RESEARCH METHODOLOGY: QUANTITATIVE

ASSIGNMENT 2:

Mathematical perceptions

Full Name:
1. Introduction

SPSS (Statistic Package for the Social Sciences) provides a powerful statistical analysis and data

management system in a graphical environment. With the assistance of SPSS, researchers can

perform most of the regular data management and statistical analysis functionalities such as

descriptive statistics, commonly used tests, linear regression, ANOVA, etc. In this study, the data

were entered into SPSS.

Six scales including Generalization, Induction, Deduction, Use of symbols, Logical relations,

and Mathematical proof were used to access students’ mathematical thinking. Each scale was

measured by five test items. Each item in the mathematical thinking test was scored out of three.

Students who did not answer a question were taken as not being able to do the question and were

given a score of zero for that item. The total score for each scale was calculated by adding up the

scores for the 5 items in that scale.

2. Participants

The sample consisted of 576 eleventh-grade students attending 9 schools. Of these, 176 students

(30.6%) were from two schools in an urban area, 111 students (19.3%) were from two schools in

a suburban area, and 289 students (50.2%) from five schools in a rural area. There were 279

males (48.4%) and 297 females (51.6%) in the sample. Table 1 shows student numbers by school

and school location, and Table 2 shows student numbers by gender.

Table 1 Student numbers by school and school location

1
 School location Total
Urban Suburban Rural
1 0 0 54 54
2 0 67 0 67
3 0 0 58 58
4 0 0 53 53
School
5 0 44 0 44
Number
6 71 0 0 71
7 105 0 0 105
8 0 0 77 77
9 0 0 47 47
Total 176 111 289 576
(30.6%) (19.3%) (50.2%) (100%)

Table 2 Student numbers by gender

Frequency Percent Valid Percent Cumulative

Percent
Male 279 48.4 48.4 48.4
Valid Female 297 51.6 51.6 100.0
Total 576 100.0 100.0

3. Mathematical proof scale:

Five of the six scales of mathematical thinking (Generalization, Induction, Deduction, Use of

symbols, Logical relations) had been constructed. It was required to construct the sixth scale

score (Mathematical proof). This was done using the COMPUTE VARIABLE option within

TRANSFORM. Then “Totalproof” was inserted as the TARGET VARIABLE. From the

FUNCTIONS on the right hand side of the Compute Variable box, SUM was selected using the

up arrow. The five Mathematical proof scale items were inserted within the parentheses after

SUM, separated by commas. Then, OK was clicked to complete the compute. The new

2
Totalproof variable which was labeled Total Mathematical proof was now added to the end of

the data file.

It is important to determine the number of missing data before proceeding in the development of

a new variable. However, in this circumstance, the problem of missing data was not of concern

because, as mentioned above, students who omitted a question were considered not able to

answer the question and were given a score of zero.

Table 3 shows the summary statistics for the Mathematical proof scale. Table 4 in Appendix 1

shows the frequency distribution for the Mathematical proof scale for all students.

Table 3 Summary statistics for the Mathematical proof scale

Valid 576
N
Missing 0
Mean 4.84
Std. Deviation 3.66
Skewness .53
Std. Error of Skewness .10
Kurtosis -.39
Std. Error of Kurtosis .20
Minimum .00
Maximum 15.00

As can be seen in Table 4 in Appendix 1, only 4 students (0.7%) could answer all the

Mathematical proof test items. At the other extreme, 83 students (14.4%) could not answer any

test items. The mean score for the Mathematical proof scale is 4.84, and the Std. Deviation is

3
3.66 (See Table 3). The histogram of the distribution of the Mathematical proof scale scores

below shows the scores are non-normally distributed (See Figure 1). This variable has a positive

skewness of .53 (Std. Error of Skewness = .10) and kurtosis of -.39 (Std. Error of Kurtosis = .20)

Figure 1 Distribution of scores for the Mathematical proof scale

4. Means, Standard Deviations, and Reliabilities for mathematical thinking scales

Means, standard deviations and reliabilities for the six sub-scales of mathematical thinking and

for the scale of total mathematical thinking score scale are presented in Table 5.

