3577041
Descripción
Fichas por seonapalmer, actualizado hace más de 1 año
|
Creado por seonapalmer
hace alrededor de 9 años
|
|
| Pregunta | Respuesta |
| What is Mediation? | Mediation refers to a situation when the relationship between a predictor variable and an outcome variable can be explained by their relationship to a third variable (the mediator) |
| What is moderation? | Moderation occurs when the relationship between two variables depends on a third variable. So a moderator variable is one that affects the relationship/strength between the IV and DV. |
| Are we predicting a continuous or categorical outcome in mediation and moderation? | Continuous |
| Mediation and Moderation are forms of? | Multiple Regression |
| Moderation tests for? Mediation tests for? | Moderation - Interactions Mediation - Indirect effects |
| What is an Interaction in Multiple regression? | Interaction is the multiplication of two or more variables. This happens in Moderation. |
| What is one of the problems with interactions? | * RAW interactions correlate strongly with their source variables causing multicollinearity (Two variables are highly related to each other). *B weights for RAW interactions are often difficult to interpret. *WAY TO FIX IS CENTERING |
| What is Centering? | * When you take the mean away from each value. This transforms a variable into deviations around a fixed point (the mean). Centering Removes multicollinearity (removes correlation). * Centering also makes differences easier to interpret * Centering sets the mean value to 0 |
| If an interaction is significant what should you do? | If it's significant it must be interpreted first. You MUST interpret both variables simultaneously. (Simple slopes analysis) |
| Can skewed Data be centred? | No Skewed data cannot be centred, If data is skewed it must be cleaned first. Must remove outliers. |
| What does the non additivity of interacting variables refer to? | With non-additivity (Curved data), the two variables don't add up, they might increase linearly or don't necessarly increase linearly. Because they are multiplication they can adopt curvelinear functions. (Explain variance) so the effect of X on Y depends on the value of a third variable. (To check this you plot the interaction term against the DV. |
| What are simple slopes? | If you get a significant interaction term you need to check simple slopes. Is the interaction constant for all values of IV's/variables. So we work out the regression equations for the predictor and outcome and high, low and average levels of the moderator. |
| What are the requirements to run a Mediation? | * Must be a sig relationship between IV and DV (path c) *Must be a sig relationship between IV and MV (path a) *Must be a sig relationship between the MV and DV (path b) *SO PATH A<, B AND C MUST BE SIG |
| How do you choose which to use, Mediation or Moderation? | If you theoretically believe the variables add together you use Mediation, if you believe the variables interact (multiply together such as gender and age) you use moderation. |
| What are the three outcomes of Mediation? | *If path c' is still significant when path b is included in the model, the mediation is only partial. *If path c' is not significant while path a and b are significant, then there is full mediation. *If path a or b and c' are not significant, there is no mediation. |
| How many multiple regressions are undertaken in Mediation? | Three. |
| How do we test significance in Mediation? | b weight for path a multiplied by the b weight for path b divided by standard error. This gives us a Z score. (Sobel Test). |
| How do you interpret Mediation effect size? | * The ration of indirect to direct effect (shows how much variance is indirect compared to how much is direct). * Preacher and Kelly Kappa squared (measures 0 -1) closer to 1 is better. |
¿Quieres crear tus propias Fichas gratiscon GoConqr? Más información.