7 Ways to Simplify Life as a Statistician/Data Analyst

By: Wamukoya Marilyn, Data Analyst. 7th June 2012.

After 3 years of working as a statistician/data analyst at the African Population and Health Research Center (APHRC), I have had easy and hard days. Reflecting on these days, I have come up with 7 ways in which to simplify my life and make most of my days easy ones.

APHRC runs the Nairobi Urban Health & Demographic Surveillance System (NUHDSS) which collects health and demographic data on the same population every three months. In addition, there are dozens of other projects that collect data of their own. After 11 years, the Center has accumulated mountains of data that require cleaning, analysis, documentation, dissemination and storage. These data are critical because they impact policy promulgation and/or revision. This makes the work of a statistician at the Center especially pertinent to its mission, which is: To be a global center of excellence, consistently generating and delivering scientific evidence for policy and action on population, health, and education in Africa.

 

 7. Make Order Out of Chaos

The work of a statistician mostly involves making order out of chaos. I believe that this is very noble work because it emulates The Creator. We can simplify our lives by putting our best foot forward every day so that by the end of each day, we can sit back and say ‘It is good’. An important question to answer as a statistician is what it would take to do good work. Doing good work in my view is being meticulous in documenting what we do (see # 3), being patient with those we work with (see # 1), maintaining a healthy work-life balance (see # 4), recognizing the value of the data that we handle (see # 6) and being ethical in our handling of  data (see # 5 and 2).

 

6. Back It Up

Over the course of time, we have all accumulated data, do-files, as well as supporting files of one kind or the other on our hard drives and other storage devices. If there is no system in place to back-up the work that we do, we can end up in a situation in which we lose everything. While it may seem inconceivable that this might occur, consider these situations: you drive over your laptop bag, your vehicle is broken into, you are mugged, carjacked, involved in an accident, etcetera. While some organizations have formal back-up systems in place, it is always wise to have a personal back-up system.

 

 5. Just Say No

As with any profession, a certain amount of ethical aptitude is required in order to make things simple. I am sure that we have all come across individuals during the course of our work, who have directly or indirectly, hinted that they wanted certain results, regardless of what the data actually say. This particular situation can become very challenging if we are dealing with an individual who is in a position to make decisions that have a direct impact on your job security. If you know that you must say ‘NO’ -- because cooking belongs in a kitchen not in a do-file --, then just say ‘NO’ because that ‘NO’ will have a direct impact on your integrity and that of the organization.

 

4.  Sleep on It

There are times when as a statistician, you know what you want to accomplish but it seems impossible to write the code that would get the desired result. While the adage of taking a step back from your work to let your brain solve the problem might seem cliché, it actually works. If you must, step away from your desk and do something else or go home and sleep.

 

3.  Document Your Do-File

There are two things to consider here. First, write a do-file – this is a plain text file into which you can type Stata commands. Sometimes, for one reason or the other, we might be tempted to write commands directly into the command window of Stata and then copy them into a do-file later. What seems like simplicity itself transforms into a complicated situation when the machine crashes before you are able to transfer the commands into a do-file and save them. Second, when you write a do-file, document each and every step that you take. This ensures that the next time you run that do-file, your decision-making process will be clear to you and any recipients of your do-files. I have found myself in two kinds of frustrating situations. The first one was when I wrote a beautiful do-file that did a walloping amount of work in just a few lines of code but then I neglected to write why I had written any of those beautiful lines. The second situation is when I inherited do-files from a predecessor who did not provide any comments in their do-file. In both cases, I had to rewrite the do-files while running each line of code and physically observing what it was doing to the dataset – not an easy thing to do.

 

2.  Accept ‘Missingness’ & Outliers

Sometimes as a data analyst you can feel pressured to produce perfect data. However, there is no such thing as perfect data. There will always be the outliers that seem to throw off all the other results and that, you cannot explain. Similarly, there will always be missing data that you cannot impute to satisfy everyone. It is far simpler to accept that you have missing data and outliers and document them as they occur.

 1. Trust Your Supervisor(s)

There are times when your supervisor will make requests that do not seem to make sense or has a management style that conflicts with your working style. For example, you are a statistician who prefers to be assigned tasks with clear time frames and left alone to accomplish them but you have a supervisor(s) who would prefer to meet every so often to discuss the progress you are making. This can be frustrating because you may feel that the progress meetings take time away from actual work and accomplish very little, or that your supervisor does not trust you. What if your supervisor’s management style is their way of buffering you from his/her own supervisor, who is skeptical about your abilities and/or your contribution?