My class inventory-short form as an accountability tool for elementary school counselors to measure classroom climate.

A psychometric study with more than 2,800 upper-elementary-age students examined the reliability and factorial validity of the My Class Inventory-Short Form (MCI-SF). Factor analytic and structural equation modeling results suggested that the original measure is a less than satisfactory approach to appraise various dimensions of classroom climate. Researchers subsequently tested a revised version of the MCI-SF, showing that the 18-item measure with four scales (Cohesion, Competitiveness, Friction, and Satisfaction) was psychometrically sound. Implications for elementary school counseling programs and practices are discussed.

**********

In accord with the recent legislation (e.g., No Child Left Behind [NCLB], U.S. Department of Education, 2001) mandating increased educational accountability in public schools, leaders in the school counseling profession (American School Counselor Association [ASCA], 2003, 2004; Carey & Dimmitt, 2004; Erford, House, & Martin, 2003; Johnson & Johnson, 2003; Lapan, 2001, 2005; Paisley & Hayes, 2003; Ripley, Erford, Dahir, & Eschbach, 2003) have made it a point to encourage additional counselor accountability and efficacy research on comprehensive school counseling programs (CSCPs) and interventions. In fact, reflecting on a key element of NCLB (U.S. Department of Education)--doing what works based on scientific research--it has been suggested that school counseling will not thrive if it lacks a data-driven and results-based orientation to the profession (e.g., Adelman & Taylor, 2002; Bemak, 2000; Brigman & Campbell, 2003; Carey & Dimmitt; House & Hayes, 2002; Hughes & James, 2001; Isaacs, 2003; Myrick, 2003; Shepherd Johnson, 2000). Taking up this challenge, the National Center of School Counseling Outcomes Research has become an important repository of outcomes-based research. In brief, evaluation has become a centerpiece of the school counseling profession (e.g., Erford et al., 2003; Sink, 2005).

More specifically, counselors must take it upon themselves to meet not only student developmental competencies, but also program goals (ASCA, 2003; Lapan, 2001, 2005). For instance, comprehensive programs should lead to improved school and classroom climates that should, in turn, enhance student academic performance (e.g., ASCA; Bemak, 2000; Brown, 1999; Hernandez & Seem, 2004; Lapan, 2001; Littrell & Peterson, 2001; Sink, 2005; Sutton & Fall, 1995). Early on in the CSCP movement, Gerler (1985) stressed the need for school counselors to pay closer attention to improving the learning environments in elementary schools. Over time, school counselor program effectiveness has been shown to influence school climate (Lapan, 2005; Sutton & Fall). One concern, however, is the lack of psychometrically sound instruments to measure climate in elementary CSCP classrooms. The purpose of this article then is to report on a psychometric reanalysis of an elementary-level classroom climate survey, My Class Inventory-Short Form (MCI-SF; Fraser, 1982, 1989), which could be utilized as a CSCP accountability tool.

SCHOOL AND CLASSROOM CLIMATES

Definitional Issues

There is a multiplicity of definitions of classroom climate and school climate (e.g., Baker, Dilly, Aupperlee, & Patil, 2003; Hernandez & Seem, 2004; Kaplan & Geofffoy, 1990). In general, the latter concept reflects how children feel about and experience the essential characteristics of the school environment (i.e., entire physical and psychological milieu) as well as the school's faculty, staff, and administrators. Moreover, climate (a) focuses on the "quality of life phenomenon in school and classrooms" (Dunn & Harris, 1998, p. 100), (b) relates to the organizational milieu for teaching and learning, and (c) is linked to the aggregation of particular classroom climates (Johnson & Johnson, 1999).

Although the effects of classroom and school environments are interdependent and cumulative, Fraser (1991) has differentiated these settings in terms of their climates. Classroom climate obviously concerns the dynamics of classrooms or smaller learning environments, including how children feel and experience the characteristics of this milieu. Judgments as to the nature of the classroom climate are based on a student perceptual consensus about the educational, psychological, social, and physical aspects of the environment (Dunn & Harris, 1998).

General School Climate Literature

It is beyond the scope of this article to summarize the decades of research on this topic; however, a perusal of the school and classroom climate literature indicates that the stability and efficacy of elementary schoolchildren's social interactions influence their academic and social development (e.g., Baker et al., 2003; Dunn & Harris, 1998; Freiburg, 1999; Haynes, Emmons, & Ben-Avic, 1997; Left, Power, Costigan, & Manz, 2003; Lehr & Christenson, 2002; Moos, 1996; Morrison, 1985). The secondary school-classroom climate literature yields similar conclusions (e.g., Brand, Felner, Shim, Seitsinger, & Dumas, 2003; Diamantes, 2002; Hoge, Smit, & Hanson, 1990; Trickett, Leone, Fink, & Braaten, 1993; Trickett & Moos, 1973).

Elementary School Counseling Programs and Climate

Albeit the school counseling research base is sparse in support of the conclusion that counselors' services and interventions can positively affect elementary classroom and school climates (see, e.g., Lapan, 2005; Whiston, 2003; Whiston & Sexton, 1998, for reviews), there are several studies warranting mention. For example, Stewart and McKay (1995) showed that a group counseling program with upper elementary-age students is an effective mode to teach and facilitate prosocial, interpersonal coping skills. As the frequency of aggressive behaviors was reduced, the school environment was positively influenced. Using an inventory of classroom management style as the outcome variable, Martin and Baldwin (1996) subsequently demonstrated that school counselors, in part, could generate healthier classrooms through assisting nascent elementary school teachers with their classroom management strategies and styles. On the other hand, Otwell and Mullis's (1997) research failed to document that school counseling programs with effective counselor-led teacher consultation training could significantly improve classroom environment ratings on the My Class Inventory. Finally, a recent study reconfirmed that group counseling that focused on teaching elementary-age students prosocial skills diminished problem behaviors at recess time and helped create a more healthy school environment overall (Anderson-Butcher, Newsome, & Nay, 2003).

With some notable exceptions (e.g., Lapan, Gysbers, Multon, & Pike, 1997), school counseling researchers have not constructed good scales to assess key aspects of the classroom or school environment. A recent school counseling intervention study at the elementary school level could not be located that used a psychometrically sound school or classroom climate instrument to measure actual or perceived environmental changes following programmatic activities. The more general school and classroom environment literature was therefore scrutinized in an effort to uncover the best possible measure to deploy within elementary-level CSCPs.