4
 Table 5 Means, standard deviations, and internal consistency reliabilities for the six sub-scales

and for the total mathematical thinking scale

Scale Mean Standard Deviation Cronbach’s Alpha

Generalization 8.02 3.50 0.62
Induction 8.50 3.75 0.66
Deduction 7.32 4.30 0.61
Use of symbols 7.45 3.76 0.68
Logical relations 7.33 2.77 0.55
Mathematical proof 4.84 3.66 0.62
Total Mathematical Thinking 43.60 14.42 0.83

The internal consistency reliability for each scale was calculated separately using the items in

that scale as the scale variables. For example, to calculate the Cronbach’s Alpha for

Generalization, go to ANALYZE, then SCALE, then RELIABLITY ANALYSIS. Then, 5 items

Gen1, Gen2, Gen3, Gen4, Gen5 were selected for the ITEMS box. Then click the STATISTICS,

select the SCALE IF ITEM DELETED in Descriptives for box, then click CONTINUE and OK.

It should be noted that to calculate the alpha reliability for the Total Mathematical Thinking

scale, all 30 items from the six sub-scales were selected for the ITEMS box.

As can be seen in Table 5, the scale reliability for Total Mathematical Thinking was satisfactory

with a Cronbach alpha coefficient of 0.83. Considering the reliability statistics for the six sub-

scales (Generalization, Induction, Deduction, Use of symbols, Logical relations, and

Mathematical proof), the reliability for Logical relations was low, 0.55. The reliability for the

other sub-scales ranged from 0.61 to 0.68. These reliabilities are not particularly strong and raise

concerns about the internal consistencies of the scales.

5
In terms of the standard deviation for each of the six sub-scales, the Deduction scale had the

highest standard deviation of 4.30 while the Logical relations scale had the lowest standard

deviation of 2.77. This suggests that for the Deduction scale, students’ answers were widely

distributed while for Logical relations, students’ answers clustered more closely around the

mean score.

As shown in Table 5, the sub-scale of Mathematical proof had the lowest mean score (4.84)

indicating that this is students’ weakest mathematical area. Students did best at Induction (M=

8.50) and Generalization (M= 8.02), and not as well at Deduction (M= 7.32) or Use of symbols

(M= 7.45), or Logical relations (M= 7.33). However, given that the mean score for each sub-

scale would be 7.50, students were performing under the mean for four of the six scales.

5. Gender differences

To determine if there were any differences between male and female students in any of the

mathematical thinking scales or in mathematics achievement, t-tests for independent samples for

each of the six sub-scales, for the total mathematical thinking scale, and for the mathematics

achievement scale were conducted.

Table 6 in Appendix 2 presents the results of T-tests for independent samples for the

mathematical thinking scales and the mathematics achievement test. For the variables Induction,

Deduction, Use of symbols, Mathematical proof, Total mathematical thinking, and Mathematics

Achievement, Levene’s tests of equal variance were not significant (p>0.05), so the t-value, df

6
and two-tail significance for the equal variance estimates were used to determine whether gender

differences exist.

As can be seen in Table 6 and Table 7, although the male students had a higher mean score for

Induction than the female students, there was no significant difference between male and females

students (t= 1.10, df= 558, p= .274). Likewise, there was no statistically significant gender

difference in Use of symbols (t=0.24, df= 558, p= .814). Although the female students’ score was

higher than that of the male students for Deduction, the difference was not statistically

significant (t= .04, df= 558, p= .972). However, for the Total mathematical thinking, female

students had a statistically significantly higher mean score than male students (t=1.98, df= 558,

p= .048).

Statistically significant differences were found in Mathematical proof between male and female

students. Females had a statistically significantly higher mean score than males for Mathematical

proof (t= 3.53, df= 574, p< 0.001). Similarly, female students had a higher mean score than male

students for Mathematical achievement, and the difference was statistically significant (t= 4.05,

df= 541, p< 0.001).

