Preparing for the CASA Statistical Analysis Certification Exam

Why I Decided to Take It

Let me show off this certificate first, so the experience I am about to share has at least a tiny bit of credibility. XDDD

A CASA certification? What even is that? Never heard of it. (scratches ear)

I imagine that is how most people react the first time they hear about it. It was exactly my reaction when I first decided to take this certification. XD

You might wonder why I would take such an obscure certification. Do I collect certificates, or do I need one to strengthen or prove my ability?

No—neither. I am not interested in collecting certificates, and I do not really need one to prove what I can do.

I simply find that I study a little more seriously when I am preparing for an exam, so I deliberately set up an exam to force myself to learn the related knowledge properly.

Without an exam or the pressure to pass, I might skim many things and move on. Too much would remain half-understood, leaving an unstable foundation.

I first encountered this idea during my second master’s degree. My adviser asked whether I planned to take the lawyer or judge examinations. I originally saw no need: I was only in the graduate law program to write my thesis.

My adviser told me, ‘Preparing for a national examination will make you study and learn more seriously. I still recommend preparing for one.’

The more I thought about it, the more sense it made. Since then, whenever I feel that a subject deserves serious study, I deliberately look for a related certification to prepare for.

I had studied statistics for a long time, but I had never systematically run through the whole body of knowledge. I had learned a little here and there, and it did not feel solid. After finishing my thesis, I wanted to prepare properly and make my statistical knowledge a little more complete.

‘Is there a certification that could push me to study statistics properly?’

After some searching, I found the Chinese Applied Statistics Association’s CASA Statistical Analysis Certification and decided to use it as my target for tackling statistics thoroughly.

For detailed exam information, see the official CASA Data Science Professional Competency Certification website.

You may ask: why not prepare for a SAS certification instead? Would that not be more valuable and more credible?

Because I wanted to prepare statistical theory rather than practise coding, I did not want the focus to be SAS or R programming exercises.

That does not rule out taking a SAS certification later—if I have time.

The scary part is not having time. 555

About the Exam

This exam is one certification within the CASA Data Science Professional Competency Certification run by the Chinese Applied Statistics Association.

The certification included a Social Media Word-of-Mouth Analysis certificate and a Statistical Analysis certificate. Because I only researched and prepared for the latter, everything below concerns the Statistical Analysis certificate only.

Registration and payment were online, with several convenient options such as convenience stores, post offices, and banks.

The exam lasted one hour and had 40 questions. They were all single-choice questions, with both conceptual and calculation questions.

For a statistics exam, you may bring a calculator that meets national-examination requirements—that is, one without formula-memory functions. For approved brands and models, see Approved Models of Calculators for National Examinations.

For this certification, 70 is a passing score; 80–89 earns an Excellent Award, and 90 or above earns an Outstanding Award.

The association’s official website also has an examination guide with other candidate information. Check it if you need it; I will not reproduce everything here!

How Much Statistics Background Do You Need?

The scope of statistics is a huge question. In short: it has no boundaries.

In terms of depth, statistics can be very basic—calculating an average or reading a chart—or it can extend through using hypothesis tests to the mathematical theory underneath them. All of that belongs to statistics.

There is a considerable gap between knowing how to use something and knowing where it comes from, and statistics makes that gap especially obvious.

In breadth, it ranges from descriptive statistics such as means and standard deviations through hypothesis testing, multivariate analysis, regression, and time-series analysis. It is enormous; many parts could each become a semester- or year-long course.

My conclusion first: for me personally, preparing for this certification was a little challenging.

You can treat my high-school mathematics as if I had never learned it. When I graduated, I was a failing high-school student who could not even keep matrices and determinants straight.

People who know me are amazed by how bad my maths was: ‘How can someone this bad at maths teach physics?’

It is fine—I have been teaching it just fine, right? XD It only shows that high-school physics does not necessarily require exceptionally strong mathematics.