POTENTIAL SCHOOL OR CLASSROOM CLIMATE SCALES

Because numerous measures of school and classroom climates have been developed over the past few decades, the search was narrowed to only those self-report surveys that were (a) multidimensional in nature; (b) designed especially for use in elementary classrooms and schools with less than proficient readers; (c) well documented in the elementary school, psychological, and educational literature; (d) supported by at least some reliability and validity data; and (e) readily available at a low cost. Although various instruments fit several of the criteria--for example, the Children's Classroom Environment Scale (Humphrey, 1984), Barclay Classroom Assessment System (Barclay, 1983), Classroom Environment Scale (Moos & Trickett, 1986), Inventory of School Climate-Student (Brand et al., 2003), Learning Environment Inventory (Fraser, Anderson, & Wahlberg, 1982, 1991), School Climate Survey (van Horn, 2003), and Tennessee Classroom Climate Inventory (Valesky, 1990)--only the My Class Inventory (e.g., Fraser ct al., 1991; Fraser & Fisher, 1986; Moos, 1994) and its corresponding abbreviated version (MCI-SF; Fraser, 1982, 1989; Fraser & Fisher) appear to meet each of them.

The MCI and MCI-SF are brief enough to avoid fatigue in younger students, logistically simple to administer and score, germane especially to what CSCPs are designed to enhance, relatively well researched, and widely cited in the school-classroom climate literature (e.g., Diamantes, 2002; Fugua, 1989; Morrison, 1979, 1985; Otwell & Mullis, 1997; Richardson, 1990; Wright & Lombardi, 1991). Because the shorter version of the MCI can be administered to children as young as 8-year-olds (Grade 3) and has some demonstrated reliability and validity (e.g., Fraser, 1982, 1989, 1991; Fraser et al., 1982, 1991; Fraser & Fisher, 1982a, 1982b, 1986; Fraser, Malone, & Neale, 1989; Fraser & O'Brien, 1985), it appeared to be a good candidate for a thorough psychometric reassessment.

OVERVIEW OF MY CLASS INVENTORY--SHORT FORM

This self-report measure is a 25-item derivative of the original 38-item MCI (Fraser, 1982, 1989) and is further detailed below in the Methods section. Assuming the MCI-SF was found to be psychometrically sound, the goal of school counseling interventions or classroom guidance lessons using this measure from pretest to posttest would be to generate higher Satisfaction and Cohesion scores and lower scores on the other three scales.

Numerous studies authored primarily by the instrument's developers and associates have examined the MCI's psychometric properties, including its reliability and, to a lesser extent, its validity. However, although the MCI-SF has been used in research, only a few investigations have reported on its psychometric properties. Fraser (1982) and Fraser and Fisher (1983) reported, using a very large sample (N = 2,305) of seventh-grade Australian students, the following internal consistency reliability coefficients (Cronbach alphas) for the longer and shorter MCI versions, respectively: Satisfaction, .88 and .78; Friction, .75 and .71; Competitiveness, .81 and .71; Difficulty, 73 and .65; and, Cohesiveness, .80 and .67. The concurrent validity comparing the long and short forms across each of the five scales was strong (rs = .91 to .97). These researchers attempted to demonstrate the MCI-SF's discriminant validity by correlating MCI-SF subscales scores with other school climate subscales (the Individualized Classroom Environment Questionnaire and the Classroom Environment Scale). The mean correlations with the other scales ranged from .13 (Difficulty) to .30 (Friction).

Two other investigations suggest that the MCI-SF possesses at least some internal consistency. Fraser and O'Brien (1985) reported largely adequate coefficient alphas for class means with an Australian sample of 758 Grade 3 students in 32 classes in eight schools in the Sydney metropolitan area (Satisfaction, .68; Friction, .78; Competitiveness, .70; Difficulty, .58; and Cohesiveness, .81). Using a sample of 1,675 Australian students in grades 7, 9, and 11, Byrne, Hattie, and Fraser (1986) found that the shorter version of the MCI yielded more than acceptable alpha coefficients for each of the scales, ranging from .84 to .93. Regrettably, there is little or no published evidence to date on the reliability and validity of the MCI-SF with American elementary schoolchildren.

In short, the construct and the factorial validity of the MCI-SF are equivocal with American school-age samples. To our knowledge, results from confirmatory factor analyses on the MCI or its short form have yet to be published in a juried publication. In addition, there are no published formal evaluations of the MCI-SF, but two technical reviews of the MCI exist, presenting largely similar conclusions. The instrument is a promising measure, but the initial psychometric data need to be replicated with U.S. schoolchildren and its validity requires further documentation (Barclay, 1985; Reed, 1985).

RESEARCH FOCUS

Given the paucity of strong empirical research conducted with American elementary schoolchildren demonstrating the reliability and validity of the MCI-SF, before it could be recommended to school counselors as a viable measure of classroom climate within CSCP contexts, the instrument needs to be thoroughly analyzed psychometrically. Therefore, the present study addressed the following research question: Does the MCI-SF demonstrate adequate internal consistency and factorial validity to justify its use with upper elementary-age students? To answer this question, the following methods were implemented.

METHODS

Participants and Sampling

Data were purposely and voluntarily drawn from 67% of the fourth-, fifth-, and sixth-grade elementary students (N = 2,835) attending one of 20 participating K-6 elementary schools in a largely lower-to middle-class urban school district in Washington state. Specifically, data were garnered in relatively equal proportions from students in fourth (n = 882, 31%), fifth (n = 1,023, 36%), and sixth (n = 912, 32%) grades. Overall, the students sampled from the three grade levels reflected an ethnically diverse sample (approximate percentages: European American, 44%; Asian American, 21%; Hispanic, 18%; African American, 15%; and Native American, 2%). The number of girls (n = 1,435, 50.6%) and number of boys (n = 1,373, 48.4%) were similar. Only 1% of the students did not specify a gender (n = 27) or grade level (n = 18). A very small percentage of students with special needs was included in the sample, but precise numbers were unattainable due to the constraints of confidentiality. Finally, on average the proportion of participants from each of the K-6 schools making up the entire sample was largely comparable.

