Psychometric Properties

Reliability and Validity

Reliability and validity are concepts, which are very useful and can be arrived at

statistically for psychometrically oriented objective instruments. Any projective test,

however, cannot give numerical values for all of its parameters, on a single scale, and

projective tests are also considered to be more of a measure of state rather than traits.

Therefore, any slight difference in the environment or self can make some change in

the precept or responses. Because of these limitations, the reliability and validity of

projective tests are in true sense, a little difficult to determine. Nevertheless, in terms of

consistency of responses, some scientific evidence is possible over a period of time,

with the assumption that the time interval will not be so long that it will influence the

perception of individual. Intra and inter-scorer reliability may be considered more in

terms of interpretation or the concepts verbalized onto the images presented to the

individual.

Table 1

Agreement of Responses on Repeat Administration of SIS-II

on Psychosomatic Subjects (n=100)

Image Number Total % of Subject

Images giving same

response

A12, A30, A31, B5, B22, B29 6 100 %

B4, B15, B21 3 90 %

A7, A13, A20, B8, B9, B16 ,B25, B31 8 80 %

A1, A4, A6, A10, A16, A23, A25, A28, B7, B12, B23, B27, B28 13 70 %

A9, A17, A29, B6, B18, B24, B26, B30 8 60 %

A5, A14, A18, A21, A24, B2, B11 B14, B20 9 50 %

A8, A11, A15, A19, B13 5 40 %

A22, B3, B10, B17 4 30 %

A2, A3, A27, B12 4 20 %

A26 1 10 %

B1 1 10 %

Response Consistency

In psychometric terms, it is known as score consistency. Since, in the Somatic Inkblot

Series, there are no scores in true sense, we are calling it response consistency. Unlike

Rorschach and similar to Holtzman, the length of the protocol does not vary considerably

from one administration to another. It has been found from various sources that the

agreement in total number of responses when the test is repeated after a short time

was very high. The detailed response consistency was undertaken for SIS-II images on

Indian subjects (Dubey and Cassell, 1991).

The absolute response consistency was found to be above 70 percent for the images

A1, A4, A6, A7, A10, A12, A13, A16, A20, A23, A25, A28, A30, and A31 and B4, B5, B7,

B8, B9, B12, B15, B16, B21, B22, B23, B25, B27, B28, B29, B31 for the remaining

images it was 60 percent and less as shown in Table-1.

The correlation for the content categories over a period of four weeks, on repeat testing

was found to be more than .69 as shown in Table - 2.

Table 2

Test Re-test Reliability of the Contents of SIS-II (n=100)

Content categories Correlation (Rho)

Human responses 0.88

Animal responses 0.71

Anatomical responses 0.76

Most Typical responses 0.69

High correlation indicates consistency in the pattern of responses and supports the

high reliability of the test.

Split-half Reliability

Split-half reliability requires splitting of the test into two equal halves. Inkblot images

cannot be split into two equal halves because they are heterogeneous. The task of

categorizing the test into two equal halves is theoretically not only difficult but impossible,

provided that the test is projective in its true sense. Many researchers, however, have

tried to establish split-half reliability of the Rorschach test, not in terms of material,

rather in terms of responses elicited on the test (Vernon, 1933; Hertz, 1934).

The split-half reliability of SIS-II and SIS-Video was established by splitting the test into

two equal halves. These equal halves are the natural divisions of the test that consists

of two series, A and B. Cassell, claims that the SIS is a structured projective test, where

images are selected, by and large, to elicit anatomical response. Therefore, it can be

assumed that it is possible that Cassell might have something in mind to develop two

parallel forms. However, in view of rigorous field testing, he chose to combine both sets

“A” and “B” into one single test. 20 Psychometric Properties

Table 3

Means and Correlation between Series A and Series B (n=220)

Set A Set B t r

SD 7.92 5.60 p=.01

SD 1.68 1.39 n. s.

Human Responses M 8.49 8.27 0.71 0.61

SD 3.59 3.08 n. s.

Animal Responses M 5.25 3.63 5.79 0.71

SD 3.28 2.56 p=.01

Anatomical Responses M 5.59 4.56 3.41 0.66

SD 3.53 2.75 p=.01

Sex Responses M 2.15 1.66 2.52 0.67

Movement Responses M 4.69 4.12 1.85 0.66

SD 3.62 2.84 n. s.

Rejection of Images M 0.73 1.38 2.86 0.76

The test re-test reliability reported hereafter might be considered as parallel form

reliability as well. It was tested on 220 subjects (100 normal, 41 schizophrenics, 39

neurotics and 40 American neurotics) considering both set A and set B. Both the tests

SIS-II and SIS-Video are alternate forms. The means and correlation of the two sets for

the Total number of responses, Human responses, Animal responses, Anatomical

responses, Sex responses, Movement responses, Most Typical responses, and

Rejection of images are presented in Table 3.

