TN-2E	CP	E	E%	FR
MMST, point value	209	0.62*	0.62*	0.62*	0.62*	-0.29*	-0.45*	0.17*
SKT, error points	208	-0.67*	-0.67*	-0.66*	-0.66*	0.09	0.31*	-0.17*
Benton Test, Form C, correct	69	0.69*	0.71*	0.72*	0.73*	-0.40*	-0.55*	-0.05
Benton Test, Form C, errors	69	-0.71*	-0.74*	-0.75*	-0.75*	0.42*	0.58*	0.06
Block Design of WAIS-III, Raw score	188	0.77*	0.78*	0.79*	0.78*	-0.40*	-0.59*	0.08
Digit Span WAIS-III	113	0.21*	0.20*	0.19*	0.20*	0.02	-0.11	0.06
Digit Span backwards WAIS-III	112	0.48*	0.47*	0.46*	0.46*	-0.08	-0.25*	0.28*
SPM Raven, raw score correct	62	0.57*	0.60*	0.61*	0.62*	-0.49*	-0.60*	-0.09
Trail Making Test, time TMT-A	24	-0.70*	-0.71*	-0.70*	-0.69*	0.21	0.44*	-0.12
Trail Making Test, time TMT-B	24	-0.80*	-0.80*	-0.80*	-0.78*	0.06	0.34	-0.40
Visual Organization Test, correct	68	0.49*	0.50*	0.50*	0.51*	-0.35*	-0.49*	0.15
MWT-A, Correct	155	0.41*	0.42*	0.43*	0.43*	-0.18*	-0.28*	0.07
WST, Correct	52	0.48*	0.48*	0.48*	0.47*	-0.13	-0.22	-0.05
LPS-1/2, Correct	193	0.64*	0.63*	0.63*	0.63*	-0.12	-0.31*	0.19*
LPS-3, correct on 1st trial	213	0.74*	0.76*	0.77*	0.77*	-0.46*	-0.64*	0.07
LPS-4, correct on 1st trial	206	0.68*	0.70*	0.72*	0.72*	-0.48*	-0.64*	0.02
TLT, average reaction time	214	-0.61*	-0.62*	-0.62*	-0.61*	0.20*	0.36*	-0.06
TLT, SD of reaction time	214	-0.54*	-0.56*	-0.57*	-0.57*	0.32*	0.43*	0.02
SZT, performance rate 1st-3rd series	214	0.88*	0.89*	0.89*	0.89*	-0.32*	-0.56*	0.21*
SZT, error rate 1st-3rd series	214	-0.05	-0.07	-0.09	-0.09	0.28*	0.27*	0.06
SZT, error percentage	214	-0.26*	-0.28*	-0.29*	-0.30*	0.34*	0.40*	0.02
JZT, average reaction time	76	-0.84*	-0.85*	-0.85*	-0.84*	0.22	0.44*	-0.25*
Tower of Hanoi, time 5th-tower	58	-0.59*	-0.61*	-0.62*	-0.63*	0.39*	0.53*	0.05
Tower of Hanoi, number of moves 5th-tower	58	0.05	0.02	-0.01	-0.02	0.42*	0.42*	0.14
MEMO-Test, Correct 1st series	158	0.50*	0.50*	0.50*	0.50*	-0.20*	-0.32*	0.10
MEMO-Test, correct 1st-5th series	158	0.53*	0.53*	0.53*	0.53*	-0.22*	-0.35*	0.11
MEMO-Test, delayed reprod.	116	0.31*	0.31*	0.30*	0.31*	-0.09	-0.19*	0.16
VEG, Correct 1st-4th series	94	0.54*	0.55*	0.55*	0.55*	-0.27*	-0.39*	0.12
Performance difficulties (c.I. Scale)	198	-0.31*	-0.30*	-0.30*	-0.30*	0.03	0.14	-0.15*
Depressive Symptoms (CRS-D)	59	-0.21	-0.21	-0.20	-0.20	0.14	0.20	-0.12