In relation to Generalization and Logical relations, given that Levene’s tests have probabilities

less than .05, as shown in Table 6, the unequal variance estimates are interpreted. For the scale of

Generalization, although females had a higher mean score than males, the difference was not

statistically significant (t= 1.11, df= 555.78, p= .267). In contrast, there were statistically

significant gender differences found in Logical relations (t= 5.65, df= 530.95, p< .001).

7
Table 7 Gender differences in the Mathematical thinking scales and the Mathematics achievement test

Student gender N Mean Std. Deviation Std. Error Mean

Male 274 7.85 3.30 .20
Generalization
Female 286 8.18 3.67 .22
Male 274 8.68 3.70 .22
Induction
Female 286 8.33 3.79 .22
Male 274 7.32 4.44 .27
Deduction
Female 286 7.33 4.16 .25
Male 274 7.49 3.89 .23
Use of symbols
Female 286 7.41 3.64 .22
Male 274 6.67 2.94 .18
Logical relations
Female 286 7.96 2.44 .14
Male 279 4.29 3.62 .22
Mathematical Proof
Female 297 5.36 3.63 .21
Total Mathematical Male 274 42.37 14.68 .89
Thinking Female 286 44.78 14.10 .83
Mathematics Male 268 25.50 8.90 .54
Achievement Female 275 28.51 8.44 .51

6. School differences

To determine if there were differences among schools in any of the six aspects of mathematical

thinking, in mathematical thinking overall, or in mathematical achievement, one-way ANOVA

tests were performed.

Table 8 in Appendix 3 shows the results of One-way ANOVA for school differences in the

mathematical thinking scales and the Mathematics achievement

Generalization:

8
As shown in Table 8, there was a statistically significant difference in Generalization among

schools (F (8,551) = 6.99, p< .001). A Scheffe post-hoc test (See Table 9) revealed that school 1

with a mean of 10.31 differed statistically significantly from schools 7, 3, and 9 with means of

6.50, 7.20, and 7.36 respectively. These differences are illustrated in Figure 2.

Table 9 Scheffe results: Mean scores for the Generalization scale by school

School Number N Subset for alpha = 0.05

1 2
7 104 6.50
3 56 7.20
9 45 7.36
8 74 8.18 8.18
6 69 8.25 8.25
2 66 8.47 8.47
5 43 8.62 8.62
4 51 8.68 8.68
1 52 10.31
Sig. .140 .170
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 58.084.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are not guaranteed.

Figure 2 Mean Scores for the Generalization scale by school

9
Induction

As can be seen in Table 8, there was also a significant difference in Induction among schools (F

(8, 551) = 10.13, p< .001). A Scheffe post-hoc test revealed that school 1 with a mean of 10.63

differed statistically significantly from schools 2, 7, 6 with means of 6.77, 6.94, and 7.96

respectively. Both schools 2 and 7 differed significantly from schools 4, 3, 5 with means of 9.56,

9.69, and 10.35 respectively. Table 10 and Figure 3 below show these differences graphically.

Table 10 Scheffe results: Mean scores for the Induction scale by school

School Number N Subset for alpha = 0.05

1 2 3
2 66 6.77
7 104 6.94
6 69 7.96 7.96
8 74 8.27 8.27 8.27
9 45 8.99 8.99 8.99
4 51 9.56 9.56
3 56 9.69 9.69

10
 5 43 10.35 10.35
1 52 10.63
Sig. .175 .102 .112
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 58.084.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are not guaranteed.

Figure 3 Mean scores for the Induction scale by school

Use of symbols

As shown in Table 8, Use of symbols differed significantly among nine schools (F (8, 551) =

8.59, p < .001). As can be seen in Table 11, a Scheffe post-hoc test of comparisons of the nine

groups indicates that while students at school 8 had a significantly lower mean score than

students at schools 1, 2, 4, 5, 7, 9, students at school 5 had a significantly higher mean than

students at schools 2, 7, and students at school 1 had a significantly higher mean than students at

school 7. These differences are illustrated in Figure 4 below.

