Quantitative Methods – I

Answer 1:
Introduction
Business is now data-driven, and today’s competitive marketplace demands quantitative
analysis for effective decision making. Managers and analysts rely on statistical methods to
overcome guesswork and develop research-based strategies. One of the most common methods
for this is interval estimation, in which organizations make estimates of a population parameter
based on sample data and incorporate uncertainty into those estimates. Unlike point estimation,
however, with interval estimation you don't focus on a single number from the sample; instead
it provides a range within which the true population parameter should fall. This interval is
stated to a certain degree of confidence, such as 90%, 95%, or 99%.

In the current situation, a telecommunication company is launching a new online service and
it has conducted in-depth interviews with 250 customers chosen at random. Of those, 162
requested to subscribe. With 90% confidence, the company would like to estimate what portion
of the entire customer base is odicted to subscribe to the service. We cannot simply report that
64.8% of the sample is interested, for this percentage might slightly differ if we would take
another sample. Therefore, a confidence interval for proportions is the most suitable technique
to yield an accurate and reliable estimate for this high level strategic decision making.

In the following paragraphs I will elaborate the methods of the Interval Estimation in terms of
Proportion, its calculation step by step, how to interpret as a manager and where this method
becomes inadequate.

Principles and Practice of Confidence Intervals

Sample Proportion

To compute the confidence interval, we first create the sample proportion, denoted by p̂.

Formula:

p̂ = x / n

Where,

x = number who respond in the affirmative

1
n = sample size

Here, x = 162 and n = 250.

p̂ = 162 / 250 = 0.648

What this of course translates into is that 64.8% of customers who were polled made mention
that they would like to be subscribing to the service.

Why Confidence Intervals Are Needed

However, if the analyst presents only the sample proportion (64.8%), the manager becomes
lucky in that he/she might expects to be exactly same as population. But due to sampling error,
different samples would yield slightly differing results. Confidence intervals take this into
consideration by producing a range of likely values for the population proportion. So managers
have a more dependable range in which to make their decisions, instead of just one number.

General Formula for Confidence Interval of proportions

The formula is:

CI = p̂ ± Z × √( p̂ (1 – p̂) / n ) Where CI is the confidence interval, CIs are the two values that
mark off an interval of \(100\% - \alpha\) between them.

Where,

CI = confidence interval

p̂ = sample proportion

n = sample size

Z = z-value for the selected confidence level

√( p̂ (1 – p̂) / n ) = std error of the proportion

For a 90% level of confidence, the z-value is 1.645.

Step-by-Step Calculation

Step 1: Standard Error (SE)

SE = √( 0.648 × (1 – 0.648) / 250 )

SE = √( 0.648 × 0.352 / 250 )

2
SE = √( 0.228096 / 250 )

SE = √0.000912

SE = 0.0302

Step 2: Margin of Error (ME)

ME = Z × SE

ME = 1.645 × 0.0302

ME = 0.0497

Step 3: Construct Confidence Interval

Lower Limit= p̂ – ME = 0.648 – 0.0497 = 0.5983 (59.8%)

Upper Limit = p̂ + ME = 0.648 + 0.0497 = 0.6977 (i.e.,69.8%)

Thus we had the 90% confidence interval:

59.8% ≤ p ≤ 69.8%

Interpretation of Results

This interval provides the analyst with 90% confidence that the actual proportion of customers
interested in subscribing to new internet service falls between 59.8 and 69.8%. That is, if the
company conducted random surveys of the same size multiple times, we would expect about
90 per cent of the calculated intervals to contain the true proportion.

This way of interpreting the input allows decision-makers to appreciate both the potential
upside (almost 70% adoption) and the lower-bound risk (about 60% adoption).

Practical Implications of the Confidence Interval

Infrastructure Planning: It has to build additional capacity of internet infrastructure – servers,

Revenue Forecasting With 60–70 % adoption range, finance teams can create conservative,
moderate and optimistic revenue forecasts. (Note: if there are 50,000 customers for example)

Lower bound = 59.8% × 50,000 = 29,900 emails to the list

3
Upper bound = 69.8% × 50,000 = 34,900 subs Maybe!

Risk Mitigation: Managers are taking less risk by looking at both bounds. It means like, getting
ready for at least 29900 customers guarantees that the service level will be at a specific quality
and increasing if there's demand (and reaching like, 34+900) in order to not run out of some
supply.

Strategic Marketing: Marketing strategies can be adjusted according to what they predict where
the actual adoption will lie within the interval. More fervent campaigns could nudge interest
toward the high end.

Role of Confidence Level

The company opted for a 90% confidence level, not 95% or 99%. Greater confidence levels
will result in wider intervals, with less precision. In business decision making for example, a
90% confidence narrow interval may be more useful. It is a good compromise between
certainty and usability.

