[R] calculate response probabilities using sem-analysis
Hi everyone, I just conducted a structural equation model for estimating a response model. This model should predict the probability that someone is responding to a direct mailing. I used the sem package for this. When I have my coefficients I want to know how well my model predicts the probability of response. How can I calculate these probabilities? I tried to use the unstandardized coefficients, just like a regression coefficient in the following formula: Y = b1*x1 + b2*x2 But then I have values larger than 1, so that aren't probabilities. Does anyone dealt with this problem before? You can be of great help to me!! Kind regards, Tryntsje [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] calculate response probabilities using sem-analysis
Hi everyone, I just conducted a structural equation model for estimating a response model. This model should predict the probability that someone is responding to a direct mailing. I used the sem package for this. When I have my coefficients I want to know how well my model predicts the probability of response. How can I calculate these probabilities? I tried to use the unstandardized coefficients, just like a regression coefficient in the following formula: Y = b1*x1 + b2*x2 But then I have values larger than 1, so that aren't probabilities. Does anyone dealt with this problem before? You can be of great help to me!! Kind regards, Tryntsje [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Confidence/prediction interval for choice probabilities mlogit
Hi, I have estimated a multinomial logit model with use of the mlogit-package for my choice data. I predicted the probabilities for each choice, using the formula: pred1 - exp(B1*X1.1 + B2*X2.1 + B3*X3.1) pred2 - exp(B1*X1.2 + B2*X2.2 + B3*X3.2) pred3 - exp(B1*X1.3 + B2*X2.3 + B3*X3.3) sumpred - pred1+pred2+pred3 pred1 - pred1/sumpred pred2 - pred2/sumpred pred3 - pred3/sumpred Then for each choice I have a prediction of the choice probabilities. Now I want to calculate a confidence interval of the choice probablities. I'm not sure if it is called a confidence interval or a prediction interval. At the internet I haven't found anything useful. Most examples are about the confidence interval of a certain regression coefficient, for example the intercept. But I want to know the interval around my choice probabilities. The predict() function does not work with mlogit, I think. Can anybody help me? Kind regards, Tryntsje __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Difference MNP-package and rmnpGibbs from bayesm-package
Hi all, For my research I have to use a Multinomial Probit model. I saw that there are two packages, that include a method to estimate my parameters. The first one is the MNP-package of Imai and van Dyk. The second one is part of the bayesm-package of Rossi. The results for both packages are not the same using the same data. Does anybody know what the difference is between these two approaches? Kind regards, Tryntsje __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] LCA (e1071 package): error
Hello, I will use the lca method in the e1071 package. But I get the following error: Error in pas[j, ] - drop(exp(rep(1, nvar) %*% log(mp))) : number of items to replace is not a multiple of replacement length Does anybody know this error and knows what this means? Kind regards, Tryntsje __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Output mlogit package (multinomial logisti c regression)
Hello, For my master thesis I conducted a conjoint analysis. Using the mlogit-package of Yves Croissant I should be able to analyse my choice data. I read the whole program of mlogit, but I do not understand how my coefficients must be interpreted. Are they a result of P(Y=i) or are they P(Y=i)/P(Y=1) where 1 is my base category, or are they a result of ln(P(Y=i)/P(Y=1)). I can not find an answer to this in the package. Hence, are they logits, odds or just a ratio? Can anyone help me with this? Kind regards, Tryntsje __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.