That is only true for high-school physics, though. University physics is another story…

So my first real encounter with statistics and combinatorics began with biostatistics in my second year of public-health studies.

Biostatistics is one of the most important courses in public health. Oddly, although I had always hated maths, I found combinatorics and statistics unexpectedly interesting.

What a public-health biostatistics class covers depends heavily on the instructor. At minimum, it usually touches chart reading, means and standard deviations, hypothesis testing, multivariate analysis, and regression. The emphasis is usually on how to use them; the mathematical principles behind them may receive only a little attention, or none at all.

I was not satisfied, so I took statistics in the mathematics department and got thoroughly beaten up for a whole semester. But it finally gave me a first, basic understanding of many ideas whose methods I had learned without knowing why they worked.

Because I loved statistics, I later took Environmental Data Analysis in the Department of Bio-Environmental Systems Engineering and Experimental Design in Life Sciences, and spent additional time reading about regression and time-series analysis.

I thought preparing for this certification would be easy. It absolutely was not.

This is exactly what I mentioned earlier: without the pressure of an exam or certificate, knowledge does not sink in as firmly. Too often, we swallow it whole, skim it, and move on.

For example, I had never really understood when to use a bar chart and when to use a histogram, or what the difference between them was.

My earlier self thought they were both vertical. Why did I need to distinguish them?

Later in school, when topics such as ANOVA and regression came up, I usually did not calculate by hand; I just gave them to software to run.

But in an exam, you do not have a computer!

That creates the problem of knowing what to do but not being able to set up the calculation. I might know that I needed a correlation coefficient and how to interpret it, but sadly I could not remember its exact formula…

From my own experience, if you have not once studied statistics seriously and thoroughly enough to understand both the principles and calculations, I recommend starting again from the beginning. Do not be overconfident just because you have seen it before and go in cold. XDDDD

How I Prepared

Given how broad statistics is, where do you begin? What kind of reference book should you buy?

The official exam site listed one reference book, so naturally I bought it.

It was the only listed book! XDDDD If there had been many books, maybe not—but there was only one, so why not buy it? XD

Since it is called a question-type question bank, the whole book is basically questions. There is no summary of key points or concepts. But the earlier parts provide questions by concept, while the later parts contain several sets of mock exams.

Does that still make the book worth reading?

Of course. Let me explain.

Statistics is broad, but the exam scope is limited. The table of contents lets us roughly identify the chapters and topics that are likely to be tested.

That scope includes data-collection methods, descriptive statistics, probability and combinatorics, distributions, estimation, and hypothesis testing; a quick look at the appendices adds ANOVA and regression.

That gives us a workable preparation boundary. At least we can set aside overly difficult multivariate analysis and time-series analysis.

Next, confirm the direction of the question types. A quick look showed that most questions focus on how to use concepts—calculating 95% confidence intervals, means, standard deviations, correlation coefficients, and so on—with few proof-style questions.

So we can ruthlessly drop the most theoretical material and focus on knowing how to use and interpret these statistics and methods.

But if you start doing questions with no conceptual foundation at all, you get stuck on every question and may end up memorising answers. You still need suitable concept summaries or textbooks as a base.

When I prepared for this certification, I used the biostatistics slides from my public-health undergraduate years as a foundation, together with the statistics notes from my mathematics courses.

Those materials covered the breadth reasonably well, but perhaps not enough depth.

So for each topic, I searched online for resources such as course handouts and slides. For particular chapters I also watched online statistics courses from different universities. Only then did the preparation feel more solid.

Once each unit seemed reasonably clear, I began working through the questions in the book.

Do not let the small book fool you. Each chapter has hundreds of questions, so it is a real workload. While preparing, I required myself to do at least about 150 questions a day. Orz

Each unit also has detailed solutions, so when I truly could not solve a problem I had something to consult instead of being completely lost when correcting it.

Because many explanations are mainly calculations, though, if you do not understand why a particular expression is set up that way, you really have to look it up yourself. There is no other way. Orz

Why, then, did I say earlier that you should definitely buy this question bank?

