Daily research tips
One practical, general tip every day across question design, literature, methodology, statistics, writing, submission, and SCFHS documentation habits.
Previous tips
Keep raw data immutable. Do all cleaning and recoding in copies or scripts so any value can be traced back to its original source.
List plausible confounders from the literature at design time and make sure each one is collected. You cannot adjust for a variable you never recorded.
When measuring anything subjective, prefer an instrument already validated in your language and population over questions you wrote yourself. Developing a new scale is a separate project in its own right.
Give the primary outcome an explicit measurement window. "Mortality" is not an outcome; "all-cause mortality within 30 days of admission" is.
State how your sampling frame differs from your target population. Knowing who could never enter your sample tells you where your findings do not apply.
Log every protocol deviation on the day it happens, with the reason. A dated deviation log turns a weakness into transparent reporting.
Blind outcome assessment where you can, even in observational work. Keeping the person who classifies outcomes unaware of exposure status removes a common source of bias.
The full searchable archive is free with an account
Create a free account to search every published tip by keyword and filter by category.