Reliability Analysis

 SOUTHERN CONNECTICUT STATE UNIVERSITY

501 Crescent Street • New Haven, Connecticut 06515

TO WHOM IT MAY CONCERN:

 

I conducted Item Analysis, Reliability, and Validity Studies of the Conflict Analysis Battery which was constructed by Dr. Albert Levis.

The reliability study of the battery indicates that the different scales have high reliability coefficients to be used in individual diagnosis. The Dominant Cooperative, Dominant Antagonistic, Submissive Cooperative, Submissive Antagonistic, and Psychic Conflict Tension scales have reliability coefficients which range from .88 - .96. The Antagonistic/Cooperative Scale has a reliability coefficient of .79. This is due to the fact that it is the shortest scale. It has only ten items. The different reliability coefficients are much higher than those of the MMPI scales.

The different validity studies of the battery indicate that the scales have enough evidence of validity for its purpose.

In addition, factor analysis was conducted on the battery items. The factors which were extradited highly overlapped the original scales. The results of factor analysis are indicative of the factorial validity of the different battery scales.

The battery is comparable to the best available personality inventories in construction, reliability and validity. Also, it measures important personality dimensions based on Dr. Levis' Theory of Behavior. By continuing research, the battery will gain prominence in personality assessment.

Shawky F. Karas, Ed.D.

Professor & Director of the Research & Measurement Program

 

Reliability Analysis

Reliability

As far as the reliabilities of the scales are concerned, the Conflict Analysis Battery scales are much better than the MMPI scales. "Several studies among psychiatric groups have reported coefficient within the range of .11 (Welsh, 1952) to .96 (Winfield, 1952)." (Kleinmuntz, 1967). The reliabilities of the Conflict Analysis Battery scales are comparable to the carefully constructed the California Psychological Inventory scales whose reliabilities are generally high, "in the upper .80's and lower .90's (Kleinmuntz, 1967). The Conflict Analysis Battery scales reliabilities are generally higher than those of the Guilford-Zimmerman Temperament Survey scales whose split-half reliabilities range from .75 to .87 (Kleinmuntz, 1967).

Validity

SA of the "Conflict Analysis Battery" has a significant correlation with SA of the "Interpersonal Check List." The correlation is noticeably a high correlation (.778). While the correlations between SA of the "Conflict Analysis Battery" and other scale in the “Interpersonal Check List” is not significantly higher than zero. This is a very clear evidence of the concurrent validity and discriminant validity of SA.

Factorial Validity

To find the factorial structure of the Conflict Analyis Battery items, the intercorrelations among the items were computed, factor analysis of the correlation matrix was carried out using the Principle Component method, and the factors were rotated using the Varimax method. Three factors were extracted using the criterion that the sign values should not be less than one. The items of the three factors are highly overlapping with the items of the major three scales. When six factors were extracted, the six battery scales were almost reproduced. The results are clearly indicative of the factorial validity of the six scales of the battery.

Conclusion

There is enough evidence of the construct validity of the battery scales besides its ability to discriminate between the clinical groups. While "the greatest limitation of the MMPI, as critics have repeatedly indicated (Adock, 1965; Lingoes, 1965), is its lack of sensitivity in discriminating within abnormal or normal group themselves" (Kleinmuntz, 1967).

The Conflict Analysis Battery scales ere constructed around the framework of a new theory and have enough evidence of reliabilities and construct validity for each scale to identify clinical groups and discriminate among them. While the Conflict Analysis Battery is theoretical the MMPI is atheoretical. In addition, the Conflict Analysis Battery scales have enough high reliabilities to be used for the interpretation of individual scores; and enough evidence of construct validity for the purpose of each scale.

