Table 4.

Main effects panel mixed multinomial logit model model: preference estimates

Attribute	Attribute level	Whole sample		Kilifi		Bungoma
		β	SE	β	SE	β	SE
ASC	ASC	Ref. (0)
	ASC µ	0.138	0.192	−0.191	0.277	0.555	0.320
	ASC σ	−0.089	0.115	−0.064	0.151	−0.238	0.941
Recognition	Recognition at community level	Ref. (0)
	Recognition at facility µ	−0.039	0.175	0.350	0.259	−0.466	0.300
	Recognition at facility σ	1.415**	0.247	1.521**	0.350	1.464**	0.456
	Award mechanism for best performing CHV µ	−0.682**	0.186	−0.585*	0.278	−0.819**	0.277
	Award mechanism for best performing CHV σ	1.489**	0.211	1.628**	0.299	1.532**	0.367
Transport	Bicycles for CHV	Ref. (0)
	Motorcycle for CHV/CU µ	−2.430**	0.253	−2.309**	0.348	−2.728**	0.422
	Motorcycle for CHV/CHU σ	2.098**	0.290	2.351**	0.371	1.851**	0.507
	Motorcycle for CHEW and bicycle for CHV µ	0.285	0.299	−0.506	0.423	1.337*	0.556
	Motorcycle for CHEW and bicycle for CHV σ	0.433	0.394	0.164	0.373	−0.891	1.035
Tools of trade	Supplies and commodities+non-pharmaceutical+job aids/IEC materials	Ref. (0)
	Supplies and commodities+non-pharmaceutical+job aids/IEC materials+identification µ	1.330*	0.161	1.424*	0.242	1.386**	0.271
	Supplies and commodities+non-pharmaceutical+job aids/IEC materials+identification σ	−0.562*	0.283	−0.060	0.140	0.924*	0.359
	Supplies and commodities+non-pharmaceutical+job aids/IEC materials)+Identification + Safety gears µ	2.281**	0.225	2.746**	0.346	1.836**	0.306
	Supplies and commodities+non-pharmaceutical+job aids/IEC materials)+Identification + Safety gears σ	1.149**	0.255	1.160**	0.346	1.113*	0.516
IGA	KES 100 000	Ref. (0)
	KES 150 000 µ	−0.119	0.107	−0.109	0.170	−0.146	0.147
	KES 150 000 σ	0.114	0.106	0.184	0.333	0.141	0.106
Monthly stipend (thousands of KES)	Monthly stipend µ	0.543**	0.269	0.470**	0.175	0.697**	0.175
	Monthly stipend σ	0.766**	0.153	0.865	0.984	0.753**	0.236
Decision makers		211		109		102
Observations		5064		2616		2448
Log likelihood		−1088.899		−554.010		−521.411
McFadden’s R2		0.104		0.129		0.085
Akaike’s information criterion		2213.798		1144.02		1078.822
Bayesian information criterion		2331.336		1249.669		1183.276

All attributes were random and normally distributed except monthly stipend which was restricted to a lognormal distribution. Values of monthly stipend are in thousands of KES. Statistical significance: **at 0.01 level, *at 0.05 level. β is the coefficient, SE is the robust SE, µ is the mean, σ is the SD and ASC is alternative specific constant for alternative. All models used 1000 Halton draws

ASC, alternative specific constant; CHEW, community health extension worker; CHV, community health volunteer; CU, community unit; IEC, information, education and communication; IGA, income-generating activities.