All the correlation between sets A and B of the SIS images were found to be statistically,

highly significant for the eight variables considered. The lowest correlation was .51 for

the “Most Typical’ responses. The highest correlation was .89 for the length of responses

independent series for quicker assessment of the patient by the busy clinician. The two

series can be used independently whenever a repeat assessment is suggested, after

a short interval.

The length of the stimulus material may produce feelings of fatigue and monotony in

the subjects. The analysis presented in Table 3 clearly suggests that the total number of

Responses, the number of Animal responses, the number of Anatomical responses,

and the number of Sex responses were significantly more in set “A” than set “B.” It might

be assumed that the motivation of the subject deteriorates in set B, because the nature

of the stimulus was not much different in the two sets. The hypothesis that the number

of responses will be significantly reduced in the second set and is not determined by

set A or B, is being tested at different centers. The initial observations, however, confirm

the correctness of the hypothesis.

Inter-scorer Reliability

Inter-scorer reliability for the indices reported above was determined. The same protocol

was scored separately by a trained and experienced clinical psychologist, having

experience with the SIS, and by a fresh psychologist who was provided with a set of

definitions for the variables as reported in a subsequent chapter. It was discovered that

there was no difference between their two independent scorings. It can be argued that

it was not a scoring, but merely a counting of the responses under different categories

(Total number of Responses: Human responses, Animal responses, Anatomical

responses, Sex responses, Movement responses, Most Typical responses and Rejection

of images).

Validity

The validity of a psychological test is an essential characteristic that must be

demonstrated beyond any doubt. Its validity can be based on theoretical assumptions,

clinical experiences and its utility in the field. The division of tests into objective and

projective requires different standards for measuring validity. The projective tests, such

as the SIS, cannot be tested for validity, like objective tests. The SIS is a method of

observation and appraisal. With regards to a test that measures some aspects of

observation, the appropriate questions are: “Is it useful?”, “Is it productive?”, “Does it

conform to some construct?”.

The SIS is an objective procedure for observation; therefore validity in psychometric

terms cannot be determined. The question then arises; can we still call it a valid

procedure? The clinical interview and the diagnostic procedures are based on the

information elicited from the patient and his observations. Has anyone doubted the

validity of such clinical interviews and diagnostic procedures? If not, the validity of the

projective test, the SIS in particular, cannot be doubted if its utility is adequately

established. It is therefore necessary to demonstrate empirically the content and

construct validity of the SIS. The other types of validity generally given in textbooks

cannot be determined.

The SIS is a test that elicits mostly somatic responses and is based on the theories of

body perception of Schema (body orientation) and body symbolization. The content

validity of the test can be demonstrated if the patient with somatic problems is found to

22 Psychometric Properties

give more somatic responses. This was assessed in a study of pregnant women and

non-pregnant controls. The pregnant women projected more responses of a pregnancy

theme than non-pregnant women did (Cassell, 1990). In a number of studies, it has

been demonstrated beyond a doubt that relationships exist between an individual’s

physical disturbances and his/her projected anatomical responses (Cassell, 1969,

1971, 1972, 1977, 1980; Dubey and Cassell, 1993, and Green grass and Adamson,

1977). In a current study on a heart patient in India, the interim results suggested that

his somatic preoccupation was not verbalized any more than compared with a healthy

control. It was interpreted that his body disturbances were consciously denied (Goel et

al, 1990).

Table 4

Distribution of Atypical Responses

Number of Percentage of Subjects in Significance

Atypical Responses

Normal Neurotics Schizo. Dis.

(n = 100) (n = 39) (n = 41)

0 - 2 93 53 7 X 2 = 2.03

3 & more 7 47 93 p = .01

*These were the adult subjects studied in India.

To have construct validity on the basis of existing literature on inkblot technique, it can

be assumed that Most Typical responses and Atypical responses have to be different in

broad psychiatric diagnostic categories compared to healthy controls. This hypothesis

was tested on Neurotics, Schizophrenics, and healthy Normal subjects. The following

tables clearly confirm that Atypical responses were significantly higher in Psychotics

than Neurotics and Normals. Likewise, Most Typical responses were significantly higher

in normal subjects compared to psychiatric patients.

