4.1 Two Sample test
Key question of the day: I know how to test for a single parameter, but what if I have two
samples and two parameters and want to compare?
Example:
To know whether young people are less likely to buy your product than older people.
What is the issue? Example 1:
- You measured sales in 65 stores in prima A locations: 140k per store per month.
- You measured sales in 70 of your stores in B locations: 135k per store per month.
You get asked:
- Are sales in prima A locations higher than in B locations?
- Do the differences warrant the additional costs of these locations?
What is the issue: Example 2:
- You measured sales in your 125 stores, average sales is 130k.
You did a marketing campaign -> average is now 131K
- You get asked -> Did sales go up, was this due to the campaign, wat were the
costs of the campaign.
Short recap of what we did previously:
,Comparing two means: two rare u’s
Examle from peter dalley: https://www.youtube.com/watch?v=nGPCI-
dFZaM&list=PLd605q1Prvz-oZckPw6XxBl1dH_QU0TCz&index=2
a-> de ho: m1-m2 kleiner dan gedeelte-> WORDT IN OEFEN EXAMEN GEVRAAGD
b-> 2.11
c-> 0.0174 -> na z formule
D-> 0.05 =1.645, ook reject
E-> zie grafiek
Excersize 2 from peter dalley:
,For answer a-> See upright corner
B-> 1.15
C-> P value 0.1251 not reject cause bigger than 0.05 (get from stat tables)
D-> critical value approach-> Z alpha = 1.645. 1.15 is to the left, left skewed so not
reject.
Lower tailed test -> m1-m2 < 0
Upper tailed test -> m1-m2 > 0
, Comparing two means:
Key question of the day: I know how to test for a single parameter, but what if I have two
samples and two parameters and want to compare?
Example:
To know whether young people are less likely to buy your product than older people.
What is the issue? Example 1:
- You measured sales in 65 stores in prima A locations: 140k per store per month.
- You measured sales in 70 of your stores in B locations: 135k per store per month.
You get asked:
- Are sales in prima A locations higher than in B locations?
- Do the differences warrant the additional costs of these locations?
What is the issue: Example 2:
- You measured sales in your 125 stores, average sales is 130k.
You did a marketing campaign -> average is now 131K
- You get asked -> Did sales go up, was this due to the campaign, wat were the
costs of the campaign.
Short recap of what we did previously:
,Comparing two means: two rare u’s
Examle from peter dalley: https://www.youtube.com/watch?v=nGPCI-
dFZaM&list=PLd605q1Prvz-oZckPw6XxBl1dH_QU0TCz&index=2
a-> de ho: m1-m2 kleiner dan gedeelte-> WORDT IN OEFEN EXAMEN GEVRAAGD
b-> 2.11
c-> 0.0174 -> na z formule
D-> 0.05 =1.645, ook reject
E-> zie grafiek
Excersize 2 from peter dalley:
,For answer a-> See upright corner
B-> 1.15
C-> P value 0.1251 not reject cause bigger than 0.05 (get from stat tables)
D-> critical value approach-> Z alpha = 1.645. 1.15 is to the left, left skewed so not
reject.
Lower tailed test -> m1-m2 < 0
Upper tailed test -> m1-m2 > 0
, Comparing two means:
