http://aprendeconalf.

es

Statistical formulas
Formulas
Statistics Excel function Sample Freq. table Interpretation
∑
xi ∑ The value that best represent the values of the
Mean x̄ AVERAGE(sample) xi fi
n sample (except when there are outliers).
The value in the middle of the ordered sample.
Median M e MEDIAN(sample) 50% of values of the sample are above and 50%
below it.
Mode M o MODE(sample) The most common value in the sample.
Minimum M in MIN(sample) min{xi } The minimum value of the sample.
Maximum M ax MAX(sample) max{xi } The maximum value of the sample.
25% of the values of the sample are lower or equal
First quartile Q1 QUARTILE(sample,1)
to it.
50% of the values of the sample are lower or
Second quartile Q2 QUARTILE(sample,2)
equal to it.
75% of the values of the sample are lower or equal
Third quartile Q3 QUARTILE(sample,3)
to it.
i ∗ 10% of the values of the sample are lower or
Decile i Di PERCENTILE(sample, i/10)
equal to it.
i% of the values of the sample are lower or equal
Percentile i Pi PERCENTILE(sample, i/100)
to it.
Range MAX(sample)-MIN(sample) M ax − M in M ax − M in Measures the overall spread of the sample.
Measures the spread of the 50% central values of
Interquartile Range IQR QUARTILE(sample,3)-QUARTILE(sample,1) Q3 − Q1 Q3 − Q1
∑ the sample.
(xi − x̄)2 ∑ Measures the average spread with respect to the
Variance s2 VAR.P(sample) (xi − x̄)2 fi
n mean in square units.
√ √ Measures the average spread with respect to the
Standard deviation s STDEV.P(sample) s2 s2
mean in the units of the variable.
Measures the relative spread with respect to the
s s
Coef. variation cv STDEV.P(sample)/ABS(AVERAGE(SAMPLE)) mean. It has no units. The lower the dispersion,
|x̄| |x̄|
the more representative is the mean.
∑ ( xi − x̄ )3 ∑ ( xi − x̄ )3 Measures the asymmetry of the sample distribu-
Coef. skewness g1 SKEW(sample) fi tion (g1 = 0 symmetry, g1 > 0 right-skewed, g1 < 0
s s
left-skewed).
Measures the peakness or flatness of the sample
∑ ( xi − x̄ )4 ∑ ( xi − x̄ )4 distribution compared to a normal distribution
Coef. kurtosis g2 KURT(sample) −3 fi − 3 (g2 = 0 normal kurtosis, g1 > 0 leptokurtic or
s s
peaked distribution, g1 < 0 platykurtic or flat dis-
tribution).