Table 5

Mean Number of Most Typical Responses

Normal Indian Schizophrenic American

Neurotics Disorder Neurotics

Number (n=) 100 39 41 40

Mean 13.68 11.41 11.54 13.35

SD 1.98 3.76 2.29 1.93

F = 12.67; df = 3/216; p= .01

A number of other categories of responses like Total number of responses, Human

responses, Animal responses, Anatomical responses, Sex responses, Movement

responses, Most Typical responses, and Rejection of Images were also analyzed with

the assumption that Neurotics and Schizophrenics should differ from Normal subjects

because of their different psychopathology. To test this assumption, 220 subjects

consisting of 100 Indian normals, 39 Indian neurotics, 41 Indian schizophrenics, and

40 American neurotics, were administered the SIS-Video and analyzed. It was found

that the normal subjects differed significantly from clinical groups on all the variables

except for the mean number of Sex responses (Table 6).

Table 6

Comparison of Healthy Controls with Clinical Groups on

Different Categories of Responses on SIS-II/Video

Indian American

Indices Normal Neurotics Schizophrenic Neurotics F

Disorders

(n=100) (n=39) (n=41) (n=40)

Total Responses M 74.08 63.92 64.71 61.60 20.34

SD 12.40 9.85 6.88 6.16 p=.01

Movement M 9.60 5.15 7.66 10. 62 8. 46

Human Responses M 14.66 17. 26 18.17 19. 77 10. 23

Animal Responses M 8.65 8.82 10. 56 7. 30 3.18

Anatomical Responses M 9.18 9.15 11.12 12.08

3. 72

Sex Responses M 4.12 3.15 3.59 3. 30 1.08

Rejection of Images M 0.82 3.49 2.12 1. 78 5. 33

The above table clearly indicates that the healthy controls had significantly different

patterns of responses than the clinical groups. However, amongst clinical groups,

there were not many significant differences. This might be interpreted to mean that the

SIS, as viewed by its mentor is more a therapeutic instrument. Its real utility lies in its

content analysis, which helps in understanding the psychopathology of the patient.

This understanding can help in the psychotherapeutic procedure. No projective test

can provide a categorization of the thought processes of the individual, because it is

based on the perception of the sufferer and the circumstances around him. There is no

set pattern in different diagnostic categories. This provides strength to the content

validity of the test (Dubey et al, 1992).

24 Psychometric Properties

Table 7

Mean Scores and SD on Nine Indices of the SIS-II for Normal and

Pathological Groups

Score Indices Normal Neurotics Schizophrenics

Total M 64. 34 64. 28 63.16 61.68 61.76

Human M 21. 86 20.80 20.88 19.96 17.88

Animal M 7. 80 11.12 10.44 10.12 11.36

responses SD 3. 23 4.46 4.13 4.93 5.17

Anatomical M 12.20 13.44 14.40 12.44 16.04

Sex M 1.12 1.16 1.28 2.08 1.44

responses SD 1.32 1.62 1.43 2.85 1.75

Human M 3.46 2.44 3.48 3.12 2.24

Movement SD 2.58 1.90 2.90 3.00 2.08

Animal M 1.14 1.32 1.20 1.44 0.72

Movement SD 1.23 1.22 1.20 2.37 0.87

Most M 16.30 14.56 14.72 11.20 10.84

Typical SD 1.32 0.85 1.00 1.30 1.32

Rejection M 0.58 1.56 1.44 2.56 2.28

of Images SD 1.40 1.24 1.36 2.53 2.16

Mishra & Dwivedi (1997)

In another study conducted by Mishra and Dwivedi (1997) on a sample of 150 subjects

(50 normal, 25 each of anxiety neurosis, neurotic depressive, paranoid schizophrenics

and undifferentiated schizophrenics) using SIS-II observed that the trend of mean

scores for pathological groups was almost similar where the highest mean score was

amassed by the human responses category, followed by the most-typical, anatomical,

animal, human movement, rejection, animal movement and sex categories. The

depressive neurotics, paranoid, and undifferentiated schizophrenic groups, however,

followed a slightly different pattern where the mean sex responses outnumbered animal

movement responses.

A comparison of means of the pathological and normal groups shows that the SIS can

clearly distinguish the patients on the indices of rejection, animal movement and sex

responses. The mean rejection scores of normal subjects was 0.58, SD=1.40, while the

mean scores for pathological groups were invariably higher. The highest mean rejection

has been obtained by the paranoid schizophrenic group followed by the undifferentiated

schizophrenics, anxiety neurotics, and depressive neurotics. A similar picture emerged

for the other two indices where pathological groups overshadowed normals. However,

the mean scores for the animal movement by undifferentiated schizophrenic group

was slightly lesser than that of the normal group.

It can be safely concluded that the indices of animal responses, most-typical responses

and rejection of images very well differentiate normals from four pathological groups.

Hence, these can be taken as hard-core indicators. Other indices such as total number

of responses, anatomical responses, human movement responses, and human

responses have been found to differentiate normals from undifferentiated

schizophrenics. The schizophrenic patients are known to have thinking disturbances.

