Return to Commission IX Clearinghouse Index
The Parker Cognitve Development Inventory
John C. Parker and Albert B. Hood
Perry (l970) has developed a theory of cognitive development which traces the intellectual and ethical development of college students through nine sequential and hierarchial developmental positions. Each position describes a particular cognitive "form" in which individuals construe the diversity of the world around them. Perry's first three positions deal with cognitive forms that are characterized by a dualistic (right-wrong, we-they, black-white) world view, the next three positions by the individual's growing awareness and acceptance of the relativistic nature of life and the last three positions by the development of personal commitments in a relativistic world. The Parker Cognitive Development Inventory (PCDI) was developed to yield an objectively scored measure of cognitive development according to Perry's theory.
Instrument Development
As with the development of the other Iowa instruments, the PCDI was constructed in three phases: (l) development of the item pool, (2) pilot testing of the item pool, and (3) validation of the revised instrument. Creation of the initial items for the item pool was carried out within the framework of the following considerations of Perry's theory.
l. His theory of cognitive development emphasizes qualitative differences in patterns of thought rather than quantitative differences in behavior. Therefore the items were written with the intent of eliciting responses that could be interpreted as representing "how they think" rather than "what they do."
2. Perry found no students who expressed the structure of his position one and only occasionally did an individual judge assign a position nine to a student. Subsequent research has also failed to produce appreciable numbers of ratings at this positional extremes. Therefore items representing the middle seven positions were written and items representive of positions one and nine were not included in the item pool.
3. Perry's theory attempts to understand the cognitive structures that individuals employ to construe (l) the nature and origin of knowledge, (2) the nature and origin of values, and (3) the nature and origin of responsibility. Therefore an attempt was made to anchor each item in one of these three conceptual "themes."
4. Items representative of three content areas--those of education, career, and religion were included in the item pool. Items were written to represent (1) the various positions, (2) content areas, and (3) themes. Thus, the item "When it comes to religious issues, you have to learn to trust your own judgment," is at position three with relativism subordinate, it has a content area of religion, and has a theme of responsibility. The item "I can best describe my educational commitments as being established but still developing," is at position eight dealing with implications of commitment, its content is education and its theme deals with the nature of values.
As items were written, the readability and content base of each item were checked by several individuals. In addition, "expert" judges assessed the positional content validity for each statement. To accomplish this, 228 items were submitted to the "Syracuse Rating Group." These individuals have had extensive training and experience in the rating of Perry materials. They have rated well over 3,000 pieces of data since the formation of the group in l978. Two sets of the 220 items were sent to the rating group, each typed on an individual card, identified only by a code number. Two members of the rating group worked independently of each other in rating the items according to the Perry position they thought was best represented by the reasoning contained in the individual statement. At the conclusion of their ratings, the items were submitted to a third judge who attempted to resolve any discrepancies in the ratings. Those items where discrepant ratings could not be resolved or items that could not be rated at all were labeled "confusing." The rated items were then returned.
Subsequently l09 additional items were submitted to the rating group. These items represented new items and reformulations of items previously rated as "confusing." The second set of items was rated following the same procedures as for the first set. From the 337 total rated items, 300 were selected for inclusion in the pilot form of the PCDI.
The pilot form was administered to 53 students--30 undergraduate and graduate students at The University of Iowa and 23 students at a nearby Bible college. All participants were volunteers who completed the PCDI either as a part of a small group or on an individual basis. Use of the Bible college sample was for the purpose of exploring the content format of the PCDI. Students at the Bible college major in church related vocations and could be described as religiously evangelical. Their religious expression can be described as evangelical and their life style as conservative. Responses to the form were based on a forced choice format employing a four position Likert scale-- "strongly agree", "agree", "disagree", and "strongly disagree".
Initial plans for the development of the PCDI anticipated the creation of individual positional sub-scales for each Perry position two through eight within each of the three content scales. However, reliability estimates based on Cronbach's (l95l) alpha coefficient of internal consistency were quite low for several of the individual positional sub-scales. Due to low reliability estimates generated for these sub-scales it was decided to reorganize the sub-scales into the three more general positional groupings described by Perry. These are dualism (positions one to three), relativism (positions four to six), and commitment in relativism (posions seven to nine). Alpha coefficients were then recalculated for each of the revised sub- scales. In addition, items with low or negative item-to-total- sub-scales were eliminated until l6 items remained in each sub- scale. The alpha coefficients generated by the revised sub-scale format were sufficiently high as to indicate the utility in adopting the three sub-scale format for the validation phase of the study. These alpha coefficients are shown in Table G-1.
Validation
The revised l44-item PCDI was administered to additional students at both The University of Iowa and at the Bible college. Usable inventories were obtained from 257 University students and from 47 Bible students. In addition, the inventories from the 53 students involved in the pilot phase were rescored to the l44- item format which yielded a total validation sample of 357 students. A descriptive breakdown of the validation sample by college, grade level, and sex is shown in Table G-2.