11
Table 11 Scheffe results: Mean scores for the Use of symbols scale by school

School Number N Subset for alpha = 0.05

1 2 3
8 74 5.21
7 104 6.72 6.72
2 66 7.04 7.04
6 69 7.40 7.40 7.40
3 56 7.77 7.77 7.77
9 45 7.90 7.90
4 51 8.29 8.29
1 52 9.24 9.24
5 43 9.71
Sig. .064 .072 .147
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 58.084.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are not guaranteed.

Figure 4 Mean scores for the Use of symbols scale by school

12
Logical relations

Table 8 shows that there was also a significant difference in Logical relations among schools (F

(8, 551) = 7.54, p< .001). A Scheffe post-hoc test revealed that school 5 with a mean of 9.14

differed statistically significantly from schools 2, 7, 3 with means of 6.00, 6.50, and 7.07

respectively. Both schools 2 and 7 differed significantly from schools 1, 9 with means of 8.19,

8.23 respectively. Table 12 and Figure 5 show these differences.

Table 12 Scheffe results: Mean scores for the Logical relations scale by school

School Number N Subset for alpha = 0.05

1 2 3
2 66 6.00
7 104 6.50 6.50
3 56 7.07 7.07
6 69 7.25 7.25 7.25

13
 8 74 7.36 7.36 7.36
4 51 7.89 7.89 7.89
1 52 8.19 8.19
9 45 8.23 8.23
5 43 9.14
Sig. .066 .139 .065
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 58.084.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are not guaranteed.

Figure 5 Mean Scores for the Logical relations scale by school

Mathematical proof

There was a significant difference in Mathematical proof among schools (F (8, 567) = 3.17, p =

.002) (See Table 8). A Scheffe post-hoc test revealed that school 4 with a mean of 6.65 differed

significantly from schools 6 with a mean of 3.98. Table 13 and Figure 6 show these differences.

14
Table 13 Scheffe results: Mean scores for the Mathematical proof scale by school

School Number N Subset for alpha = 0.05

1 2
6 71 3.98
3 58 4.34 4.34
7 105 4.38 4.38
8 77 4.54 4.54
2 67 4.56 4.56
9 47 5.26 5.26
1 54 5.36 5.36
5 44 5.69 5.69
4 53 6.65
Sig. .564 .139
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 59.939.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are not guaranteed.

Figure 6 Mean scores for the Mathematical proof scale by school

15
Mathematical thinking overall

As shown in Table 8, Mathematical thinking overall differed significantly among the schools (F

(8, 551) = 8.55, p < .001). As can be seen in Table 14, a Scheffe post-hoc test indicates that

while students at school 7 had a significantly lower mean score of Mathematical thinking overall

than students at schools 1, 4, 5, students at schools 1, 5 had a significantly higher mean than

students at schools 2, 8. Figure 7 shows these differences graphically.

Table 14 Scheffe results: Mean scores for the Total Mathematical thinking scale by school

School Number N Subset for alpha = 0.05

1 2 3
7 104 38.22
8 74 39.54 39.54
2 66 40.00 40.00
6 69 43.22 43.22 43.22
3 56 43.47 43.47 43.47
9 45 46.07 46.07 46.07
4 51 48.69 48.69
1 52 51.11
5 43 52.24
Sig. .302 .117 .129
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 58.084.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are not guaranteed.

Figure 7 Mean scores for the Total Mathematical thinking scale by school

16
Mathematics achievement

As shown in Table 8, there was a significant difference in Mathematical achievement among

schools (F (8, 534) = 6.89, p < .001). A Scheffe post-hoc test revealed that school 7 with a mean

of 22.67 differed significantly from schools 5, 6, 1 with means of 29.85, 29.93, and 30.77

respectively. Table 15 and Figure 8 show these differences.

Table 15 Scheffe results: Mean scores for Mathematics achievement by school

School Number N Subset for alpha = 0.05

1 2
7 96 22.67
3 52 25.09 25.09
8 77 25.63 25.63
2 62 27.25 27.25
9 46 27.74 27.74
4 53 28.57 28.57
5 42 29.85
6 63 29.93
1 52 30.77

17
 Sig. .087 .118
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 56.860.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are not guaranteed.