Therefore, their imagination for SIS images is likely to differ greatly with normal subjects.

The present findings support the contention of Cassell et al (1995).

Table-8 displays the ‘t’ values obtained from the intergroup comparisons for various

scoring indices of the SIS-II test. The normal subjects elicited more number of total

responses on SIS-II than the four pathological groups. The paranoid and undifferentiated

schizophrenics have given significantly lesser number of total responses than either

normal or anxiety neurotic groups.

Normal subjects gave the lowest number of anatomical responses than the four

pathological groups, which goes hand in hand with the animal response category. In

the anatomical response category the normal subjects and undifferentiated

schizophrenics have shown a statistically reliable difference (t=2.49, p <0.05) while

other groups have failed to do so. This pattern of results is contrary to the often-voiced

emphasis placed on the SIS test. It is however, possible that qualitatively anatomical

responses might be different in five groups. A normal subject is likely to elicit anatomical

responses on the images structured to elicit anatomy. The clinical groups divided

somatic imagery on images meant to elicit it and perceived anatomy on the structures

generally not stimulating any anatomy. This process of evading reality and the theory of

somatic repression support such findings (Cassell, 1979, 1980b; Croog et al, 1971;

Savage, 1995, 2001).

The normal subjects have emitted less number of sex responses as compared to the

four pathological groups. Higher mean number of sex responses by the paranoid

schizophrenics with reference to their delusions of infidelity supports the observation

of most researchers on the SIS. The mean number of human movement responses by

the normal subjects has been found to be quite high and it is almost similar to that of

26 Psychometric Properties

depressive neurotics. A similar pattern of responding was evident for animal movement

category where the mean of normal group was less than most of the pathological

groups. However, none of the comparisons reached a level of statistical significance.

The mean number of most-typical responses by the normal subjects was higher than

that of the four pathological groups. Significant differences were found among most of

the groups. Normal subjects rejected significantly lesser number of images than did

the four pathological groups. The two schizophrenic groups rejected more number of

inkblot images than did their neurotic counterparts.

Table 8

Inter Group Comparisons for Different Scoring Categories of the SIS-II

Scoring NvsA NvsD NvsPS NvsUS AvsD AvsPS AvsUS DvsPS DvsUS PSvsUS

Indices

Human 0.94 1.07 1.47 3.79** 0.07 0.57 2.30 0.63 2.75** 1.46

responses

Animal 3.32** 2.78** 2.13* 3.15** 0.56 0.75 0.18 0.25 0.70 0.87

responses

Anatomical 1.11 1.92 0.31 2.49* 0.59 0.56 1.34 1.12 0.86 1.77

responses

Sex 0.07 0.46 1.71 0.80 0.20 1.24 0.44 1.33 0.11 1.00

responses

Human 1.96 0.03 0.49 2.22* 1.46 0.96 0.35 0.42 1.70 1.21

movement

Animal 0.62 0.21 0.60 0.62 0.35 0.23 0.86 0.42 0.71 0.89

movement

Most-typical 6.96** 5.64** 15.94** 17.06** 0.62 10.84** 12.00** 10.67** 11.76** 0.76

responses

Rejection 3.16** 2.53* 3.60** 3.62** 0.33 1.79 0.24 1.93 1.65 0.42

These results suggest that most-typical responses, rejection of images, animal

responses and total number of responses are hard core indicators in differentiating

normal subjects from the four pathological groups. Human and human movement

responses can also be considered as second order indicators, which discriminate

undifferentiated schizophrenics from normal subjects. However, sex responses, and

animal movement responses could not prove to be diagnostic indicators in this study.

Undifferentiated schizophrenics and paranoid schizophrenics gave lower number of

most-typical responses, less number of total responses and human responses with

higher number of rejection of images. These four indices i.e. low R, low H, low MT and

high rejection of images may be taken as indicators of undifferentiated and paranoid

schizophrenia (Dubey et al, 1995). Anxiety neurotics and depressive neurotics gave

more animal responses, rejected more images and gave less number of most-typical

responses than the comparable normal controls. These results suggest that these

three indices can help in discriminating anxiety neurotics and depressive neurotics

from the normal subjects.

Table 9

Multiple Discriminant Function Analysis in Terms of U-Statistic & Corresponding

F-Ratios for Identifying the best Predictors of SIS- II

Scoring Indices U- Statistic Approximate F- Degrees of

Statistic Freedom

Total No. of Responses 0.2979 29.957*** 8 288

Anatomical Responses 0.2830 19.290*** 12 379

Animal Responses 0.2652 14.774*** 16 435

Human Responses 0.2552 11.936*** 20 469

Sex Responses 0.2473 10.049*** 24 490

Animal Movement Responses 0.2422 8.649*** 28 503

Human Movement Responses 0.2369 7.621*** 32 511

Most Typical Responses 0.2273 6.983*** 36 515

Mishra (1996)

The diagnostic certainty of the SIS-II and the Rorschach was also ascertained by

applying the statistics of multiple discriminant function analysis and shown in table 9.

