In what instances is it better to use Zone of indifference instead of inverse distance for hot spotting?

[StephnieWatson]

I have ridership data for transit that I am trying to hotspot to see if there are significant spatial clusters of riders in certain neighbourhoods. The data unfortunately is not tied to bus stops so I have had to sum rides for each neighbourhood and add it to the shapefile that way. I am now trying to determine which conceptualization to use. Based off of the descriptions I believe I have it narrowed down to Zone of indifference or Inverse distance. Any thoughts as to how you would go about hot spotting this data or choosing the proper Conceptualization? I have run them all as well to get a feel for the data but it has just made it more confusing. 

Thank you for your thoughts!

[DanPatterson_Retired]

Modeling spatial relationships—Help | ArcGIS Desktop Zone of Indifference example, and the suggestions here

Hot Spot Analysis (Getis-Ord Gi*)—Help | ArcGIS Desktop 

Based on the example in the first link, the question would go back to you as to what your thoughts are on the data you have?  What have you observed?  You have aggregated you data on a neighborhood level, what about the stops within those neighborhoods? How many etc.  Do you have any visuals?

[StephnieWatson]

Unfortunately due to the way the data was collected we can't tell which stops the riders got on at so we have had to sum the riders to each neighborhood polygon. I am not sure how else to put the data together to be honest and still be able to see how the neighborhoods ridership cluster. I can input each rider at a lat/long but then each point represents a rider which then only has a value of 1 which does not work for the analysis. 

I believe I am getting confused because I keep thinking that distance does not really affect where the rider gets on it hence the setting of the Zone to cover the city. But then at the same time it could affect hence the inverse distance. So trying to see which works. Or I was trying to figure out how to create a spatial matrix file. But same problem with choosing that as well. 

What I am trying to do is figure out where our highest significantly different rider usage is at a neighborhood level. Which is proving difficult with that data that was made available to us. 

[DanPatterson_Retired]

I see the problem.  Basically a person is going to go to the closest bus stop, so IDW wouldn't be a concern... unless stop # 2 (a bit further away) had a starbucks and they just might have an attractant to make the effort.  The only time I could see distance being a consistent issue if the stops were not at an acceptable walking distance, but most transit services have already accounted for that in route design and stop placement.  There will, however, be certain neighborhoods that regardless of the provided access, the ridership is low... probably due to personal preference

[StephnieWatson]

Yeah that is what I was thinking too. So do you think that the Zone of indifference would be most acceptable choice if set at the parameters of the city? Or would you go inverse distance? 

[DanPatterson_Retired]

For me? Zone of indifference.  But someone is going to validly argue the opposite case.  I know that my city is currently looking at ridership.  Sometimes it isn't where you put things, it is 'when' you put them.  The routes and stop locations aren't an issue at all, ridership stagnation sometimes boils down to the prevalence of crowded buses or perceived frequency of service. Everyone wants transportation available to them, where they want it, when they want it and with a comfort level that may not be available within a sustainable cost structure.  If they can afford to avoid public transit, they will, unless other factors compel them to use it.  My guage of how well a system works is when no one talks about it.  

[StephnieWatson]

Yeah that is what I was leaning towards as well. I just have to find a way to defend it as Inverse can also be defended depending on perspective. Haven't been able to find published articles on using this type of analysis for this type of data either. Ahh got to love thesis work!