Attribute-level balance with mfederov

[MirtheN]

Hi,

I have written the following code:

Code: Select all

? Efficient design
Design
;alts = location1, location2, route1, route2, none
;rows = 10
;eff = (mnl, d)
;alg = mfederov(stop=noimprov(30 mins))
;reject:
location1.x1 >= location2.x1 and location1.x2 >= location2.x2 and location1.x3 >= location2.x3 and location1.x4 >= location2.x4 and location1.x5 >= location2.x5,
location1.x1 <= location2.x1 and location1.x2 <= location2.x2 and location1.x3 <= location2.x3 and location1.x4 <= location2.x4 and location1.x5 <= location2.x5,
route1.x6 >= route2.x6 and route1.x7 >= route2.x7 and route1.x8 >= route2.x8 and route1.x9 >= route2.x9,
route1.x6 <= route2.x6 and route1.x7 <= route2.x7 and route1.x8 <= route2.x8 and route1.x9 <= route2.x9
;model:
U(location1) = a_location + b1[0.001] * x1[0,1] + b2[0.001] * x2[0,1] + b3[0.001] * x3[0,1] + b4[0.001] * x4[0,1] + b5[0.001] * x5[0,1] /
U(location2) = a_location + b1        * x1      + b2        * x2      + b3        * x3      + b4        * x4      + b5        * x5      /
U(route1)    = a_route    + b6[0.001] * x6[0,1] + b7[0.001] * x7[0,1] + b8[0.001] * x8[0,1] + b9[0.001] * x9[0,1]                       /
U(route2)    = a_route    + b6        * x6      + b7        * x7      + b8        * x8      + b9        * x9                   
$


I am not sure whether this design is balanced. After half an hour of running, I got a B-estimate of 99.999624, but is this the right measure for attribute-level balance?

If my design lacks attribute-level balance, how can I improve it? Or is attribute-level balance unnecessary for an efficient design?

[Michiel Bliemer]

B-estimates refer to utility balance, and it is near-100% because your priors are near-zero and hence choice probabilities are balanced. I generally ignore B-estimates as they are not important.

Attribute-level balance is not needed but your design will likely be quite attribute-level balanced since you only have 2 levels for each attribute.

I would increase the number of rows, 10 rows is quite minimal for estimating a model. Perhaps use 20 rows with 2 blocks.

Michiel