Hypothesis tests for the difference between two means

Hypothesis tests for the difference between two means Flashcards

Spec 3.7–8 · Edexcel International A Level Further Maths: Flashcards

Studying this course? Keep your progress in one place.

Flip each card. Be honest: the ones you mark Not yet come back first.

14 cards

Tap a card to flip it. Swipe right if you knew it, left if not.

See all 14 cards
Hypotheses for testing whether two means are equal?
H0:μx=μyH_0:\mu_x=\mu_y against H1:μx≠μyH_1:\mu_x\neq\mu_y (or >> or << for a one-tailed test).
Distribution of Xˉ−Yˉ\bar X-\bar Y for independent Normal samples?
N(μx−μy, σx2nx+σy2ny)N\left(\mu_x-\mu_y,\ \frac{\sigma_x^2}{n_x}+\frac{\sigma_y^2}{n_y}\right).
Do the variances of Xˉ\bar X and Yˉ\bar Y add or subtract?
They add, even though the means are subtracted.
Test statistic for two means with variances known?
Z=Xˉ−Yˉσx2/nx+σy2/nyZ=\frac{\bar X-\bar Y}{\sqrt{\sigma_x^2/n_x+\sigma_y^2/n_y}} under H0H_0.
Standard error of Xˉ−Yˉ\bar X-\bar Y?
σx2nx+σy2ny\sqrt{\frac{\sigma_x^2}{n_x}+\frac{\sigma_y^2}{n_y}}.
What if the variances are unknown and both samples are large?
Use sx2s_x^2 and sy2s_y^2 in place of σx2\sigma_x^2 and σy2\sigma_y^2; Z≈N(0,1)Z\approx N(0,1).
What justifies the Normal approximation for large samples?
The Central Limit Theorem, and S2S^2 being close to σ2\sigma^2 for large nn.
One-tailed 5%5\% critical value for zz?
1.6451.645 (or −1.645-1.645 in the lower tail).
Two-tailed 5%5\% critical values?
±1.96\pm1.96.
One-tailed 1%1\% critical value?
2.3262.326 (or −2.326-2.326 in the lower tail).
Why must the samples be independent?
The variance formula σx2nx+σy2ny\frac{\sigma_x^2}{n_x}+\frac{\sigma_y^2}{n_y} only applies to independent samples.
A test shows μA<μB\mu_A<\mu_B. Does it show every AA value is smaller?
No. It concerns population means, not individuals.
How do you word a conclusion when H0H_0 is not rejected?
'There is insufficient evidence at the ...% level that ...', never 'H0 is true'.
Which order of subtraction fits H1:μA<μBH_1:\mu_A<\mu_B?
Use xˉA−xˉB\bar x_A-\bar x_B, which gives a negative zz in the rejection region.

Flashcards