MGMT2263 final exam practice solutions
Question 1
a) Ho: m1 = m2 = m3 = m4
Ha: not all means are equal
Reject Ho if F > 2.95
Here is the ANOVA table:
|
Source
of Variation |
SS |
df |
MS |
F |
|
Between
Groups |
8179.625 |
3 |
2726.5417 |
10.6383 |
|
Within
Groups |
7176.25 |
28 |
256.2946 |
|
|
Total |
15355.875 |
31 |
|
|
Reject Ho. Conclude the average annual expenditures are difference for at least 2 of the categories.
b) P-value = P(F > 10.6383). The 1% critical value is 4.57. Since P(F > 4.57) = 1%, this means P(F > 10.6383) < 1%. By inference, this means the p-value < 5%. Therefore, reject Ho.
c) Q = 3.85; D = 3.85sqrt(256.2946/8) = 21.79. Label the groups as follows:
|
Under 10 |
10 to 49 |
50 to 99 |
100+ |
|
group 1 |
group 2 |
group 3 |
group 4 |
Their means are as follows:
|
group 1 |
group 2 |
group 3 |
group 4 |
|
80.875 |
94.25 |
113.5 |
121.625 |
The following groups are significantly different:
group 1 from groups 3 and 4
group 2 from group 4
d) Lower limit = (121.625 80.875) 2.048sqrt(256.2946/8 + 256.2946/8) = 40.75 16.39 = 24.36
Upper limit = 40.75 + 16.39 = 57.14
24.36 < m4 - m1 < 57.14
Note that m4 - m1 = 0 does not fall in the confidence interval. From this we would conclude that m1 and m4 are significantly different. This is consistent with ANOVA since we concluded that at least two means are significantly different.
Question 2
a) Ho: median 1 = median 2 = median 3 = median 4
Ha: not all are equal
Reject Ho if KW > 7.8147
This is the rankings for each group (total at the bottom):
|
Class 1 |
Class 2 |
Class 3 |
Class 4 |
|
16 |
18 |
6 |
1 |
|
17 |
22 |
25.5 |
32 |
|
7 |
23 |
27 |
2 |
|
28 |
24 |
10.5 |
3 |
|
14 |
5 |
29.5 |
31 |
|
9 |
19.5 |
8 |
25.5 |
|
12 |
19.5 |
10.5 |
4 |
|
15 |
21 |
13 |
29.5 |
|
118 |
152 |
130 |
128 |
We see that 118 + 152 + 130 + 128 = 528 = 32(33)/2 so the rank sums check out.
KW = [12/(32(33))](1182/8 + 1522/8 + 1302/8 + 1282/8) 3(33) = 0.875
Do not reject Ho. Conclude there is no significant difference among the pedagogies.
b) The p-value = P(c2 > 0.875). From the chi-square table, we see the 10% critical value is 6.2514. Thus, the p-value is greater than 10% and under the general rule of thumb we would not reject Ho.
Question 3
a) Sxx = 9(6.397916327)2 = 368.4; Syy = 9(2.442880813)2 = 53.709; Sxy = 1,398.5 10(16.4)(7.69) = 137.34
b) b1 = 137.34/368.4 = 0.3728; b0 = 7.69 (0.3728)(16.4) = 1.5761
income = 1.5761 + 0.3728(hours)
c) r = 137.34/(sqrt(368.4)sqrt(53.709)) = 0.9764. This indicates a strong direct relationship between the number of hours cold-calling and income. (Based on what the others teach, a correlation between 75% and 100% is strong, between 50% and 75% is medium and below 50% is weak.)
d) r2 = 95.33%
e) Ho: B1 = 0
Ha: B1 Ή 0
Reject Ho if F > 5.32
Here is the ANOVA table:
|
|
df |
SS |
MS |
F |
|
Regression |
1 |
51.2005 |
51.2005 |
163.2885 |
|
Error |
8 |
2.5085 |
0.3136 |
|
|
Total |
9 |
53.7090 |
|
|
Reject Ho. Conclude the model is significant.
f) Ho: B1 £ 0
Ha: B1 > 0
The degrees of freedom for the critical value is 8.
Reject Ho if t > 1.86.
s(b1) = sqrt(0.3136/368.4) = 0.0292
t = 0.3728/0.0292 = 12.7671 (Note that should be the square root of the F statistic in part e)
Reject Ho and conclude there is a significant positive relationship between hours and income.
g) Lower limit = 0.3728 2.306(0.0292) = 0.3728 0.0673 = 0.3055
Upper limit = 0.3728 + 0.0673 = 0.4401
0.3055 < B1 < 0.4401
We would reject Ho since the hypothesis slope of zero does not fall in the confidence interval.
h) income = 1.5761 + 0.3728(25) = 10.896 which we multiply by 1000 to get $10,896.
i) Lower limit = 10.896 2.306sqrt(0.3136)sqrt(1/10 + (25 16.4)2/368.4) = 10.896 0.7082 = 10.1878 which we multiply by 1000 to get $10,188
Upper limit = 10.896 + 0.7082 = 11.6042 which we multiply by 1000 to get $11,604.
$10,188 < my < $11,604
Question 4
a) Ho: median 1 = median 2 = median 3
Ha: not all are equal
Reject Ho if FR > 5.9915
Here are the rankings within each block with the rank sums at the bottom:
|
Judge 1 |
Judge 2 |
Judge 3 |
|
1 |
2.5 |
2.5 |
|
3 |
1.5 |
1.5 |
|
3 |
1.5 |
1.5 |
|
1.5 |
1.5 |
3 |
|
2 |
2 |
2 |
|
1 |
2.5 |
2.5 |
|
11.5 |
11.5 |
13 |
We see that 11.5 + 11.5 + 13 = 36 = 6(6) so the rank sums are correct.
FR = [12/(6(3)(4))](11.52 + 11.52 + 133) 3(6)(4) = 0.25
Do not reject Ho. Conclude there is no significant difference among the judges in how they rate the couples.
b) Ho: median 1 = median 2 = = median 6
Ha: not all are equal
Reject Ho if FR > 11.0705
Here are the rankings within each block with the rank sums at the bottom:
|
Pair 1 |
Pair 2 |
Pair 3 |
Pair 4 |
Pair 5 |
Pair 6 |
|
2 |
6 |
3.5 |
1 |
5 |
3.5 |
|
3 |
6 |
2 |
1 |
4.5 |
4.5 |
|
3 |
6 |
2 |
1 |
4.5 |
4.5 |
|
8 |
18 |
7.5 |
3 |
14 |
12.5 |
We see that 8 + 18 + 7.5 + 3 + 14 + 12.5 = 63 = 3(21) so the rank sums are correct.
FR = [12/(3(6)(7))](82 + 182 + 7.52 + 32 + 142 + 12.52) 3(3)(7) = 13.71
Reject Ho and conclude that at least 2 pairs are rated differently.
c) The p-value = P(c2 > 13.71). We see the test statistic is between the 2.5% critical value of 12.8325 and 1% critical value of 15.0863. Thus, the p-value is between 1% and 2.5%.
Question 5
a) Ho: m1 = m2 = m3 = m4
Ha: not all means are equal
Reject Ho if F > 3.29
Here is the ANOVA table:
|
Source
of Variation |
SS |
df |
MS |
F |
|
Factor |
20.635 |
3 |
6.8783 |
1.1296 |
|
Block |
8756.255 |
5 |
1751.2510 |
287.6090 |
|
Error |
91.335 |
15 |
6.0890 |
|
|
Total |
8868.225 |
23 |
|
|
F = 1.1296.
