Computer attitude and achievement: is time an intermediate variable?

Studies have suggested that students' computer attitudes are related to their success in learning computer technology. The current study investigates how positive attitudes are "transferred" into higher learning achievement by determining an intermediate variable between computer attitudes and computer learning and achievement--Time spent on learning technology. Participants of the study are 609 teacher education students. Data analyses results demonstrate that (a) four computer attitude variables (enjoyment, motivation, importance, and freedom from anxiety) have linear relationships with Time, and (b) Time has a linear relationship with computer achievement, suggesting that computer attitudes influence computer learning mediated by Time. The design method of this study demonstrates research strategies to detect intermediate variables as well as direct and indirect relationships in the field of education.

**********

A review of studies on attitudes toward learning and using information technology in education has revealed that most studies have shown that students' attitudes toward technology are critical (Francis & Evans, 1995; Freedman & Liu, 1996; Liu & Johnson, 1998; Mitra, 1998; Mitra & Steffensmeier, 2000; Houtz & Gupta, 2001).

In the past, attitude studies have focused on investigating computer users' attitudes measured by one or more variables such as: Enjoyment--the degree to which students enjoy learning and using technology (Temple & Lips, 1989; Cooper & Stone, 1996; Liu & Johnson, 1998; Christensen & Knezek, 2001); Motivation--the degree to which students are willing to learn and use technology (Clariana, 1993; Kellenberger, 1996; Liu & Johnson, 1998, 2001; Christensen & Knezek, 2001); Importance--the extent to which students see learning and using technology as important (Pelton & Pelton, 1996; Corston & Colman, 1996; Liu & Johnson, 1998, 2001); and Computer Anxiety--the degree of fear that students feel while learning and using technology (Schumacher, Morahan-Martin, & Olinsky, 1993; Ayersman, 1996; Liu, 1997; Christensen & Knezek, 2001). Findings from these studies have suggested that such attitude variables are related to students' success in learning to use computer technologies.

A previous study (Liu & Johnson, 1998) found that the four computer attitude variables--Enjoyment, Motivation, Importance, and Freedom from Anxiety--had a linear relationship with computer achievement, and could be used to predict students' success in learning to use computer technologies. One question following that study was whether or not the relationship between attitudes and success is causal, and if so, how positive attitudes are "transferred" into higher learning achievements. While there might be numerous explanations for this, the current study focused on time spent learning and using computer technologies.

Time has been found to be related to learning and achievement in subject areas (Carroll, 1963; Parkerson, Lomax, Schiller, & Walberg, 1984; Reynolds & Walberg, 1992; DeBlois, 1997; Wilson, 2001; Hofferth & Sandberg, 2001). Previous research also suggests a causal relationship between time and learning (Carroll, 1963; Fredrick & Walberg, 1980; Parkerson, Lomax, Schiller, & Walberg, 1984; Slavin, 1995). Time has been studied as an internal variable when it referred to the time an individual spends on learning (DeBlois, 1997; Hofferth & Sandberg, 2001), or as an external variable when it referred to the instruction time or learning time provided by schools (Brosnan, 2001; Quilter & Chester, 2001). However, nothing was found concerning the relationships among time and computer attitude, computer learning and achievement.

The purpose of the current study was to determine whether computer attitude influences computer learning mediated by Time. Specifically, Time was treated as an internal variable (time that students spent on learning and using computer technologies) and examined in relation to both computer attitudes and computer achievement.

Underlying Logic of the Research Design

Computer-attitude studies in the past have often been based on the assumption that attitude is an internal factor that stimulates learning from within, and therefore contributes to learning outcomes (Carroll, 1963; Tarpy, 1997; Liu & Johnson, 1998). However, it is possible that attitude is a many-faceted variable, and therefore should not be measured or treated as a single variable. Accordingly, we wanted to use a set of internal attitude measures.

A core concept in the research design of the current study is that of an "intermediate variable." We define an intermediate variable (variable B) in a context that contains two other relevant variables (variables A and C). It can be illustrated as follows: if A is related to B, and B is related to C; then B is an intermediate variable between A and C. In another words, if A has an impact on B, and B has an impact on C; then we can say that A has an impact on C through B, and B is the intermediate variable between A and C.

