You are on page 1of 6

Student Name ID: Student

1.0 INTRODUCTION
A short paragraph intro about what the data is and what sort of questions you are asking about
the data.
Another short paragraph about the database and what data is contained, some descriptive
statistics like variables are …, n=, mean =.
Finally the Intro normally finishes with a proposition statement like … “The paper analyses the
collected data and presents findings on the issue under investigation.”

1.1 Research questions


By performing hypothesis tests, the following questions under this study are investigated:

State your questions here from the Business Report guide PDF you have been given. An example is given
below from a previous semester:

1. Does the representation of female directors on the board of directors depend


on industry sectors?
2. Do companies with gender-diversity boards have a lower cost of debt
(CoD), compared to those without female representation on the board?
3. Is there any difference in the cost of debt (CoD) among the three industry
sectors, namely Energy, Health Care, and Consumer Staples?

2.0 STATISTICAL ANALYSES

Please note that the following example is from a previous semester’s student. In this space they
have written the explanation of the Question 1 problem, what data is in the dataset that relates
to this problem and which statistical technique was appropriate to use with this data. They they
use section 2.2 to demonstrate the statistical problem solving technique to be used to find a
conclusion about the population based on the hypothesis made from the sample data collected.

Please make special note that you are not to copy this text from a previous student as that would
be Plagiarism and be Misconduct. However, you can use this as an excellent example on which
to model your own answers.

Example of a great answer from previous semester’s student:

2.2 Gender-diversity boards and cost of debt (CoD) (Question 1)


The problem objective is whether the cost of debt (CoD) of firms with gender-diversity boards
of directors is lower compared to companies without female representation in the boards. As

Page 1 of 6
Student Name ID: Student

the sample data is quantitative (i.e., CoD is measured in numerical values), the problem
objective can be determined by comparing the difference between the means of the 2 sample
datasets. These are the average CoD of firms with gender diversity boards µ1 and average CoD
of firms with non-gender diverse boards µ2.
• The parameter of interest is µ1 - µ2;
• The populations of interest are two, firms with gender diversity boards and
organisations without female representation in corporate executive boards (i.e., non-
gender diversity boards).

Given the quantitative sample data, 2 populations of interests, and population variances being
unknown, the t-test is the most appropriate technique for this problem objective. Indeed, z-test
cannot be applied as it performed with the variance values obtained directly from the entire
population not a sample and one-way ANOVA is useful when more than 2 populations exist
(Black et al. 2019). However, before applying the t-test, the following assumptions need to be
considered (Black et al. 2019):
• observations of the sample are random and independent of each other;
• the sample sizes are large;
• each sample dataset is drawn from normally distributed populations.
Table 7. Descriptive statistics of each sample dataset (CoD was divided by 100 to obtain numerical
values instead of percentages)
Descriptive Statistics Gender-diversity CoD Non gender-diversity CoD
Mean 0.034134639 0.061027093
Standard Error 0.002903874 0.003740892
Median 0.0310709 0.0479806
Mode 0 0
Standard Deviation 0.033988972 0.047760513
Sample Variance 0.00115525 0.002281067
Kurtosis 9.165691197 0.453242761
Skewness 2.134274893 1.058988278
Range 0.2361863 0.1957
Minimum 0 0
Maximum 0.2361863 0.1957
Sum 4.6764455 9.9474162
Count 137 163

As discussed above, the observations of the samples under investigation were selected
randomly and are independent. Referring to “count” in Table 7, the sample size of firms with
gender-diversity boards and those without female representation are 137 (n1) and 163 (n2)
respectively. As these values are greater than 30, the sample sizes are large (Black et al. 2019).
Figure 1 and 2 provides a graphical representation of the distribution of both sample datasets.
From both histograms, it can be noticed that the data points are clustered towards the left rather
than the centre of the distribution and with fewer observation values towards the right of the
distribution. The distributions are not symmetrical but positively skewed, thus not
resembling Page 2 of 6
Student Name ID: Student

the normal distribution. Ultimately, the underlying assumptions of the t-test are all satisfied
except for the assumption of normal distribution. Nevertheless, the t-test will still be performed
regardless of the shape of both distributions.
Figure 1 and 2. Graphical representation of the distribution of the cost of debt (CoD) of
gender-diversity and non gender-diversity boards

Histogram of CoD of gender-diversity boards


55
45
50
40
45
40 35
35 30

Frequency
Frequency

30 25
25 20
20 15
15 10
10 5
5
0
0
1 3 5 7 9 11 13 15 17 19
1 3 5 7 9 11 13 15 17 19 21 23
Class midpoint (%)

While the two population variances are unknown, it can be assumed that they are equal as the
ratio of the larger sample variance to the smaller sample variance does not exceed 3 (i.e., from
Table 7, ratio is 0.00228/0.00116= 1.97) (Dean, Voss, and Draguljic 2017). Hence, the t-test
assuming equal variances will be utilised.

2.2.1 Hypothesis testing procedure:


Step 1: Determining null and alternative hypotheses
H0: µ1 - µ2 ≥ 0
Ha: µ1 - µ2 < 0 (since the alternative hypothesis is testing whether µ1 < µ2)

While the null hypothesis refers to the status quo remaining in place (= sign), the ≥ sign
was used instead as all possible outcomes need to be considered in the hypotheses
(Black et al. 2019).

