ARMS studiepad specifiek deel: experimentele psychologie
Practical test 1: Know Your Data
Alle soorten analyses
&
Aflezen van de output
Auteurs: Kaylee Lauffer & Merel Boonstra
, Opdracht 1 (1 conditie in 2 groepen -> between subjects design)
Researchers have discovered a new study pill, which they claim can help people
learn faster. Before this ‘smart pill’ can be brought on the market, and even
before it can be tested on students, they would like to have a proof of principle.
To this end, they examine how quickly rats learn to find their way through a
maze. The researchers compare the speed at which a group of rats to
which the active ingredient has been administered are able to find the
exit to that of a group of rats that have received a placebo.
Let op bij between design in spss: grouping variabele maken
spss: eerst assumpties testen ->
> Descriptive statistics -> explore en dan bij plots ‘histogram’ en ‘normality
plots with tests’ aanklikken.
> Dan kijken naar de smirnov p-waarde wanneer sample >2000,
> Kijken naar shapiro-wilk wanneer sample <2000.
> Wanneer de p-waarde significant is, is er niet voldaan aan de voorwaarde van
normale verdeling.
Wat te doen wanneer het dus niet normaal verdeeld is? Geen t-test mag je dan
doen, dus dan moet je: non-parametrisch toetsen
non-parametrisch toetsen:
> Analyze > nonparametric tests > legacy dialogs > 2 independent samples >
waarden invullen > define groups > Mann-whitney
u test aanklikken & exact > exact
aflezen:
> Verschil in medianen (bereken in excel)
Mann-Whitney U test:
Z-waarde = -2.025
P-waarde = 0.043
Effect-size (Z / wortel N) = 2.025 / wortel 42
Practical test 1: Know Your Data
Alle soorten analyses
&
Aflezen van de output
Auteurs: Kaylee Lauffer & Merel Boonstra
, Opdracht 1 (1 conditie in 2 groepen -> between subjects design)
Researchers have discovered a new study pill, which they claim can help people
learn faster. Before this ‘smart pill’ can be brought on the market, and even
before it can be tested on students, they would like to have a proof of principle.
To this end, they examine how quickly rats learn to find their way through a
maze. The researchers compare the speed at which a group of rats to
which the active ingredient has been administered are able to find the
exit to that of a group of rats that have received a placebo.
Let op bij between design in spss: grouping variabele maken
spss: eerst assumpties testen ->
> Descriptive statistics -> explore en dan bij plots ‘histogram’ en ‘normality
plots with tests’ aanklikken.
> Dan kijken naar de smirnov p-waarde wanneer sample >2000,
> Kijken naar shapiro-wilk wanneer sample <2000.
> Wanneer de p-waarde significant is, is er niet voldaan aan de voorwaarde van
normale verdeling.
Wat te doen wanneer het dus niet normaal verdeeld is? Geen t-test mag je dan
doen, dus dan moet je: non-parametrisch toetsen
non-parametrisch toetsen:
> Analyze > nonparametric tests > legacy dialogs > 2 independent samples >
waarden invullen > define groups > Mann-whitney
u test aanklikken & exact > exact
aflezen:
> Verschil in medianen (bereken in excel)
Mann-Whitney U test:
Z-waarde = -2.025
P-waarde = 0.043
Effect-size (Z / wortel N) = 2.025 / wortel 42