Figure 1 illustrates the deductive procedure employed in this study to determine whether Time is an intermediate variable between computer attitude and computer achievement. It is known (Liu & Johnson, 1998) that four computer attitude variables--Enjoyment, Motivation, Importance, and Freedom from Anxiety--have a linear relationship with computer achievement (A relates to C). Therefore,

if--It can be demonstrated that:

* four attitude variables are related to Time (A relates to B), and

* Time is related to computer achievement (B relates to C),

then--It follows that:

* Time (B) is an intermediate variable between A and B.

Research Questions

There are two research questions in this study:

1. Can time spent on learning and using computer technologies be predicted from the computer attitude measures (Enjoyment, Motivation, Importance, and Freedom from Anxiety)?

2. Can computer achievement be predicted from the time spent on learning and using computer technologies?

[FIGURE 1 OMITTED]

METHOD

Sample

For the purpose of making predictions, the larger the size of the sample, the better the data from that sample will represent the population, and the more powerful the data will be in producing a useful predictive model.

The population from which the sample for this study was selected was the 1,061 undergraduate teacher education students enrolled in the college of education at a state university in the Western US. Participants in this study were 609 teacher education students who had been enrolled in an introductory computer technology course over three semesters. The course was required for all teacher education students in the college of education. The participants' ages ranged from 18 to 51 (the mean age was 23.26, and the median was 22), including 381 females and 228 males. Approximately 90% of the participants only had computer experiences of using a word processor, e-mail, or web search.

Measurements, Instruments, and Scoring

The data was analyzed using multiple regressions. The measurement of computer achievement was the students' final scores earned in the computer technology course. The final scores were the sum of 10 test scores and 10 module assignment scores. The 10 tests examined concepts and applications of current technologies, and theories and designs of technology integration. The 10 modules were projects focusing on the integration of the following basic technologies into the classroom: Windows basics, word processing, graphic designs, multimedia and hypermedia, web applications, web page designs, spreadsheets, databases, educational software, and comprehensive technology-based instructional design. The highest possible score for this course was 100. The term "computer achievement" has a broad scope; and for people in different fields it is measured with different standards or criteria. In the current study, it reflects college students' learning achievement in basic knowledge and skills of educational computing and technology integration.

The measurement for the "Time spent on computer" variable was the self-reported weekly average time in minutes an individual spends on learning or working with a computer. We 'used students' self-reported time as the measure of this variable for two reasons. First, in the current study, Time was treated as an internal variable--that is, time that students spent on learning and using computer technologies, which did not include any instruction time. Second, in this course, students controlled their own learning pace, and determined how much time they needed to complete the course-work at a given quality level; they could spend as much or as little time as they deemed necessary.

The four computer attitude variables are qualitative variables. In this study, the four qualitative attitude variables were measured quantitatively with a computer attitude instrument based on Aiken's (1979) attitude instrument. The instrument was a Likert-style questionnaire consisting of 24 statements (Liu & Johnson, 1998) sorted into four categories with three positive statements and three negative statements in each category. The six statements in each category measured one attitude variable (Table 1). The Likert scaling implied that each of the items had the same "level of difficulty"; that is, respondents found them equally easy or difficult to endorse. Also, this instrument had been found to reliably measure each of the variables independently (Aiken). Coefficient alpha for this instrument was 0.835 as reported in a previous study (Liu & Johnson, 1998).

As shown in Table 1, for example, the Motivation variable was measured by items 2, 10, 18, 6, 14, and 22. Items 2, 10, and 18 were positive statements and 6, 14, and 22 were negative statements. The answer for each statement must be chosen from: strongly disagree (SD), disagree (D), undecided (U), agree (A), or strongly agree (SA). For the positive statements, the score for answer SA was the highest (5 points); for negative statements, the score for answer SD was the highest (5 points). The score for each attitude variable was the sum of the six statements, and the highest possible score was 30. Higher scores represented a more positive attitude.

Data Collection

The data was collected during a period of three semesters. The questionnaire was sent to all students enrolled in the computer technology course--approximately 250 students each semester. A total of 609 of 740 students (82.3%) returned their questionnaires.

