click below
click below
Normal Size Small Size show me how
Res Mtds - 2-2 (inc)
Question | Answer |
---|---|
What is a good way to define internal validity when compared to experiments only? | You can also think of internal validity, at least for experiments, as the extent to which there are no confounds or the extent to which nothing is confounded with the independent variable. |
What does it mean for two variables to be changing in parallel? | It means that they are associated in some way, but, for present purposes, you can simplify this to mean that they are correlated, since the correlation coefficient is our favorite way (in psychology) of quantifying associations. |
What is the Experimental Control Hierarchy? | Experimental control is the main way that experimenters avoid and eliminate confounds. Thus, a hierarchy of methods. |
What must you do before giving the levels of hierarchy in an experiment? | You must apply the rules of experimental control to each extraneous variable separately, using the highest measure in the hierarchy when possible. |
What is at the top of the Experimental Control Hierarchy? | Hold it constant; that is, you don’t have to allow extraneous variables to change at all, as it is the same value for every subject in every condition. |
What is the second level of the Experimental Control Hierarchy? | Allow the extraneous variable to vary across subjects in the experiment, but make sure that the mean value is the same for every condition; a good name for this tier of the hierarchy is “equalize on average.” |
What is the third level of the Experimental Control Hierarchy? | The third-best approach to dealing with an extraneous variable is to measure it and, if it turns out to be correlated with the independent variable, somehow remove its effect on the data. This is done through covariance analysis. |
What is the measure-and-remove method? | Occurs on the 3rd level of the ECH and through covariance analysis. As long as the correlation between the confound and the indpt. variable is closer to zero than ±.70, you can subtract out the confound's effect and see if there’s anything left. |
What is the 4th level of the Experimental Control Hierarchy? | Manipulate the variable that isn’t of interest while holding the variable of interest constant. You subtract the effect of the confound from the control experiment from the results of the main experiment & the result is the effect of variable of interest. |
What is an important rule regarding independent variables? | They must be completely under the control of the experiment. |
What is an important rule regarding dependent variables? | They should be a labile (& non-qualitative) measured variable. |
What is an important rule regarding extraneous variable(s)? | They are anything other than the IV that could influence the DV. |
What is the difference between the independent variable, and the levels of the independent variable? | The IV is the variable; the entire range of values, whereas the levels of the IV are specific values; those used we rarely include all possible values of the IV in a single experiment. |
What must something do in order to be considered a confound? | The extraneous variables must change in parallel with the independent variables; that is, they must be correlated. |
What must a confound be in order to be considered a problem? | The extraneous variable must be capable of influencing the dependent variable. |
What is experimental control? (NOT the Experimental Control Hierarchy) | When experimenters hold everything other than the IV constant across the conditions of an experiment. This is the best way of getting high internal validity. If an EV is constant, it can’t be a confound, so it can’t be a way of explaining the results. |
What are the 4 levels of the Experimental Control Hierarchy? (No definition needed) | 1. “hold it constant” 2. “equalize on average” 3. “measure and remove” 4. run a “control experiment” |