* p < 0.05

The highest correlations of the parameters are again attained for the construct-similar parameters on the SZT (rS  =  0.88 and 0.89), reaction time on the JZT (rS  =  -0.84 and -0.85), and completion time on the TMT-B (rS  =  -0.80 and -0.78). However, also the point score of the Block Design of the WAIS-III, which is very sensitive to the influence of organic brain impairments on performance, correlates highly with the numerical parameters of the d2-C (rS  =  0.77 to 0.79). Relatively high correlations are obtained with a series of other test scores, even with memory tests and such approximate estimates of cognitive impairments as the point score of the MMST or the error score of the SKT. In contrast to the healthy sample, the accuracy level for the patients appears to represent a more interpretable parameter.

As is consistently shown with the practical administration of the test, in the interpretation of the performances on the d2-C of patients with neuropsychological symptoms and impairments, it should be taken into account that low performance scores as well as poor accuracy can arise due to very diverse disorders, and the poor performance cannot simply be attributed to an impairment of selective attention.

Alternative causes of a poor performance can also be due to; a general mental slowing down (in this case, for example, the psychomotor reaction time is delayed for simple demands that require no decision process); visual-spatial perceptual disorders (poor differentiation between the letters b and p, and poor registration of the number of dashes), visual neglect of the left visual field (in this case, in particular target items at the beginning of the row would be consistently overseen), serious memory failures (repeated questioning about which item should be found; repeated forgetting of the activation of the space bar at the end of the row); and also functional disorders of the hand.

CRITERION-RELATED VALIDITY: Healthy vs Patient samples
Finally, the construct validity of the test was further investigated with a comparison of the performances of the healthy and the patient samples. As, on the whole, the healthy subjects are younger and more highly educated than the patient groups (even if the differences are not statistically significant, this is the case for many of the investigations), a pair-wise comparison of the two groups was undertaken for the analysis at hand. Out of the total group of examined healthy and patient subjects, subjects were paired according to same gender, age (age deviations allowed up to a maximum of one year), and number of educational years (deviations allowed up to a maximum of one year). In this way, 97 healthy-patient pairs were formed, with 56 male and 41 female subjects, with an average age of 37.6 years for the two groups (SD  =  13.3), and an average 13.6 years of education (SD  =  2.4) .

The results of the d2-C for the two groups are presented opposite each other in Table d2-C.13. In addition to the parameters of the d2-C, the results of the construct similar SZT as well as the TLT measurement of psychomotoric reaction speed are presented. As the majority of parameters did not show approximation to a normal distribution after a visual check, Mann Whitney U tests were carried out for the group comparisons; only for the numerical and error corrected parameters of the d2-C were t tests carried out (each with 192 degrees of freedom with homogeneity of variance between the groups). Figure d2-C.2 shows the distribution of the parameters TN-E for the two groups.

As the table and figure show, the parameters of the two groups (TN, TN-E, CP und TN-2E) disperse very well with a good homogeneity of variance: the difference is larger than the expected difference. The raw score distribution favourably shows that the median of the error-corrected parameters of the two groups are very near to average. With higher performance scores, the healthy group does not make significantly more errors. With regards to the percentage error, there are no group differences. The higher values of the patient group are due to a greater number of extremely high E% values.

A similar difference between the groups regarding performance score is seen on the SZT, although the overlap is quite large. Here the error rate is larger for the healthy group. Healthy and patient subjects differ from one another purely on the level of psychomotor speed, including greater intra-individual variability (see also Sturm, 1983), as can be seen from the group comparison of the TLT.