 

Southern Connecticut State University

Item Analysis, Reliability, and Validity Studies
Conflict Analysis Battery

I. The Different Scales in the Battery

1.     Dominant Cooperative Scale (DC)

2.     Dominant Antagonistic Scale (DA)

3.     Antagonistic/Cooperative Scale (AC)

4.     Submissive Cooperative Scale (SC)

5.     Submissive Antagonistic Scale (SA)

6.     Psychic Conflict Tension Scale (PC)

II. Item Analysis

For each scale item analysis was conducted by obtaining the correla­tion between each item scores and the total scale scores.

1.       Dominant Cooperative Scale (DC)

The correlations between item scores and the total scale scores range from .244 to .632. None of the items should be excluded. There are 19 items in this scale.

2.       Dominant Antagonistic Scale (DA)

Thirty seven items out of forty five have correlations with the total scale scores which range from .261 to .696; and eight items have correlations with total scale scores below .2. These are items 3, 6, 7, 31, 37, 38, 39, and 45. They should be excluded or be included in other appropriate scales.

3.       Antagonistic/Cooperative Scale (AC) 

Eight items out of ten items have correlations with the total scale scores which range .37 to .69; and two items have correlations below .2. These are items 2 and 6. They should be excluded from this scale or included in other appropriate scales.

4.       Submissive Cooperative Scale (SC)

Twenty four items out of thirty have correlations with the total scale scores which range from .26 to .67; and six items have correla­tions below .2. These are items 4, 6, 11, 18, 26, and 28. They should be excluded from this scale or included in other appropriate scales.

1.Submissive Antagonistic Scale (SA)

Thirty eight items out of forty one have correlations with the total scale scores which range from .30 to .68; and three items have correlations below .2. These are items 2, 6, and 12. They should be excluded from this scale or included in other appropriate scales

              6.    Psychic Conflict Tension Scale (PC)

Thirty one items out of thirty five have correlations with the total scale scores which range from .24 to .76; and four items have correlations below .2. These are its 9, 23, 31, and 35. They Should be excluded from this scale or included in other appropriate scales

III. Reliability

For each scale the split-half reliability coefficient was computed. Also, the intercorrelations among the different scales were computed which are important to estimate the reliabilities of the difference scores between the scales.

A.   The Split-Half Reliabilities
No. of Items
Reliability
The Scales
Dominant Cooperative
19
.88
Dominant Antagonistic
45
.93
Antagonistic/Cooperative
10
.79
Submissive Cooperative
30
.88
Submissive Antagonistic
41
.96
Psychic Conflict Tension
35
.93
 

The reliabilities of all the scales except the third one range from .88 to .96. These reliabilities are high enough to use their respective scales in interpreting individual scores. The third scale has a reliability of .79. This is due to the fact that it is the shortest scale. It has only ten items. If the scale is lengthened to twenty items by adding comparable items measuring the same function and of the same quality the reliability will be .88.

B.  The Intercorrelations Between the Scales

The correlations between DC and DA, and SA and DA are quite high. The other correlations are moderate or low enough for reasonable reliabilities of the difference scores. The correlation between DC and DA is due to the fact that both scales have a dominant characteristic.

As far as the reliabilities of the scales are concerned, the Conflict Analysis Battery scales are much better than the MCI scales. "Several studies among psychiatric groups have reported coefficient within the range of .11 (Welsh, 1952) to .96 (Winfield, 1952)." (Kleinmuntz, 1967). The reliabilities of the Conflict Analysis Battery scales are comparable to the carefully con­structed the California Psychological Inventory scales whose reliabilities are generally high, "in the upper .80's and lower .90's" (Kleinmuntz, 1967). The Conflict Analysis Battery scales reliabilities are generally higher than those of the Guilford-Zimnerman Temperament Survey scales whose split-half reliability range from .75 to .87 (Kleinmuntz, 1967).                                                                                                                                                                                     

IV. Validity

A. Concurrent Validity & Discriminant Validity

To find out the concurrent validity and discriminant validity of the different scales of the Conflict Analysis Battery, the different scales were correlated with the different scales of the Interpersonal Check List by Timothy Leary

The correlations between DC of the "Conflict Analysis Battery" and DC 4 DA of the "Interpersonal Check List" are. significantly higher than zero; and relatively higher than the correlations between DC and BA or SC of the "Interpersonal Check List." These correlations are indicative of the concurrent validity and discriminant validity of DC of the "Conflict Analysis Battery."