Instrumentation

To recap, it has been suggested that the My Class Inventory-Short Form (Fraser, 1982, 1989; Fraser & Fisher, 1986) is well suited for use in elementary schools. The measure and its items are (a) written at a low reading level, (b) brief, (c) easily administered and hand scored, and (d) simple for children to answer. For the American respondents, the term pupils used in many items was changed to students. In addition, rather than using the MCI's 4-point Likert scale, the shorter format requires children to merely circle "yes" or "no" representing either "agreement" or "disagreement" with each item's content. The condensed format with 25 items, like the 38-item MCI, asks respondents about their perceptions of five different dimensions of their actual classroom environment: Satisfaction (items 1, 6, 11, 16, 21), Friction (items 2, 7, 12, 17, 22), Competitiveness (items 3, 8, 13, 18, 23), Difficulty (items 4, 9, 14, 19, 24), and Cohesiveness (items 5, 10, 15, 20, 25). The underlying scale meanings might be best described as follows: Cohesiveness--the degree to which students understand, collaborate, and are friendly with one another; Friction--the extent of tension and conflict among students; Difficulty--the level of difficulty students have with the classroom work; Satisfaction--the extent to which students feel satisfied with or like their class; and Competition--the perceived amount of classroom competition.

Moreover, each scale score is a sum of the five items composing the scale. Twenty of 25 items are scored in this manner: "Yes" = 3 points, "No" = 1, and omitted or invalidly answered (e.g., student circles both "yes" and "no") items = 2. A score of "2" is figured into the total score for each scale and interpreted as if the student was "uncertain" about whether the statement was an accurate or inaccurate reflection of the actual classroom environment. In a sense, then, a "2" score can be viewed as if the student was conflicted about the specific question. The remaining five items are reverse-coded: 6, 9, 10, 16, and 24 (i.e., "No" = 3 and "Yes" = 1). Each scale has a total possible score of 15 points. There is no overall score for the whole test.

Procedures

Elementary school counselors, with some assistance from relevant teachers, group-administered the MCI-SF early on in January and February 2003 to participants in grades 4, 5, and 6. Students who needed help were allowed to ask questions. In some cases, the items were read to the students. Trained graduate-level students reviewed all the respondents' inventories for accuracy, calculated subscale scores for each valid test, as well as assisted with data entry. Nearly all of the children required no more than 15 minutes to complete the inventory.

Data Analyses

Assuming that the scaling of the items approximated a 3-point Likert scale, internal consistency reliabilities (alpha coefficients) were computed for each of the derived factors as specified in Fraser (1989). Factorial validity and adequacy of fit for the dimensionality of the MCI-SF were assessed through principal component analyses (PCAs) with direct oblimin rotations ([DELTA] set at 0 to limit the amount of intercorrelation among derived components; Nunnally & Bernstcin, 1994; Pert, Lackey, & Sullivan, 2003) and confirmatory factor analyses (CFAs) using the Analysis of Moments Structure (AMOS; Arbuckle & Wothke, 1999), respectively. AMOS is a structural equation modeling (SEM) procedure that tests for the goodness-of-fit between the researcher's obtained or observed data and the hypothesized structure. To examine this fit, AMOS calculates maximum likelihood estimates from the covariance and several fit indexes. Given space limitations, detailed results could not be included (e.g., inter-item correlation matrices).

RESULTS

The results from item and reliability analyses using Fraser's (1982, 1989) original five-scale MCI-SF model are first summarized. Thereafter, the most germane PCA and CFA findings are presented.

Initial Item and Reliability Analyses

Descriptive statistics, item-scale correlations, and internal consistency estimates were calculated prior to computing the PCAs and CFAs. Item means were relatively similar across scales, ranging from 1.15 (SD = .52) to 2.46 (SD = .99). The item-scale correlations were largely in the acceptable range (.27 to .61, [r.sub.M] = 45) for all scales except for the problematic Difficulty measure (rs = .15 to .47, [r.sub.M] = .29). The range of inter-item correlations for each of the scales was as follows: Satisfaction, rs = .18 to .46; Friction, rs = .17 to .59, Competitiveness, rs = .17 to .40; Difficulty, rs = .06 to .63; and Cohesiveness, rs = .25 to .58. K-R 20 coefficients for Difficulty (r = .52), Competitiveness (r = .65), Satisfaction (r = .69), Friction (r = .71), and Cohesiveness (r = .72) indicated low to moderate internal consistency, respectively.

Principal Component and Secondary Reliability Analyses

To determine the MCI-SF's underlying dimensionality, the 25-item intercorrelation matrix was submitted to a series of PCAs. Oblique rotations (i.e., the derived components are thought to be interdependent or co-vary with each other), as opposed to orthogonal rotations (i.e., the derived components are assumed to be independent and noncorrelated), were utilized due to the suspected covariance among the inventory's scales, and they provided better component interpretability (Pett et al., 2003). The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett's test of sphericity were used, in part, to determine the adequacy of the data for factor analysis (Pett et al.). (1) By convention, components were marked by item loadings of approximately .30 or greater (Pett et al.).

Forced five-component solution. Although the Kaiser criterion, as well as the initial scree test, suggested a six-component solution (see below), a forced five-component PCA was computed on the data in order that Fraser's (1982, 1989) five-dimension model could be compared to the findings derived from the current sample. The resulting pattern matrix is available from the first author. Two of the resulting five components--Competitiveness and Cohesion--were composed of the same items presented in Fraser's original model. The Friction and Difficulty components were the most problematic in composition, with items from both dimensions loading on Fraser's Satisfaction and Difficulty components. The five-component solution explained 46.21% of the variance of the intercorrelation matrix. The rotated eigenvalues for components 1 to 5 were 2.66, 3.19, 2.90, 1.81, and 2.93, respectively. The magnitude of the intercomponent correlations ranged from .02 to .32. Because this solution was unsatisfactory--that is, the resulting factor structure was not clearly interpretable--coefficient alphas were not computed for each component.

