Analysis Write Up

Sample Write up for

  Data Analysis

This section covers respondent profile, reliability testing of multi-item scales, and hypothesis testing via a number of statistical tools.

e.g., “First, descriptive analysis was performed to understand respondents’ profile including ethnic, gender  …... Second, reliability test was performed on six multi-item scales which were money orientation value, …... Finally, five hypotheses were tested using multiple statistical tools such as t-test, ANOVA, spearman correlation and hierarchical regression analysis. Analysis used to test each hypothesis is elaborated in detail as follow.”

Respondent’s Profile

Descriptive analysis of gender, Age, …………..were performed to understand respondent’s profile. First, of the 170 respondents (see Table 4.1), there were about equal number of male and female respondents participated in this study. Second, ……

Table 4.1: Summary of Demographic Profile of Respondents

Reliability Analysis

Cronbach Alpha’s test of reliability was conducted to determine the reliability of the multi-item scales. As shown in table 1, the scales’ reliabilities ranged from 0.60 to 0.87. Since they were at least 0.60, thus all the six scales were reliable for use in subsequent analysis as suggested by Nunnally (1978). Full number of items were used to perform reliability test for money orientation value, religiosity value, external locus of control value, relationship orientation value and ethical level. As for social desirability response bias, the alpha obtained from full 10 items included in questionnaire was too low (i.e. 0.2). Researchers have warned the use of reverse-worded items (e.g. Wong et. al., 2003) because respondents may tend to get confused. Thus, four reverse items (item 2. 3. 4 and 6) were dropped and true enough, Cronbach’s alpha of social desirability response bias improved significantly to 0.60.

Table 4.2: Reliability

Hypothesis Tesing

There were five hypotheses developed for testing in this research. H1 was tested using independent sample t-test, H2 used ANOVA, H3 used spearman rank correlation, H4 used one sample T-test while H5 used hierarchical regression analysis. Steps taken to test each hypothesis were elaborated in detail below.

H1: Females’ ethical level are significantly higher than that of males.

An independent sample t-test was performed to test H1 as gender was a two-category nominal variable. As can be seen from table 3a, mean ethical perception of male and female were 5.47 and 5.13 respectively. Since significance level of Levene’s test was 0.91 (greater than 0.05), therefore assumed equal variance. Significance level of t-test for equal variances assumed in table 3b was 0.00 (<0.05), therefore H1 was supported. That was, female students’ ethical level (mean = 5.47) are significantly higher than male (mean = 5.13).