1. Monthly overtime hours (in hundreds) for six branches are 12, 15, 15, 18, 20 and 70. Which measure best describes the typical branch, and what is its value?

A) The median, 18, because it is the middle value of the six
B) The median, 16.5, because the outlier of 70 inflates the mean of 25
C) The mean, 25, because it uses every value and is unaffected by extreme values
D) The median, 15, because it is the most frequently occurring value

Correct: B difficult Measures of central tendency

Mean = 150 / 6 = 25, pulled up by 70. The median of six values is the average of the third and fourth: (15 + 18) / 2 = 16.5, which is resistant to the outlier. If you chose the mean you ignored that means are sensitive to extremes. If you chose 15 you confused median with mode. If you chose 18 you took the fourth value instead of averaging the middle two.

2. The values 4, 6, 8, 10 and 12 are the whole population being studied. What is the population standard deviation?

A) 3.16
B) 2.40
C) 8.00
D) 2.83

Correct: D difficult Standard deviation

Mean = 8. Squared deviations: 16, 4, 0, 4, 16 = 40. Population variance = 40 / 5 = 8, so the standard deviation = square root of 8 = 2.83. If you chose 3.16 you divided by n - 1 (the sample formula). If you chose 8.00 you gave the variance. If you chose 2.40 you gave the mean absolute deviation.

3. A cost model y = 1,200 + 2.5x was fitted using data for 400 to 900 units. A manager uses it to estimate the cost at 1,100 units. Which answer is correct?

A) Tk 2,750, less reliable because 1,100 units lies outside the range of the data
B) Tk 5,150, less reliable because 1,100 units lies outside the range of the data
C) Tk 3,950, fully reliable because a linear relationship holds at any volume
D) Tk 3,950, but less reliable because 1,100 units lies outside the range of the data

Correct: D complex Regression and extrapolation

y = 1,200 + 2.5 x 1,100 = 3,950. Because 1,100 is beyond the observed range, the relationship may change (extrapolation), so reliability is lower. If you chose the fully reliable answer you assumed linearity holds everywhere. If you chose Tk 2,750 you omitted the fixed element. If you chose Tk 5,150 you added the fixed element twice.

4. A customer-satisfaction survey was sent to all customers. Only 12% replied, and most replies came from customers with strong complaints. Which bias is most likely to affect the results?

A) Measurement bias, from a flawed survey instrument
B) Survivorship bias, from studying only survivors
C) Non-response (self-selection) bias
D) Confirmation bias, from favouring supporting evidence

Correct: C difficult Bias in data

When a small, self-selected group responds, their views may differ systematically from those who did not. Survivorship bias concerns looking only at successes that remain. Confirmation bias is the analyst favouring evidence that supports an existing belief. Measurement bias arises from flawed collection instruments. If you chose any of these you described a different mechanism from the one in the scenario.

5. Which TWO are examples of structured data? (Select TWO.)

Multiple response
A) Customer emails
B) A sales ledger held in fixed fields
C) Photographs of damaged stock
D) Call-centre recordings
E) A relational database of invoice lines

Correct: B, E difficult Structured data

Structured data has a defined format, such as rows and columns. Emails, photographs and recordings are unstructured. If you chose A, C or D you chose unstructured data.

6. Which THREE are the original 'three Vs' of big data? (Select THREE.)

Multiple response
A) Validity
B) Volume
C) Velocity
D) Value
E) Variety
F) Visibility

Correct: B, C, E difficult Big data characteristics

The original three Vs are volume, velocity and variety. Veracity and value were added later, and validity and visibility are not part of the model. If you chose A, D or F you chose terms that are not in the original three.

7. Which TWO displays are most suitable for showing a trend over time and the relationship between two variables? (Select TWO.)

Multiple response
A) A line chart of monthly sales over five years
B) A pie chart with 14 categories
C) A three-dimensional exploded pie chart
D) A scatter diagram of advertising spend against sales
E) A table of 500 rows

Correct: A, D difficult Data visualisation

Line charts show change over time and scatter diagrams show relationships. Pie charts with many slices and 3-D effects make comparison hard, and a 500-row table does not show patterns. If you chose B, C or E you chose a display that hides the pattern.

8. Consider correlation and outliers. For each statement, select Yes if it is correct and No if it is not.

Yes/No grid

1. Correlation proves that one variable causes the other.   No

2. A correlation coefficient of -0.9 indicates a strong negative linear relationship.   Yes

3. Outliers distort the mean more than the median.   Yes

Correct: N, Y, Y difficult Correlation

(1) No: correlation does not prove causation. (2) Yes. (3) Yes: the median depends on position, not magnitude.