Figure 8 Mean scores for Mathematics achievement by school

Deduction

A Scheffe post-hoc test revealed that there were no significant differences in Deduction among

schools as can be seen in Table 16.

Table 16 Scheffe results: Mean scores for the Deduction scale by school

School Number N Subset for alpha = 0.05

1
8 74 5.80
2 66 7.10
18
 7 104 7.15
1 52 7.17
3 56 7.26
4 51 7.35
9 45 8.10
6 69 8.27
5 43 8.60
Sig. .127
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 58.084.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are not guaranteed.

7. Separate and Joint Effects of School Location and Gender

To look at the separate and joint effects of school location and student gender on Generalization,

a two-way ANOVA was conducted. Gender and school location were considered as independent

variables with Generalization as the dependent variable.

As shown in Table 17 and Table 18, a main effect of gender was found (F (1, 554) = 4.01, p =

.046), indicating that the Generalization mean was significantly higher for female students (M =

8.18, SD = 3.67) than male students (M = 7.85, SD = 3.30). There was also a main effect of

school location (F (2, 554) =6.96, p= .001). This effect showed that students in the suburban

area (M = 8.53, SD = 3.89) and rural area (M = 8.34, SD = 3.43) had higher means for

Generalization than those in the urban area (M = 7.19, SD = 3.21).

Table 17 Two-way ANOVA: Generalization scale by Gender and School location

19
Source Type III Sum of Df Mean Square F Sig.
Squares
Corrected Model 275.40a 5 55.08 4.66 .000
Intercept 30891.34 1 30891.34 2610.87 .000
gender 47.41 1 47.41 4.01 .046
location 164.65 2 82.33 6.96 .001
gender * location 71.24 2 35.62 3.01 .050
Error 6554.84 554 11.83
Total 42862.50 560
Corrected Total 6830.24 559
a. R Squared = .040 (Adjusted R Squared = .032)

Table 18 Means for the Generalization scale by Gender and School location

Student gender School location Mean Std. Deviation N

Urban 7.37 3.39 69
Suburban 7.65 3.46 55
Male
Rural 8.15 3.19 150
Total 7.85 3.30 274
Urban 7.08 3.09 104
Suburban 9.43 4.13 54
Female
Rural 8.55 3.69 128
Total 8.18 3.67 286
Urban 7.19 3.21 173
Suburban 8.53 3.89 109
Total
Rural 8.34 3.43 278
Total 8.02 3.50 560

As can be seen in Table 17, there was a significant interaction between the effects of gender and

school location on Generalization, F (2, 554) = 3.01, P = .05. It should be noted that the

significance of the interaction here is 0.05. As such, it may not reach statistical significance if the

usual level of significance is less than 0.05. These interaction effects are shown graphically in

Figure 9.

20
Figure 9 Mean scores for the Generalization scale by Gender and School location

The figure shows that the female students in suburban and rural areas had the higher scores than

male students in these areas, but this was not the case for students in the urban area.

Similarly, a two-way ANOVA was conducted to examine the effect of gender and school

location on Induction. Gender and school location were considered as independent variables with

Induction as the dependent variable.

Table 19 Two-way ANOVA: Induction scale by Gender and School location

21
Source Type III Sum Df Mean Square F Sig.
of Squares
Corrected Model 599.41a 5 119.88 9.16 .000
Intercept 33092.82 1 33092.82 2528.89 .000
gender 36.24 1 36.24 2.77 .097
location 425.14 2 212.57 16.24 .000
gender * location 154.27 2 77.13 5.89 .003
Error 7249.59 554 13.09
Total 48343.00 560
Corrected Total 7848.99 559
a. R Squared = .076 (Adjusted R Squared = .068)

Table 20 Means for the Induction scale by Gender and School location

Student gender School location Mean Std. Deviation N

Urban 7.69 3.53 69
Suburban 9.12 3.56 55
Male
Rural 8.98 3.77 150
Total 8.68 3.70 274
Urban 7.12 3.57 104
Suburban 7.22 3.49 54
Female
Rural 9.79 3.59 128
Total 8.33 3.79 286
Urban 7.35 3.56 173
Suburban 8.18 3.64 109
Total
Rural 9.35 3.70 278
Total 8.50 3.75 560

As shown in Table 19, there was a main effect of school location (F (2, 554) =16.24, p<.001).