Items that made up the Friction scale (MCI-SF) produced a relatively high inter-item reliability (r = .78) yet dispersed loadings over different components. Three of the five items for this scale--items 2, 12, and 22--addressed student perceptions of classroom conflict (e.g., Item 12: "Many students in my class like to fight"). These items loaded together, whereas the other two items (7 and 17) that did not mention fighting (i.e., Friction) loaded on other components. For example, Item 7 ("Some students in my class are mean") loaded with Cohesiveness questions, and Item 17 ("Certain students always want to have their own way") loaded with Competitiveness items. Close inspection of the three Friction items that loaded together revealed a specific component (i.e., a statistical artifact of factor analysis) rather than a coherent, comprehensible factor (Kline, 1994). Items included in Satisfaction, Cohesiveness, and Competitiveness also produced relatively strong loadings on their respective components.

Six-component solution. Although the "eigenvalue greater than one rule" can be misleading in that it can underestimate or overestimate the number of components to rotate, the six-component structure also met the primary criteria outlined by Kline (1994) for obtaining a simple structure or a readily comprehensible component structure. Like the previous PCA reported above, in comparison to the structure matrix, the pattern matrix (available from the first author) was more interpretable. In other words, the MCI-SF's questions loaded more clearly on different components, where each item had at least moderate to high loadings (i.e., at least .36) with one component and low loadings (i.e., below .22) with the remaining components. The items' means and standard deviations for this solution were identical to the five-component solution. Excluding items 9 and 24 from the Difficulty component, correlations between each scale's item scores and that scale's total score were adequate, ranging from -.34 to .61, with the majority of correlations falling between .43 and .49. Correlations between rotated components ranged from .03 to .32.

A six-component solution was obtained using a rotation procedure that assumes a small correlation between components (oblique rotation). This rotation procedure showed that about half (50.6%) of the total variance of the original intercorrelation matrix among MCI-SF questions was reproduced. The rotated eigenvalues (i.e., amount of total variance accounted for by specific items that make up a particular component) for components 1 to 6 were 3.06, 3.11, 2.81, 1.79, 3.04, and 1.51, respectively.

For the six-component solution, the items marking the Satisfaction component (coefficient = .69) were identical to those Satisfaction items on the MCI-SF as reported in Fraser (1982). The derived components of Competitiveness (coefficient = .68) and Cohesiveness (coefficient = .51) largely reflected Fraser's model of classroom climate; however, two additional items, 7 and 17, which originally loaded on Fraser's Friction component, loaded on Competitiveness and Cohesiveness, respectively. With these two items loading on different factors, Friction comprised only three items (2, 12, and 22; coefficient = .78). Three Difficulty items (9, 14, and 24) loaded on Component 6 (coefficient = .35).

Psychometric concerns related to the Difficulty scale were found. For both PCA solutions, two items requiring students to self-assess the level of classroom difficulty (i.e., Item 4, "In my class the work is hard," and Item 19, "School work is hard to do") loaded on the same component, while the other three Difficulty items asking students to make judgments regarding their peers (e.g., Item 24, "Most students in my class know how to do their work") loaded with comparable magnitudes on another component. Given these issues, Difficulty as modeled in MCI-SF appears in need of revision.

Confirmatory Factor Analyses

In order to test the goodness-of-fit of" Fraser's (1982, 1989) five-factor inventory and the derived six-factor classroom climate model, a SEM was formulated and examined through a series of CFAs. With the caveat that the observed data, which could be considered ordinal (i.e., because students only had three response options, the reliability of the results may be compromised), were treated as interval in nature, the following key assumptions of the CFA maximum likelihood approach were met: (a) the sample was large (asymptotic), (b) the distribution of observed variables was multivariate normal, (c) the hypothesized model was valid, and (d) the scale of the observed variables was assumed to be continuous (Byrne, 2001; Pettet al., 2003).

The conservative estimates of goodness-of-fit for SEMs, chi-square indexes ([chi square] and CMIN), represent essentially a comparison between the covariance matrices of Fraser's (1982) five-scale model and the covariance matrix of the sample, with lower and nonsignificant chi-squares indicating a better fit. However, because these estimates are highly sensitive to large sample sizes and the manner by which items are scored, they often yield spurious results (Thompson, 2004). Along with the chi-square estimate, therefore, several other more salient comparative indexes of fit were reported (Tabachnick & Fidell, 2001).

Although not reproduced here (see the first author for details), a five-dimension and a six-dimension SEM were conducted on the data, yielding two unsatisfactory models. The SEM findings together with the PCA results above indicate that with revisions (i.e., deletion of problematic items and the Difficulty scale) to some of the MCI-SF scales, a modified four-dimension inventory would be more reliable and the factorial validity of the instrument enhanced. This proposition was explored and confirmed in the follow-up PCA and CFA, respectively.

Follow-Up Four-Component PCA and CFA

To examine whether refinements to the MCI-SF scales would produce a psychometrically sound inventory, (a) problematic items discerned from previous analyses were eliminated, (b) an exploratory four-component PCA with direct oblimin rotation ([DELTA] = 0) was computed, and (c) a confirmatory four-dimension SEM was evaluated for goodness-of-fit. As such, of the original MCI-SF's 25 items, five items making up the original Difficulty scale and two of five items (7 and 17) from the Friction scale were removed, leaving a revised MCI-SF with 18 items and four scales (Cohesion, Competitiveness, Friction, and Satisfaction).

Explaining 50.19% of the variance in the intercorrelation matrix, the findings for the exploratory PCA revealed four coherent components with moderate coefficient alphas and reflecting the following four MCI-SF dimensions: Cohesion (items 5, 10, 15, 20, 25, [lambda] = 2.95, item-scale correlations ranged .41-.57, coefficient [alpha] = .72); Competitiveness (items 3, 8, 13, 18, 23, [lambda] = 2.44, item-scale correlations ranged .33-.47, coefficient [alpha] = .65); Friction (items 2, 12, 22, [lambda] = 2.70, item-scale correlations ranged .59-.67, coefficient [alpha] = .78); and Satisfaction (items 1, 6, 11, 16, 21, [lambda] = 2.94, item-scale rs ranged .37-.51, coefficient [alpha] = .69). The PCA pattern matrix showing how each MCI-SF item loaded on a particular component can be seen in Table 1. Intercomponent correlations were within the acceptable range for oblique rotated factors (rs = .02 to .31).

