What can you conclude from the output of an alternative-specific design?

Question

The description of alternative specific designs in the manual is helpful, but I need help envisioning the output/and what you can conclude from it. Using this example from the manual...
Car
Parking fee $5.00/day
Parking fee $8.00/day
Parking fee $10.00/day

Bus
Picks up every 20 min.
Picks up every 15 min.
Picks up every 10 min.
Picks up every 5 min.

25 cents per one-way trip
50 cents per one-way trip
75 cents per one-way trip
$1.00 per one-way trip

What would you be able to comment on regarding the alternative-specific attributes based on the output?

For example, could you determine the "importance" of pick up time, compared to the "importance" of bus trip cost when car parking fees are $5 vs when car parking fees are $8 ?

Could you also estimate the probability of mode of transportation outcome, in each scenario? i.e. ...
When car parking fee is $5, the bus picks up every 20mins and the bus costs 25cents, X% would choose car, X% would choose bus.

When car parking fee is $5, the bus picks up every 20mins and the bus costs 50 cents, X% would choose car, X% would choose bus.

and so on, to determine which combination of attributes across transportation types makes it most likely people will choose car and which combination makes it most likely people will choose bus?

I hope these questions make sense. Thanks for your help!

Answer 1

Cassidy,

Well, the importances of attributes is unaffected by the particular levels at which other attributes are present in a choice set- importance is a static calculation, not a choice-set-specific one.  

But your other two questions are spot-on:  in a simulator you would get share predictions for any combination of the attributes and levels.

Comments

Thank you for the speedy response! I'm a soon-to-be user, new to the software and terminology. Great to hear about the share predictions.

To follow-up on the first question, I understand what you're saying.
If I suspected the relative importance of bus pick up time and bus trip cost to change, depending on car parking fee ... is there any way to get at that? Could I run one CBC with car parking fee always fixed at $5 against the bus options and then another CBC with car parking fee fixed at $10 against the bus options and compare the utility or importance output for bus wait time and bus trip cost across the two CBC tasks?

Thanks again!

---

Or if I'm using the wrong words...
If I suspected the utility* of bus pick up time levels* and bus trip cost levels* to change, depending on car parking fee ... is there any way to get at that?

---

Cassidy,

Typically the utilities of Attribute A and Attribute B don't change if the levels of Attribute C change, but of course the share of a combination of these in a competitive situation will change.  I say "typically" because there are situations (called interactions) when the levels of one attribute DO affect the utilities of levels on other attributes.  This is something you can test for when you're running the analysis in Lighthouse Studio.

---

Got it. And hypothetically, interactions can occur/you can look for them between two conditional attributes?

---

Correct - as long as the interaction isn't between two attributes that are mutually exclusive.

---

Just to clarify... you could look for an interaction between car parking fee and bus pick up time utilities? Falling under different primary attributes doesn't make them mutually exclusive, right? Maybe, could you give an example of two attributes that are mutually exclusive?

Thanks so much for your help!

---

This is a different kind of interaction called a "cross-effect."  You can do it in our software (specifically, you can estimate the effect of car parking fee on bus share) to measure deviations from "fair share" draws assumed by the logit model.  It takes some extra effort (exporting the design file, adding some coding for the cross-effects, then analyzing your data with the modified design file) but it's something you can (and that I have) done.

---

I really appreciate it. These terms will help me find additional resources. Thanks!