Quiz Regression points Flashcards
What is df for the regression?
What happens to df with dummy coding?
One categorical variable vs more IVs?
- The degrees of freedom from the regression tell us how many independent variables we have in the regression equation
- This can be one dummy coded variable), not about the levels of the IVs.
- Although, given that we only have one categorical variable, in this example, we could infer this information. In general, however, with more IVs, such inferences are incorrect.
The correlations between Age and Years of Experience (n = 3000) are not adequate as we forgot to exclude the 89 cases for which we have no hourly salary information (n = 2911).
Are the correlations between two variables where we have no info for some people in one of these IVs totally inadequate?
This statement is indeed false. EACH bivariate correlation within the table would be calculated between two variables only. Other correlations within this table could be based on smaller or larger numbers of people.
The actual estimates of the correlations between two variables are not affected by the differences in the n’s, but if these differences are large, the results of the regression analysis are likely to be affected
where in a frequency table do you find the percentage of that sample overall that fall into a specific group?
cumulative percent column
df for regression is?
The degrees of freedom for the Regression line indicates the number of predictors.
If you have an inisignificant result in a sequential regression is it unimportant as a predictor of the DV?
While the IV2 is not a significant predictor of the number of DV over and above IV1 (when IV1 is included in the model), it is possible that the IV2 is a significant predictor of the DV in a simple linear regression, or while controlling for OTHER variables apart from age.
If you have coding of 1 and 2 does this cancel out the possobility that someone can score a 0?
In the Descriptives table, the minimum value is 1, but this is just a nominal value. The frequency tables will clearly show that 4 people have 0 drinks per week.
if we have gender as 1 of the ivs and faculty (three levels) as the other IV and gender is not sig, is it true that gender is not a sig predictor?
We cannot tell for sure whether there is a significant difference between the genders in starting salary from this output. Even though gender is not a statistically significant predictor in this model, we are controlling for the faculty in which they are enrolled.
can we use anova to test starting salary for graduate students based on the faculty (three levels)
Yep. Since both IVs are categorical, ANOVA is an appropriate analysis.
can we interpret in real world a constant of 0 for age with an IQ score for recall?
This statement is indeed false. While it’s interpretable—it’s the predicted recall for someone with age and IQ of 0—this does not refer to anyone in the real world.
Say we have age and faculty and age is dummy coded, faculty is not, do we code it or leave labels?
Code it. Categorical variables like faculty mean random large numbers will appear when we really want to know which group they are in. The Faculty variable is categorical with three levels. Thus, two dummy coded variables are required.
if age is sig in the firts and second box of a sequential regression what do we conclude (even though it’s not entered last)
From the coefficients table, age is a significant predictor of the DV in both models, but we are particularly interested in Model 2 (age is not coded last but it is still sig in both models (Rsqaure change is not sig.)