The four-dimension MCI-SF model also was evaluated through a SEM procedure (see Figure 1). The goodness-of-fit indexes suggested a relatively close model specification (e.g., CFI = .91, normed fit index = .90, and RMSEA = .05 [PCLOSE = .13]). The modifications to the MCI-SF therefore produced a better overall instrument to measure classroom climate in elementary schools. This model improves factor interpretability and reliability while, at the same time, reducing the number of items from 25 (MCI-SF) to 18 (MCI-SF revised).

[FIGURE 1 OMITTED]

To summarize, the findings suggested that the MCI-SF without amendment is insufficient to measure cogently the five dimensions of classroom climate as specified by Fraser (1982, 1989). Because items on the Difficulty dimension were problematic, these were deleted from the MCI-SF. A four-scale model (i.e., MCI-SF revised) of classroom climate is an adequate substitute. What follows is a discussion of these results in light of previous research on the instrument and their potential application to school counseling practice.

DISCUSSION

In an attempt to recommend to elementary school counselors a reliable and valid instrument to measure classroom climate within the context of CSCPs, the authors psychometrically reexamined Fraser's (1982, 1989) 25-item MCI-SF using a sample of nearly 3,000 third-, fourth-, and fifth-grade students from a large and ethnically diverse urban school district in Washington state. To reiterate, the results from the principal component analyses and the confirmatory factor analyses indicate that the five-dimension model suggested by Fraser yielded an inadequate representation of the data. Low to moderate coefficient alphas for each of the five scales were found as well. Furthermore, the small communality values for the five-component solution were perhaps an indicator of measurement error (Thompson, 2004). The CFA findings also revealed a less than adequate model specification. The six-component solution that emerged from the PCA was marginally interpretable. Finally, the authors, based on the previously discussed statistical analyses, modified Fraser's original inventory in an attempt to produce a more viable measure of classroom climate. As a result, an easy-to-use, 18-item, four-dimension solution emerged with adequate reliability coefficients for each scale. With the same scoring and administration procedures as the original MCI-SF, the Satisfaction, Cohesion, Competitiveness, and Friction scales from the revised MCI-SF are useful as an accountability tool for elementary CSCPs.

Comparison of Findings with Prior MCI-SF Research--Reliability and Factorial Validity

Several large psychometric studies (e.g., Byrne et al. 1986; Fraser, 1982) conducted with secondary students generally demonstrated stronger scale internal consistencies than those found in the current study with elementary-age students. Dunn and Harris's (1998) study, however, with Grade 4 American students (N = 758) reported, for the five-factor MCISF, alphas more in line with those found in the present study. Although cultural differences among the Australian and American students exist, the stronger coefficient alphas for secondary students could be due to cognitive developmental differences among samples.

Similar to this investigation, across each of the cited studies, the Difficulty scale produced the lowest reliability coefficients of the five components. Furthermore, in the present study, items from Fraser's (1982, 1989) Difficulty scale, for both the five- and six-component solutions, loaded on separate components. In particular, contrary to earlier studies mentioned above, the six-component model showed that the Difficulty scale is more accurately represented as two subscales, one that assesses the perceived level of difficulty for self and one that evaluates the perceived level of difficulty for classroom peers. Based on the follow-up PCA and CFA and to guard against unnecessarily increasing the complexity of the assessment tool (Reise, Waller, & Comrey, 2000), the revised (four-scale) MCI-SF appears to be the best alternative among contemporary options to measure elementary school classroom climates.

Implications for Elementary School Counseling Practice

Coupled with the results from Dunn and Harris (1998), the findings here suggest that the use of the MCI-SF in its current form with American elementary-age students warrants caution. Because items on the Difficulty scale seem to be less than readily comprehensible, employing instead the revised four-dimension, 18-item MCI-SF is most advisable. Practical suggestions for classroom use are discussed next.

Allowing for about 15 minutes to complete the survey, the MCI-SF can be administered and scored as specified in the manual (Fraser, 1989). In this way, items (4, 9, 14, 19, and 24) making up the Difficulty scale can be examined and then a decision to use or discard the scale can be made. If retained, caution should be used when interpreting this scale's mean scores. If eliminated, the scoring is the same for the Cohesion, Competitiveness, and Satisfaction scales (i.e., for each student, add up his or her score on each question for each of the four dimensions). For the Friction scale, however, only items 2, 12, and 22 are totaled (items 7 and 17 are ignored).

Elementary school counselors can use the resulting four-dimension instrument to assess how they are potentially influencing students' perceptions of their learning environment. Specifically, by reviewing over time trends in student responses to the questions posed on the Satisfaction, Cohesion, Friction, and Competitiveness scales, counselors should be able to judge, in part, how students report their level of satisfaction and cohesion before and after a 6-week series of guidance lessons on, for example, the topic of "kids working better together." What should occur following the classroom guidance activities is that the students will experience a more satisfying and cohesive classroom environment while, at the same time, experiencing less friction and competitiveness among students in the classroom. Thus, interpreting the students' average scores for the four MCI-SF dimensions from pretest to posttest should be straightforward: higher Cohesion and Satisfaction scale averages provide school counselors and teachers with positive indicators of a healthy classroom environment, whereas higher mean scores for the Competitiveness and Friction scales suggest the need for further collaborative classroom interventions. Oftentimes a graphic presentation of the scale mean scores is a good way to display the results.

Perhaps the best way to use the MCI-SF (original or revised version) is in the context of action research, in which the goal is to demonstrate to internal and external audiences that classroom guidance does, in part, positively affect the classroom climate (see, e.g., Isaacs, 2003; Sagor, 2000, for useful suggestions). For example, suppose bullying is occurring in several classrooms--conducting pretest and posttest analyses before and after classroom interventions can provide some indication of how students feel about their learning environment. Adding, perhaps, a few age-appropriate, open-ended questions to the survey, which would later be coded into coherent themes, could enhance the interpretability of the scale means. Examples: "Name three of the best things about your class?" "What are three things you would like to change about your class?" "When do you feel most happy in your class?" "In what three ways are kids nice in your classroom?" "Give me three ways kids are mean in your classroom?"

Fostering Safe, Healthy Classroom-School Environments

Though it is beyond the scope of this article to address fully the role that CSCP school counselors can play in collaboratively promoting safe, caring, and healthy learning environments, here are broad guidelines to consider:

* Emphasize that the school is like a neighborhood (community) where people enjoy, learn from, and help each other, as well as peacefully resolve their differences.