Reliability coefficients for the nine sub-scales on the PCDI were obtained in the vicinity of .70--ranging from .64 to .87 (Table G-3). The highest alpha coefficients were found for the religion sub-scales followed by the career sub-scales, then the education sub-scales. This arrangement is probably created by the specificity of the response set tapped by the three content areas. Overall items that create a focused response will be more reliable than those that involve a more diffuse response set. It is probable that the religion sub-scales are more specific in nature than those in the other two areas.
Perry's theory is descriptive of a continuous process of development involving the individual's progressive replacement of less adaptive forms of reasoning by more adaptive forms. Therefore individuals should "occupy only one developmental position at a given time." In doing so, their highest positional scores should dictate lower scores for the other two positional sub-scales. Within each content scale the correlation of dualism with relativism and commitment should be negative and substantial while the relationship between relativism and commitment should be low but positive. Positional sub-scale correlations for each content sample are shown for the University sample and the Bible college sample in Table G-4. Expectations regarding the relationship of the sub-scales were only partially supported. In each content scale for both samples, the expected relationship of dualism with relativism and commitment was present. A high score for dualism was followed by lower scores for both relativism and commitment. However, the expected relationship between relativism and commitment was not supported. In each case, relativism and commitment were highly correlated in a positive direction.
When the Bible college sample was compared with the University sample, differences which would be expected between these two groups were found. Nine different analysis of variance tests were needed in order to account for all possible combinations involving the three positional sub-scales across the three content scales. In seven of the nine tests significant mean differences were found. Bible college subjects were more dualistic regarding all three content areas, less relativistic in all three areas, and more committed in the area of religion. There were no significant differences between the two samples in terms of their scores on the education and career commitment sub- scales.
If the PCDI is measuring development along the lines of the Perry theory, positional level scores should vary as a function of grade level with dualism scores decreasing with advancing grade level while relativism and commitment scores increase. Since there were the obvious sample differences mentioned above, subsequent comparison of grade level differences were based on a separate analysis of variance procedures for the University and Bible college samples. Support for this expectation was marginal at best for although the means were often ordered in the expected direction, the differences were significant in only five of the l8 individual analysis of variance procedures. These five were career dualism and religion relativism for the Bible college sample and education dualism, career commitment, and religion dualism for the University sample. The Scheffe Test was used to examine differences in group means in those cases where the overall F was significant. These differences were as follows:
l. Freshmen in the University sample were more dualistic regarding education than were graduate students.
2. Juniors in the Bible college sample were more dualistic regarding career than were seniors.
3. Freshmen in the University sample were more dualistic regarding religion than both juniors and graduates.
4. Freshmen in the Bible college sample were more relativistic concerning religion than were juniors.
Of particular importance in assessing the construct validity of the PCDI was the determination of the degree to which the developmental profiles conformed to the sequentiality predicted from Perry's theory. From his theory, it would be expected that a subject's highest and second highest scores would be in the adjacent developmental positions. Profiles were defined as "admissible" by Davison, King, Kitchener, and Parker (l980), when an individual's highest and second highest scores were found in adjacent developmental positions. Otherwise the profile was termed "inadmissible." For example, an individual who scored highest on the dualism sub-scale and second highest on the relativism sub-scale would have had an admissible profile. However, individuals with their highest score in dualism and their second highest on commitment would have an inadmissible profile. Of the six possible types of profiles, four would be admissible and two inadmissible. Therefore as a matter of chance one would expect that two-thirds of the actual profiles would be admissible and one-third would be inadmissible. Since the results for any subject can be classified as admissible or inadmissible the sampling distribution of the results is binomial. Therefore the hypothesis that the percentage of admissible profiles observed was different from that expected due to chance was tested through the development of standard (z) scores. The profile percentages and z scores for both the University and the Bible college samples are shown in Table G-6. With the exception of the religion profiles for the Bible college sample, expectation that the percentage of admissible profiles would differ significantly from that expected due to chance was supported. There are two possible explanations for the exception of the religion profiles for the Bible college sample. The first involves the possibility of a "dualistic commitment." This is where an individual takes on many appearances of commitment but is really operating from a type of dogma that has been adopted but not personalized or internalized. Therefore it is really a commitment based in dualistic thought. The second possibility was offered by Perry when explained that some types of religious commitment might have a very dogmatic expression. The commitment however is very personalized and internal to the individual. These results obviously do not offer an explanation as to which description best fits the Bible college sample.
The construct validity of an instrument can be examined by correlating the scores on that instrument with a second instrument that is designed to measure the same theoretical construct. For this purpose, 2l6 students (l96 University and 20 Bible college) completed both the PCDI and the Scale of Intellectual Development (SID). The SID was developed by Erwin (1983) and consists of l0l statements that on the basis of their wording, are each representative of one of the nine Perry positions. The SID represents a revision of the Scale of Ethical and Intellectual Development (SEID) developed by Roberts (l977). The SEID represented the initial item pool from which SID was developed. Initially testing of the SID was conducted at a large public university and involved 3,32l entering freshmen. Four factors were identified through a factor analysis of the scores (l) dualism, (2) relativism, (3) commitment, and (4) empathy.