9. Consider types of data analytics. For each statement, select Yes if it is correct and No if it is not.

Yes/No grid

1. Predictive analytics uses historical patterns to estimate future outcomes.   Yes

2. Descriptive analytics explains why an event occurred.   No

3. Prescriptive analytics recommends actions.   Yes

4. Diagnostic analytics only summarises what happened.   No

Correct: Y, N, Y, N difficult Types of analytics

(1) Yes. (2) No: descriptive analytics summarises what happened; diagnostic explains why. (3) Yes. (4) No: that is descriptive analytics.

10. Consider the use of personal data. For each statement, select Yes if it is correct and No if it is not.

Yes/No grid

1. Personal data should be collected for specified and legitimate purposes.   Yes

2. Anonymised data can never be re-identified.   No

3. Customer consent is unnecessary once a company already holds the data.   No

Correct: Y, N, N difficult Data ethics and governance

(1) Yes. (2) No: poorly anonymised data can be linked back to individuals. (3) No: use must still be lawful and consistent with the purpose for which it was collected.

11. In hypothesis testing, which description of a type I error is correct?

A) Failing to reject a null hypothesis that is actually true
B) Rejecting a null hypothesis that is actually false
C) Failing to reject a null hypothesis that is actually false
D) Rejecting a null hypothesis that is actually true

Correct: D complex Type I and type II errors

A type I error is a false positive: the null hypothesis is true but is rejected. A type II error is a false negative: the null hypothesis is false but is not rejected. The other two options describe correct decisions (rejecting a false null; not rejecting a true null) or the type II error.

12. Cells C2:C201 of a worksheet hold the taka value of 200 customer orders. Which formula counts the orders larger than Tk 50,000?

A) =SUMIF(C2:C201,">50000")
B) =COUNTIF(C2:C201,">50000")
C) =AVERAGEIF(C2:C201,">50000")
D) =COUNT(C2:C201)

Correct: B difficult Spreadsheet functions

COUNTIF counts cells that meet a stated condition. SUMIF adds the qualifying values instead of counting them. COUNT counts all numeric cells regardless of size. AVERAGEIF returns the mean of qualifying values.

13. A bank compares 'monthly customer visits' for two branches. Branch A counts both in-person and mobile-app sessions; Branch B counts only in-person visits. What is the main problem with comparing the two figures?

A) The measure is defined differently in each branch, so the data are not comparable
B) The comparison has too large a sample, because data from every customer visit is included in both figures
C) The comparison commits a type I error, because a true difference in visits is rejected without evidence
D) The comparison suffers from non-response bias, because only customers who completed the survey are counted

Correct: A difficult Comparability of data

Data can only be compared when it is defined and measured consistently. Here the two branches count different things. Type I errors relate to hypothesis testing, non-response bias to surveys, and counting every visit is not a sampling problem.

14. A bank's customers are 60% urban and 40% rural. To reflect this, it randomly selects 600 urban and 400 rural customers from the two groups for a survey. Which sampling method is this?

A) Quota sampling
B) Stratified random sampling
C) Simple random sampling
D) Systematic sampling

Correct: B complex Sampling methods

The population is divided into strata (urban and rural) and a random sample is taken within each in proportion to its size. Quota sampling also sets group targets but selects respondents non-randomly until the quota is met. Simple random sampling does not use groups. Systematic sampling selects every nth item.

15. Which THREE are risks of big data and data analytics? (Select THREE.)

Multiple response
A) Faster decision-making
B) Breaches of data privacy or security
C) Biased conclusions from unrepresentative data, leading to unfair decisions
D) Improved targeting of customers
E) Lack of transparency over how personal data is used
F) Improved forecasting accuracy

Correct: B, C, E difficult Risks of big data and analytics

Risks include privacy and security breaches, bias in data and algorithms, and ethical issues such as lack of transparency. Faster decisions, better targeting and improved forecasts are benefits.

16. Which TWO measures best help to protect commercially sensitive information from deliberate or accidental employee action? (Select TWO.)

Multiple response
A) Limiting access to those who need the data for their roles
B) Allowing all staff to download full datasets to personal storage devices
C) Training staff and requiring confidentiality agreements
D) Publishing reports on the intranet with no access restrictions
E) Switching off the system's activity logs to save storage

Correct: A, C difficult Protecting commercially sensitive information

Role-based access and training with confidentiality commitments reduce deliberate and accidental disclosure. Unrestricted downloads, open publication and disabling logs all weaken protection.