* Uncover core positive values that most members of the learning community can uphold.

* Appreciate unity and diversity (i.e., everyone has gifts) among students, teachers, and others, spotlighting the skills and successes of all children.

* Emphasize collaboration and cooperation over competitive goals and learning activities.

* Foster cross-grade learning and mentoring partnerships.

* Provide students with a "say" in the day-to-day class-school activities by conducting regular class and school-wide meetings and allowing for a limited array of healthy choices.

* Stress classroom management approaches that focus on fostering prosocial behavior. (Baker et al., 2003; Donald, 1989; Freiburg, 1999; Hernandez & Seem, 2004; Sink & Rubel, 2001)

Limitations of the Study and Suggestions for Future Research

Several research limitations and recommendations for future study warrant some attention. First, because the MCI-SF has been largely psychometrically tested within Australian school systems, direct comparisons with American samples can be problematic. Second, scale scores were assumed to reflect continuous data, but this supposition is tenuous especially when conducting item-level factor analyses (Reise et al., 2000). Third, using self-report data from elementary-age students is generally seen as less than reliable. Fourth, although the convenience sample was very large and drawn from a diverse population, this sampling procedure inherently diminishes the generalizability of the findings. Finally, research should be conducted with elementary-age American students to establish the construct validity, scale stability, and factor invariance among disparate age and ethnic groups of the four-dimension MCI-SF.

Conclusion

One of the major goals of comprehensive school counseling programs is to help nurture safe schools and healthy learning environments (ASCA, 2003; Hernandez & Seem, 2004; Lapan, 2001). Yet, in our results-driven educational world, it is imperative that school counselors assess whether their activities and interventions do in fact achieve this end. In our view, elementary counselors operating within the framework of a comprehensive program (e.g., ASCA's National Model, 2003) can more readily test this proposition, for various evaluation mechanisms should already be in place. It is perhaps best to evaluate environmental change at the classroom level rather than at the school level, for counselors are often in these settings conducting guidance curriculum units and meeting with students. Moreover, younger children's perceptions are probably most consistent when they appraise those climate situations with which they are most familiar.

To evaluate their buildings' learning environments, elementary school counselors require an easy-to-use measure that is also reliable and valid. Numerous classroom climate surveys have been utilized, but most are better suited to secondary-level students. In this article, we reported on the utility of the My Class Inventory-Short Form for this important task. Although the original measure has questionable psychometric properties, a good information-gathering tool for school counselors to consider is the revised MCI-SF detailed above. It is appropriate for upper-elementary-age students and uncomplicated to administer and score. This four-scale, 18-item inventory can assist counselors in gauging whether their classroom work is fostering a higher level of student satisfaction with the learning environment, building more cohesiveness among students, as well as reducing the perceived level of classroom friction and competitiveness.

References

Adelman, H. S., & Taylor, L. (2002). School counselors and school reform: New directions. Professional School Counseling, 5, 235-248.

American School Counselor Association. (2003). The ASCA national model: A framework for school counseling programs. Alexandria, VA: Author.

American School Counselor Association. (2004). Ethical standards for school counselors. Retrieved July 25, 2005, from http://www.schoolcounselor.org/content.asp?contentid=173

Anderson-Butcher, D., Newsome, W. S., & Nay, S. (2003). Social skills intervention during elementary school recess: A visual analysis. Children & Schools, 25, 135-146.

Arbuckle, J. L., & Wothke, W. (1999). AMOS 4.0 user's guide. Chicago: SPSS.

Baker, J. A., Dilly, L. J., Aupperlee, J. L., & Patil, S. A. (2003).The developmental context of school satisfaction: Schools as psychologically healthy environments. School Psychology Quarterly, 18, 206-221.

Barclay, J. R. (1983). Barclay Classroom Assessment System. Los Angeles:Western Psychological Services.

Barclay, J. R. (1985). Review of My Class Inventory. In J.V. Mitchell, Jr. (Ed.), Ninth mental measurements yearbook (p. 1025). Omaha, NE: Buros Institute of Mental Measurements.

Bemak, F. (2000).Transforming the role of the counselor to provide leadership in educational reform through collaboration. Professional School Counseling, 3, 323-331.

Brand, S., Felner, R., Shim, M., Seitsinger, A., & Dumas, T. (2003). Middle school improvement and reform: Development and validation of a school-level assessment of climate, cultural pluralism, and school safety. Journal of Educational Psychology, 95, 570-588.

Brigman, G., & Campbell, C. (2003). Helping students improve academic achievement and school success behavior. Professional School Counseling, 7, 91-98.

Brown, D. (1999). Proven strategies for improving learning and achievement. Greensboro, NC: ERIC Clearinghouse on Counseling and Student Services and CAPS.

Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming. Mahwah, NJ: Lawrence Erlbaum.

Byrne, D. B., Hattie, J. A., & Fraser, B. J. (1986). Student perceptions of preferred classroom learning environment. Journal of Educational Research, 80, 10-18.

Carey, J., & Dimmitt, C. (Eds.). (2004, June). Evidence-based school counseling interventions: Rules of evidence, outcomes and outcome measures. Symposium papers presented at the annual meeting of the American School Counselor Association, Reno, NV.

Diamantes, T. (2002). Improving instruction in multicultural classes by using classroom learning environment. Journal of Instructional Psychology, 29, 277-282.

Donald, T. M. (1989).The counselor in effective schools. The School Counselor, 36, 249-252.

Dunn, R. J., & Harris, L. G. (1998). Organizational dimensions of climate and the impact on school achievement. Journal of Instructional Psychology, 25, 100-115.

Erford, B.T., House, R., & Martin, P. (2003).Transforming the school counseling profession. In B.T. Erford (Ed.), Transforming the school counseling profession (pp. 1-20). Upper Saddle River, NJ: Merrill Prentice Hall.

Fraser, B. J. (1982). Development of short forms of several classroom environment scales. Journal of Educational Measurement, 3, 221-227.

Fraser, B. J. (1989). Assessing and improving classroom environment. In What research says to the science and mathematics teacher (No. 2). Perth, Australia: Curtin University of Technology, Key Centre for School Science and Mathematics.

Fraser, B. J. (1991). Two decades of classroom environment research. In B. J. Fraser & H. J. Walberg (Eds.), Educational environments: Evaluation, antecedents, and consequences (pp. 3-27). Oxford, UK: Pergamon Press.

