Writing is at least 25% of research time

Many people will leave the writing part of their research at the last after they have done experiments and data analyses. This way is wrong and we should avoid this paradigm.

Writing is at least 25% of research time

Many people will leave the writing part of their research at the last after they have done experiments and data analyses. This way is wrong and we should avoid this paradigm. In this post, we will show why we should not leave the writing at the last and how we should do the writing correctly by an example.

Writing is learning

Writing is a learning process because writing is a cognitively demanding activity. Writing trains and constructs a systematic mind in our brain. Also, writing requires a good planning activity.

The process of writing involves many activities, for examples:

  • Searching for ideas to write
  • Organising data
  • Checking and verifying the data
  • Revising the data
  • Presenting the data in a meaningful and systematic way

All of the above activities involve thinking processes. During these thinking processes, the neurons in our brain start to make new connections and/or adjusted their weight representing new knowledge.

Writing is difficult but, of course, can be learned by any people with practice and repetition. A good written article can articulate and convey meaning with good combinations of, for example, sentences and wording choices, and sentence cohesions. A good writing can pose a dynamic view of a subject by transforming information using the writer’s knowledge.

Writing is important

Here are some of the importance of writing (at least):

1. Writing is one of the two most important research tool. Only with writing, we can communicate the results of our research via, just to mention a few, journal articles, conference papers, magazines and newspapers.

2. Writing is a learning process (as mentioned above). During a writing process, we learn how to analyse, criticize, evaluate, combine and select information as well as draw conclusions from the information.

3. Writing is exposing our thinking to other people or audiences. With writing, we can convey what we think or information that we have to other people. Also, writing help us organising our mind when we want to communicate our thinking verbally.

Writing is a trained skill

Writing requires continuous practices and repetitions as well as individual commitments to keep practice (writing). We should practice writing since very young age, but, there is no excuse! We must start learning now (if we haven’t started yet) no mater how old we are.

It is always better to have something written with no structures (right) rather than starts from a blank sheet (left).

Some ways to learn writing:

  • Start to learn writing as early as possible
  • Write consistently within a specific duration of time every day
  • Select various topics to write (such as engineering, travelling, food, diary and other topics). With various topics of writing, we can avoid boredom in one topic and learn different styles of writing
  • Put the writing first and then we can revise the writing later on. When we have already something written, it will be easier to add on top of it rather than starting from a plain paper (see the figure above)
  • Involve in the community of people who like writing
  • Learn to write by both handwriting and typing
  • Learn grammar and sentence structures
  • Keep reading new materials so that we can write them in our own way
Do not leave writing at the last moment!

Writing is at least 25% of the total research time

If we have, let's say, a research period of one year from start to finish. Then, from this one year, we have to put at least full three months (or even sometimes more) only for writing, for example, a final technical report or a journal article.

Also, we should NOT put the three-month writing period at the last moment after finishing all experiments and data analyses. Instead, we must insert and distribute the three-month writing period within the total one-year research period. If we put the writing period at the last moment, we will have many data cluttered in our notes or other devices and will have difficulties to re-arrange and analyse them. Very often, we will miss and forget many important things that we should write in the report. Our brains will be overwhelmed with too much collected data and results to process.

So, do not think that we have one year to do experiments and data analyses. Instead, we have only nine months total for the experiments and data analyses. When we schedule a research project, we should reduce the total available time with the writing time to determine the research and data analyse timeline. So, be careful about our time estimation for the research.

To clarify the above information, an example of why we need to write at the same time (in parallel) or even before research activities. When we have a written paragraph of what we want to do in our experiment, this written paragraph will help our brain to prepare for the experiment and analysis as well as to expect certain results.

By preparing our brain, we will be more structured in doing the experiment and also, we prepare our brain to react when the results of the experiment are not as expected or do not follow theories. This situation significantly increases the productivity of our research.

We will use an example of a research activity to improve the baking softness of breads. Here, before experiments, we write what we want to do and the expected results (see below). With this writing ahead of the experiment, we prepare our brain in a systematic way to conduct the experiments. Also, our brain will be programmed to be ready to digest and process data during experiments and prepare to synthesise information received.

Writing before experiment:

The experiments will test several parameter settings of an oven so that baked breads will have the expected softness level. We will vary two parameters: baking temperature and time. We consider, based on baking theories, that these two parameters are the ones that significantly affect the softness of breads.

