disagree

Rapport1

11.3

8.7

29.6

30.4

20

Rapport2

15.7

20.9

23.5

25.2

14.8

Rapport3

8.7

10.4

25.2

38.3

17.4

Rapport4

9.6

13

21.7

35.7

20

Rapport5

39.1

33

11.3

9.6

7

Rapport6

24.3

8.7

13.9

32.2

20.9

Rapport7

21.7

20.9

24.3

20.9

12.2

Rapport8

32.2

28.7

13

13.9

12.2

Rapport9

7

3.5

10.4

38.3

40.9

Rapport10

1.8.3

14.8

31.3

12.2

23.5

Now in the table 6 the results of descriptive statistics are shown for the whole samples and the cheating questions:

Table 6 Descriptive statistics

cheating

Mean

Median

SD

Cheating1

2.48

2

1.13

Cheating2

2.51

3

1.17

Cheating3

2.51

2

1.31

Cheating4

2.43

2

1.29

Cheating5

2.27

2

1.19

Cheating6

2.72

3

1.35

Cheating7

2.54

2

1.31

Cheating8

1.86

1

1.22

Cheating9

1.77

1

1.23

Cheating10

1.61

1

0.96

Cheating11

1.81

1

1.02

Cheating12

1.56

1

0.98

Percentages

cheating

never

rarely

sometimes

usually

always

Cheating1

24.3

27.8

26.1

19.1

2.6

Cheating2

25.2

23.5

30.4

16.5

4.3

Cheating3

28.7

24.3

24.3

12.2

10.4

Cheating4

32.2

21.7

24.3

13.9

7.8

Cheating5

33

31.3

14.8

17.4

3.5

Cheating6

25.2

19.1

27

15.7

13

Cheating7

28.7

24.3

20

18.3

8.7

Cheating8

56.5

20

10.4

7

6.1

Cheating9

64.3

13.9

8.7

7

6.1

Cheating10

64.3

18.3

10.4

6.1

0.9

Cheating11

50.4

27.8

14.8

4.3

2.6

Cheating12

68.7

14.8

11.3

2.6

2.6

Inferential statistics

In the table 6, the research hypotheses are tested.

Hypothesis 1: There is a significant relationship between student-teacher relationship and students’ cheating.

The statistical hypothesis test is as follows:

Null hypothesis: There is not a significant relationship between student-teacher relationship and students’ cheating.

The opposite hypothesis: There is a significant relationship between student-teacher relationship and students’ cheating.

Given that the questionnaires’ items are qualitative and of ordinal scale, nonparametric Spearman’s correlation coefficient was used. The results are as follows:

Table 7 Correlations

cheating

rapport

Spearman’s rho

cheating

Correlation Coefficient

1.000

-.093

Sig. (1-tailed)

.

.162

N

115

115

rapport

Correlation Coefficient

-.093

1.000

Sig. (1-tailed)

.162

.

N

115

115

Interpretation:

Since in table 7 the probability of error is higher than 5%, so the assumption is not significant. That is, with 95% confidence there is not a significant relationship between student-teacher relationship and students” cheating. Because the correlation coefficient of the two variables is negative, the relationship is inverse type. It means that the stronger the relationship between student and teacher, the less cheating in the exam. However, in this case, the correlation is not so strong that the hypothesis is confirmed.

Correlation graph between relationship and cheating

Scatter plot

The Correlation chart

The above chat shows the Scatter Plot of the relationships between teacher-student relationships and students’ cheating. If the points on the chart be on or near the first and the third quadrant bisecting line, there will be a strong correlation. This feature is not seen in the above chart and the points are scattered quite chaotically. Therefore, as the correlation coefficient was calculated, the correlation coefficient is not significant.

ANOVA test

In order to conduct ANOVA test, the distribution of the data should be normal. Therefore, firstly, using Kolmogorov – Smirnov test, the normality of data is investigated. This test has the following assumptions:

Null hypothesis: The distribution is normal.

Converse hypothesis: The distribution is not normal.

Oneway

Table 9 ANOVA

rapport

Sum of Squares

df

Mean Square

F

Sig.