Fraser, B. J., Anderson, G. J., & Wahlberg, H.J. (1982).Assessment of learning environments: Manual for Learning Environment Inventory (LEI and My Class Inventory). Perth, Australia: Australian Institute of Technology.

Fraser, B. J., Anderson, G. J., & Wahlberg, H.J. (1991). Assessment of learning environments: Manual for Learning Environment Inventory (LEI and My Class Inventory). Perth, Australia: Curtin University of Technology, Science and Mathematics Education Center.

Fraser, B. J., & Fisher, D. L. (1982a). Effects of classroom psychosocial environment on student learning. British Journal of Educational Psychology, 52, 374-377.

Fraser, B. J., & Fisher, D. L. (1982b). Evaluation studies: Predictive validity of My Class Inventory. Studies in Educational Evaluation, 8, 129-140.

Fraser, B. J., & Fisher, D. L. (1983, April). Assessment of classroom psychosocial environment: Workshop manual. Paper presented at the annual meeting of the National Association far Research in Science Teaching, Dallas, TX.

Fraser, B. J., & Fisher, D. L. (1986). Using short forms of classroom climate instruments to assess and improve classroom psychosocial environment. Journal of Research in Science Teaching, 23, 387-413.

Fraser, B. J., Malone, J. A., & Neale, J. M. (1989). Assessing and improving the psychosocial environment of mathematics classrooms. Journal for Research in Mathematics Education, 20, 191-201.

Fraser, B. J., & O'Brien, P. (1985). Student and teacher perceptions of the environment of elementary-school classrooms. Elementary School Journal, 85, 567-580.

Freiburg, J. H. (Ed.). (1999). School climate: Measuring, improving and sustaining healthy learning environments. New York: RoutledgeFalmer.

Fugua, D. J. (1989). How are we doing? Why not ask the kids: Measuring climate in kindergarten and primary classrooms. Paper presented at the annual meeting of the Southern Association an Children Under Six, Richmond, VA. (ERIC Document Reproduction Service No. 313162)

Gerler, E. R. (1985). Elementary school counseling research and the classroom learning environment. Elementary School Guidance & Counseling, 20, 39-48.

Haynes, N. M., Emmons, C., & Ben-Avie, M. (1997). School climate as a factor in adjustment and achievement. Journal of Educational Psychological Consultation, 8, 321-329.

Hernandez, T. J., & Seem, S. R. (2004). A safe school climate: A systemic approach and the school counselor. Professional School Counseling, 7, 256-262.

Hoge, D. R., Smit, E. K., & Hanson, S. L. (1990). School experiences predicting changes in self-esteem of sixth--and seventh-grade students. Journal of Educational Psychology, 82, 117-127.

House, R. M., & Hayes, R. L. (2002). School counselors becoming key players in school reform. Professional School Counseling, 5, 249-256.

Hughes, D. K., & James, S. H. (2001). Using accountability data to protect a school counseling program: One counselor's experience. Professional School Counseling, 4, 306-309.

Humphrey, L. L. (1984). Children's self-control in relation to perceived social environment. Journal of Personality and Social Psychology, 46, 178-188.

Isaacs, M. L. (2003). Data-driven decision making: The engine of accountability. Professional School Counseling, 5, 235-248.

Johnson, D. W., & Johnson, R. T. (1999). Learning together and alone: Cooperative, competitive, and individualistic learning (5th ed.). New York: Allyn & Bacon.

Johnson, S. K., & Johnson, C. D. (2003). Results-based guidance: A systems approach to student support programs. Professional School Counseling, 6, 180-184.

Kaplan, L. S., & Geoffroy, K. E. (1990). Enhancing the school climate: New opportunities for the counselor. The School Counselor, 38, 7-12.

Kline, P. (1994). An easy guide to factor analysis. London: Routledge.

Lapan, R. T. (2001). Results-based comprehensive guidance and counseling programs: A framework for planning and evaluation. Professional School Counseling, 4, 289-299.

Lapan, R. T. (2005). Evaluating school counseling programs. In C. A. Sink (Ed.), Contemporary school counseling: Theory, research, and practice (pp. 257-295). Boston: Houghton Mifflin.

Lapan, R. T., Gysbers, N. C., Multon, K. D., & Pike, G. R. (1997). Developing guidance competency self-efficacy scales for high school and middle school students. Measurement& Evaluation in Counseling and Development, 30, 4-16.

Leff, S. S., Power, T. J., Costigan, T. E., & Manz, P. H. (2003). Assessing the climate of playground and lunchroom: Implications for bully prevention programming. School Psychology Review, 32, 418-430.

Lehr, C. A., & Christenson, S. L. (2002). Best practices in promoting a positive school climate. In A. Thomas & J. Grimes (Eds.), Best practices in school psychology IV (pp. 929-947). Bethesda, MD: National Association of School Psychologists.

Littrell, J. M., & Peterson, J. S. (2001).Transforming the school culture: A model based on an exemplary counselor. Professional School Counseling, 4, 310-319.

Martin, N. K., & Baldwin, B. (1996). Helping beginning teachers foster healthy classroom management: Implications for elementary school counselors. Elementary School Guidance & Counseling, 31, 106-113.

Moos, R. H. (1994). The social climate scales: A user's guide. Palo Alto, CA: Consulting Psychologists Press.

Moos, R. H. (1996). Understanding environments: The key to improving social processes and program outcomes. American Journal of Community Psychology, 24, 193-201.

Moos, R. H., & Trickett, E. (1986). Classroom Environment Scale manual (2nd ed.). Palo Alto, CA: Consulting Psychologists Press.

Morrison, T. L. (1979). Classroom structure, work involvement, and social climate in elementary school classrooms. Journal of Educational Psychology, 71, 471-477.

Morrison, T. L. (1985). Personal and professional boundary attitudes and effective group leadership in classrooms. Journal of Psychology, 119, 101-111.

Myrick, R. D. (2003). Accountability: Counselors count. Professional School Counseling, 6, 174-179.

Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York: McGraw-Hill.

Otwell, P. S., & Mullis, F. (1997). Counselor-led staff development: An efficient approach to teacher consultation. Professional School Counseling, 1, 25-30.

