In: Statistics and Probability
Question 2: The Office of the Superintendent of Bankruptcy of Canada (OSBC) is developing a new index to measure the vulnerability of firms in the new technology industry. The index is a ratio of current assets to current liabilities adjusted for various factors specific to this industry. The OSBC wants to compare the index among healthy and failed firms for validation purposes. They expect that failed firms should have a lower index than the healthy ones. Based on a Canadian business registry, they draw a random sample of 68 firms still in operations and another random sample of 33 firms which failed in the last 3 years. Index data for the sampled firms based on their latest financial statements are included in the Minitab file Bankruptcy_Index.mtw and in the Excel file Assign2.xlsx under the Bankruptcy_Index tab. Data is below:
a) Use Minitab or other appropriate software to produce boxplots of the index values for the two groups of firms and comment on their distribution. 2
b) Use an appropriate statistical test to determine, at the 1% significance level, whether the data provide evidence of a higher average index for the healthy firms. Make sure you provide your manual calculations using the critical value approach.
c) Calculate manually a 99% one-sided confidence interval for the difference in the average index of healthy and failed firms and compare your results with b) above.
d) Use Minitab or other appropriate software to perform the calculations in b) and c) above and comment on any differences
Here is the data for Bankruptcy_Index.mtw
Healthy | Failed |
1.5 | 0.82 |
2.08 | 0.05 |
2.23 | 1.68 |
0.89 | 0.91 |
1.91 | 1.16 |
1.2 | 0.42 |
1.95 | 0.88 |
2.73 | 1.11 |
1.62 | 2.03 |
1.71 | 0.92 |
1.03 | 0.73 |
1.96 | 0.89 |
0.1 | 0.83 |
1.43 | 0.99 |
2.5 | 0.52 |
0.23 | 1.32 |
1.67 | 0.48 |
2.17 | 1.1 |
2.61 | 0.19 |
1.56 | 0.51 |
1.76 | 0.26 |
1.02 | 0.88 |
1.8 | 1.31 |
1.81 | 0.9 |
1.76 | 0.62 |
0.68 | 1.45 |
2.02 | 1.17 |
1.2 | 0.93 |
1.87 | 0.75 |
2.61 | 0.13 |
1.11 | 1.12 |
2.73 | 1.15 |
2.22 | 0.71 |
2.5 | |
0.67 | |
1.14 | |
3.15 | |
1.44 | |
2.16 | |
1.21 | |
3.05 | |
0.95 | |
0.9 | |
2.8 | |
1.55 | |
2.44 | |
1.84 | |
1.24 | |
1.39 | |
1.8 | |
2.05 | |
1.52 | |
0.96 | |
2.12 | |
1.85 | |
1.69 | |
2.3 | |
2.21 | |
2.03 | |
1.64 | |
1.87 | |
1.06 | |
1.93 | |
2.25 | |
1.42 | |
0.96 | |
1.64 | |
2.21 |
Question.a) Use Minitab or other appropriate software to produce boxplots of the index values for the two groups of firms and comment on their distribution.
Box plot from minitab is given below .
you can get the same from paste the data on minitab and use menu graph-boxplot and select the data and put in ggraph variables and click OK.
Interpretation of Distribution :
Key things to look .
one can check the parameters for both
Median 1.78 Healthy 0.89 failed -- Q1 : 1.21 healthy 0.57 failed -Q3 2.17 healthy 1.14
Q2. Use an appropriate statistical test to determine, at the 1% significance level, whether the data provide evidence of a higher average index for the healthy firms. Make sure you provide your manual calculations using the critical value approach.
Answer 2 sample t test is the appropriate test as data is continuous , data sets are independent and we want to comapre mean , population standard devation is not known,
Null Hypothesis H0 : Mean Healthy Index(mu1) - Mean of Failed Index(mu2) =0
Alternate Hypothesi : H1 : Mean Healthy Index(xbar) -Mean of Failed Index(ybar) >0
sample size n1 = 68 , n2 =33 so degrees of freedom =68+33-2 =99
Level of significance alpha =0.01 ,
Criterion for rejecting Ho if t calculated > t 99, 0.01 (t critical ) = 2.365
tcalculated = (xbar - ybar)/Sp(1/n1 +1/n2)
whrere Sp = pooled standard deviation
Sp2 =( (n1-1) S12 + (n2-1) S22 )/ (n1 +n2 -2)
Calculations are as follows
healthy failed
Mean | 1.729559 | 0.876364 |
Std Deviation | 0.635082 | 0.431906 |
variance | 0.40333 | 0.186543 |
Sample Size | 68 | 33 |
Sp = 0.5773 t critical : 6.97 ,
Hence Null Hypotheis is rejected , i.e Index for Healthy is higher than Failed
c) Calculate manually a 99% one-sided confidence interval for the difference in the average index of healthy and failed firms and compare your results with b) above.
One sided confidence interval for differeence of mean at 99% CL = (xbar-ybar) - (t 99, 0.01 (t critical ))*Sp*SQRT(1/n1 +1/n2)
= 0.853- 2.365*0.577*sqrt(0.015+0.030) = 0.853 -0.289 = 0.563 lowe bound
Output from Minitab (select stat-Basic stat-2 sample t ,select the samples in diffrent column put values in sample 1 and 2 , in options put ci 99% , test difference 0 and alternative greater than, select Assum equal variance)
Two-Sample T-Test and CI: Healthy, Failed
Two-sample T for Healthy vs Failed
N Mean StDev SE Mean
Healthy 68 1.730 0.635 0.077
Failed 33 0.876 0.432 0.075
Difference = mu (Healthy) - mu (Failed)
Estimate for difference: 0.853
99% lower bound for difference: 0.564
T-Test of difference = 0 (vs >): T-Value = 6.97 P-Value = 0.000
DF = 99
Both use Pooled StDev = 0.5773
There is no difference