The U-values and their corresponding F-values demarcating the diagnostic

compatibility of the SIS with the Rorschach test showed the popular responses {U =

0.2231, F (4,145) = 126.25, p < 0.001} of Rorschach as the best indicator followed by

human movement and total number of responses. On the other hand, the anatomical

responses of SIS stepped out as the best predictor {U = 0.0971, F (24,490) = 19.408, p

< 0.001} followed by animal responses of Rorschach and animal movement responses

of the SIS-II. All the indices of the SIS-II were found to be as powerful discriminator as

were found for the Rorschach test, which establishes the diagnostic certainty of the

SIS-II. The contention of Cassell et al. (1995) that SIS-II is an extension of the Rorschach

thus gets statistically supported (Cassell et al,1995). It also validates the discriminating

power of SIS-II, and as SIS fares quite well with the Rorschach, the parallel status of the

SIS is established by the present findings (Mishra, 1995).

Table 10 presents the percentile norms for the nine SIS indices. These norms extend

some interesting features of responding by the normal control subjects. These norms

also vindicate the contention that this test can easily discriminate normal subjects from

28 Psychometric Properties

other pathological groups. Although this holds true for all the nine indices used in the

present study, yet some factors clearly stand out as hard core markers which need to

be briefly mentioned. Total responses emerge as a key marker in as much as it is

Table 10

Percentile Norms Based on the Responses of Normal Controls for

Nine Scoring Indices of the SIS-II

Percentile Total H A Anato- Sex Human Animal MT Rej

Points R mical Move- Movement ment

p100 79 31 22 35 10 14 6 18 7

p95 71 29 15 25 9 8 3 18 4

p90 69 28 14 21 9 7 3 18 3

p85 69 27 13 19 9 5 2 17 2

p80 68 26 12 18 9 5 2 17 2

p75 68 25 11 17 9 5 2 17 1

p70 67 24 10 16 9 5 2 17 0

p65 67 23 10 15 9 4 1 16 0

p60 67 23 9 14 8 4 1 16 0

p55 66 22 9 13 8 3 1 16 0

p50 66 21 8 12 8 3 1 16 0

p45 65 21 8 11 8 3 1 16 0

p40 65 20 7 10 8 2 0 15 0

p35 64 20 7 9 8 2 0 15 0

p30 63 19 6 9 8 2 0 15 0

p25 62 18 6 8 7 2 0 15 0

p20 61 17 5 7 7 2 0 14 0

p15 60 16 4 6 6 1 0 14 0

p10 59 15 3 6 6 0 0 14 0

p5 55 14 2 5 5 0 0 13 0

Mishra & Dwivedi (1997 )

noticed that at p5 there are 55 responses, which reach up to 79 at p 100. The median

point of this index is 66. In other words, this index clearly discriminates normal subjects

from other groups. The index of human responses is another potent marker where the

scores range from 14 to 31 with median at 21. The scores of pathological groups on

this index is much less. The animal responses are lesser in normal controls, which

range from 2 to 20 with a median at 8. The index of rejection shows zero response up

to p 70 while the pathological groups evince much higher rejection of images. The

norms derived on the basis of the normative sample of the present study have, however,

drastically limited value and it needs to be further enriched by administering SIS-II on

a much larger group.

From Table-11, it is evident that alcohol dependent cases who ‘improved’, differed

significantly from the cases who ‘failed to improve’ in respect to most of the SIS-II

indices i.e. ‘Total number of responses’, ‘Movement’, ‘Animal’, ‘Human’, ‘Sex’ and

‘Rejection of Images’. The ‘improved’ category of alcohol dependent cases were significantly more likely than ‘failed to improve’ cases to give less number of ‘Total’, ‘Movement’, ‘Animal’, ‘Sex’ and ‘Rejection of Images’ categories of responses. On the other hand, ‘improved alcohol dependent cases gave significantly more number of ‘Human’ responses in both the groups, did not differ significantly in respect to ‘Anatomical’ and ‘Most typical’ categories of responses.

Table 11

Comparison of SIS Indices for Alcohol Dependent Group (n=50)

SIS Indices No. of Patients Patients Failed Level of

Resp. Improved to Improve Significance

(n = 32) (n = 18)

Total Responses < 70 90. 62% (29) 33. 33% (06) .01

Movement < 15 81. 25% (26) 38. 88% (07) .05

Animal < 9 81. 25% (26) 33. 33% (06) .01

Human > 15 90. 62% (29) 55. 55% (10) .01

Anatomical > 12 84. 37% (27) 61.11% (11) N.S.