DATA ANALYSIS AND RESULTS

There were six variables included in the multiple regressions:

1. Computer Achievement

2. Time spent on computer

3. Enjoyment

4. Motivation

5. Importance

6. Freedom from Anxiety

The data analyses were performed in two steps (Table 2). In the first step, time was treated as the response variable regressed by the four computer attitude variables; and in the second step, time was treated as a predictor variable of computer achievement. From each step, a regression model was developed.

The plots obtained from the data exploration showed that there were no significant outliers, and the equal variance assumption and the normality assumption were not violated.

Results from First Regression Analysis 

In the first regression analysis, the predictor variables were the four attitude variables--Enjoyment (X1), Motivation (X2), Importance (X3), and Freedom from Anxiety (X4); and the response variable was Time (Y)--time spent on learning or using technologies. This analysis showed the following:

The F ratio for the model was significant (F = 370.849, p<0.0001). This indicated that at least one of the coefficients in the regression equation was not zero, and indicates that a linear relationship exists between the response variable (time) and the set of predictor variables (enjoyment, motivation, importance, and freedom from anxiety). The linear regression trend was also significant (F = 553.1, p<0.0001). This F ratio indicated that the linear model was the desired model that represented the data better than other regression models. According to this, the model should include only linear terms (of the predictor variables) that significantly contribute to the response variable (Time).

The next step was to determine which linear term(s), or predictor variable(s), should be included in the model. The t statistics for each variable were examined and were: Enjoyment (t = 6.493, p< 0.0001), Motivation (t = 7.248, p< 0.001), Importance (t = 4.724, p<0.0001), and Freedom from Anxiety (t = 18.889, p< 0.0001). The t statistics are used to test hypotheses about the individual parameters ([H.sub.0]: [B.sub.n] = 0). This could be explained in terms of comparing the fit of full and reduced models. In this case, the full model contained all four variables, and the reduced model contained all except the one being tested. For example, the t statistic of 7.248 for testing the null hypothesis that [B.sub.2] = 0 (there is no effect due to the variable Motivation) was actually testing whether the full, four-variable model fits the data better than the reduced three-variable model, without the variable Motivation, containing only the other three variables--Enjoyment, Importance, and Freedom from Anxiety. In other words, the test indicated whether there was variation in Time due to Motivation that is not due to the other three. The p value of 0.0001 for this test indicated that there was an effect due to Motivation. Therefore, Motivation should be included in the model. The t statistics for all four variables were significant. Therefore, the linear terms of all four variables should be included in the model.

The correlation coefficients showed consistent relationships: r = 0.7902 (Enjoyment and Time), r = 0.6405 (Motivation and Time), r = 0.5535 (Importance and Time), and r = 0.7568 (Freedom from Anxiety and Time). The four attitude variables all related to Time.

The regression analysis generated a set of coefficients that were used to formulate the regression equation:

Y = -428.15 + 15.22 (X1) + 3.34(X2) + 16.02(X3) + 5.57 (X4)

According to this equation, a one-unit increase in X2 (computer attitude Motivation score), for example, would increase 3.34 units (minutes) on Y (Time spent on learning and using computer technologies). The equation also displays the interesting fact that coefficients for X1 (Enjoyment) and X3 (Importance) are substantially larger than the other two. These two variables appear to be much better predictors of the Time score than are the other two variables. As can be seen from inspection of the equation, a one-unit increase in X1 (computer attitude Enjoyment score) would be accompanied by an increase of 15.22 units (minutes) on Y, and a one-unit increase in X3 (computer attitude Importance score) would be accompanied by an increase of 16.02 units (minutes) on Y. The increases for X2 (Motivation) and X4 (Freedom from Anxiety) are only 3.34 and 5.57 units respectively.

The equation indicates a positive linear trend of the data. Notice that the intercept of the equation has a negative value. This does not influence the overall trend of the data--when attitude scores increase/decrease, time spent on learning and using technology increases/decreases; and the value of the response variable will not be negative unless all the attitude variables are scored zero--which is beyond the range of the data.

