What is Logistic Regression?
Logistic regression is a statistical method that models the probability of a binary outcome (yes/no) from one or more predictor variables. It outputs a probability (0–1) and is commonly used to predict categories like respond/didn't respond or agree/disagree.
Logistic regression estimates how predictor variables (numeric or categorical) relate to the likelihood of a specific outcome. Unlike linear regression, which predicts a continuous number, logistic regression uses the logistic (sigmoid) function to map predictions to probabilities between 0 and 1. Coefficients are usually reported as odds ratios: a coefficient tells you how the odds of the outcome change when a predictor increases by one unit, holding other variables constant. The method requires reasonable sample size and assumes a linear relationship between predictors and the log-odds of the outcome; it is robust and widely used for classification and explanation in non-expert settings.
Usage example
Use logistic regression to predict the probability that a respondent completes a survey based on their chosen language, device type (mobile/desktop), and number of questions — e.g., English speakers on mobile with long surveys may have lower completion probabilities.
Practical application
For Hearo users, logistic regression helps identify which factors reliably increase or decrease the chance of outcomes you care about (survey completion, consent given, positive response). That lets teams prioritize changes (shorter surveys, targeted reminders in specific languages, mobile-friendly design), allocate outreach resources, and measure whether changes actually improve participation across different language communities.
FAQ
Can logistic regression handle more than two outcomes (e.g., multiple response categories)?
Standard (binary) logistic regression predicts two outcomes. For more than two categories you can use multinomial logistic regression or ordered logistic regression (if categories have a natural order).
Do logistic regression results prove cause and effect?
No — logistic regression shows associations in your data, not causation. To claim causality you need experimental design or strong controls for confounding factors; otherwise treat coefficients as indicators of relationships to investigate further.
What do I do if I have many predictors or small sample sizes?
With many predictors relative to your sample size, consider regularization (like L1/L2), feature selection, or combining categories to reduce variables. Also ensure you have enough events (outcomes) per predictor — otherwise coefficients can be unstable.