Sex < 3 78.12% (25) 50. 00% (09) .05

Rejection of Images < 1 87. 5% (28) 61.11% (11) .05

Most Typical > 13 78.12% (25) 55. 55% (10) N.S.

Dubey and Singh(1997) Figures in parenthesis indicate number of cases

From Table-12, it is apparent that at least four indices of SIS-II could discriminate

significantly both categories ‘improved’ and ‘failed to improve’ of drug dependent cases.

The ‘improved’ category of cases are likely to give significantly more number of ‘Total R’

and ‘Anatomical’ responses on one hand and significantly lesser number of ‘Movement’

and ‘Sex’ responses on the other hand than the ‘failed to improve’ category. Unlike

30 Psychometric Properties

alcohol dependent group, the ‘Animal’, ‘Human’ and ‘Rejection of images’ did not

discriminate significantly both the categories of drug dependent cases.

Table 12

Comparison of SIS Indices for Drug Dependent Group

SIS Indices No. of Patients Patients Failed Level of

Responses Improved to Improve Significance

(n1 = 32) (n2 = 18)

Total Responses > 66 85.18% (23) 60. 86% (14) .05

Movement < 10 74. 07% (20) 39.12% (09) .05

Animal < 10 66. 66% (18) 52.17% (12) N.S.

Human > 15 81. 48% (22) 60. 86% (14) N.S.

Anatomical > 13 81. 48% (22) 39.12% (09) .01

Sex < 3 88. 88% (24) 65. 21% (15) .05

Rejection of images < 1 81. 48% (22) 73. 91% (17) N.S.

Most Typical > 10 74. 07% (20) 65. 21% (15) N.S.

Dubey and Singh (1997) Figures in parenthesis indicate number of cases.

Table 13

Pathological Indices in Normal and Clinical Groups

Pathological Indices Normal Neurotic Schizo- Chi-square

phrenic and p

Disorder

Perseveration Present 0 7 13 X 2 = 16.06

Absent 15 8 2 p = .01

Pathological Anatomy 0-1 15 15 15 X 2 = 0

Score 2 & Above 0 0 0 p = n s

Hostility Aggression 0-2 11 13 11 X 2 = 0.02

Score 3 & Above 4 2 4 p = n s

Depression Score 0-1 13 15 13 X 2 = 0. 05

2 & Above 2 0 2 p = n s

Paranoia Score 0-1 15 14 9 X 2 = 7. 84

2&Above 0 1 6 p = 0. 01

X2 = was calculated combining normal, neurotics and schizophrenics together.

Table-13 reveals that, using the available definitions of pathological scales, only a

limited number of responses qualified to indicate the Pathological Anatomy scale, the

Hostility and Aggression scale, Depression and Paranoia. Even then, in Schizophrenics,

a significantly large number of subjects obtained Paranoid score above the cut off

points of 2, compared to Neurotics and Normals. Perseveration was clearly seen more

in psychotic cases than neurotics and normals.

The results of validity studies suggested that the test has significant construct, content

and discriminating validity to a great extent. Further studies are definitely required for

more detail on different groups from different countries.

Normative Data

To give a better understanding to one having less exposure to SIS procedure, images

assessing different areas are shown in table 14.

Table 14

SIS Images Assessing Different Areas

Area Images

Interpersonal Relationship A - 9,21,24,28,31

B - 4,5,28,31

Ego/Self Image A - 5,6,7

B - 22,27,28,29

Healthy Body Imagery A - 10,12,13,14,15,16,17,18,19,20,22,23

B - 9,11,16,17,20,30

Erotic Imagery A - 4,8,22,25,27

B - 18,25

Hostility / Aggression B - 15,21

Insecurity A - 13,24,26 - B - 19

Thinking In Group Conformity A - 7,9,13,20,23,27,30,31

B - 4,5,9,22,27,28,29,30,31

Reproduced from Dubey, Cassell, Pershad & Dwivedi (1995)

Some more data on SIS Video and its comparison with SIS-II and Rorschach test has

been shown in further tables for its compatibility and better understanding.

The SIS-Video stimulates the subjects more than SIS-II leading to more number of total

responses (R=74.08). The percentage of responses under the category of Movement,

Human, Animal, Anatomical and Most-Typical responses had been different in the two

sets. Movement responses were more in SIS-Video images whereas images are static

in SIS-II. The table does not indicate much difference in the scoring categories except

for the percentages of response in Movement, Human and Anatomical categories as

shown in Table 15.

