Q1 Statistical Significance in Decision-Making: Validating Student Food Preferences

The Fallacy of Anecdotal Evidence

In today’s business world good leaders need to tell the difference between noise and important signals.
When making decisions like changing university food services leaders can't just rely on instincts feedback
from a few vocal people or casual observations. A change in a group of people might seem promising but
leaders need to figure out if its representative of the whole group or just a coincidence.

For example, let’s say a university tracks what students like to eat. Usually, 30% of students prefer food. A
recent survey of 100 students shows that 40% now prefer it. To some people a 10% increase might seem like
a deal enough to change the food staff and ordering. A manager who uses data would start a hypothesis test.
When making decisions university leaders should look at student food preferences. Use data to guide them.
They should not just rely on evidence or simple observations. By doing they can make informed decisions
that are in line with actual student behaviour and preferences. This approach helps to avoid mistakes and
ensures that resources are allocated effectively.

By using frameworks university leaders can gain a better understanding of student food preferences. They
can identify trends and patterns that might not be immediately apparent. This approach helps to ensure that
decisions are based on facts, then intuition or guesswork.

In conclusion statistical significance plays a role, in organizational decision-making. It helps university

leaders to validate student food preferences and make decisions. By using tools and frameworks leaders can
ensure that decisions are data-driven, transparent and aligned with actual student behaviour.
The Statistical Decision Process is a five-step framework. We need to use this to see if student preferences
have really changed a lot. This way our final answer is not just what we think, What the numbers tell us.
1. Stating the Hypotheses

We have to start with two ideas that're opposite: the Null Hypothesis and the Alternative Hypothesis.

The Null Hypothesis is like saying nothing has changed. It says the number of students who like food is still
the same as it was before.

The Statistical Decision Process says we have to look at this Null Hypothesis.

Null Hypothesis is that the number of students who like food is thirty percent.

If the Null Hypothesis is true then any difference between what we found and what we expected is because
of luck.

The Alternative Hypothesis is what we are trying to prove. We want to know if the number of students who
like food has changed a lot.

The Statistical Decision Process helps us with the Alternative Hypothesis.

We are looking for a change either more students like Chinese food or less students, like Chinese food.

The Statistical Decision Process is important here because it helps management know what to do.

If the number of students who like food has changed then management needs to change how they do things.
The Statistical Decision Process and the Null Hypothesis and the Alternative Hypothesis all work together to
give us the answer.

2. Why We Chose the Z-Test

Before we start let’s make sure, we're using the tool for the job. For this problem a Z-test for proportions is
the choice for a few reasons:

* The data is simple: it's either yes (Prefers food) or no (Prefers something else). We're looking at
proportions, not averages.

* The sample size is big enough: for a Z-test to work we need a sample that’s large enough to be like a
distribution. We check this with the success/failure condition:

* n \times p0 = 100 \times 0.30 = 30(That's than 5)

* n \times (1. P0) = 100 \times 0.70 = 70 (That's also, then 5)

* We're comparing a sample proportion (p = 0.40) to a known population proportion (p0 = 0.30$).

3. Calculating the Standard Error

The Standard Error tells us how much our sample proportion might vary from the population proportion.
This is important because we're only looking at a subset of 100 students, not the university. The Standard
Error helps us understand that uncertainty.

The formula for the Standard Error of a proportion is:

SE = sqrt {{p0 (1 – p0)}/{n}}

Plugging in our values:

SE = sqrt {{0.30 \ 0.70}{100}} = sqrt {{0.21}{100}} = sqrt {0.0021} (approx. 0.0458)

4. Computing the Z-score

The Z-score is essentially a measure of "distance in units of standard error." It calculates how many standard
errors our observed sample proportion (0.40) is away from the historic mean (0.30).

The formula for the Z-score is:

Z = {{p} – p0}/{SE}
Calculating the score:

Z = {0.40 - 0.30}/ {0.0458} = {0.10}/ {0.0458} (approx. 2.18)

A Z-score of 2.18 indicates that our sample result is 2.18 standard deviations away from what we would
expect if the preference were still truly 30%. In statistics, anything beyond 2 standard deviations is generally
considered "unusual."

Conclusion

We looked at the numbers. Since 2.18 is bigger than 1.96 we do not believe the null hypothesis. The 10
percent preference surge for food is real and it is not just because of random chance. This means that
students really do want Chinese food and it is not just a fluke.

Strategic Takeaways:

* Procurement: We should buy ingredients for Chinese dishes.

* Staffing: We need to train our staff to make food or hire people who already know how.

