F test calculator - for the equality of two variances, with solution https://www.statskingdom.com/220VarF2.

html

Statistics Kingdom
Reporting results in APA style
Results of the F-test for equality of variances indicated that there is a non-significant large difference between CHB without Anahaw Fibers (SD₁
= 3.7) and CHB with (2%) Anahaw Fibers (SD₂ = 10.6),
F(2,2) = 0.1, p = .215.

Parameter CHB without Anahaw Fibers CHB with (2%) Anahaw Fibers

Sample size (n) 3 3

Standard deviation (S) 3.6916 10.6405

Sample Variance (S²) 13.6278 113.2205

Normality p-value 0.8591 0.3074

Skewness 1.259 1.668

Skewness Shape Potentially Symmetrical (pval=0.304) Potentially Symmetrical (pval=0.173)

Excess kurtosis NaN NaN

Kurtosis Shape Potentially Mesokurtic, normal like tails (pval=NaN) Potentially Mesokurtic, normal like tails (pval=NaN)

Outliers

How to do with R?

P-value d SW1 P-value SW1 P-value Power

0 1 0 1 0 1 0 1 0 1

0.215 8.308 0.859 0.307 0.05

F-test for equality of variances-test, using F(df₁:2, df₂:2) distribution (two-tailed)[Validation]

1. H0 hypothesis
Since the p-value > α, H0 can not be rejected.
The CHB with (2%) Anahaw Fibers population's variance ratio is considered to be equal to the CHB without Anahaw Fibers population's variance
ratio.
In other words, the sample difference between the variance ratios of CHB with (2%) Anahaw Fibers and CHB without Anahaw Fibers is not big
enough to be statistically significant.
A non-significance result can not prove that H0 is correct, only that the null assumption can not be rejected.

2. P-value
The p-value equals 0.2149, ( P(F ≤ 0.1204) = 0.1074 ). It means that the chance of type I error, rejecting a correct H0, is too high: 0.2149 (21.49%).
The larger the p-value the more it supports H 0.

3. Test statistic
The test statistic F equals 0.1204, which is in the 95% region of acceptance: [0.02564, 39].
The S₁/S₂ (0.3469), is in the 95% region of acceptance: [0.1601, 6.245].
The 95% confidence interval of σ₁²/σ₂² is: [0.003086, 4.6942].

4. Effect size
The observed effect size F is large, 8.31. This indicates that the magnitude of the difference between the variances is large.

1 of 3 4/6/2024, 4:27 PM
F test calculator - for the equality of two variances, with solution https://www.statskingdom.com/220VarF2.html

Density
0.6F=0.12

H1(0.025) ──► H1(0.025)

0.4

0.2

0.026
0.0
0 2 4 6 8 10 12 14

Significance level (α) Show annotations

Calculation steps
S₁²
F(df₁,df₂) =
S₂²
df₁ = n₁ - 1 = 3 - 1 = 2. (Numerator)
df₂ = n₂ - 1 = 3 - 1 = 2. (Denominator)

3.6916²
F(2,2) = 10.6405²
F(2,2) = 0.1204
p = p(x ≤ 0.1204) = 0.1074
p-value = 2 * Min(p, 1 - p) = 2 * Min(0.1074, 0.8926) = 0.2149
p-value = 2 * Min(p, 1 - p) = 2 * Min(0.1074, 0.8926) = 0.2149

Max(S₁², S₂²) 113.2205

Effect Size = Min(S₁²,
= S₂²) 13.6278
=
8.3081

Variance Histogram

150 S² 3

100 2

50 1

0 0
CHB without Anahaw CHB with (2%) 10 20 30 40 50
Fibers Anahaw Fibers

Validation
The data doesn't meet all the F-test assumptions.

◉
Outliers

Outliers' detection method: Tukey Fence, k=1.5.

The data doesn't contain outliers.

◉
Normality

The assumption was checked based on the Shapiro-Wilk Test. (α=0.05)

It is assumed that CHB without Anahaw Fibers does follow the normal distribution (p-value is 0.8591), or more accurately, you can't
reject the normality assumption.
It is assumed that CHB with (2%) Anahaw Fibers following a normal distribution (p-value is 0.307), or more accurately, you can't reject
the normality assumption.

2 of 3 4/6/2024, 4:27 PM
F test calculator - for the equality of two variances, with solution https://www.statskingdom.com/220VarF2.html

◉
Test power

The priori power is low 0.05045, hence the test may not reject an incorrect h0

It is suggested to improve the test power by:

◦ sample size: use a larger sample.
◦ σ: check if the standard deviation can be reduced by eliminating noises that are not relevant to the tested measurement.
◦ effect size*: when planning the research it was possible to increase the required effect size, at the price of the ability to identify
smaller effect sizes.
◦ test tail: if only one of either the positive or negative changes is relevant, change to the one-tailed test.
◦ α*: when planning the research it was possible to increase the significance level (α), at the price of increasing the probability of a
type I error.
*Note: determining the test power, sample size, effect size and the significant level (α) should be done before collecting the data.

3 of 3 4/6/2024, 4:27 PM