For example, a 95% interval could stretch it to between 58 and 71%, so planning is less certain.
Therefore, 90 per cent confidence is justified here.

Illustrative Example

Now, pretend that instead of 250, the company had surveyed only 50 customers. The sample
proportion will still be about 65%, but the standard error would be larger and the confidence
interval would be much wider than it is now. That would also make the estimate less reliable.

This helps to explain why larger sample sizes result in less error, and tighter (more reliable)
intervals. The selection of 250 participants by the company, reinforces our analysis.

Limitations of the Confidence Interval

Sampling Bias: If the 250 customers sampled are not truly random (i.e. highly skewed city-
wise), the interval might not represent the actual population proportion.

Evolution of Customer Preferences: Interest in internet services may shift due to competitive
promotions, pricing schemes or technology advancements. Therefore, the confidence interval
is only valid for at the time of the survey.

4
Confidence Intervals Reading: Managers also need to know that a 90% confidence interval
does not mean there is a 90% chance the true value will be in this interval. By this I mean that
the technique must be 90% successful in long-term use.

Managerial Insights

The analyst has converted raw survey data into actionable insights, thanks to confidence
intervals. Instead of the use of only a sample proportion, however, managers can have a realistic
range that they can work with. This supports:

More reliable financial models

Smarter investment decisions

Better marketing campaigns

Reduced operational risks

It also assured shareholders and stakeholders that the company’s choices are based on rigorous
statistical algorithms.

Conclusion

The analyst at the telecommunications company calculated confidence interval for proportion
to determine customer interest in a new internet service. Based on the survey of 250 customers,
a sample proportion of 64.8% was obtained and with a confidence level of 90%, an interval
from 59.8% to 69.8%. In other words, the analyst is 90% sure that the actual proportion of
interested customers falls within that range.

This finding holds important implications for businesses. It permits management to make more
precise infrastructure, revenue and risk plans and marketing decisions. Confidence intervals do
not remove uncertainty, but rather a framework for dealing with it when making decisions.

In summary, the application of interval estimation can turn uncertain data into strategically
useful knowledge. Acknowledging minimum and maximum interest from customers the
company is prepare to cope for demand in a flying start of the new service. This shows that
quantitative methods can directly improve managerial judgment and firm performance.

5
Answer 2(A):

Introduction

Client satisfaction is an important component of long term success in the financial services
industry. Businesses are always on the hunt for a means to measure, monitor and elevate
satisfaction because it so straightforwardly impacts customer retention and brand perception.
Bayes’ theorem is a powerful statistical tool that can assist in this type of work. This theorem
lays out a means of revising probabilities in light of new evidence. To simplify the financial
planner problem, initial probabilities are calculated based on the number of clients that each
planner serves. As more information comes in (e.g. a client giving a high satisfaction report!)
the firm can re-calculate the probabilities to see which advisor is now more likely to have
actually served that client.

From the practical point of view, this method becomes very important for performance
evaluation and resource allocation. But before they use this kind of application, management
has to evaluate Bayes’ theorem as an appropriate tool for the needs of their organization and
weigh some of the factors that impact how reliable and helpful those new probabilities actually
are.

Concepts and Application

Explaining Bayes’ Theorem in This Context

Bayes’ theorem states:

P(A|B) = [ P(B|A) * P(A) ] / P(B)

Where:

P(A) = Prior probability of an event (e.g., probability that Advisor X served the client,
considering distribution in clients).

P(B|A) = How likely we are to see the evidence if the event is true (e.g., how likely Advisor
X’s client says that they have high satisfaction.)

P(B) = Probability of observing the evidence (the probability that any client gives high
satisfaction).

P(A|B) = Posterior probability (new belief of the likelihood that Advisor X did serve satisfied
client).

6
The firm can then update its beliefs with the new satisfaction information by applying this
formula.

Benefits of Applying Bayes’ Theorem

Performance Insights: It helps to management can see what advisors are more strongly linked
with the satisfied client. This helps in evaluating effectiveness.

Resource Distribution: Advisors may have access to more valuable clients if they consistently
satisfied many (j) of their high satisfaction requests. And to those developers for whom it's less
likely, you can provide extra training.

Predictive Updating: Bayes’ theorem doesn’t compute fixed probabilities but updates them
upon receipt of new data, making it choice for settings where client feedback is ongoing.

Factors to Consider

Bayes’ theorem can provide useful insights that are actionable when some conditions hold:

Prior Probabilities: The initial allocation of clients to advisors has to be reliable. Mis-specified
priors will issue into biased updated estimates.

