Warning:
JavaScript is turned OFF. None of the links on this page will work until it is reactivated.
If you need help turning JavaScript On, click here.
The Concept Map you are trying to access has information related to:
missing data, reinforces the relationshiops present in the samplele data, which then become less generneralizable, out of bounds correlations, more accurate estimate of mising values, not feasible if a variable w missing observationsdoes not covary at least moderately w other variables in the data set , ????, random imputation (sequential hot deck imputation) (little & rubin, 1987) , scores from regressions fit together better than they should because the estimates have been based on teh other variables and are likley to be more consistetn with them than actual scores would be (tabachnick & fidell, 1996), complex, 2 groups, (not) missing - difference? (hair et al, 1995) , does not systematically affect variance, missing data is associated w few cases, 15% of subjects are missing data (hertel, 1976), more random variablity, effective sample size can vary from sample to sample, mean substitution, problems cannot be fixed, not large, missing data, maximum likelihood imputation , if missing data is associaetd e with a variable(s), nonposiitve definite (singular), ????, not sensitive to subjects' patterns of scores on tother variables, not at random, no, sem, artifically reduced variance reduced correlation between variables, delete cases w missing data, a missing observation is replaced w a predicetd score generated for each subject by suusing mr based on nonmissing scores on other variables , cold deck imputation, simplicity , traditional options, sem options, imputation, pair wise deletion, list wise deletion, delete cases when missing data is in dvv (hair et al, 1995) , = 10%, missing at random (mar), teh process of data loss on some variables is unrelated to subjects' true status on the variable, all anlaysis are conduced with the same cases, dichotomized correlations (hair et al, 1995) , yes, reduced power via subject loss, eliminate variables, missning completely at random (mcar), regression imputation , most conservative - mean of the distribution of that variable does not change