Excel users who move to Google Sheets often look for the Data Analysis ToolPak. The usual answer is the XLMiner Analysis ToolPak add-on, but its Marketplace rating is low and many users report it not starting. The good news: most of what the ToolPak does can be done with built-in Sheets functions, no add-on needed.
Descriptive statistics
| Statistic | Formula (data in A2:A51) |
|---|---|
| Mean, median | =AVERAGE(A2:A51), =MEDIAN(A2:A51) |
| Sample standard deviation | =STDEV.S(A2:A51) |
| Standard error | =STDEV.S(A2:A51)/SQRT(COUNT(A2:A51)) |
| Skewness, kurtosis | =SKEW(A2:A51), =KURT(A2:A51) |
t-test
=T.TEST(A2:A21, B2:B21, 2, type) returns the two-tailed p-value. Use type 1 for paired samples, 2 for two samples with equal variance, 3 for unequal variance (Welch). Use tails = 1 for a one-tailed test.
F-test and z-test
=F.TEST(A2:A21, B2:B21) gives the two-tailed p-value for equal variances. =Z.TEST(A2:A51, 100) gives the one-tailed p-value that the mean is greater than 100 (when the population standard deviation is unknown, Sheets uses the sample's).
Correlation and regression
=CORREL(A2:A51, B2:B51) gives Pearson's r. For a full regression, use =LINEST(B2:B51, A2:A51, TRUE, TRUE): with the last argument TRUE it returns slopes and intercept, their standard errors, R², the standard error of the estimate, the F statistic, degrees of freedom and sums of squares, in a 5-row array. With several X columns, pass them as one range, e.g. A2:C51.
To get p-values for the coefficients: t = coefficient / standard error, then =T.DIST.2T(ABS(t), df).
One-way ANOVA with formulas
Sheets has no ANOVA function, but it is three steps. Suppose three groups are in columns A, B and C (rows 2–11):
- Between-groups sum of squares: for each group
=COUNT(A2:A11)*(AVERAGE(A2:A11)-AVERAGE($A$2:$C$11))^2, then add the three. - Within-groups sum of squares:
=DEVSQ(A2:A11)+DEVSQ(B2:B11)+DEVSQ(C2:C11). - F = (SSB / (k − 1)) / (SSW / (N − k)), and p =
=F.DIST.RT(F, k-1, N-k), where k is the number of groups and N the total count.
Chi-square test of independence
=CHISQ.TEST(observed_range, expected_range) returns the p-value. You must build the expected table yourself: row total × column total / grand total for each cell.
When formulas are not enough
Formulas are fine for one-off checks, but the ToolPak's value is the formatted output table. We are building StatPak for Sheets, an add-on that writes Excel-style output tables (descriptive statistics, t-tests, ANOVA, regression, Goal Seek, and a Solver for linear and integer problems) and is checked against scipy and statsmodels. It is not yet available on the Google Workspace Marketplace; this page will link to it when it is.
Sources
- Google Docs Editors Help – Google Sheets function list
- Google Workspace Marketplace – XLMiner Analysis ToolPak listing
Published 2026-10-02 by Karuna Labs. Our tools check file structure and checksums; always review outputs (and payment files in your bank's preview) before relying on them. This is general information, not financial, tax or legal advice.