32 Psychometric Properties

Table 15

Comparison of Responses on SIS-II & SIS-Video in Healthy Normals

Frequency per Protocol

Categories SIS-II* (n=100) SIS -Video* (n=100)

Total Number of Responses 69. 50 74. 08

Movement Responses 5. 00 (7.19) 9. 60 (12. 96)

Human Responses 17. 40 (25.02) 14. 66 (19. 79)

Animal Responses 9. 20 (13.22) 8. 65 (11. 68)

Anatomical Responses 12. 20 (17.55) 9.18 (12. 39)

Most-Typical Responses 12. 00 (17. 26) 13. 68 (18. 47)

* Reproduced from Dubey and Cassell (1993) Mean Percentages are given in parenthesis

Table-16 suggests that the responses in series A and series B are almost proportionate.

The correlation for the different scoring categories are significantly high except for

Most-Typical responses where the range of scores is less hence correlation was low

but still significant beyond .01 level.

Table 16

Comparison of Responses on Series A & Series B of SIS-Video

Categories Series A Series B r

Total Number of Responses 40.10 33. 92 . 86

Human Responses 18. 60 21. 46 . 56

Animal Responses 12. 64 10. 11 . 60

Anatomical Responses 12. 42 12. 17 . 59

Movement Responses 14. 01 11. 82 . 59

Most-Typical Responses 17. 46 19. 28 . 39

Table-17 suggests that 62 images of SIS-II/Video are able to discriminate normal

subject from neurotic and schizophrenic disorder. Certain indices are able to differentiate

within the clinical groups, neurotics & schizophrenic disorders. As evident the total

number of responses is more in normal than the other two groups. Normal subjects

give more of Most-Typical responses and schizophrenic disorders the least. Neurotics

are in between. Similar trend is found in Typical responses also. Atypical responses are

more in schizophrenic disorder and least in normal subjects. Neurotics are again in

between. Human responses have not been able to differentiate the groups, except that

the quality of human responses is poor in schizophrenic disorder. Animal and Anatomical

responses are more in schizophrenic disorders than in the normal group. Movement

responses are unable to discriminate the groups. Rejections of responses are much

higher in clinical groups and lowest in the normal group. Pathological responses are

almost absent in the normal group. Neurotics give more of Hostility Aggression score

and Depression Score where as the Schizophrenics give more of paranoid and

Pathological Anatomy Score. On the basis of the results, Most-Typical responses,

Rejection, Pathological Scores and Atypical indices are found to be hard-core indicators

and H, A, Anatomical and Sex as soft-core indices. The findings are shown in Table-17

Table 17

Diagnostic Indicators of SIS-II / Video

Indices Normal Neurotic Schizo Disorder

n = 300 n = 100 n = 100

Total Response 6 1-75 53-65 50-65

<62 <59 <62

Most Typical Response(out of 18) >13 <11 <08

Typical Response >50% 35-50% <30 %

Atypical Response <12% 15-20% >20 %

Human Content 20-25% >25% >25 % With

Poor Form

Animal content <11% >12% >16 %

Anatomical Content <15% >15% >18 %

Sex Content About 6% About 4% About 6 %

Movement <13 <08 <10

Rejection of Images <01 Image >05 >04

& PAS

Reproduced from Dubey, Cassell, Pershad & Dwivedi (1995) Hard-core Indices: MT, Rejection, Pathological,

Atypical Soft Indices: H, A, AT, Sex Analysis Based on Data Published in SIS Jr. of Proj. Psy. & Mental Health -

(1994) and Brit. Jr. Proj. Psy (1993) on over 500 cases.

Keeping the above indices in view the SIS was tried during the selection of employees

as a screening test. The findings are very encouraging. The subjects who did not reject

34 Psychometric Properties

any images, gave more than 13 MT responses, more then 20 percent of Human

responses, less then 15 percent of Animal and Anatomical responses and no

pathological responses were recommended for inclusion. However, none was rejected

on the basis of psychological test. Selection was finalized on the basis of overall

performance in the interview. Indices found successful are shown in table-18.

Table 18

Successful SIS Indices

Indices Scores

Rejection of Images Less Than 1 Percent

Most Typical Responses More Than 13 (out of 18)

Human Responses More Than 20 Percent

Animal Responses Less Than 15 Percent

Anatomical Responses Less Than 15 Percent

Pathological Responses Almost Absent

Reproduced from Dubey, Cassell, Pershad & Dwivedi (1995)

Table 19

SIS Video Responses of Executives in India

Indices Telephone Private Public Public

Cables Sector Sector Sector

(n=90)* (n=300)** (n=400)** (n=102)*

Total No. of Responses 65.57 70.25 66.50 68

Human Responses 19.21 14.20 20.30 19

(30.96) (28)

Animal Responses 7.31 8.45 10.65 8

(12.75) (11)

Anatomical Responses 12.38 11.40 11.80 14

(19.97) (19)

Sex Responses 2.24 4.20 3.30 3

(3.62) (4)

Most Typical Responses 15.24 14.55 14.55 15

(85) (83)

Movement Responses 6.31 8.35 5.75 5

(10.18) (7)

Rejection of Images 0.35 0.75 0.85 0.40

(0.20) (0.60)

Pathological Responses 0.19 Almost Almost 0.14

(0.21) Nil Nil (0.20)

*Dubey et al (2001) Percentage given in paranthesis. **Data presented in APA Annual Conference Aug. 2001

Sanfrancisco and Aug. 2002 Chicago.