Table. d2-C.13: Comparison of the two stringently parallelled subgroups

Test / Variable		Healthy (N  =  97)			Patient (N  =  97)			U Test	t Test
		m	SD	Median	m	SD	Median	(Z)	(t)
d2	TN	353.1	82.0	362	260.1	82.3	255	6.95**	7.88**
	TN-E	346.0	82.3	346	253.4	83.8	251	6.85**	7.76**
	CP	148.2	36.4	149	106.6	39.7	108	6.71**	7.61**
	TN-2E	338.8	82.9	336	246.6	86.0	249	6.74**	7.60**
	E	7.18	5.44	6	6.78	7.37	5	1.65+	-
	E%	2.14	1.69	1.71	3.28	4.49	1.84	0.19	-
	FR	13.2	4.7	13	9.5	3.5	9	6.47**	-
SZT	Amount	195.1	37.2	196	151.3	51.9	155	5.86**	-
	Errors	3.56	3.04	3	2.26	2.17	2	3.29*	-
	E%	1.85	1.58	1.38	1.85	2.81	1.08	1.68+	-
TLT	m (ms)	29.25	4.42	29	35.57	10.78	32	5.03**	-
	s	7.21	2.44	7	9.91	4.83	9	4.62**	-

** p < 0.001; * p < 0.01; ⁺p < 0.10

Duration

Test form		Duration, ca.	No. of items
	0	0.00 min	%_NUMBEROFITEMS_%
Durations are absolute, from start to end of the test process (incl. instruction phase etc).

References

Bäumler, G. (1991). Auf dem Weg zur operationalen Definition von Aufmerksamkeit. In J. Janssen, E. Hahn, H. Strang (Eds.), Konzentration und Leistung, pp. 11-26. Göttingen: Hogrefe.

Belter, O. (1995). Eichung und Validierung der Computerversion des Aufmerksamkeits-Belastungs-Tests (Test d2). Diplomarbeit, Humboldt-Universität zu Berlin, Naturwissenschaftlich-Mathematische Fakultät II, Fachbereich Psychologie.

Benton, A. L. (1990). Der Benton-Test. Handbuch. German version of the Benton Test by Otfried Spreen. 6th unabridged Edition. Bern; Switzerland: Huber.

Brickenkamp, R. (1991). Fehlinterpretationen von Testleistungen? Anmerkungen zum Beitrag "Konzentrationsleistung ohne Konzentration?" Diagnostica, 37, 52-57.

Brickenkamp, R. (1993). Zur Lösung der Problematik von Gesamttestwerten in Konzentrationstests. Report Psychologie, 18 (3), 24-26.

Brickenkamp, R. (1994). Test d2 Aufmerksamkeits-Belastungs-Test. Handanweisung. 8th expanded and revised edition. Göttingen, Germany: Hogrefe.

Erzigkeit, H. (1989). SKT - Ein Kurztest zur Erfassung von Gedächtnis- und Aufmerksamkeitsstörungen. Manual. Weinheim, Germany: Beltz Test.

Fay, E., & Meyer, M. (1991). Ein computergestützter Test zur Erfassung der Konzentrationsfähigkeit: Das Verfahren, seine Vorteile gegenüber "Papier-Bleistift-gestützter" Diagnostik und empirische Ergebnisse. In H. Schuler & U. Funke (Eds.), Eignungsdiagnostik in Forschung und Praxis, Volume 10, pp. 85-88. Stuttgart, Germany: Verlag für Angewandte Psychologie.

Fay, E., & Meyer, M. (1993). Bonner Konzentrationstest BKT. Ein computergestütztes Verfahren zur Messung der Konzentrationsfähigkeit. Version 2. Göttingen, Germany: Hogrefe.

Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1990). Mini-Mental-Status-Test MMST. German version: J. Kessler, H. J. Markowitsch, & P. Denzler. Weinheim, Germany: Beltz.

Keller, I., & Grömminger, O. (1993). Aufmerksamkeit. In D.Y. von Cramon, N. Mai, & W. Ziegler (Eds.), Neuropsychologische Diagnostik, pp. 65-90. Weinheim, Germany: VCH.

Merten, T. (1995). Vorläufige Ergebnisse von Eichung, Validierung und Normierung des Aufmerksamkeits-Konzentrationstests d2-Computerversion (d2-C). Unpubllished manuscript.

Merten, T., & Hänsgen, K.-D. (1994). Konzeption zur Entwicklung einer Computerversion des Aufmerksamkeits-Belastungs-Tests (Test d2) by R. Brickenkamp for the Leistungsdiagnostische Labor (LEILA). Unpublished manuscript. Berlin.