* Efficiency: We should use numbers to make decisions not just guess. This will help us make sure that the
changes we make to the dining hall are good and will save us money.
This plan will help us make sure that we are spending our money on the things and that students are happy,
with the food we are serving. Chinese food is what students want so we should give it to them.

Q2 (A): Probability Theory in Customer Segmentation

In data-driven retail a contingency table is more than a list of numbers. It is a guide for deciding how to use
resources. By looking at the relationship between customer attributes, such as gender and shopping habits
like the number of purchases management can create ads based on evidence. Understanding the details of
Probability Theory in Customer Segmentation, including joint and conditional probabilities is essential for
finding the customer segments that will give the highest return on marketing investment.

Application
Evaluation of Probabilities:

* Marginal Probability: This measures the chance of one thing happening like the chance that a customer's
Female. It is easy to calculate. It does not provide much insight into behaviour making it too general for
specific campaign targeting.

* Joint Probability: This looks at two things happening at the time like the chance that a customer is Male
and makes more than five purchases every week. It is useful for being specific. It can be misleading because
it does not show the likelihood of that behaviour within a specific group, such as the Male customer
segment.

* Conditional Probability: This calculates the chance of something happening given a trait like the chance
that a customer makes more than five purchases given that she is Female. Its major advantage is that it
isolates the behaviour of a segment allowing for targeted -selling strategies, which is a key aspect of
Probability Theory in Customer Segmentation.
Independence vs. Dependence:

The question of whether gender and purchasing behaviour related is important for creating a budget.

* Assuming Independence: This means that gender has no impact on how someone shops. If the retail chain
assumes this, they might waste their budget on a campaign that does not work for either group.

* Assuming Dependence: This means that different genders may have shopping habits.

Recommended Approach:

The retail chain should use Conditional Probability and assume that gender and purchasing behaviour are
related, until testing shows otherwise. This is the reasonable choice because it acknowledges that customer
behaviour is complex which is a key consideration in Probability Theory in Customer Segmentation. If
women shop often but spend less per trip or the other way around Conditional Probability allows
management to create specific incentives, such as discounts for one group and loyalty rewards for the other
to maximize the return on investment of the campaign.

Using marginal or joint probabilities gives a limited view of the customer base. By focusing on probabilities
and looking at the relationship between gender and purchasing habits the retail chain can create highly
targeted marketing campaigns, which is a key application of Probability Theory, in Customer Segmentation.
This approach ensures that budget allocation is based on behavioural trends rather than oversimplified
averages ultimately making the campaign more effective and increasing conversion rates, which is the
ultimate goal of using Probability Theory in Customer Segmentation.

Q2 B Analysis of Sales Benchmarking through Distribution

The Bell Curve as a Strategic Map

The Normal Distribution is a great way to look at how well a company is doing. For a retail chain with 250
stores sales numbers usually gather around a middle point making a bell shape. In this case sales are 150,000
on average and most numbers are within 20,000 of that. This gives us a picture of what is normal and what
is really good. If we want to know if a store can reach 180,000 in sales we can figure out the probability.
This helps leaders decide if their idea of a "premium store" is realistic or not.

Application: Decoding the Probability

1. Statistical Calculation

To see if 180,000 is a goal we need to calculate the Z-score. This score shows how many steps away from
the average our target's

Z = (X- Μ) / σ = (180,000- 150,000) / 20,000 = 30,000 / 20,000 = 1.5

We look at a table and find that the probability of getting a Z-score of 1.5 or less is 0.9332. To find the
probability of a store doing better than 180,000 we look at the end of the curve:

P(X > 180,000) = 1. P(Z < 1.5) = 1. 0.9332 = 0.0668

This means there is a 6.68% chance that any store will do better than 180,000.

2. Managerial Context

In a network of 250 stores this probability means about 17 stores will qualify for the "premium" category.
This shows that the current goal is very exclusive for the best stores that do 1.5 times better than average.

Evaluation of the $180,000 Threshold

The goal has two main implications:

* High Prestige: Only 6-7% of stores can reach this goal so it is very special. It helps find the stores and
learn from them.

* Risk of Demotivation: If this goal is used for all stores the other 93% might feel it is too hard to reach and
get discouraged.

Strategic Balancing

The 180,000 goal is good mathematically. Leaders need to decide if they want a exclusive or inclusive
"premium" category. If they want to motivate stores they might lower the goal to 170,000. This would
include 15.8% of stores making the goal easier to reach for good managers while still being special. The
distribution helps balance how special the goal is, with what the company wants to achieve.