The SID gives an overall positional rating without defining separate content areas and thus the composite sub-scales of the PCDI were used in this study. The correlations between the PCDI and SID scales are shown in Table G-7. As expected, the respective positional sub-scale correlations were positive and significant with each PCDI sub-scale showing its highest correlation with its respective SID sub-scale.
The PCDI contains items in three different content areas and according to Perry's theory, individuals cognitive-developmental positions should be relatively similar across content areas. This similarity across areas is called horizontal decalage. One of the purposes of including different content areas in the development of this instrument was to examine the possibility that an individual might vary developmentally across different content areas. Ideally, a direct comparison of an individual's mean score for one content area with that of a second content area would serve as a basis for evaluating this question. However, such a comparison would require that the separate content scales represent equivalent forms for measuring the same factor. Although the separate content scales of the PCDI were designed to address the particular factor of cognitive development, it cannot be assumed that the three content scales are equivalent. Without the ability to use a direct comparison of differences between means, the examination of horizontal decalage had to be based on the examination of differences between the correlations of the means. This was done on a within sample basis and it was expected that for the Bible college sample, the correlation of religious scores with education and career scores would be different from the correlation between education and career scores. It was expected that such differences would not be found within the University sample. These correlations are shown in Table G-8. As would be expected, the positional sub-scales correlations for each content scale were positive and significant within each sample. However, it is also evident in Table G-8 that the content scales are not highly correlated for dualism for the Bible college sample. Since scores within these samples are not independent, correlated t- tests (Ferguson, l976) were used to test for differences between content area correlations. Within each sample nine different t- tests were conducted and for the University sample five reached significance while only three were significant for the Bible college sample. Thus dualism scores for the Bible college sample were significantly less useful in predicting scores across content areas than within the University sample.
Conclusion
The alpha coefficients of internal consistency for the PCDI were acceptable ranging from a low of .64 for education/commitment to .87 for religion/relativism. The highest alpha coefficients were found for religion followed by career and then education. The construct validity of the PCDI was supported by certain of the analyses conducted in its development and validation, e.g., the high correlations with the SID, received less support particularly in regard to the commitment scores in others, e.g., the less than substantial differences between grade levels. Means and standard deviations for each of the nine scales on the PCDI are shown for both the University and Bible College samples in Table G-9.
References
Cronbach, T. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrica, 16, 297-334.
Davison, M. L., King, P. M., Kitchener, K. S., & Parker, C. A. (1980). The stage sequence concept in cognitive and social development. Developmental Psychology, 16, 121-131.
Erwin, T. D. (1983). The Scale of Intellectual Development: Measuring Perry's scheme. Journal of College Student Personnel, 24, 6-12.
Ferguson, G. A. (1976). Statistical analysis in psychology and education. Monterey, CA: McGraw-Hill.
Perry, W. G. (1970). Forms of intellectual and ethical development in the college years: A scheme. New York: Holt, Rinehart, & Winston.
Roberts, N. J. (1977). Construction and validation of the Scales of Ethical and Intellectual Development (SEID). Unpublished manuscript, The University of Iowa.
**************************************************
Table G-1
Pilot Form Alphas Based on
General Positional Subscales
________________________________________________________________
Positional CONTENT SCALE
Subscale Education Career Religion
________________________________________________________________
Dualism .81 .85 .87
Relativism .87 .92 .89
Commitment .81 .82 .83
________________________________________________________________
N = 53
**************************************************
Table G-2
Description of the Validation Sample
________________________________________________________________
University Bible College
Men Women Total Men Women Total
________________________________________________________________
Freshmen 24 40 64 13 17 30
Sophomores 16 67 83 6 5 11
Juniors 18 61 79 6 3 9
Seniors 11 20 31 9 11 20
Graduates 11 19 30 -- -- --
________________________________________________________________
TOTAL 80 207 287 34 36 70
________________________________________________________________
**************************************************
Table G-3
PCDI Validation Sample:
Subscale Alpha Coefficients
________________________________________________________________
Positional CONTENT SCALE
Subscale Education Career Religion
________________________________________________________________
Dualism .67 .66 .77
Relativism .72 .80 .87
Commitment .64 .71 .79
________________________________________________________________
N = 357
**************************************************
Table G-5
Percentage of Admissible
Profiles: University Sample
__________________________________________________________________
Procedure 1 Procedure 2
Content Scale % z % z
__________________________________________________________________
Education 98.5 11.25*** 98.5 ll.25***
Career 99.7 11.68*** 99.7 11.68***
Religion 94.8 9.93*** 94.2 9.71***
Composite 99.3 11.54*** 99.3 11.54***
__________________________________________________________________
Procedure 1 = Ties broken in favor of admissibility.
Procedure 2 = Ties broken in favor of inadmissibility.
***p<.001
Return to Iowa Index Return to Commission IX Clearinghouse Index