search

JestonBlu / Neighbor-Works

Masters Project: Statistical Research
0 stars 0 forks source link

Pulling things together #8

Closed profgeraci closed 5 years ago

profgeraci commented 7 years ago

Hi, gang. I've posted an initial draft of what I propose as our Project Document under "Code". Please feel free to pull down a copy, turn on track changes and comment on anything you like. I'm kind of waiting now for Joseph to figure out exactly what type of model we're going to use and which variables we need.

Once we have reached consensus on that, I can modify the document to reflect this - or anyone else can if you get to it first.

I'll be in and out this weekend, but ready to pull this together next week.

JestonBlu commented 7 years ago

Also, the contrasts in the report should be updated to the new contrast table, since we only have 4 neighborhood IDs we can only do 3 contrasts, Initially I did 4 separate contrasts in SAS, but that was inappropriate. Still they were all not significant.

rmglazner commented 7 years ago

Thank you for the extra analysis, Joseph! Just so I understand the table correctly, what does a positive coefficient for PoliceRatingCat[medium-high] indicate? It is significant and positive, so does that mean as police rating moves from high to medium, years of residence increases?

I think that this extra information looks good, but I am worried that if we replaced what we have with ordinal analysis she may not like that our presentation and report are different. The 5 page limit makes this difficult, but could we add these tables in addition to what we already have to show that we approached the data in multiple ways? I think an extra sentence similar to what you said about the comparisons and then including these two tables would help a lot. If the results are similar, we can say that our conclusions are more greatly supported by multiple approaches maybe?

rmglazner commented 7 years ago

I think if we made the map images smaller we would have room for the extra tables.

nitroys commented 7 years ago

I'll give it a shot at fitting it all in. I like the ordinal analysis too, though I'm surprised the interactions were giving you trouble. But I guess it depends on how the variables are structured.

JestonBlu commented 7 years ago

Yeah, im not sure... not a lot of time to dig into it though... so that way I would interpret the PoliceRating of High compared to Medium is that on average people who rated Police as high have lived in their residence 30% longer than those who rated Police as medium. And then of course there is no significant difference between those who rate police medium vs low.

nitroys commented 7 years ago

Sorry, where is the new contrast table?

JestonBlu commented 7 years ago

It looks a little different in the JMP output, its kind of buried, but here it is

image

nitroys commented 7 years ago

Ah okay. That's what I thought, but wanted to make sure. Thanks! Almost done making changes, will upload it soon for a final look through from you all.

nitroys commented 7 years ago

Okay, it's called postpresentationwriteup and can be found in Code.

JestonBlu commented 7 years ago

I think it looks good, I dont have any edits

rmglazner commented 7 years ago

I agree!

nitroys commented 7 years ago

Great! I'll give Anne some time to look over it before I submit. Joseph, are your final dataset and JMP code versions uploaded? I can submit those when I submit the report, I just want to make sure I have the right ones.

rmglazner commented 7 years ago

If the interaction is insignificant, should we have the ordinal data be presented without the interaction term? Otherwise we are comparing a model with an interaction to a model without an interaction.

rmglazner commented 7 years ago

I just thought of that. I don't know if it is that important, but I just wanted to check!

nitroys commented 7 years ago

I think it's fine the way it is, given the time constraint.

rmglazner commented 7 years ago

Okay!

profgeraci commented 7 years ago

Hi, guys. Unfortunately I will not be able to look at it before it's due. I'm going through thyroid cancer and have an appt with my doctor today. Timing! Please go ahead without me.

On Mar 3, 2017, at 12:20 PM, rmglazner notifications@github.com wrote:

I just thought of that. I don't know if it is that important, but I just wanted to check!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

JestonBlu commented 7 years ago

You can just submit the ReducedData.jmp data set. It has the models saved as scripts inside the data table and they all have the same titles as the report names I generated. The last thing I ran was with the data changed to ordinal so if she runs ANCOVA--log(Years) Model 3 (Best with interaction) she is going to get the ordinal version.

nitroys commented 7 years ago

Oh Anne, I'm so sorry to hear that. No worries, I hope your doctor's appointment goes well.

Thanks Joseph! I'll go ahead and submit all this stuff now.

JestonBlu commented 7 years ago

Sorry to hear that Anne. I think the report is high quality. Im comfortable submitting.

rmglazner commented 7 years ago

Anne, I am really sorry to hear that. I agree what we have is good, thank you for submitting it for us.

profgeraci commented 7 years ago

Thanks for your thoughts all. I appreciate all your efforts to get the project submitted before the deadline. The report looks really good (IMHO) and reflects all of your hard work. I'm impressed that we were able to say what we wanted to say in 5 pages! If we all lived closer I would buy everyone a beer!