17. Which TWO chart choices are appropriate? (Select TWO.)

Multiple response
A) A line chart showing monthly revenue over 24 months
B) A pie chart showing monthly revenue over 24 months
C) A bar chart comparing sales by region for one quarter
D) A line chart showing each region's share of a single quarter's total
E) A pie chart with 25 slices of near-equal size

Correct: A, C difficult Choosing charts

Line charts show trends over time and bar charts compare categories. A pie chart is for the proportions of a single whole using few categories, and is poor for trends or many slices.

18. Consider basic spreadsheet functions. For each statement, select Yes if it is correct and No if it is not.

Yes/No grid

1. The AVERAGE function returns the sum of the selected cells.   No

2. COUNTIF counts the cells that meet a stated condition.   Yes

3. What-if analysis shows how changing assumptions affects results.   Yes

4. SUM adds the values in the selected cells.   Yes

Correct: N, Y, Y, Y difficult Spreadsheets

(1) No: AVERAGE returns the mean, i.e. the sum divided by the number of values. (2) Yes. (3) Yes. (4) Yes.

19. Consider surveys and data quality. For each statement, select Yes if it is correct and No if it is not.

Yes/No grid

1. Good survey questions avoid leading wording.   Yes

2. A very long survey can reduce the response rate and the quality of answers.   Yes

3. A larger sample size removes bias caused by a flawed selection method.   No

4. Professional scepticism means accepting management's data without challenge.   No

Correct: Y, Y, N, N difficult Surveys and scepticism

(1) Yes. (2) Yes. (3) No: a bigger biased sample is still biased. (4) No: professional scepticism means questioning and challenging data.

20. Consider the use of data in business. For each statement, select Yes if it is correct and No if it is not.

Yes/No grid

1. Competitor annual reports are an internal data source.   No

2. The sales ledger is an internal data source.   Yes

3. Control involves comparing actual performance with targets.   Yes

4. Planning uses only historical data and never external information.   No

Correct: N, Y, Y, N difficult Sources and uses of data

(1) No: they are external. (2) Yes. (3) Yes. (4) No: planning also uses external information such as market data.

21. A study of 40 Dhaka retailers finds a correlation coefficient of -0.85 between the number of competing shops within 500 metres and monthly sales. What is the best interpretation?

A) A strong negative relationship, though it does not by itself prove that competitors cause lower sales
B) A weak positive relationship, which shows that more competitors raise sales
C) No relationship, because correlation coefficients below zero mean the data are invalid
D) A strong positive relationship, which proves that competitors cause lower sales

Correct: A complex Correlation

A coefficient near -1 indicates a strong negative association: more competitors accompany lower sales. Correlation does not prove causation; other factors (location quality, rents) might drive both.

22. Weekly sales (Tk million) for seven weeks are 4, 6, 6, 7, 9, 12 and 30 (an exceptional bulk order). Which measure of central tendency best describes a typical week and what is its value?

A) The median, Tk 7 million
B) The range, Tk 26 million
C) The mean, about Tk 10.6 million
D) The mode, Tk 6 million because it is the lowest

Correct: A complex Central tendency with outliers

The median is not distorted by the outlier of 30; the middle (fourth) value is 7. The mean is pulled up to about 10.6. The mode is 6 but is based on only two observations, and the range is a measure of spread.

23. Cells B2:B6 contain 10, 20, 30, a blank cell, and 40. What does =AVERAGE(B2:B6) return, and what would it return if the blank cell contained 0?

A) 20 in both cases
B) 25 with the blank cell, and 20 if the cell contained 0
C) 25 in both cases
D) 20 with the blank cell, and 25 if the cell contained 0

Correct: B complex Spreadsheet averages

AVERAGE ignores blank cells: (10 + 20 + 30 + 40) / 4 = 25. A zero is a value: 100 / 5 = 20.

24. A survey of 100 customers has a standard error of the mean of 2 (population standard deviation 20). If the sample is increased to 400, what is the standard error, assuming the same standard deviation?

A) 4
B) 1
C) 2
D) 0.5

Correct: B complex Sampling error

Standard error = sigma / square root of n. For n = 100 it is 20 / 10 = 2; for n = 400 it is 20 / 20 = 1. Quadrupling the sample halves the standard error.

25. Which survey question is least likely to bias the answer?

A) Given the heavy traffic we face, was our delivery acceptable?
B) How much do you love our fast and friendly delivery service?
C) Don't you agree that our delivery service is excellent?
D) How satisfied are you with our delivery service, from very dissatisfied to very satisfied?

Correct: D difficult Survey questions

A neutral question with a balanced scale avoids leading wording. The others suggest the desired answer.