As part of my job, I coach and guide student consultants to work on interdisciplinary projects. From experience, I have found most consultants have strong technical skills to perform an analysis but lack various soft skills. One of them is communication - written and verbal. Today, I will focus on some common mistakes in their written work.
The consultants are always eager to demonstrate everything. They write about all the steps taken to clean and analyze the data. While we appreciate the effort put into the analysis, the codes and intermediate steps can be confusing and intimidating for those with minimal statistics training.
On the other hand, some consultants do not share any information about their modelling choices. They output many tables and graphs without explaining their lines of thought. During the meeting to discuss the results, the consultant may need to scroll or execute many lines of code to review their reasonings.
Understanding what to include in a report is essential. We need to know the readers, their backgrounds, and goals. Based on experience, I like to draft the report to:
- Remind the readers about the data
- When was the data received?
- Did we alter the data in any way?
- Did we incorporate other data or information based on previous email exchanges/meetings?
- Briefly mention the modelling choices.
- What model did we choose?
- Why did we choose this model?
- Did we use tuning parameters or cut-off values different from the literature? Why?
- Present the result in the context of the topic.
- What variables were significant/insignificant?
- Avoid long sentences and complex jargon.
- Let the experts explain beyond the context of the statistical significance.
Other little things can also improve the readability of a report. I will leave it for next time.