Last, the R-Square of the model ([R.sup.2] = 0.7871) shows that a major portion (around 78%) of the variation of the response variable Time was explained by this model, or by the variation in the four attitude variables in the model.

In summary, from the first set of regression analyses, the relationship "A relates to B" (Figure 1) was approved: Time (spent on learning and using computer) was a function of computer attitudes.

Results from the Second Regression Analysis

In the second regression analysis, the predictor variable was Time (X)--time spent on learning or using technologies; and the response variable was Computer Achievement (Y). The following results were obtained:

The F ratio for the model was significant (F = 9339.85, p<0.0001). This F ratio was used to test the null hypothesis ([H.sub.0]:[B.sub.1] = 0; since Time was the only predictor variable, only one coefficient was tested). The associated p value of 0.0001 led to the rejection of this hypothesis and indicated that the coefficient for Time was not zero, that is, a linear relationship exists.

The linear regression trend was significant (F = 18596, p<0.0001). This F ratio indicated that the linear model was the desired model that represented the data better than other regression models. Since there was only one predictor variable in a simple regression analysis, the model selection was not to determine which predictor variables or what combination of them to include as in multiple regression analysis, but to compare with higher level models such as a quadratic model. Interestingly, the quadratic trend was also significant (F = 83.6, p<0.0001). However, the parameter estimate (the coefficient) for the quadratic term (Time*Time) was "-0.0001" which indicates very little contribution to the variance in the response variable. Therefore, to avoid confusion and simplify the model, the quadratic term was not included. Only the linear term was included. Since the linear trend was significant, we know the predictor variable Time must be a significant contributor to the variance in the response variable (Computer Achievement).

The t statistic for the Time variable provided consistent results (t = 31.165, p<0.0001). As discussed in the first regression analysis, a significant t statistics indicates that there is variation in Computer Achievement due to Time. The correlation coefficient showed that Time and Computer Achievement were highly related (r = 0.88). The regression analysis generated the coefficients that were used to formulate the regression equation:

Y = 42.35 + 0.24 (X)

According to this equation, for example, a one-unit increase in Time (minutes spent on learning and using computer technologies) would result in an increase of 0.24 units (achievement scores) on Y (Computer Achievement). The R-Square of the model ([R.sup.2] = 0.7705) shows that a major portion (around 77%) of the variation of the response variable Computer Achievement was explained by this model, or by the variation in the predictor variable Time.

In summary, from the second set of regression analyses, the relationship "B relates to C" (see Figure 1) was verified. Computer Achievement varies as a function of Time.

Model Functions

The analyses produced two sets of relationships shown in the two regression equations. The results and relationships can be summarized in the two model functions:

T = f[E, M, I, F] ______ (1)

CA = f[T] ______ (2)

Where:

T = Time spent on computer

E = Enjoyment

I = Importance

CA = Computer Achievement

M = Motivation

F = Freedom from Anxiety

f[] indicates a significant linear relationship ("a linear function of ...")

Function (1) reads "Time spent on learning and using computer technologies is a linear function of the four computer attittude variables--Enjoyment, Motivation, Importance, and Freedom from Anxiety;" and function (2) reads "Computer achievement is a linear function of the time spent on learning and using computer technologies."

CONCLUSIONS AND DISCUSSIONS

In conclusion, the two hypothesized conditions discussed in the underlying logic section have been supported, and the results from data analyses lead to the following conlusions:

1. Time spent on learning and using computer technologies can be predicted by a linear combination of the four computer attitude variables--Enjoyment, Motivation, Importance, and Freedom from Anxiety. Students who have more positive attitudes tend to spend more time on learning and using technologies.

2. Time spent on learning and using technology has a positive linear relationship with computer achievement. Students who spend more time learning about, or using, technology tend to have higher computer achievement scores.

3. Therefore, Time (spent on learning and using computer computer technologies) is an intermediate variable between computer attitdutes and computer achievements (as shown in Figure 2).