Evidentiary Reliability: Reported satisfaction must be measured with some level of consistency
– whether via well-crafted surveys, independent rating systems or vetted feedback.

Independence of Observations: If our satisfaction reports are affected by external factors (e.g.,
promotions, market movements, service policies), the independence assumption may not be
met.

Interpretability for Managers: While the strength of Bayes’ theorem is in its handling of (strong)
probabilities, managers have to understand what it means in practical terms - performance or
directions for improvement.

Critical Viewpoint

Bayes theorem is good for a lot of things but it has limitations. Overemphasis on revised odds,
without inclusion of qualitative criteria like advisor-client relationship or communication style
and market factors can result in underestimation. Furthermore, satisfaction is subjective and
can depend on individual client expectations. Hence, Bayes’ theorem can be only a
complementary means of measuring advisor performance and not the sole one.

7
Conclusion

Lastly, the use of Bayes’ theorem may be very suited within a financial advisory company to
update probabilities concerning link advisers and client satisfaction. It is rigorous in that it
integrates previous information with new evidence for decision-making. Yet, in order for the
findings to have real value and be actionable, the firm will need good priors, sound data on
satisfaction and insight into what outcomes matter most. In combination with qualitative
insights, if employed appropriately Bayes’ theorem can strongly contribute to performance
assessment and resource management. This well-rounded perspective helps the company to
evaluate satisfaction and drive client service standards in a methodical, data-driven manner.

Answer 2(B):

Introduction

The probability analysis is an important aspect for the risk management in financial institutions.
Analysts frequently have to model probabilities using the normal distribution model, which is
widely used in finance for modelling returns, risk and client behavior. Historically, these
computations have been accomplished through z-tables that give the probability of standard
normal values. But now that technology has advanced, tools like Excel’s NORM. DIST and
NORM. INV functions have become popular.

In the instance of the major financial institution, this inconsistency is caused by certain
departments implementing Excel while others are still using z-tables. Leadership wanting you
precise, fast and easily trained analysts. In order to select a "best" method, we must critically
evaluate the trade-offs between Excel's computational tools and the traditional z-table
technique.

Concepts and Application

The Traditional Z-Table Method

The z-table allows us to determine the cumulative proportion of a standard normal distribution
with values that are less than or equal to the z-score. To apply it, the data analyst first of all
needs to scale up or standardize the data using the following formula:

z = (x – μ) / σ

8
Where,

x = observed value

μ = mean

σ = standard deviation

Then, given the z-score, the table is used to look up the appropriate probability.

Advantages:

Develops good understanding of the concept of probability.

Is not technology dependent, and is effective in teaching and as a measure in testing.

Limitations:

Time-consuming when calculations are frequent.

Human induced inaccuracies when manually searching.

Precision limited due to rounding the numbers in the table.

Excel’s NORM. DIST and NORM. INV Functions

Excel makes our life easier by computing probabilities for us.

NORM. DIST(x,mean, standard_dev,cumulative): It returns the probability at a given value.

NORM. INV(probability, mean, standard_dev): Looks up the input value from the given
probability.

Advantages:

Efficient: Helps to automate tasks and performance time-consuming calculations faster.

Accurate – More than 6 decimal places.

Scalability: It can process large datesets with ease — must-have for financial institutions.

Access: Easy to find in industry, meaning it can be applied between teams.

Limitations:

Needs to learn the Excel functions, which could be difficult for some beginners.

9
Relying too much on software without a good background in programming can produce weak
conceptual understanding.

Dependent on access to technology.

Balancing Efficiency, Accuracy, and Interpretability

How you can decide between which method to use:

Compute Efficiency – Excel is unbeatable the less your need to do manually and if you intend
to perform complex analysis.

Precision: With Excel you get much more precision, z-tables are being approximations behind
the scenes.

Interpretability: Tables aid beginners to understand the concept underneath, while Excel
numbers are just a “black box” if users do not know how they’re getting their answer.

You should use a mix of both: Learn to use z-tables for conceptual understanding and then
move onto Excel for professional work. If you've 100 aff units fluctuations and need responses
within milliseconds use excel as standard but keep z-tables for academic stuff.

Conclusion

Overall, there is a time and place for both Excel functions and z-tables. For a big bank the
NORM. DIST and NORM. INV functions are a more efficient, accurate and scalable approach
so this is the more portable solution for common application of probabilities across
departments. But to prevent blind dependence on software, a training could still include z-
tables so that analysts retain understanding of the concept. By integrating powerful theoretical
understanding with state-of-the-art computational tools, the institution is able to deliver both
in precision of risk analysis and operational efficacy, while enforcing a standard approach
across desks and creating lasting analytical depth.

10