DA of the "Conflict Analysis Battery" has a significant correlation with DA of the "Interpersonal Check List" which is relatively higher than the correlations between DA of the battery and other scales in the list. This is a clear evidence of the concurrent validity and discriminant validity of the DA scale in the battery.

SC of the "Conflict Analysis Battery" has a significant correlation with SA of the "Interpersonal Check List" which is higher the correlations between SC of the battery and DC and DA of the list. However, there is a comparable correlation between SC of the battery and BA of the list. This is an indicative evidence of the concurrent validity and discriminant validity of the SA scale of the battery.

SA of the "Conflict Analysis Battery" has a significant correlation with SA of the "Interpersonal Check List." The correlation is noticeably a high correlation (.778). While the correlations between SA of the "Conflict Analysis Battery" and other scale in the "Interpersonal Check List" are not significantly higher than zero. This is a very clear evidence of the concurrent validity and discriminant validity of SA.

B.  The different clinical individuals who completed the battery were classified into four different clinical groups - Dominant Aggressive, Dominant Cooperative, and Submissive Antagonistic, and Submissive Cooperative groups. Analysis of Variance was carried out for the different scale scores for each clinical group. The level of significance was .05 for each analysis.

For each clinical group there are significant differences between the scale means.

In the Dominant Aggressive group, the means of the Dominant Aggressive scale is higher than the means of the other scale scores. This is an indication of the construct validity of the Dominant Aggressive scale.

For the Dominant Cooperative group, the mean of the Dominant Aggressive scale scores is the highest mean (.67). However, the mean of the Dominant Cooperative scale scores should have been the highest mean according to the Conflict Analysis theory.

For the Submissive Antagonistic group, the highest means are for the Submissive Cooperative scale (.80) followed by the Antagonistic Cooperative scale (.44) and the Submissive Antagonistic scale (.23). According to the Conflict Analysis theory, the Submissive Antagonistic scale mean should have been the highest.

For the Submissive Cooperative group, the highest means are for the Dominant Aggressive (-.60) followed by the Antagonistic Cooperative scale mean (-.61) and the Submissive Cooperative scale mean (.60). According to the Conflict Analysis theory, the Submissive Antagonistic scale mean should have been the highest mean.

C. Factorial Validity

To find the factorial structure of the Conflict Analysis Battery items, the intercorrelations among the items were computed, factor analysis of the correlation matrix was carried out using the Principle Component method, and the factors were rotated using the Vatimax method. Three factors were extracted using the criterion that the sign values should not be less than one. The items of the three factors are highly overlapping with the items of the major three scales. When six factors were extracted, the six battery scales were almost reproduced. The results are clearly indicative of the factorial validity of the six scales of the battery

V. Conclusion

There is enough evidence of the construct validity of the battery scales besides its ability to discriminate between the clinical groups. While "the greatest limitation of the MMPI, as critics have repeatedly indicated (Mock, 1965; Lingoes, (1965), is its lack of sensitivity in discriminating within abnormal or normal group themselves" (Kleinmuntz, 1967).

The Conflict Analysis Battery scales were constructed around the frame­work of a new theory and have enough evidence of reliabilities and construct validity for each scale to identify clinical groups and discriminate among them. While the Conflict Analysis Battery is theoretical the MMPI is atheoretical. In addition, the Conflict Analysis Battery scales have enough high reliabilities to be used for the interpretation of individual scores; and enough evidence of construct validity for the purpose of each scale.

References

Kleinmuntz, Benjamin. Personality Measurement. Dorsey, 1967. 

Dr. Shawky F. Karas, Ed.D. Professor and Director of the Research Program 

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *