DEPARTMENT OF MASTER OF BUSINESS ADMINISTRATION

I MBA - 24FC101

STATISTICS FOR MANAGEMENT

ASSIGNMENT – SET 3 (Solved)

Name: ____________________________ Reg. No: ____________________________

Course: I MBA - 24FC101 - Statistics for Management Year/Sem: I/I

Submitted to: ____________________________

Date of Submission: ____________________________

 06 November 2025

Statistics for Management

Detailed Solutions — Assignment Set 3

Q1. Bayes’ theorem — Dual test results (detailed)

Question

A manufacturing plant produces electronic components with a historical defect rate of

5%. The quality control department uses two tests: Test A (sensitivity 90%, false positive
8%) and Test B (sensitivity 85%, false positive 5%). A randomly selected component
tests positive on both tests. (a) Find probability it is actually defective. (b) If the manager
wants at least 95% confidence before scrapping, should it be scrapped?

Theory / Explanation (detailed)

Bayes' theorem gives the posterior probability of a hypothesis (here: defective = D) given
evidence (here: positive results on tests A and B). When multiple test results are
observed, and if the tests are conditionally independent given the true state, the joint
probability of both positives given the state equals the product of individual conditional
probabilities.

Notation

Let D = component is defective, D' = component is not defective. A+ = positive on Test

A, B+ = positive on Test B.

Given

P(D) = 0.05 (prior defect rate)

P(D') = 0.95
P(A+ | D) = 0.90 (sensitivity of A)
P(A+ | D') = 0.08 (false positive rate of A)
P(B+ | D) = 0.85 (sensitivity of B)
P(B+ | D') = 0.05 (false positive rate of B)

Assumption

We assume Test A and Test B results are conditionally independent given whether the
component is defective or not. This means P(A+ and B+ | D) = P(A+|D)·P(B+|D), and
similarly for D'.
Solution — step by step

1) Compute joint likelihoods:

P(A+ and B+ | D) = 0.90 × 0.85 = 0.765

P(A+ and B+ | D') = 0.08 × 0.05 = 0.004

2) Multiply by priors to get numerator and denominator of Bayes formula:

Numerator = P(A+ B+ | D) × P(D) = 0.765 × 0.05 = 0.03825

Denominator = Numerator + P(A+ B+ | D') × P(D') = 0.03825 + 0.004 × 0.95 =

0.03825 + 0.0038 = 0.04205

3) Posterior probability:

P(D | A+ B+) = Numerator / Denominator = 0.03825 / 0.04205 ≈ 0.9100

Interpretation and Conclusion

The probability that the component is actually defective given positive results on both
tests is approximately 91.0%. Since the manager requires at least 95% confidence before
scrapping, this 91% is below that threshold. Therefore, based on only these two test
results and the stated requirement, the component should NOT be scrapped.

Practical note: If scrapping is costly and manager insists on 95% certainty, consider
repeating tests, using an alternative confirmatory test with higher accuracy, or performing
a physical inspection.
Q2. Proportion estimation — Android users (detailed)

Question

From a random sample of 250 smartphone users, 140 preferred Android. (i) Construct a
99% confidence interval for the true proportion. (ii) Does the interval support the industry
benchmark of 55%? (iii) If firm wants to reduce margin of error by 40% in future, what
sample size is required?

Theory / Explanation

For a sample proportion p̂ from a large sample, the sampling distribution of p̂ is

approximately normal with mean p and standard error sqrt(p(1-p)/n). For a confidence
level (1−α), the CI is: p̂ ± z_{α/2} × SE. To plan sample size for a desired margin of error
(ME), use ME = z × sqrt(p(1-p)/n) and solve for n.

Given

n = 250, x = 140 → p̂ = 140/250 = 0.56. Use 99% confidence (z_{0.005} ≈ 2.5758).

Solution — step by step

1) Compute sample proportion:

p̂ = 140 / 250 = 0.56

2) Standard error (SE):

SE = sqrt(p̂ (1−p̂ )/n) = sqrt(0.56×0.44/250) = 0.031394

3) Margin of error for 99%:

ME = z × SE = 2.5758 × 0.031394 = 0.080865

4) Confidence interval:

99% CI = p̂ ± ME = 0.56 ± 0.080865 = [0.479135, 0.640865]

Interpretation of (ii)

The industry benchmark is 0.55 (55%). Since 0.55 lies inside the 99% CI, the sample data
do not contradict the benchmark — the interval supports the benchmark.

Part (iii) — sample size to reduce ME by 40%

Current ME = 0.080865. Reducing by 40% means new ME = 0.6 × current ME.

Using p̂ for planning, required n ≈ 695 (rounded up). If conservative estimate used
(p=0.5), n would be slightly larger.

Conclusion

99% CI ≈ [0.4791, 0.6409]. The benchmark 0.55 is inside the interval, so supported. To
reduce the margin of error by 40%, need about 695 respondents.
Q3(i). One-sample z-test (pricing strategy) — detailed

Question

Before implementing new pricing strategy, average daily revenue per customer was $45
(population SD known to be $12). After 30 days with the new strategy, sample mean =
$49.50. Test at 1% significance whether mean increased.

Theory / Explanation

When population standard deviation σ is known, use z-test: z = (x̄ − μ₀) / (σ/√n). For one-
sided test H₁: μ > μ₀, p-value = P(Z ≥ observed z). Compare p-value to α.

Given

μ₀ = 45, σ = 12, n = 30, x̄ = 49.5, α = 0.01 (one-sided test).

Hypotheses

H₀: μ = 45 (no increase)

H₁: μ > 45 (mean increased)

Solution — step by step

1) Standard error SE = σ/√n = 12/√30 = 2.1909

2) Test statistic z = (49.5 − 45) / 2.1909 = 2.0540

3) One-sided p-value = P(Z ≥ 2.0540) ≈ 0.019990

Decision and conclusion

Compare p-value with α = 0.01. Since p ≈ 0.019990 > 0.01, we fail to reject H₀ at 1%
level. That means the evidence is not strong enough at 1% significance to conclude the
mean revenue increased. However, at the 5% level (α = 0.05), p < 0.05, so the increase
would be considered significant at 5% but not at 1%.
Q3(ii). One-way ANOVA — productivity across shifts (detailed)

Question

Compare mean productivity (tasks/hour) across Morning, Afternoon, Night shifts using
ANOVA at α = 0.05. Data:
Morning: 18,22,20,19,21
Afternoon: 16,15,18,17,14
Night: 12,14,13,15,11

Theory / Explanation

One-way ANOVA tests equality of group means by partitioning total variance into
between-group variance (SSB) and within-group variance (SSW). F = MSB / MSW
where MSB = SSB/(k−1) and MSW = SSW/(N−k). If F is greater than critical F or p-
value is small, reject H₀ that all group means are equal.

Hypotheses

H₀: μ₁ = μ₂ = μ₃ (means equal across shifts)

H₁: At least one mean differs

Calculations — step by step

1) Grand mean = 16.3333

2) SSB (between groups) = 123.3333

3) SSW (within groups) = 30.0000

4) df_between = 2, df_within = 12

5) MSB = 61.6667, MSW = 2.5000

6) F = MSB/MSW = 24.6667

Decision and conclusion

Critical F for df(2,12) at α = 0.05 ≈ 3.885. Here F ≈ 24.6667 > 3.885 ⇒ reject H₀. There
is a significant difference in mean productivity among shifts. We can follow up with
post-hoc tests (like Tukey) to find which pairs differ, but based on means, Morning >
Afternoon > Night in productivity.
Q4(i). Chi-square test for independence — Age group vs Payment method (detailed)

Question

A sample of 300 customers produced the table:

Age Group / Payment: 18-30 (25,35,40), 31-45 (30,45,25), 46-60 (45,30,25). Column
totals: Cash=100, Credit Card=110, Mobile Pay=90. Test independence at 5%.

Theory / Explanation

Chi-square test compares observed counts Oij with expected counts Eij under
independence: Eij = (row_i total × col_j total) / N. The statistic χ² = Σ (Oij − Eij)^2 / Eij
with df = (r−1)(c−1).

Given & expected counts

Each row total = 100. So expected counts for each row are:

E(Cash) = 100×100/300 = 33.33; E(Credit) = 100×110/300 = 36.67; E(Mobile) =

100×90/300 = 30.00

Solution — calculations

Compute χ² = Σ (O−E)^2 / E = 14.6818

Degrees of freedom = (3−1)(3−1) = 4. Critical χ²(4,0.05) = 9.488.

Decision and interpretation

Since χ² = 14.6818 > 9.488, we reject independence. Conclusion: choice of payment

method depends on age group. For example, older customers (46–60) show higher cash
usage (45 observed vs 33.33 expected).

Recommendation: tailor payment promotions by age group; encourage digital/mobile

payments among older age groups through incentives.
Q4(ii). Sign test — Wait times before and after (detailed)

Question

Wait times (minutes) for 10 customers before and after new queuing system are given.
Use Sign test at 5% to check if wait times reduced.

Theory / Explanation

Sign test for paired data counts positive signs (before > after) vs negative signs. Under
null of no change, positives ~ Binomial(n, 0.5). For one-sided test expecting reduction,
compute P(X ≥ observed positives).

Data and computation

Differences (before−after): [2, 2, 3, 2, 3, 2, 3, 2, 3, 2]

Signs: ['+', '+', '+', '+', '+', '+', '+', '+', '+', '+'] → positives = 10, negatives = 0, ties = 0

Effective n (ignoring ties) = 10

One-sided p-value = P(X ≥ 10) under Binomial(10, 0.5) = 0.000977

Decision and conclusion

Since p ≈ 0.000977 < 0.05, reject H₀. There is strong evidence that the new system
reduced wait times. Practical implication: the queuing changes were effective; consider
rolling out across branches.
Q5(i). Regression — Experience vs Sales (detailed)

Question

Experience (years): 1,2,3,4,5,6,7; Quarterly sales ($000s): 50,65,75,82,88,95,100. (a)

Find regression equation predicting sales from experience. (b) Predict sales for 8 years.
(c) Calculate and interpret R².

Theory / Explanation

Simple linear regression fits a line Ŷ = a + bX minimizing squared errors. b = Σ(x−x̄ )

(y−ȳ)/Σ(x−x̄ )², a = ȳ − b x̄ . R² = SSR / SST measures proportion of variance explained by
model.

Calculations — step by step

Mean X = 4.0000, Mean Y = 79.2857

Slope b = 7.9643

Intercept a = 47.4286

Regression equation: Ŷ = 47.4286 + 7.9643X

Prediction for X=8: Ŷ = 47.4286 + 7.9643×8 = 111.1429

Total SST = 1839.4286, SSR = 1776.0357, R² = 0.9655 (≈ 96.55%)

Interpretation and conclusion

The fitted model shows a strong positive relationship between experience and sales. Each
additional year of experience is associated with an average increase of about $7964.29k
in quarterly sales. R² ≈ 96.55% indicates the model explains most of the variation in
sales.
Q5(ii). Correlation — Price vs Demand (detailed)

Question

Price ($): 10,12,15,18,20,22,25; Demand (units): 200,180,150,120,100,80,60. (a)

Compute correlation coefficient. (b) Interpret strength and direction. (c) Business insight.

Theory / Explanation

Pearson correlation r = Σ(x−x̄ )(y−ȳ) / sqrt(Σ(x−x̄ )² Σ(y−ȳ)²). r ranges from −1 to +1. R² =

r² measures proportion of variance explained by linear relationship.

Calculations

Computed Pearson r = -0.998354

R² = 0.996711 (≈ 99.67%)

Interpretation and business insight

There is a very strong negative linear relationship between price and demand (r ≈ -0.998).
This means as price increases, demand decreases almost perfectly for this dataset.
Business insight: price reductions are likely to raise demand substantially, but the firm
must consider profit per unit vs volume to determine optimal price. Use elasticity
estimates and profit models to choose price.
Final Notes and How to Write by Hand

• Copy each question fully, then write the detailed solution step-by-step. Underline final
answers.
• Use Times New Roman, 12 pt, 1.5 spacing. Write on one side of A4 only.
• For hypothesis tests always state H0 and H1, test statistic, decision rule, p-value (or
compare with critical value), and conclusion in simple language.
• If you need, add a short references section: textbooks on Statistics for Management,
lecture notes, and class data sources.