[FIGURE 2 OMITTED]

This may explain, at least from one perspective, how positive attitudes are "transferred" into higher learning achievement. As is known, computer attitude variables have a linear relationship with computer achievement (Liu & Johnson, 1998). However, computer attitudes are internal variables reflecting how people think or feel about computer technologies, and they will not produce computer achievement until some actions are taken. The current study, following up on the Liu and Johnson study, found that Time spent on the computer is one action that produces computer learning achievement. The major distinction between the findings of this and earlier studies is that previous studies describe what the relationship is, while the current study explores how and why the relationship exists. Accordingly, two more extended conclusions were drawn from the findings:

4. Time spent on computer influences computer achievement directly.

5. Computer attitudes influence computer achievement indirectly, through intermediate variable(s).

Because motivation and anxiety have had solid theoretical foundation in the literature and have for years been studied as important factors related to learning, we expected that they would predict the Time score more efficiently than would the other two. However, Enjoyment and Importance appear to contribute more substantially to the variance in Time scores than do Motivation and Freedom from Anxiety. This is interesting, since in the past, Enjoyment and Importance have not received as much attention in the literature as Motivation and Freedom from Anxiety. The current study indicates that these variables may be more important than previously suspected.

When we plan computer education for students, we should take into consideration not only factors related to establishing positive attitudes toward computer technologies, but also those related to increasing quality computer-learning time. To improve students' computer learning achievement, for example, we should provide students convenient access to computers so they can have sufficient time working on the computers. We should offer quality help to students so they can use their computer time efficiently; and we should plan the course work in different formats (e.g., electronic version, or online version) so students can work on computers at convenient times or locations.

This study suggests there are complex relationships, known or unknown, among the related factors that influence computer learning. Notice that conclusion five implies another question: what other intermediate variable(s) might exist that influence computer achievement? In further studies, we might continue to explore:

1. how factors such as computer access, computer help, or course delivery format relate to students' computer attitudes, time spent on computers, or computer achievements;

2. whether these factors can be considered intermediate variables, and

3. whether the relationships are direct or indirect.

Furthermore, this study could be a contribution to the literature from three perspectives. First, it provides new evidence concerning relationships between computer attitudes and achievement. Second, it suggests there may be practical strategies to improve students' learning of information technology. Third, it demonstrates a way to detect intermediate variables. In educational research, we often find that some factors or relationships are more complex than we expected, and cannot be explained with simple relationships. The design method we introduced here enables us to explore direct and indirect relationships, or relationships at different layers.

Limitations of the Study

The study has several limitations. First, participants were from one university and were all students in a teacher education program. Second, computer achievement was measured only at the level of basic computing skills. And third, Time was only measured as an internal variable without the inclusion of instruction time and time that the school provides for learning. These limitations may limit the generalization of the findings. It is hoped that other researchers will continue exploring the potential of improving learning achievement, and use the present findings to further refine research on this important topic.

Table 1 Computer Attitude Scoring

                         Positive Statement  Negative Statement Items
                         Items

Computer   Enjoyment     5, 13, 21           1, 9, 17
Attitude   Motivation    2, 10, 18           6, 14, 22
Variables  Importance    3, 11, 19           7, 15, 23
           Freedom from  8, 16, 24           4, 12, 20
             Anxiety
Answer                      SD  D  U  A  SA     SD  D  U  A  SA
Scores                      1   2  3  4  5      5   4  3  2  1

Table 2 Procedures of the Data Analysis

        Achievement  Time  Enjoyment  Motivation  Importance  Freedom
                                                              from
                                                              Anxiety

Step 1               Y     X1         X2          X3          X4
Step 2  Y            X

References

Aiken, L.R. (1979). Attitudes toward mathematics and science in Iranian middle schools. School Science and Mathematics, 79(3), 229-234.

Ayersman, D.J. (1996). Effects of computer instruction, learning style, gender, and experience on computer anxiety. Computers in the Schools, 12(4), 15-30.

Brosnan, M. (2001). On happiness and high achievement. Independent School, 60(3), 38-45.

Carroll, J.B. (1963). A model of school learning. Teachers College Record, 64(8), 723-733.

Christensen, R., & Knezek, G. (2001). Instruments for assessing the impact of technology in education. Computers in the Schools, 18(2/3), 5-25.