Table 19 suggests that Executives have shown high productivty, creativty and good

interpersonal relationships. The aggressive responses were within normal range. They

had healthy body imagery and erotic perception. Rejection of images and Pathological

responses were found to be negligible. As a whole they showed healthy normal profile.

Table 20

Comparison of Adults with Adolescents

Percentage of Responses in

Categories Adults ( n = 100 ) Adolescents ( n = 48 )

Human 19. 79 34. 46

Anatomical 12. 39 22. 08

Most Typical 18. 47 11. 97

Mean Response 74. 08 59. 31

Reproduced from Dwivedi, Mishra and Cassell (1994)

Table 20 suggests that adolescents differed significantly on different SIS indices when

compared with adults. This is possible because of the differences in their maturity level

and interests.

Table 21

Comparison of Series A & B in Adolescents

Scoring Series- A (n=48) Series- B (n=48)

Categories Mean Percent (%) Mean Percent (%)

No. of Responses 29.79 29.52

Human Responses 9.56 32.09 10.88 36.86

Animal Responses 2.29 7.69 1.50 5.08

Anatomy Responses 7.06 23.70 6.04 20.46

Sex Responses 0.83 2.79 0.63 2.13

Movement 0.19 0.64 0.50 1.69

Most Typical Responses 3.77 12.66 3.33 11.28

Rejection 1.25 4.20 1.56 5.28

Dwivedi, Mishra and Cassell (1994) (None of the responses between series A & B differed significantly)

36 Psychometric Properties

None of the indices between Series A and Series B differed significantly as shown in

Table 21.

Fifty Normal, fifty Neurotics (Anxiety & Depression) and fifty with Schizophrenic Disorder

were given SIS and Rorschach and their protocols were scored on nine indices. X2 test

suggests that both the test are compatible on following nine indices as shown in

Table-22.

Table 22

Compatibility of SIS-II and Rorschach Indices in Normal and Psychiatric Groups

Scoring Indices Obtained X 2 Values Significant Level

Total Number of Responses 5.02 ns

Human Responses 2.86 ns

Animal Responses 4.11 ns

Anatomical Responses 0.78 ns

Sex Responses 0.36 ns

Human Movement Responses 5.79 ns

Animal Movement Responses 1.21 ns

Most-Typical Responses 4.04 ns

Rejection of Images 1.51 ns

Mishra (1996)

Values of Kendall’s coefficient of concordance between the nine indices of SIS and

Rorschach (Table 23) could reveal significant relationship suggesting their

concordance on nine indices. The SIS appraises the same construct as the Rorschach

test. However, because of its emphasis on self-administration, it moves beyond the

Rorschach.

Table 23

Kendall’s Coefficient of Concordance between the 9 Indices of SIS-II and Rorschach

Group wn k Significance Level

Normal .85 9 2 .01**

Anxiety .82 9 2 .01**

Depression .77 9 2 .01**

Schizophrenic Disorder

Paranoid Type .77 9 2 .01**

Schizophrenic Disorder

Undifferentiated Type .74 9 2 .01**

Mishra , M - 1995 ** p = .01 Level

It can be concluded that Somatic Inkblot Series Test is a reliable and valid test. Various

types of reliability such as Test-Retest reliability, Split-Half reliability and Parallel Form

reliability were found very high. The test has high content validity. Both the series A & B

have high correlation and can be used independently for quick assessment; however,

administration of the full test is necessary for complete assessment. The test is able to

discriminate normal subjects from clinical groups (Anxiety Neurotics, Depressives,

Schizophrenics, Alcoholics, Drug Dependents and many more). The Rorschach test

and the SIS were compared on nine common indices such as R, H, A, M, FM, Anatomical,

Sex, Rejection, and Most Typical/ Popular and found compatible. The SIS appraises the

same construct as the Rorschach test. However, due to the use of self-administration

and video technology it reaches beyond the Rorschach test. The Somatic Inkblot Series

has taken care of the lacunae of Rorschach test and does the Father of Inkblots,

Hermann Rorschach, start carrying on the work.

38 Psychometric Properties