I discovered the reason only after I had finished and reviewed the whole book, then began looking at the past papers announced on the official site.

Up to that point, the past papers for this certification had all come from this question bank.

The question wording was almost identical. The order of options might occasionally change, but the numbers generally did not.

In other words, if you completed and reviewed the entire question bank, you could basically achieve a high score. I had no doubt about that.

But I need to stress that I took the December 2020 session. That pattern was still present in the session I took, but I cannot guarantee it continued in later years. Anyone preparing should check again for themselves.

For me, that was enormously encouraging, because once I had seriously worked through a question, I could usually remember what the answer was even across thousands of questions.

Still, I did not know whether the pattern would hold for my exam year, so I carefully reviewed the concepts again before the test and made sure I understood them before I went in.

Exam Day and What Happened Afterwards

My exam was from 14:00 to 15:00 at the Taipei venue, National Taipei University of Business. I knew the place well after spending so many years in public health and environmental health.

I ate something casually at noon and arrived far too early to get ready. There seemed to be another certification exam in the same session. Maybe because so few people were taking the Statistical Analysis certification, I wandered around forever without finding my room, until I finally asked a staff member.

Once inside, it looked like an ordinary certification exam: candidate numbers and exam details were posted on the desks. Then I realised that candidates for two different exams had been placed in the same room. = =

There were fewer than 15 candidates in Taipei; every other seat belonged to candidates for the other certification.

‘That few people are taking it…?’ I muttered. I knew it was not popular, but I had not expected it to be this small.

I had prepared fairly well, and because past papers drew from the question bank, passing itself felt almost secure.

But I still did not know whether the year would follow the same pattern or change the questions.

‘What if there is suddenly a reform this year?’

With that thought, I kept seriously reviewing my own notes and the question-bank items I had missed or corrected.

Wow, if only I had prepared this seriously for the lawyer and judge examinations.

When the paper was handed out, I laughed.

Sure enough, the questions were word-for-word the same as the question bank, with only the order changed a little.

So I answered largely from the memory of having worked through them before, and finished all the questions in under 30 minutes.

I used the remaining 30 minutes to check the calculation questions again on my calculator. Once I was sure there were no major problems, I handed in the paper quickly.

While writing the exam, I noticed a question whose Z-table number had been looked up incorrectly in the question bank. The explanation and answer were therefore wrong.

The same error had not been corrected in the exam options, so there should not have been a correct answer to select.

But as a candidate who wanted to answer correctly rather than memorise a question and its solution, I wrote the number I had calculated to be more correct instead of the wrong listed answer.

A few days later, the official site published the answers. I checked them and—yes—the same uncorrected wrong answer was still there.

Should I appeal the question? I thought about it for a moment.

In the end I decided to appeal. It might be the only exam in my life where I had the chance to appeal a question, and gaining some appeal experience sounded rather fun. XDDD

A statistics appeal also did not require much supporting evidence; you only had to recalculate it. It seemed much easier than law.

Later I received a very friendly response, and the two appealed questions were simply awarded points.

Why was I so confident yet still did not get a perfect score?

Because the two questions I missed were in fact the parts I had not understood while doing the book and could not find information about. Losing those ten points was fair enough. Orz

Conclusion

I only took this certification to rebuild my statistical concepts and theory, yet I ended up studying statistics and doing questions for a full month—more than ten times as seriously as I had prepared for the lawyer and judge examinations.

During that process, I suddenly realised: ‘I really do like statistics!’

Preparing for the exam thoroughly rebuilt my statistical knowledge and improved my calculation skills. I also earned what may be my first and last question-appeal achievement, so I gained a great deal.

The claim that you have to force yourself to prepare for an exam in order to learn properly really does make sense. XDDDD

Finally, here are the notes I made while preparing, for anyone who needs them. I do not guarantee they are correct, so please use them carefully alongside other textbooks. Orz

CASA Statistics Notes