CASE 2: EFFECTIVENESS
Eline Meuleman
DATA ANALYSE Radboud Universiteit
,Exercise 1
The power depends on:
• The significance level the smaller the significance level, the smaller the power
• The effect size the higher the effect size, the greater the power
• Sample size the larger the N, the greater the power
You want the power to be high, because you want the chance of a Type 2 error to be as
small as possible. This is the wrong decision that is made when a test does not reject a false
null hypothesis. In other words, someone is ill, but is not diagnosed as ill.
The researcher controls the level of significance and the effect size, but not the sample size.
, Exercise 3
Cohen argues for as few variables as possible. That is why you should not divide the IQ score,
because then it will become three variables instead of one again. You can see in sudden and
on the basis of the averages what the direction of the effect is (ie: whether a low or high
intelligence
correlates with language skills and social emotional functioning).
The design
Dependent variable 1: language skills (quantitative)
Dependent variable 2: social-emotional functioning (quantitative)
Between-subject variable: treatment (qualitative: treatment / no treatment)
Between-subject variable: patient group (qualitative: disorder A / disorder B / disorder C)
Research question: Is the effectiveness of the treatment (BS) different for the patient groups
(BS)?
Effectiviteit The effectiveness is then assessed by progress in language skills and / or social
emotional functioning.
EVT additionally, but not associated with the design:
Factor: gender (male / female)
Covariat: age (quantitative)
Covariat: intelligence score (quantitative)
Data example
Create rows
Generate data: Compute Variable Random numbers
1. Data Compute Variabele
Taalvaardigheid & Sociaal emotioneel functioneren
Behandeling (wel/niet)
1 = wel behandeling, 0 = geen behandeling ( yes treatment, no treatment)
Eline Meuleman
DATA ANALYSE Radboud Universiteit
,Exercise 1
The power depends on:
• The significance level the smaller the significance level, the smaller the power
• The effect size the higher the effect size, the greater the power
• Sample size the larger the N, the greater the power
You want the power to be high, because you want the chance of a Type 2 error to be as
small as possible. This is the wrong decision that is made when a test does not reject a false
null hypothesis. In other words, someone is ill, but is not diagnosed as ill.
The researcher controls the level of significance and the effect size, but not the sample size.
, Exercise 3
Cohen argues for as few variables as possible. That is why you should not divide the IQ score,
because then it will become three variables instead of one again. You can see in sudden and
on the basis of the averages what the direction of the effect is (ie: whether a low or high
intelligence
correlates with language skills and social emotional functioning).
The design
Dependent variable 1: language skills (quantitative)
Dependent variable 2: social-emotional functioning (quantitative)
Between-subject variable: treatment (qualitative: treatment / no treatment)
Between-subject variable: patient group (qualitative: disorder A / disorder B / disorder C)
Research question: Is the effectiveness of the treatment (BS) different for the patient groups
(BS)?
Effectiviteit The effectiveness is then assessed by progress in language skills and / or social
emotional functioning.
EVT additionally, but not associated with the design:
Factor: gender (male / female)
Covariat: age (quantitative)
Covariat: intelligence score (quantitative)
Data example
Create rows
Generate data: Compute Variable Random numbers
1. Data Compute Variabele
Taalvaardigheid & Sociaal emotioneel functioneren
Behandeling (wel/niet)
1 = wel behandeling, 0 = geen behandeling ( yes treatment, no treatment)
