Lookin at the Data on Stata..

On Friday, after my microbiology class I walked to Westwood to Diana's office. We discussed the next steps of our proposed research--looking at the data. If we could find associations in our data through the topics that we chose then we would have something to write about. Remember, the possible topics we chose that affect physician satisfaction: gender, ethnicity, income, patient/work load, time spent with the patients, the type of practice they work in, and how insurance plas a role in this.
Diana showed me the statistics program they used for the input of all their data on this research. It is called Stata. It looks like an old program because the screen is black and the letters are green and yellow, but of course its not old and it's actually a very advanced/complex program that takes the numbers inputed from the research and can find percentages, p values, associations, etc (all those statistics terms!). It can cross two variables of research and see how they relate to each other. I think it can do more than two also, but we only looked at crossing two this day.
Another thing about Stata is that it uses codes for every category of data. Some of the codes were obvious, like the one for the race of the physician the code is "race". Their answers were also codes. Actually, all the answers to every question were numbers codes. For example, a yes or no question would have the codes 0 and 1. Or the race question ranged from 1-5 (White, Latina, African American, Asian-Pacific Islander, Other). Some other category codes weren't so easy to distinguish like mvar9 or svar9, I do not know what it means off the top of my head. I think the 9 refers to what question it was?
So what we need to look at is the variable of the physicians who were satisfied and the physicians who were the least (or not) satisfied at all and cross it with each topic we chose to look at. This will then give us a 2x2 table to show how many answered yes to each question or no to each question or yes then no or no then yes. For example, taking into account the physicians gender the boxes of the table would be: satisfied male physicians, unsatisfied male physicians, satisfied female physicians, and unsatisfied female physicians. Make sense? Crossing these variables will then give us a P value. This value tells how associated these variables are or if chance had any thing to do with the relation. If the P value is less than 0.05, than we would consider the variable for further research because less than 0.05 means that it wasnt due to chance. Sometimes researchers accept P value less than 0.10 also.
Diana and I didn't do a lot of crossing of variables, we decided we would do it on the next time we would meet. She had to go to a meeting, but I stayed in the office to read and fill out the paperwork that Khushbindar gave me at the REACH conference. I left it in Diana's box then had to go to work til 9:30p. uugh!