LAISHA
Open Laisha

Japan Statistical Society Certificate Exploratory Data Analysis & Statistical Graphics: practice questions

Japan Statistical Society Certificate (Toukei Kentei) · Japan · Japan Statistical Society (certificate operated with the Japan Statistical Association)

Original practice questions written to the published syllabus · not past papers
10worked here, free
54written to this exam in this subject
9syllabus units

About Japan Statistical Society Certificate

Japan Statistical Society Certificate (Toukei Kentei) is a professional certification run by Japan Statistical Society (certificate operated with the Japan Statistical Association), for candidates in Japan. Graded certificate exams; the graded levels are mostly computer-based.

Source: toukei-kentei.jp.

Checked against the exam body's published information. Formats and fees change; confirm the current edition there before you book.

Exploratory Data Analysis & Statistical Graphics in Japan Statistical Society Certificate

Looking at data before modelling it: summaries, robust statistics, graphs that show distributions and relationships honestly, outlier detection, and the tidy-data and reporting habits that make an analysis checkable. Deliberately practical and software-based.

Our 54 Japan Statistical Society Certificate questions in this subject: 12 easy, 29 medium, 13 hard, across 9 units.

10 practice questions, worked

Choose an answer, then open the working. Each explanation says why the right answer is right and why the others are not.

These 10 were written to Japan Statistical Society Certificate and are not shown on any other exam's page.

Question 1 · medium

Which join lists plots that have a yield record but no matching treatment record?

  1. An inner join of the yield table and the treatment table
  2. A full join of the two tables followed by removing missing values
  3. An anti-join of yields against treatments
  4. A cross join of every yield record with every treatment record
Answer and working

C. An anti-join of yields against treatments

An anti-join returns rows of the first table with no key match in the second, exactly the orphan records to investigate. An inner join silently drops them.

Question 2 · easy

Plot yields have Q₁ = 2.8 and Q₃ = 3.6 t/ha. Where is Tukey's upper inner fence?

  1. 4.8 t/ha
  2. 1.6 t/ha
  3. 6.0 t/ha
  4. 4.4 t/ha
Answer and working

A. 4.8 t/ha

IQR = 0.8; upper inner fence = Q₃ + 1.5 IQR = 3.6 + 1.2 = 4.8. 6.0 is the outer fence (3 IQR); 1.6 is the lower inner fence; measuring from the median (3.2) gives 4.4.

Question 3 · medium

Seven plot yields are 3.1, 3.4, 3.3, 3.6, 3.2, 3.5 and 8.9 t/ha. What is the median absolute deviation (MAD) about the median?

  1. 0.3
  2. 5.5
  3. 0.7
  4. 0.2
Answer and working

D. 0.2

Median = 3.4; absolute deviations are 0.3, 0.0, 0.1, 0.2, 0.2, 0.1, 5.5; their median is 0.2. The wild value 8.9 contributes only one of seven deviations, so it barely affects the MAD.

Question 4 · hard

For data from a normal distribution, about what fraction of observations fall beyond Tukey's inner fences (1.5 IQR beyond the quartiles)?

  1. 5%
  2. 0.3%
  3. 0.7%
  4. 2.5%
Answer and working

C. 0.7%

Q₃ = 0.6745σ, IQR = 1.349σ, so the upper fence is at 0.6745 + 2.0235 = 2.698σ. P(|Z| > 2.698) ≈ 0.7%. So a few points beyond the fences are expected in large normal samples.

Question 5 · hard

Why can two statistical packages report different lower quartiles for the same 10 numbers?

  1. One package sorts in descending order and so reports the upper quartile instead
  2. Quartiles depend on the random seed used by the package
  3. One package always reports the median of the whole sample as Q₁
  4. They use different sample-quantile definitions that interpolate between order statistics differently
Answer and working

D. They use different sample-quantile definitions that interpolate between order statistics differently

Hyndman and Fan catalogued nine quantile definitions in common software. They converge for large n but can differ noticeably for small samples, so reports should state the method.

Question 6 · hard

Why is a formal normality test (such as Shapiro-Wilk) often less useful than a Q-Q plot?

  1. It tests equality of variances rather than normality
  2. It assumes the data are already normal and so it always ends by accepting normality
  3. Small samples pass clear departures; large ones reject trivial ones
  4. It tests the equality of group variances rather than the normality of the data
Answer and working

C. Small samples pass clear departures; large ones reject trivial ones

Power grows with n: small samples pass skewed data, large samples reject departures that have no practical effect on t-tests or ANOVA. The plot shows the kind and size of the departure.

