Study and sheet notes

De opdracht voor de werkgroep bevatte de volgende vragen. 

1. Describe the terms of reliabilty and validity. 

• Reliability = the extent to which a measurement is free from random measurement errors. This means that the scores are independent of time, place and environment.
• Construct validity = the extent to which the instrument (operationalization) succeeds in measuring the construct and thus fits the conceptual definition.
• Internal validity = the extent to which the research method can eliminate alternative explanations for an effect/relationship.
• External validity = the extent to which the research results can be generalized to other populations, settings and times.
• Statistical validity = the extent to which the results of a statistical analysis are accurate and well- founded.

2. What are the conditions for being able to speak of a causal relationship?

• Two or more variables should be related
• The cause should precede the outcome over time
• No other explanations may be possible

3. How can a sample be used to to make inferences about a population?
Inferential statistics use probability theory to formulate general statements based on samples about the populations from which these samples were taken.

4. What are the assumptions of multiple regression?

• Independent observations
• Dependent variable should be continuous.
• Independent variable should be continuous or dichotomous. A categorical predictor can be changed into several dichotomous variables (dummy’s).
• Linear relationship between predictors and dependent variable.
• Absence of outliers (in X, Y and XY-space)
• Absence of multicollinearity
• Homoscedasticity
• Normality distributed residuals

In de rest van de werkgroep is het volgende artikel besproken:

Kanazawa, S. (2012). Intelligence, birth order, and family size. Personality and Social Psychology Bulletin, 38(9), 1157-1164. https://doi.org/10.1177/0146167212445911

6. Describe the three different explanation sgiven in the introduction for the relationship found between intelligence and birth order. Note. Do not discuss the revised confluence model. 

1. Confluence model: The average intelligence of children decreases with an increasing birth order. The higher the birth order, the less a family is a cognitively stimulating environment.
2. Resource dilution model: The average intelligence of children decreases with an increasing birth order. This is because parents have only a limited number of resources, energy and attention to distribute. The higher the birth order, the less resources the family still has available.
3. Admixture hypothesis: The relationship between intelligence and birth order is a methodological artefact. The relationship between intelligence and birth order is only found because parents with lower intelligence often have more children.

7. What are the population and sample in the article? Describe the participants that are participating. In your description also explain exactly what the different sweeps mean conceptually.

A ‘population’ of British respondents is used. Participants were all babies born in Great Britain on March 3-9, 1958. These participants were measured over a longer period of time, during which data were collected at various moments (sweeps). This study uses data collected on sweep 0 (age 0), sweep 1 (age 7), sweep 2 (age 11), and sweep 3 (age 16).

Population: Kanazawa does not define anywhere what group he wants to generalize the results of the research to.

The conclusion seems to refer to a situation that is not limited to time or place, thus generalising to all people that live, will live, or have lived. A more logical view is Western people around the time of the research, but Kanazawa never explicitly mentions this.

Sample: all babies born in Great Britain from 3-9 March, 1958

8a. In Table 1 you can see that a hierarchical regression analysis has been performed. You can also see the results of different models. What can you look at to decide which model is the best fitting model?

Usually you look at the explained variance. In this case, model 4 has the most explained variance, which makes sense, because it contains most variables. It is unfortunate that we do not see whether the change in explained variance is significant. Although you can imagine that with such a large sample size, any difference will be significant. You can ask yourself whether adding 2 variables (the difference between model 3 and 4) makes adding those variables worthwhile, because the explained variance increases by only 1%.

b. When do you use standardized, and when do you use unstandardized regression coefficients?

The non-standardized regression coefficient (B) is used to predict a score on the dependent variable.
The standardized regression coefficient (β) is used to compare which predictor is the most important.