Clariana, T.B. (1993). The motivational effect of advisement on attendance and achievement in computer-based instruction. Journal of Research on Computing in Education, 20(2), 47-51.

Cooper, J., & Stone, J. (1996). Gender, computer-assisted learning, and anxiety: With a little help from a friend. Journal of Educational Computing Research, 15(1), 67-91.

Corston, R., & Colman, A.M. (1996). Gender and /social facilitation effects on computer competence and attitudes toward computers. Journal of Educational Computing Research, 14(2), 171-183.

DeBlois, R. (1997). Using summer programs to explore the relationship between time and learning. Phi Delta Kappan, 78(9), 714-718.

Francis, L.J., & Evans, T.E. (1995). The reliability and validity of the Bath County computer attitude scale. Journal of Educational Computing Research, 12(2), 135-146.

Fredrick, W.C., & Walberg, H.J. (1980). Learning as a function of time. The Journal of Educational Research, 73(4), 183-194.

Freedman, K., & Liu, M. (1996). The importance of computer experience, learning processes, and communication patterns in multicultural networking. Educational Technology Research and Development, 44(1), 43-59.

Hofferth, S.L., & Sandberg, J.E. (2001). How American children spend their time. Journal of Marriage and the Family, 63(2), 295-309.

Houtz, L. E., & Gupta, U. G. (2001). Nebraska high school students' computer skills and attitudes. Journal of Research on Computing in Education, 33(3), 316-327.

Kellenberger, D.W. (1996). Pre-service teachers' perceived computer self-efficacy based on achievement and value beliefs within a motivational framework. Journal of Research on Computing in Education, 29(2), 124-140.

Liu, L., & Johnson, D.L. (1998). A computer achievement model: Computer attitude and computer achievement. Computers in the Schools, 14(3/4), 33-54.

Liu, L., & Johnson, D.L. (2001). Assessing student learning in instructional technology: Dimensions of a learning model. Computers in the Schools, 18(2/3), 79-95.

Liu, M. (1997). The effects of HyperCard programming on teacher education students' problem-solving ability and computer anxiety. Journal of Research on Computing in Education, 29(3), 248-262.

Mitra, A. (1998). Categories of computer use and their relationships with attitudes toward computers. Journal of Research on Computing in Education, 30(3), 281-296.

Mitra, A., & Steffernsmeier, T. (2000). Changes in student attitudes and student computer use in a computer-enriched environment. Journal of Research on Computing in Education, 32(3), 417-428.

Parkerson, J.A., Lomax, R.G., Schiller, D.P., & Walberg, H.J. (1984). Exploring causal models of educational achievement. Journal of Educational Psychology, 76(4), 638-646.

Pelton, L., & Pelton, T.W. (1996). Building attitudes: How a technology courses affects preservice teachers' attitudes about technology. In B. Robin, J. D. Price J. Willis, & D. A. Willis (Eds.), Technology and Teacher Education Annual, (pp. 167-172).

Quilter, S.M., & Chester, C. (2001). The relationship between web-based conferencing and instructional outcomes. International Journal of Instructional Media, 28(1), 13-23.

Reynolds, A.J., & Walberge, H.J. (1992). A structural model of science achievement and attitude: An extension to high school. Journal of Educational Psychology, 84(3). 371-382.

Schumacher, P., Morahan-Martin, J., & Olinsky, A. (1993). Computer experiences, attitudes, computer and mathematical anxiety, and grades of MBA students. College Microcomputer, 11(3), 183-193.

Slavin, R.E. (1995, winter). A model of effective instruction. The Educational Forum, 59, 166-176.

Tarpy, R.M. (1997). Contemporary learning theory and research. New York: The McGraw-Hill Companies.

Temple, L., & Lips, H.M. (1989). Reliability and generalizability of the Collis attitudes toward computers survey. Educational Research Quarterly, 13(4), 6-9.

Wilson, F. (2001). Effect of time and level of visual enhancement in facilitating student achievement. International Journal of Instructional Media, 28(2), 159-169.

LEPING LIU, CLEBORNE MADDUX, AND LAMONT JOHNSON 

University of Nevada-Reno

Reno, NV

USA

[e-mail]