Step 2: Identifying the hypothesis testing technique and test statistic


The t-test assuming equal variances is the most appropriate technique. The test statistic
(𝑥𝑥𝑥1− 𝑥𝑥𝑥2 )−(𝜇𝜇1−𝜇𝜇2)
is 𝑡𝑡 = 1 1
𝑠𝑠𝑝𝑝 √𝑛𝑛 𝑛𝑛2
1

However, the Excel function ‘t-Test: Two-Sample Assuming Equal Variances’ will be
used to compute the test statistic.

Step 3: Specifying the level of significance α and determining the critical value and region
With sample sizes n1 =137 and n2 = 163, the degree of freedom is
d.f. = n1 + n2 – 2 = 137 +163 – 2 = 298

The hypothesis test is (left) one-tail test (as testing whether µ1 < µ2) and performed at
three significant levels (1%, 5%, and 10%). Using Excel, the critical values are:
Page 3 of 6
Student Name ID: Student

If α = 0.10 and d.f.= 298, the critical value is - t 0.10, 298 = - 1.28440

If α = 0.05 and d.f.= 298, the critical value is - t 0.05, 298 = -1.64998

If α = 0.01 and d.f.= 298, the critical value is - t 0.01, 298 = -2.33893

Step 4: State the decision rule


Since the test is (left) one-tailed, if the test statistic t calculated is less than the critical
value of -1.28440 for 10% significance level, t falls in the rejection region and thus null
hypothesis (H0) is rejected in favour of the alternative hypothesis (Ha). This will also be
the case if the test statistic t is less than the critical value of -1.64998 for 5% significance
level or -2.33893 for 1% significance level.

Step 5: Calculating the value of the test statistic (t)


Table 8. Results of t-test performed in Excel

t-test assuming equal variances


Pooled Variance 0.001767271
Hypothesized Mean Difference 0
Degree of freedom d.f. 298
t Stat -5.519151346

Using the Excel function stated in Step 2, the observed t value is -5.51915 (Table 8).

Step 6: Conclusion
Table 9. Summary of test statistic value, critical values at each level of significance,
and the decision to reject H0 or not

Critical value at Critical value at Critical value at


10% significance 5% significance 1% significance
Test Statistic t level level level
-1.284398892 -1.649982976 -2.338926209
-5.519151346
Reject H0 Reject H0 Reject H0

Referring to Table 9, as the observed t or test statistic value of -5.51915 is less than the critical
values at each significant level (i.e., 1%, 5%, and 10%), the value of test statistic falls in the
rejection region (for all significance levels). Thus, the decision is to reject the null hypothesis
in favour of the alternative hypothesis at each significance level. This suggests that there is
overwhelming evidence to conclude that companies with gender-diversity boards of directors
have a significant lower cost of debt (CoD) compared to firms without female representation
in corporate executive boards.

Page 4 of 6
Student Name ID: Student

QUESTION 2 & QUESTION 3 ADDRESSED IN A SIMILAR MANNER

3.0 CONCLUSIONS

The following is part of a previous student’s conclusion, and demonstrated excellent writing
skills and ability to conclude about findings from the sample and inferences that can be made
about the population.

These findings are consistent with the psychologic literature regarding women being more
prone to loss aversion and thus being less inclined than men to take risks. Additionally, this study
identified that there is insufficient evidence to infer that the female representation in corporate
executive boards is dependent on the industry sector. The results are also supported by the fact
there is insufficient evidence to conclude the existence of a significant difference in the cost of
debt (CoD) among the three industry sectors. Thus, the overall findings imply that, while the
firm risk is not significantly diverse at the sector level, the firm risk varies significantly in
individual ASX firm with female directors. Indeed, observations revealed that the frequency of
female and male directors is relatively similar among sectors, ultimately balancing out any
individual differences in the firm risk.

This study acknowledges limitations. The assumption of normally distributed population was
violated when conducting t-test for two populations assuming equal variances and one-way
ANOVA. For future studies, normalising transformation or nonparametric procedures should
be used instead (Black et al. 2019). Furthermore, the focus of the industry sectors was too
narrow (i.e., only 3 sectors were considered), thus the findings may not be fully representative
of all industry sectors of the ASX market. To address this limitation, a similar study should be
conducted with broader datasets that consider all 11 industry sectors.

Page 5 of 6
Student Name ID: Student

4.0 REFERENCES
Black, Ken, John Asafu-Adjaye, Paul Burke, Nelson Perera, and Carl Sherwood. 2019.
Business Analytics and Statistics. John Wiley & Sons Australia.
https://ebookcentral.proquest.com/lib/curtin/reader.action?docID=5561257

Dean, Angela, Daniel Voss, and Danel Draguljic. 2017. Design and Analysis of Experiments.
2nd ed. Springer International Publishing. https://link.springer.com/book/10.1007/978-3-
319-52250-0

Governance Institute of Australia. 2023. “2023 Board Diversity Index.”


https://www.governanceinstitute.com.au/news-media/media-releases/2023/apr/2023-
board-diversity-index/

Special Note: This is a partial example of how one student answered one of
their three questions with a different set of company data. This is not an
example based on this semester’s data, but it is an example how to
answer a question that is relevant to the data you have been given.
Copying of the text in this answer is Plagiarism and will be detected by
Turnitin resulting in an Academic Misconduct case against you, so
please be careful in your use of this Examplar.

Page 6 of 6

You might also like