You know that moment when you're staring at a dataset in Stata and just need the average — but the syntax won't come to you? Here's the thing — yeah. We've all been there Worth knowing..
Finding the mean in Stata sounds basic. And it is. But "basic" doesn't mean obvious when you're new, or when your data has weird labels, missing values, or groups you forgot to account for Simple, but easy to overlook..
Here's the thing — most people Google this, copy one line of code, and move on. Because of that, that works until it doesn't. So let's actually talk about how to find mean in Stata without tripping over the stuff nobody warns you about That alone is useful..
What Is Finding the Mean in Stata
At its core, it's just asking Stata to compute the arithmetic average of a variable. You've got numbers in a column. Stata adds them up, divides by how many there are, and spits out a result Surprisingly effective..
But in practice, "the mean" can mean different things depending on your data. Even so, the average per group? The mean after dropping outliers? Are you looking at one variable across everyone? Stata can do all of it — the trick is telling it exactly what you want.
The Variable Has to Exist
Sounds dumb, right? Day to day, if your variable is called income and you type icome, Stata will shrug and say "not found. But a surprising number of errors come from typos or wrong label names. " Always check your variable names with describe or just ds before you start Worth keeping that in mind..
Mean vs Other Averages
People mix these up. Also, mean is the sum divided by count. Not the median (middle value), not the mode (most common). Stata keeps these separate for a reason. If you ask for summarize you get the mean. If you wanted the median, that's a different conversation It's one of those things that adds up..
Why It Matters
Why care about getting this exactly right? Because the mean quietly drives a lot of decisions. Policy briefs, business reports, academic papers — they all lean on averages That's the part that actually makes a difference. Took long enough..
And here's what goes wrong when people rush it: they compute a mean that includes missing values incorrectly, or they average across groups that shouldn't be combined. Which means i once saw a grad student report the mean income of a mixed sample that included zeros from people who didn't answer. The number was technically "right" and completely misleading Took long enough..
Real talk — a wrong mean isn't always a crashed program. On the flip side, off. Sometimes it's a confident paragraph in a report that's just... Knowing how Stata handles this saves you from that Easy to understand, harder to ignore..
How to Find Mean in Stata
Alright, the meaty part. Let's walk through the actual ways to do it, from dead simple to "okay now we're getting useful."
The Easiest Way: summarize
Open your dataset. Type:
summarize income
Boom. Stata returns the mean, along with obs, std dev, min, max. That's your mean right there in the "Mean" column.
Short version: su income does the same thing. Old Stata users love the abbreviations.
Getting Just the Mean
Don't want the whole table? Use:
tabstat income, statistics(mean)
This prints just the average. Handy when you're building output or just want less noise on screen.
Mean by Group
This is where it gets good. Say you want average income by gender. Try:
by gender, sort: summarize income
Or the cleaner version many prefer:
tabulate gender, summarize(income)
That gives you the mean income for each gender category. Turns out this is the version most people actually needed but didn't know to ask for.
Using collapse to Save the Mean
If you want to create a new dataset of group means (instead of just viewing them), collapse is your friend:
collapse (mean) income, by(gender)
Now your dataset is just the averages. Dangerous if you forget to save the old one first — but powerful for making charts.
Means with Conditions
Only want the mean for people over 30?
summarize income if age > 30
Or combine conditions:
summarize income if age > 30 & gender == 1
Stata handles these if statements cleanly. Just remember the double equals for "is equal to."
Missing Values and How Stata Treats Them
Here's what most people miss: Stata excludes missing values (.Consider this: ) from mean calculations by default. Now, that's usually what you want. But if you've coded "refused" as 999 instead of missing, Stata will happily include 999 in your average. So always check your missing-value coding before trusting a mean Most people skip this — try not to..
Storing the Mean as a Macro
Want to use the mean later in your do-file? After summarize, the result is stored in r(mean) The details matter here..
summarize income
display r(mean)
Or save it:
local avg_income = r(mean)
That's how you build loops or labels without hardcoding numbers.
Common Mistakes
Honestly, this is the part most guides get wrong — they assume you'll only ever use clean data Most people skip this — try not to..
One big mistake: running summarize on a string variable. Stata will tell you it can't. But people waste time confused why "income" won't average when it's actually stored as text. Check with describe The details matter here..
Another: forgetting by needs the data sorted. Because of that, if you write by gender: summarize income without sorting, Stata errors out. The bysort command does both at once — use that instead.
And the classic — averaging before cleaning. Consider this: you imported from Excel, there's a typo row with "N/A" converted to a weird number, and your mean is garbage. But always browse your data first. Five seconds of looking saves an hour of confusion Surprisingly effective..
Honestly, this part trips people up more than it should.
Look, I know it sounds simple — but it's easy to miss that Stata's mean command (yes, there's a command literally called mean) does something slightly different from summarize. Try:
mean income
It gives you the mean plus standard error and confidence interval. Useful for surveys. Different from summarize. People mix them up constantly.
Practical Tips
What actually works when you do this day to day?
First, get in the habit of typing codebook income before averaging. On top of that, it shows you value labels, missing, and range. You'll catch problems early.
Second, use estpost and esttab if you're writing papers. Now, you can output group means into a clean table without copy-pasting from the screen. Worth learning if you do this often.
Third, if your data is weighted (like survey data), use svy: mean income instead of plain summarize. Regular commands ignore weights. Your mean will be wrong and you won't know why.
And here's a small one: name your variables clearly. inc vs income vs hh_income — pick one style. Future you will thank past you when you're typing summarize at midnight The details matter here..
FAQ
How do I find the mean of multiple variables at once in Stata?
Just list them: summarize income age education. Stata prints a table with the mean for each. Or use tabstat income age, statistics(mean) for a tighter view Small thing, real impact. Which is the point..
Can Stata show the mean as a percentage?
If your variable is already coded 0/1, the mean is the proportion. Multiply by 100 if you want percent: display r(mean)*100 after summarizing.
Why is my mean showing as a weird decimal?
Stata displays full precision. Use format to control decimals, like format income %9.2f before summarizing, or just round mentally. It's not an error Easy to understand, harder to ignore..
How do I get the mean excluding outliers?
Use summarize income if income < 100000 or winsorize first. Stata won't auto-drop outliers — that's on you.
Is there a GUI way to find the mean in Stata? Yes. Click Statistics > Summaries > Summary statistics. Pick your variable. But learning the command is faster long-term And that's really what it comes down to..
Anyway, that's the real story on finding the mean in Stata. It's a small task that hides a few sharp edges — missing codes, grouped data, weights, wrong
commands — and knowing which tool to reach for keeps your numbers honest. Check your data, match the command to the context, and don't trust a number you haven't inspected. Worth adding: the takeaway isn't just "type summarize"; it's that the mean you report depends entirely on the assumptions you silently accept when you run it. Do that, and Stata stops being a black box and starts being what it should be: a microscope for your questions.