Paisley, P. O., & Hayes, R. L. (2003). School counseling and the academic domain: Transformations in preparation and practice. Professional School Counseling, 6, 198-204.

Pett, M. A., Lackey, N. R., & Sullivan, J. J. (2003). Making sense of factor analysis. Thousand Oaks, CA: Sage.

Reed, J. C. (1985). Review of My Class Inventory. In J. V. Mitchell, Jr. (Ed.), Ninth mental measurements yearbook (pp. 1025-1026). Omaha, NE: Buros Institute of Mental Measurements.

Reise, S. P., Waller, N. G., & Comrey, A. L. (2000). Factor analysis and scale revision. Psychological Assessment, 12, 287-297.

Richardson, A. G. (1990). Classroom learning environment: Some differences among school types. Perceptual & Motor Skills, 71, 518.

Ripley, V., Erford, B. T., Dahir, C., & Eschbach, L. (2003). Planning and implementing a 21st-century comprehensive developmental school counseling program. In B. T. Erford (Ed.), Transforming the school counseling profession (pp. 63-119). Upper Saddle River, NJ: Merrill Prentice Hall.

Sagor, R. (2000). Guiding school improvement with action research. Alexandria, VA: Association for Supervision and Curriculum Development.

Shepherd Johnson, L. (2000). Promoting professional identity in an era of educational reform. Professional School Counselor, 4, 31-40.

Sink, C. A. (Ed.). (2005). Contemporary school counseling: Theory, research, and practice. Boston: Houghton-Mifflin.

Sink, C. A., & Rubel, L. (2001).The school as community approach to violence prevention. In D. S. Sandhu (Ed.), Faces of violence: Psychological correlates, concepts, and intervention strategies (pp. 417-437). Huntington, NY: Nova Science.

Stewart, J., & McKay, R. (1995). Group counselling elementary for school children who use aggressive behaviours. Guidance & Counselling, 11, 12-15.

Sutton, J. M., & Fall, M. (1995).The relationship of school climate factors to counselor self-efficacy. Journal of Counseling & Development, 73, 331-336.

Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). New York: Harper Collins.

Thompson, B. (2004). Exploratory and confirmatory factor analysis. Washington, DC: American Psychological Association.

Trickett, E. J., Leone, P. E., Fink, C. M., & Braaten, S. L. (1993). The perceived environment of special education classrooms for adolescents: A revision of the Classroom Environment Scale. Exceptional Children, 59, 411-420.

Trickett, E. J., & Moos, R. (1973).The social environments of junior high and high school classrooms. Journal of Educational Psychology, 65, 93-102.

U.S. Department of Education. (2001). No Child Left Behind. Retrieved June 6, 2004, from http://www.ed.gov/nclb/ overview/intro/index.html

Valesky, T. (1990, November). Baseline data on school climate, classroom climate, and self concept in schools using school based decision making. Paper presented at the annual meeting of the Mid-South Educational Research Association, New Orleans, LA. (ERIC Document Reproduction Service No. ED326951)

van Horn, M. L. (2003). Assessing the unit of measurement for school climate through psychometric and outcome analyses of the School Climate Survey. Educational and Psychological Measurement, 63, 1002-1019.

Whiston, S. (2003). Outcomes research and school counseling services. In B. T. Erford (Ed.), Transforming the school counseling profession (pp. 435-447). Upper Saddle River, NJ: Prentice Hall.

Whiston, S., & Sexton, T. (1998). A review of school counseling outcome research: Implications for practice. Journal of Counseling & Development, 76, 412-426.

Wright, A. K., & Lombardi, L. G. (1991). Investigating classroom environment in British schools. Association of Educational Psychologists Journal, 7, 100-104.

(1) The KMO estimate (.83) was within tolerance and Bartlett's test revealed that the null hypothesis (i.e., the correlation matrix is an identity matrix) could be rejected ([300] = 14596.59, p < .000; determinant = .006). Moreover, the conditions for determining the number of components to be retained for rotation were (a) Kaiser's "eigenvalue greater than one" criterion, (b) Cattell's scree test, (c) percentage of variance explained by each component, and (d) the attainment of simple structure (Nunnally & Bernstein, 1994).

Christopher A. Sink, Ph.D., NCC, LMHC, is a professor and chair, School Counseling and Psychology, School of Education, Seattle Pacific University, Seattle, WA. E-mail: [e-mail]

Lisa R. Spencer, Ed.D., is an assistant professor at National University, La Jolla, CA.

Table 1. Four-Component PCA Pattern Matrix (Rotation Method:
Oblimin with Kaiser Normalization) Based on Revised 18-Item MCI-SF

                                                           Component
Item   Description                           Cohesion   Competitiveness

   5   Everybody is my friend                  .78

  10   Some students are not my friends        .77

  15   All students are close friends          .67

  20   All students like one another           .66

  25   Students like each other as friends     .43

  18   Some students always try to do                         .69
       better than others

   8   Most students want their work                          .66
       to be better than friend's

  13   Some students feel bad when
       they don't do as well as others                        .63

  23   Few students want to be first
       all the time                                           .62

   3   Students often race to see
       who can finish first                                   .58

  22   Students in my class fight a lot

   2   Students are always fighting
       with each other

  12   Many students in my class
       like to fight

  21   Class is fun

  11   Students like my class

   1   Students enjoy schoolwork

  16   Some students don't like class         -.21

   6   Some students are not happy                          -.23

Item   Description                           Friction   Satisfaction

   5   Everybody is my friend

  10   Some students are not my friends

  15   All students are close friends

  20   All students like one another

  25   Students like each other as friends                  .21

  18   Some students always try to do
       better than others

   8   Most students want their work
       to be better than friend's

  13   Some students feel bad when
       they don't do as well as others

  23   Few students want to be first
       all the time

   3   Students often race to see
       who can finish first

  22   Students in my class fight a lot        .85

   2   Students are always fighting
       with each other                         .82

  12   Many students in my class
       like to fight                           .79

  21   Class is fun                                         .79

  11   Students like my class                               .77

   1   Students enjoy schoolwork                            .66

  16   Some students don't like class                      -.53

   6   Some students are not happy                          .36

Note. Only loadings > .20 are included; items with loadings > |.36|
mark the component; items 6, 10, and 16 are reverse-coded.