Between Groups

1.748

4

.437

.508

.730

Within Groups

94.520

110

.859

Total

96.268

114

Interpretation:

NPar Tests

One-Sample Kolmogorov-Smirnov Test

rapport

Cheating

N

115

115

Normal Parametersa

Mean

3.0957

2.2696

Std. Deviation

.91894

.76905

Most Extreme Differences

Absolute

.073

.121

Positive

.050

.121

Negative

-.073

-.049

Kolmogorov-Smirnov Z

.785

1.293

Asymp. Sig. (2-tailed)

.568

.071

Interpretation:

Since table 9 showesthe probability of error for both of the variables are more than 5%, then the assumption is not significant. That is, with a 95% confidence level, for both teacher-student relationship and students’ cheating are normal and there is no problem using ANOVA.

Hypothesis 2: There is a significant relationship between teacher-student relationship and students’ GPA.

The statistical hypothesis testing is as follows:

Null hypothesis: the relationship between teacher and student is the same at the different levels of GPA.

The opposite hypothesis: There is a difference between teacher-student relationship in the different levels of GPA.

To perform this test, given that the data values are of numerical types and the number of GPA levels are more than two, the ANOVA will be used. However, since the assumption of normality of the data was not available, we use its parametric test, i.e. the Kruskal – Wallis test. This test uses an asymptotic distribution chi-square statistic. The results are as follows

Oneway

Table 10 ANOVA

rapport

Sum of Squares

df

Mean Square

F

Sig.

Between Groups

1.748

4

.437

.508

.730

Within Groups

94.520

110

.859

Total

96.268

114

Interpretation:

NPar Tests

Kruskal-Wallis Test

Ranks

moadel

N

Mean Rank

rapport

1

2

59.50

2

15

63.43

3

31

61.82

4

37

56.65

5

30

52.90

Total

115

Test Statisticsa,b

Rapport

Chi-square

1.575

df

4

Asymp. Sig.

.813

a. Kruskal Wallis Test

b. Grouping Variable: moadel

Since the probability of error is higher than 5%, then the assumption is not significant. That is, with 95% confidence level the teacher and student relationship at all GPA levels are the same. That means teacher-student relationship is not related to the students’ GPA.

Hypothesis 3: There is a significant relationship between students’ cheating and their GPA.

The statistical hypothesis test is as follows:

Null hypothesis: Students’ cheating/ not cheating is the same at different levels of GPA .

Opposite hypothesis: There is a difference between students’ cheating / not cheating at different levels of GPA.

To conduct this test, given that the data values are of numerical values and the number of GPA levels are more than two levels, the ANOVA will be used. However, since the assumption of normality of the data was not available, we use its’ paramedics’ equivalent (the Kruskal – Wallis test). This test uses an asymptotic distribution chi-square statistic. The results are as follows:

Oneway

Table 11 ANOVA

cheating

Sum of Squares

df

Mean Square

F

Sig.

Between Groups

3.029

4

.757

1.293

.277

Within Groups

64.395

110

.585

Total

67.423

114

NPar Tests

Kruskal-Wallis Test

Ranks

moadel

N

Mean Rank

cheating

1

1

2

44.25

2

15

52.00

3

31

56.61

4

37

52.38

5

30

70.28

Total

115

Test Statisticsa,b

cheating

Chi-square

6.021

df

4

Asymp. Sig.

.198

a. Kruskal Wallis Test

b. Grouping Variable: moadel

Since in table 11 the probability of error is higher than 5%, so the assumption is not significant. That is, with 95% confidence level of students’ cheating/ not cheating are the same. This means that students’ cheating / not cheating is not related with their GPA.

– Another question addressed is that whether students use old methods or advanced methods for cheating?

To investigate this question, researcher considers the frequency and percentage of responses.

Table 12. Table of response frequency

Frequency table

Cheating

Never

rarely

Sometimes

Usually

Always

Total

Cheating1

28

32

31

21

3

115

Cheating2

29

27

35

19

5

115

Cheating3

33

28

28

14

12

115

Cheating4

37

25

28

16

9

115

Cheating5

38

36

17

20

4

115

Cheating6

29

22

31

18

15

115

Cheating7

33

28

23

21

10

115

Cheating8

65

23

12

8

7

115

Cheating9

74

16

10

8

7

115

Cheating10

74

21

12

7

1

115

Cheating11

58

32

17

5

3

115

Cheating12

79

17

13

3

3

115

Of the 12 methods listed in table 12, with a little care, we can say the first 7 method refers to old methods and the last 5 methods refer to new cheating methods.

Considering the second and third column from the right of the table (always and most of the times), for the last 5 questions (items number 8-12 which refers to the new methods), the number was one digit and are less than the other scenarios. Therefore, we can conclude that the new methods are less used.

Table.13 Response percent

cheating

Never