Statistics Help @ Talk Stats Forum - Statistics
http://www.talkstats.com/
Statistics course and homework discussion. Elementary statistics.enFri, 04 Sep 2015 01:08:30 GMTvBulletin60http://www.talkstats.com/images/misc/rss.pngStatistics Help @ Talk Stats Forum - Statistics
http://www.talkstats.com/
Multinomial Logistic Regression, Main Effects do not show in SPSS
http://www.talkstats.com/showthread.php/62290-Multinomial-Logistic-Regression-Main-Effects-do-not-show-in-SPSS?goto=newpost
Thu, 03 Sep 2015 15:49:26 GMTHello everyone! I hope someone can help. I am really desperate here.

I have two categorical IV's and each IV has 2 levels, so a 2x2 Design.

The DV is also categorical and has 3 levels.

I tried to run a multinomial logistic regression analysis using with SPSS, by putting the IV's into the factor(s)-box and the DV into the dependent variable box. I then specified the model in order to include Factor A, Factor B, and A*B. They were all added by forced entry.

However, after running the analysis, I noticed that only the result for the interaction effect showed, but nothing appeared for the main effects. Any idea why this may be the case?

When I put the Factor A and Factor B into the Co-Variate Box, all three effects showed. Why is that?

Please help... :-(
]]>Statisticslost_in_statisticshttp://www.talkstats.com/showthread.php/62290-Multinomial-Logistic-Regression-Main-Effects-do-not-show-in-SPSStime series with bivariate data?
http://www.talkstats.com/showthread.php/62289-time-series-with-bivariate-data?goto=newpost
Thu, 03 Sep 2015 15:42:33 GMTI have a datetime (yearly) variable which indicates when/if a subject registered, a count of the 400,000 subjects who registered at each facility and...I have a datetime (yearly) variable which indicates when/if a subject registered, a count of the 400,000 subjects who registered at each facility and those subjects identified as being asked a particular question upon registration, a dichotomous yes/no. The requester asked that their time to registration at the facility and whether or not they were asked the question be presented. The hypothesis is that those facilities administering the question had more registrants. Would this be an appropriate approach for an analysis or is there something else I should consider? I've played around in SAS/Eg a while with this but the output isn't that interpretable. Thanks.
]]>Statisticskelevehttp://www.talkstats.com/showthread.php/62289-time-series-with-bivariate-dataWhat do you call this type of analysis?
http://www.talkstats.com/showthread.php/62285-What-do-you-call-this-type-of-analysis?goto=newpost
Thu, 03 Sep 2015 12:15:58 GMTI do a lot of statistical analysis at my job but I'm unsure of the terminology for what I do. I work for a lab where we screen every newborn in the state for 29 genetic disorders. That's 100,000 newborns a year, and sometimes we screen certain babies twice so about 125,000 samples total. I'm constantly taking the data from our information management system and answering questions such as, "How many babies were born at <2500 grams? Of those babies, how many had a positive screening for X disorder? Of this group, how many were actually diagnosed with the disorder?" Sometimes, I'm looking at the data to answer questions about compliance to our rules. Such as, "How many hospitals submitted samples which were contaminated?" Other times, I am literally looking at the data just to see what kind of patterns or anomalies exist. I use excel for all of my analyses and, again, all of the data comes from our information management system. I didn't intend on going down this path in my career (my degree is in Biology) but management has been very pleased with my work, so it is the majority of what I do now. I also really enjoy it, and would like to be able to add this work to my resume and hopefully find a similar job. I just need to be able to properly articulate it.
]]>Statisticsrsw811http://www.talkstats.com/showthread.php/62285-What-do-you-call-this-type-of-analysis