Attenuations

Question 1

observed associations vs true associations

Answer 1

Always weaker than the true scores; measurement error, can’t be perfect.

Imperfect measurement weakens/attenuates observed associations

Question 2

Implications of connection reliability and observed correlation

Answer 2

1. Observed correlation is weak than true score
2. Amount of attentuation is determined by reliability, the worse the r the more attentuation
3. reliability/error constrains the max assocaition that could be found (Due to reliability, correlation at the very best (1) is used to calculate
4. It’s possible to estimate the true score correlation (correction for attenuation)

Question 3

context to use effect sizes

Answer 3

1. The correlation is usually used to represent the association between two continuous variables
2. Cohen’s d is usually used when examining association between a dichotomous and continous variable
3. N2 values is usually used when examining association betwe categorical variable and a variable w more than two levels.

Question 4

Item discrimination

Answer 4

Concept to evaluate internal consistency. How well an item differentiates between to scores. high values are desirable.

How well does the item predict the differences in the. construct

Question 5

Item-total correlation

Answer 5

Correlation between an item and the total test score. Represents degree to which differences among responses to an item are consistent to differences in total score;. High correlation is desirable.

High correlation means it predicts total score very well. BUT you correlate it with itself bc the item is in the total

Question 6

Corrected item-total correlation

Answer 6

Resolves problem of item total by correlating 1 item to all the rest scores. The higher, the better. It best reflects differences in the measured construct