This effect showed that students in the rural area (M = 9.35, SD = 3.70) and suburban area (M =

8.18, SD = 3.64) had higher mean for Induction than those in the urban area (M = 7.35, SD =

22
3.56) (See Table 20). However, as can be seen in Table 19, the main effect of gender was non-

significant (F (1, 554) = 2.77, p = .097).

As can be seen in Table 19, there was a significant interaction between the effects of gender and

school location on Induction, F (2, 554) = 5.89, P = .003. These interaction effects are shown

graphically in Figure 10. The figure shows that the male students in the urban and suburban areas

performed better on the Induction test than female students in these areas. However, in the rural

area, males had lower scores than females.

Figure 10 Mean scores for the Induction scale by Gender and School location

23
8. Correlational relationships: Mathematics achievement with Generalization;

Mathematics achievement with Logical relations; and Generalization with Logical

relations

As part of the assignment, I am required to describe, investigate, and report one interesting

aspect of the data. I have chosen to investigate the correlational relationships between

Mathematical achievement and Generalization, Mathematical achievement and Logical relations,

and Generalization and Logical relations using the Pearson’s product-moment correlation

coefficients.

24
 Table 21 Correlations between the Mathematics achievement, Generalization scale and
Logical relations scale
Mathematics Generalization Logical
Achievement relations
Pearson
1 .63** .55**
Mathematics Correlation
Achievement Sig. (2-tailed) .000 .000
N 543 527 527
Pearson
.63** 1 .40**
Correlation
Generalization
Sig. (2-tailed) .000 .000
N 527 560 560
**. Correlation is significant at the 0.01 level (2-tailed).

As can be seen in Table 21, the correlation of Generalization with Mathematical achievement

was quite high (r= 0.63), and was also statistically significant (p <001). Likewise, the correlation

of Logical relations with Mathematical achievement was moderately high (r= 0.55), and was

statistically significant (p < .001). The correlation of Logical relations with Generalization was

moderately low (r= 0.40), and was also statistically significant (p < .001).

It is interesting to see that both Generalization and Logical relations were strongly correlated to

Mathematics achievement. Because none of the items in the scales were provided as part of the

assignment, it is difficult to interpret these correlations. The high correlation between the

Mathematics achievement test and the Generalization scale (r= .63) suggests one of two things:

the Mathematics achievement test contains items that require students to make generalizations,

or students who do well on Mathematics achievement tests also have the ability to draw

generalizations. Similarly, for the correlation between the Mathematics achievement test and the

Logical relations scale (r= .55).

25
The correlation between the Generalization and Logical relations (r= .40) may suggest that these

two measures of Generalization and Logical relations are not entirely independent measures. But

it is difficult to say more than this without access to the items that comprise the scales.

26
 APPENDIX 1

Table 4 Frequency distribution for the Mathematical proof scale

Frequency Percent Valid Percent Cumulative Percent

.00 83 14.4 14.4 14.4
.50 13 2.3 2.3 16.7
1.00 21 3.6 3.6 20.3
1.50 19 3.3 3.3 23.6
2.00 30 5.2 5.2 28.8
2.50 23 4.0 4.0 32.8
3.00 40 6.9 6.9 39.8
3.50 17 3.0 3.0 42.7
4.00 21 3.6 3.6 46.4
4.50 39 6.8 6.8 53.1
5.00 25 4.3 4.3 57.5
5.50 22 3.8 3.8 61.3
6.00 38 6.6 6.6 67.9
6.50 14 2.4 2.4 70.3
7.00 19 3.3 3.3 73.6
7.50 27 4.7 4.7 78.3
8.00 16 2.8 2.8 81.1
Valid
8.50 20 3.5 3.5 84.5
9.00 24 4.2 4.2 88.7
9.50 6 1.0 1.0 89.8
10.00 6 1.0 1.0 90.8
10.50 3 .5 .5 91.3
10.63 1 .2 .2 91.5
11.00 12 2.1 2.1 93.6
11.25 1 .2 .2 93.8
11.50 12 2.1 2.1 95.8
12.00 8 1.4 1.4 97.2
12.50 1 .2 .2 97.4
13.00 2 .3 .3 97.7
13.50 4 .7 .7 98.4
14.00 1 .2 .2 98.6
14.50 4 .7 .7 99.3
15.00 4 .7 .7 100.0
Total 576 100.0 100.0