Oehlschlägel, J., & Moosbrugger, H. (1991). Konzentrationsleistung ohne Konzentration? Zur Schätzung wahrer Leistungswerte im Aufmerksamkeits-Belastungs-Test d2. Diagnostica, 27, 42-51.

Oehlschlägel, J., & Moosbrugger, H. (1993). Bagatellisieren oder Abhilfe schaffen? Report Psychologie 18(3), 20-23.

Posner, M. I., & Boies, S. J. (1971). Components of attention. Psychological Review, 78, 391-408.

Prosiegel, M. (1991). Neuropsychologische Störungen und ihre Rehabilitation. Hirnläsionen, Syndrome, Diagnostik, Therapie. Munich: Pflaum.

Raven, J. C. (1956). Standard Progressive Matrices, Sets A, B, C, D, E. London: Lewis.

Reitan, R. M. (1958). Validity of the Trail Making Test as an indicator of organic brain damage. Perceptual and Motor Skills, 8, 271-276.

Ruch, W., & Hehl, F.-J. (1989). Psychometric properties of the German version of the EPQ-R. Poster presented at the 4th meeting of ISSID, Heidelberg, Germany, 22-25 June, 1989.

Säring, W. (1988). Aufmerksamkeit. In D. von Cramon & J. Zihl (Eds.), Neuropsychologische Rehabilitation. Grundlagen - Diagnostik - Behandlungsverfahren, pp. 157-181. Berlin: Springer.

Schwettmann, M. (1995). Der Aufmerksamkeits-Belastungs-Test d2 - Eichung und Validierung der Computerversion und vergleichende Betrachtungen zwischen dieser und der Papier-und-Bleistift-Version. Diploma dissertation, Humboldt-University of Berlin, Naturwissenschaftlich-Mathematische Fakultät II, Fachbereich Psychologie.

Siebert, K. (1995). Der d2 - Aufmerksamkeits-Belastungstest im Vergleich mit Persönlichkeits-, Leistungs-, Zustands- und Selbstbeurteilungsvariablen. Diploma dissertation, Humboldt-Universität zu Berlin, Naturwissenschaftlich-Mathematische Fakultät II, Fachbereich Psychologie.

Spreen, O., & Strauss, E. (1991). A compendium of neuropsychological tests. Administration, norms, and commentary. New York: Oxford University Press.

Sturm, W. (1983). Die neuropsychologische Relevanz einfacher und komplexer Reaktions- und Konzentrationsuntersuchungen bei Hirnkranken unter Berücksichtigung ätiologischer und lokalisatorischer Gesichtspunkte. Dissertation, University of Trier.

Sturm, W. (1989). Aufmerksamkeitsstörungen. In K. Poeck (Ed.), Klinische Neuropsychologie, pp. 314-318. Stuttgart, Germany: Thieme.

Sturm, W., Hartje, W., Orgaß, B., & Willmes, K. (1994). Effektivität eines computergestützten Trainings von vier Aufmerksamkeitsfunktionen. Zeitschrift für Neuropsychologie, 1, 15-28.

Wechsler, D. (1964). Die Messung der Intelligenz Erwachsener. Textband zum Hamburg-Wechsler-Intelligenztest für Erwachsene (HAWIE). 3rd Edition. Bern: Huber.

Western Psychological Services (1983). Hooper Visual Organization Test (VOT). Manual. Los Angeles, CA: Western Psychological Services.

Westhoff, K., & Dewald, D. (1990). Effekte der Übung in der Bearbeitung von Konzentrationstests. Diagnostica, 36, 1-15.

Wolfram, H., Neumann, J., & Wieczorek, V. (1986). Psychologische Leistungstests in der Neurologie und Psychiatrie. Methoden und Normwerte. Leipzig, Germany: Thieme.

This article is a shortened, edited version of a contribution from Merten et al. from the handbook of the "Performance Diagnosis Laboratory" (LEILA Leistungsdiagnostisches Labor).