Lecture 1, validity. Molendijk
What is mediation / a mediator?
An mechanism that causes change or that can explain an association that you observe.
Most studies investigate only changes in depression, but not changes in cognitions.
Cohen’s d is an effect size: it measures the strenghts of a certain effect. It is often cosidered to be
more inportant than the significance of a certain effect.
Experimental Clinical Psychology Lectures 1
,Formula: A difference between two means (mean 1 and mean 2) which is divided by the pooled
standaard deviation that accompanies the means.
For example: someone invents a treatment to increase intelligence. 5 people are tested: al started at
115 and after the treatment all score 130. The effectsize is a cohen’s d of 1 : (130-115):15 = 1
It is as well posible that someone started with a score of 115 and after treatment someone scores
107,5. The effectsize is a cohen’s d of -0,5 : (107,5-115):15 = -0,5
Serious science is not flawless.
Experimental Clinical Psychology Lectures 2
, Validity;the quality of being loggically or factually sound: soundness or cogency, a multi-dimensional
construct
- Internal validity: e.g., to what extent does the intervention [or what the researcher manipulates]
account for change … > Internal validity -- what was done inside the study: the integrity of the
experiment
- External validity: To what extent can the results be generalized / extended to other people, …,
settings … > External validity -- what happens outside the study [other labs, in reality: the
generalizability of the findings from the experiment
- Construct validity: What is the conceptual basis [i.e., construct] of the phenomenon under study … -
Data-evaluation validity: To what extent does the data + analyses demonstrate that what is said …
- Data-evaluation validity: To what extent does the data + analyses demonstrate that what is said.
They use a t-test, therefore the data has to be normally distribueted. In this example reaction times
have to be -4,9 (this is not possible).
They would never publish those three outliers in a paper.
Lack of internal validity (what do the researchers): Researches want to research whether criminals
have different personality traits (les empathic, more impulse etc.) than the normal population.
However there were no differences found, because everything was measured through self-reports
(Criminals are evaluated a lot of time and these are important for their free time etc). Criminals
present themselves as very normal.
Experimental Clinical Psychology Lectures 3
What is mediation / a mediator?
An mechanism that causes change or that can explain an association that you observe.
Most studies investigate only changes in depression, but not changes in cognitions.
Cohen’s d is an effect size: it measures the strenghts of a certain effect. It is often cosidered to be
more inportant than the significance of a certain effect.
Experimental Clinical Psychology Lectures 1
,Formula: A difference between two means (mean 1 and mean 2) which is divided by the pooled
standaard deviation that accompanies the means.
For example: someone invents a treatment to increase intelligence. 5 people are tested: al started at
115 and after the treatment all score 130. The effectsize is a cohen’s d of 1 : (130-115):15 = 1
It is as well posible that someone started with a score of 115 and after treatment someone scores
107,5. The effectsize is a cohen’s d of -0,5 : (107,5-115):15 = -0,5
Serious science is not flawless.
Experimental Clinical Psychology Lectures 2
, Validity;the quality of being loggically or factually sound: soundness or cogency, a multi-dimensional
construct
- Internal validity: e.g., to what extent does the intervention [or what the researcher manipulates]
account for change … > Internal validity -- what was done inside the study: the integrity of the
experiment
- External validity: To what extent can the results be generalized / extended to other people, …,
settings … > External validity -- what happens outside the study [other labs, in reality: the
generalizability of the findings from the experiment
- Construct validity: What is the conceptual basis [i.e., construct] of the phenomenon under study … -
Data-evaluation validity: To what extent does the data + analyses demonstrate that what is said …
- Data-evaluation validity: To what extent does the data + analyses demonstrate that what is said.
They use a t-test, therefore the data has to be normally distribueted. In this example reaction times
have to be -4,9 (this is not possible).
They would never publish those three outliers in a paper.
Lack of internal validity (what do the researchers): Researches want to research whether criminals
have different personality traits (les empathic, more impulse etc.) than the normal population.
However there were no differences found, because everything was measured through self-reports
(Criminals are evaluated a lot of time and these are important for their free time etc). Criminals
present themselves as very normal.
Experimental Clinical Psychology Lectures 3