27
 APPENDIX 2
Table 6 T-tests for independent samples for the mathematical thinking scales and the mathematics achievement test

Levene's Test for Equality t-test for Equality of Means

of Variances
F Sig. t df Sig. (2- Mean Std. Error 95% Confidence Interval of the Difference
tailed) Difference Difference Lower Upper
Equal variances
4.105 .043 -1.11 558 .268 -.33 .30 -.91 .25
assumed
Generalization
Equal variances not
-1.11 555.78 .267 -.33 .29 -.91 .25
assumed
Equal variances
.058 .810 1.10 558 .274 .35 .32 -.28 .97
assumed
Induction
Equal variances not
1.10 557.75 .274 .35 .32 -.28 .97
assumed
Equal variances
3.277 .071 -.04 558 .972 -.01 .36 -.73 .70
assumed
Deduction
Equal variances not
-.04 551.54 .972 -.01 .36 -.73 .70
assumed
Equal variances
1.118 .291 .24 558 .814 .08 .32 -.55 .70
assumed
Use of symbols
Equal variances not
.24 551.50 .814 .08 .32 -.55 .70
assumed
Equal variances
6.806 .009 -5.67 558 .000 -1.29 .23 -1.74 -.85
assumed
Logical relations
Equal variances not
-5.65 530.95 .000 -1.29 .23 -1.74 -.84
assumed
Equal variances
.002 .962 -3.53 574 .000 -1.07 .30 -1.66 -.47
assumed
Mathematical Proof
Equal variances not
-3.53 571.78 .000 -1.07 .30 -1.66 -.47
assumed
Equal variances
.169 .681 -1.98 558 .048 -2.41 1.22 -4.80 -.02
Total Mathematical assumed
Thinking Equal variances not
-1.98 554.17 .048 -2.41 1.22 -4.80 -.02
assumed
Equal variances
.933 .335 -4.05 541 .000 -3.01 .74 -4.47 -1.5
Mathematics assumed
Achievement Equal variances not
-4.05 537.68 .000 -3.01 .74 -4.47 -1.5
assumed

28
 APPENDIX 3

Table 8 One- way ANOVA for school differences in the Mathematical thinking scales and the
Mathematics achievement

Sum of df Mean F Sig.

Squares Square
Between Groups 629.66 8 78.71 6.99 .000
Generalization Within Groups 6200.59 551 11.25
Total 6830.24 559
Between Groups 1006.09 8 125.76 10.13 .000
Induction Within Groups 6842.90 551 12.42
Total 7848.99 559
Between Groups 339.61 8 42.45 2.35 .017
Deduction Within Groups 9972.96 551 18.10
Total 10312.57 559
Between Groups 876.32 8 109.54 8.59 .000
Use of symbols Within Groups 7029.38 551 12.76
Total 7905.70 559
Between Groups 424.15 8 53.02 7.54 .000
Logical relations Within Groups 3873.73 551 7.03
Total 4297.88 559
Between Groups 330.34 8 41.29 3.17 .002
Mathematical Proof Within Groups 7376.63 567 13.01
Total 7706.97 575
Between Groups 12835.47 8 1604.43 8.55 .000
Total Mathematical
Within Groups 103463.17 551 187.77
Thinking
Total 116298.64 559
Between Groups 3915.44 8 489.43 6.89 .000
Mathematics
Within Groups 37948.06 534 71.06
Achievement
Total 41863.51 542

29