Question 7 · hard

Which limitation of a scatterplot matrix does a conditioning plot (coplot) address?

  1. A SPLOM requires all variables to have the same units
  2. A SPLOM shows only pairs, hiding effects of a third variable
  3. A SPLOM requires all of its variables to be measured in the same units
  4. A SPLOM cannot be drawn for more than three variables at any one time
Answer and working

B. A SPLOM shows only pairs, hiding effects of a third variable

Coplots draw the y-x scatterplot separately for ranges of a third variable (for example yield against nitrogen within rainfall classes), showing how the relation changes.

Question 8 · medium

Why is log(y + 1) used with caution for counts containing many zeros?

  1. It removes all the zeros from the analysis
  2. Adding 1 to every count always turns the data into a normal distribution
  3. The added constant is arbitrary and changes the results
  4. It removes all of the zero counts from the analysis without warning
Answer and working

C. The added constant is arbitrary and changes the results

log(y + c) with c = 1, 0.5 or 0.1 gives noticeably different results for small counts. Count models (Poisson, negative binomial) avoid choosing c.

Question 9 · medium

A map shows each district's yield as a departure above or below the long-term regional average. Which colour scheme fits?

  1. A diverging palette
  2. A qualitative palette with unrelated hues for each district
  3. A sequential palette running from light to dark shades
  4. A single colour with no shading, for maximum simplicity
Answer and working

A. A diverging palette

Departures in two directions from a meaningful midpoint call for a diverging scheme (for example one hue for above, another for below, pale at zero).

Question 10 · medium

In a stacked bar chart of crop area by crop for several years, which quantities can be compared accurately across years?

  1. The bottom segment and the total
  2. Only the top segment
  3. Only the middle segments, which have the most room
  4. Only the top segment, which sits against the total
Answer and working

A. The bottom segment and the total

Only the bottom segment and the total share a common baseline. Middle segments float on different baselines, so their lengths must be compared without alignment.

A study plan for Exploratory Data Analysis & Statistical Graphics in Japan Statistical Society Certificate

This plan follows where our questions actually fall, so the time goes where the marks are likely to be. It is our reading of the syllabus, not the exam body's weighting.

  1. First, the heavy units (48% of the 54 questions): Histograms, density estimates and quantile-quantile plots (9, 17%); Scatterplots, smoothing (LOESS) and scatterplot matrices (9, 17%); Graphical perception and honest chart design (8, 15%).

    Read the manual chapters for these first and do every worked question you can find on them.

  2. Then the middle band: Five-number summaries, box plots and robust statistics (median, MAD, trimmed means) (7, 13%); Outliers and influential observations (7, 13%); Transformations to symmetry and constant variance (6, 11%).

    Aim for one sitting per unit, then mixed questions across them.

  3. Then the lighter units: Tidy data: variables, observations and tables (5, 9%); Types of data and measurement scales (2, 4%); Tables for reports and publications (1, 2%).

    Learn the definitions, classifications and standard numbers; these units are more often recalled than worked.

  4. Leave time for the hard ones: Outliers and influential observations (4 hard of 7), Transformations to symmetry and constant variance (2 hard of 6), Five-number summaries, box plots and robust statistics (median, MAD, trimmed means) (2 hard of 7).

    Come back to these after a first pass, untimed, and read every explanation, including for the options you rejected.

  5. Bank the easy marks: Types of data and measurement scales (1 easy of 2); Tidy data: variables, observations and tables (2 easy of 5).

    Do these early in revision so they are fast on the day.

  6. Finish timed. Sit the subject as a timed set without notes, mark it, and send each wrong answer back to its unit above.

Last updated · How Laisha writes and checks its pages · Report a mistake

Ask Laisha about this

Opens Laisha's chat with your question ready to send. Sign in or start free.

Unlock every question — Japan Statistical Society Certificate Exploratory Data Analysis & Statistical Graphics

US$10 a month · one price worldwide · cancel any time

  • Every practice question in this subject, with full working
  • Timed exam papers and progress by unit
  • Ask Laisha to explain any answer — up to 1,000 questions a month
Free or Smarter
FreeSmarter
Grow guides, pest pages, articlesRead allRead all
Ask Laisha5 a dayUp to 1,000 a month
Exam practiceThe sample on each pageEvery question, timed papers
Crop calendar and pest alerts—For your own fields
Calculators and tools—All
Uploads and reports—100 a month, real files
10 days free for every new account, no card.