Do not reject Ho. Conclude there is no significant difference among the office in their average monthly expenses.
b) Ho: m1 = m2 = m3 = m4 = m5
Ha: not all means are equal
Reject Ho if F > 3.26
Here is the ANOVA table:
|
Source
of Variation |
SS |
df |
MS |
F |
|
Factor |
0.1335 |
3 |
0.0445 |
0.6007 |
|
Block |
4.4630 |
4 |
1.1158 |
15.0607 |
|
Error |
0.8890 |
12 |
0.0741 |
|
|
Total |
5.4855 |
19 |
|
|
F = 15.0607
Reject Ho. Conclude there is a significant difference among the categories in average expenses.
c) Q = 4.51; D = 4.51sqrt(0.0741/4) = 0.6138
Here are the averages for each group ranked from lowest to highest:
|
Entertainment |
0.5 |
|
Office
supplies |
0.875 |
|
Transportation |
1.1 |
|
Utilities |
1.275 |
|
Miscellaneous |
1.925 |
The following groups are significantly different:
Entertainment from utilities and miscellaneous
Office supplies from miscellaneous
Transportation from miscellaneous
Utilities from miscellaneous
Question 6
a) Ho: gross income does not depend on number of employees
Ha: gross income does depend on number of employees
Reject Ho if test stat > 16.92
Here is the crosstab of expected values:
|
182.977 |
285.662 |
166.591 |
97.77 |
|
88.118 |
137.57 |
80.227 |
47.084 |
|
37.444 |
58.458 |
34.091 |
20.007 |
|
26.461 |
41.31 |
24.091 |
14.139 |
As we can see, none of the categories need collapsing since the lowest expected value of 14.139 is well above 5.
The first term of the test statistic is (286 182.977)2/182.977 = 58.006 which, by itself, is sufficiently large to reject the null hypothesis and conclude that gross income depends on the number of employees.
b) If you do all the grunge work, you will find the test statistic to be 764.844 (if we round to 3 decimals). Then Cramers V = sqrt(764.844/(1342(3))) = 43.59% which is the degree to which gross income depends on the number of employees.
Question 7
a) The model is price = -11.088 + 21.5768(sq. ft.) + 33.3956(bedroom) 30.7437(garage) + 38.951(basement). For the information given, price = -11.088 + 21.5768(15) + 33.3956(3) 30.7437(1) + 38.951(0) = 382.0071 which we multiply by 1000 to get $382,000 after rounding to the nearest thousand.
b) Suppose we call this variable x4. The reported p-value is for this test:
Ho: B4 = 0
Ha: B4 Ή 0
Since the p-value is 0.1925, we would naturally conclude that the variable does not significantly contribute to the model.
Suppose instead we want to test this hypothesis:
Ho: B4 £ 0
Ha: B4 > 0
How do we get the p-value? Recall that the p-value for a two-tail test is determined by multiplying the p-value for the appropriate one-tail test by 2. Since the p-value for the two-tail test is 0.1925, we get the p-value for the one-tail test by dividing 0.1925 by 2 to get 0.09625. Since this p-value is between 1% and 10%, the results would be inconclusive under the general rule of thumb.
c) Since the error degrees of freedom = 7, the t critical value is 2.365. The standard error for this variable is 4.593.
Lower limit = 21.5768 2.365(4.593) = 21.5768 10.8624 = 10.7144
Upper limit = 21.5768 + 10.8624 = 32.4392
(Note that these limits are slightly off the limits generated by Excel. This will be taken into account when marking the exam.)
d) The table is:
|
Variables |
Adj.
R2 |
ANOVA
p-value |
t-test:
are all variables significant? (yes or no) |
presence
of multicollinearity (yes/no) |
|
#1
All variables |
0.8338 |
0.0016 |
no |
no |
|
#2
square footage, bedrooms |
0.8295 |
0.0001 |
yes |
no |
|
#3 square footage |
0.6912 |
0.0005 |
yes |
no |
|
best |
#1 |
#2 |
#2,
#3 |
tie |
Choose model #2 since:
§ its adjusted r2 is slightly less than that of model #1
§ it has the lowest ANOVA p-value
§ all its variables are significant
§ it has no presence of multicollinearity