User talk:Karina ritter
From InterSciWiki
Contents |
[edit] Overview of Paper #2
Notes (edited by DW) from Oct 9, 07
[edit] Intro
-discuss what is a single factor whats not. common variation, two polar opposites on a scale of differences.
- one possibility that all the variables correlated on this dimension measure the same thing
- another that different features are correlated that form clusters of related variables
- or a mixture of both.
- how are you going to pick your variables, according to what ideas to start with
- paste your factor analysis tables 2 and 3 (factor structure) into your word doc and discuss those results in the paper
[edit] Types of Factors: part 1
-if you use a factor like external war, already constructed and in the database, the interest here will be in the other correlates of the factor (2nd part).
-to explore new factor dimensions there are many variables you can begin with, e.g., about women status - then cut back to get one factor. Or take a known factor build from it
- if you any of these options for a new or expanded factor, save the whole spss dataset in C:My documents. or to a flash drive.
- can put many variable in factor analysis to begin with, pare away
[edit] Correlates of Factors: part 2
-2nd part is after you construct a factor (e.g., a subdimension of woman status) to correlate the factor you construct with other variables
- take a factor you construct and correlate with other variable -correlate with variables that result from this factor or that this factor results from
- e.g., what else correlates with external war? Ex: belief in male superiority/female inferiority. Or write a paper focused on external war and its correlates, or many correlates
[edit] Preparation is crucial
-for both the 1st and second part the crucial preparation is how to choose variables for the essay -- review the on-line codebook
- relate to your opening paragraphs
-how to select variables - [of variables]
- make sure each variable has a sufficient number of cases, e.g., 80 or more
- make sure when you have constructed factors scores to check that you have 40-50 cases for which factor scores have been computed. -when keeping factor... 50 is a good number of cases
- -external war 50 societies composite variable consisting of measures (know)
[edit] Summarize your findings
[edit] Paper #3
-any pair of topics
-And correlate variables from one topic with those from the other
-think these two topics are related: here are the variables (some may be factors)
-topics must be far enough away from each other to be of interest
-then: here are some hypotheses (what you expect the relationship to be)
-see how two topics relate (similarities)
-when writing paper...what effects what?
-can use ideas from previous literature
-see distribution with map
-talk about the correlation between the variables
-think about problem in terms of what is not obvious
-evidence needs interpretation. build argument about what might be going on
-to "replicate" a finding use a "layer variable" in selecting cross-tabulation variables in Spss
- e.g., variable 200, region, or recoded "Old world"/"New world"
- e.g., state level societies/prestate
