Effect of Computer-Mediated Communications on Teachers' Attitudes Toward Using Web Resources in the Classroom - Statistical Data Included

Journal of Instructional Psychology, June, 2001 by Tiffany A. Koszalka

Attitude Instrument

Since an appropriate attitude survey could not be found, one was constructed for this study. The three constructs of attitude were measured by a 21-item questionnaire assessing beliefs, feelings, and behaviors with a possible attitude score ranging from - 42 to 42. The items on the survey were adapted from two previous surveys, Internet in the Classroom -- End of Session survey (Grabowski & Koszalka, 1997) and the Computer Use Assessment survey (Marcinkiewicz & Welliver, 1993). The final attitude instrument was processed through a series of validation procedures. Educational psychology, instructional and educational design, web technology, and teaching experts reviewed the instrument for face and construct validity as well as readability. Data were gathered from the reviewers, analyzed, and used to revise items. The attitude instrument was administered to a sample of teachers prior to in-service workshops on integrating web resources into teaching and found to have an overall Cronbach alpha coefficient of .90 with U.S. teachers (Koszalka, Prichavudhi, & Grabowski, 2000; Koszalka, in press).

Results

Response Rate

Forty-five subjects agreed to participate in the study. Twenty-five of the thirty subjects assigned to the conversation groups participated as instructed during the study, one admitted to observing and not actively participating. Fifteen of the subjects in the control group responded as requested. A total of forty of the subjects returned the attitude survey.

Attitude By Conversation Variable

All analyses were mn at a minimum significance level of 0.05. A t-test resulted in a significant difference (t=2.98; p=.0093) in overall attitude mean scores between those in the conversation groups (mean=29.69) and those in the non-conversation control group (mean=22.90), thus the null hypothesis of no difference between conversation and non-conversation groups was rejected. The null hypothesis of no difference between the three constructs of attitude for conversation and non-conversation groups was also rejected based on analysis of the difference in the mean scores in the three sub-constructs of attitude including beliefs, feelings, and behavior or intent, as depicted in Table 1.

Table 1

T-tests for Overall Attitude and Sub-construct Scores

Construct/treatment     N       Mean    t-test     p

OVERALL ATTITUDE
                                         2.98     .009
  Conversation          20     29.69
  No-conversation       15     22.90

SUB-CONSTRUCTS
Beliefs                                  2.70     .016
  Conversation          25      13.6
  No-conversation       15      8.6

Feelings                                 2.59     .020
  Conversation          25      14.1
  No-conversation       15      8.4

Behavior (intent)                        3.96     .001
  Conversation          25      1.8
  No-conversation       15      0.8

Post-hoc Analysis - Attitude By Group Size

An ANOVA was mn to look at the differences in means for the different size CMC groups and the control group: small (n=5), large (n=15) and control. Overall attitude was significantly different between the three groups at the 0.05 level (f = 6.813; p = 0.008). A Tukey-Krammer Highly Significant Difference test was mn to compare the overall attitude scores for all pairs. It was found that the significant differences were between the treatment groups and control group. No difference was found between the small and large treatment group. See Table 2.