The temperature range is set to be between 60 to 120 degrees Celsius and the baking time is set to be between five to ten minutes. The temperature is not stable outside this range.

The temperature will be set with 10 degrees increments so that the temperature setting will be 60, 70, 80, 90, 100, 110 and 120 degrees. Meanwhile, the baking time will be set with a 2.5 minutes increment, that is the time will be 5,7.5 and 10 minutes.

We expect a linear relation between oven’s temperature and bread’s softness as well as the baking’s time and bread’s softness.

During the experiments, we found that the parameter temperature and baking time cannot explain all the variations of the data of the softness level, that is our model does not sufficiently explain the softness process of breads.

Also, we found that there is a shift in the curve plot. This shift in the plot is likely due to bias introduced in our experiment due to setting the experiment parameter sequentially instead of random.

Then, based on the experiment and the results, we revise the writing to be:

Writing after experiment:

The experiments will test several parameter settings of an oven so that baked breads will have the expected softness level. We will vary two parameters: baking temperature and time. We consider, based on baking theories, that these two parameters are the ones that significantly affect the softness of breads.

However, it seems there is another parameter that affects the softness level. Because, from data analyses, the variable temperature and time cannot fit well the plot of the values of the softness level. The $R^{2}$ value of the fitting is about $0.70$. This value means that the variable temperature and time can only explain $70$ percent of the total variation of the softness level, that is we still have $30$ percent gap in the model to explain the softness level process.

It could be that the type of flour used for the bread is the other major factor that affects the softness level of the baked breads. In the next experiment and data analysis, this flour material will be included as a factor.

The temperature range is set to be between 60 to 120 degrees Celsius and the baking time is set to be between five to ten minutes. However, all breads will be over-baked at a temperature above 100 degrees Celsius. Then, the range of the temperature parameter will be limited between 60 and 100 degrees Celsius. In addition, the effect of the temperature to the softness level is not very significant with the small change of temperature increment so that we can set the temperature setting with a larger interval than the previous setting: 60, 80 and 100 degrees Celsius.

From the experiment results and after regression analyses, there seems to be a bias in the plot of results. To minimise this bias, in the next experiment, these parameter settings will be randomised, that is the setting will be not sequentially increased during the next experiments.

We expect a linear relation between oven’s temperature and bread’s softness as well as baking’s time and bread’s softness. However, there seems to be also interaction between the temperature and baking time parameters. This interaction effect between the two parameters can be seen on the surface plot of softness level versus baking temperature and time is not a plain surface, instead, it is a curved or twisted surface. In the next data analyses, interaction effects will be included in the regression model.

From the writing after the experiment above, we can expand the writing after knowing the facts about the experiment results and data analyses. Since our brains is already prepared, we can think systematically to analyse why something is not as expected during the experiment and after data analyses. We can think deeper to understand the experiment procedures and results and can process new information. Based on this new information, we can make a new plan for the next experiment.

We will repeat these steps until we reach results that are considered enough to satisfy our research goal and for writing a report or a journal article for publication.

For the final writing after several experiments and writing cycles, we will then re-write the structure, select good wordings, add more elaboration based on our knowledge to make a good story and make meaningful conclusions for the report or journal article.

Conclusion

We should consider writing as part of research and not as an additional activity after finishing all research activities. So, here are things to remember:

  • Consider the period of writing time when scheduling research experiments and data analyses by reducing the total available research time with the writing time
  • Write a plan and expected results before conducting research, such as experiment or data analysis. Write at least a paragraph long for the plan and expected results
  • Adjust or revise and re-write the plan and results after conducting the research, for example revising experiment parameters, revising methods, revising models, revising data analyses and revising plots of data
  • Complete the writing from the previous adjusted/revised versions. This is an iterative process to re-read and improve the writing

Writing is a skill to learn for our entire lifetime!


We sell all the source files, EXE file, include and LIB files as well as documentation of ellipse fitting by using C/C++, Qt framework, Eigen and OpenCV libraries in this link.

We sell tutorials (containing PDF files, MATLAB scripts and CAD files) about 3D tolerance stack-up analysis based on statistical method (Monte-Carlo/MC Simulation).