1 Experiments

Table 1.1: Participants by type of experiment and country
variable	KE	MO	TN	TZ	UG	Total
risk	249	492	390	331	111	1,573
time	355	490	466	380	331	2,022
PGG	248	260	268	0	280	1,056
trust	260	240	234	501	220	1,455
DG	260	240	234	501	220	1,455
risk.time	214	483	372	278	93	1,440
risk.PGG	121	255	206	0	57	639
risk.trust	128	237	184	331	54	934
risk.DG	128	237	184	331	54	934
time.PGG	174	258	244	0	179	855
time.trust	181	232	222	380	152	1,167
time.DG	181	232	222	380	152	1,167
trust.DG	260	240	234	501	220	1,455
risk.time.PGG	104	254	194	0	45	597
risk.time.trust	110	229	178	278	48	843
risk.time.DG	110	229	178	278	48	843
risk.trust.DG	128	237	184	331	54	934
time.trust.DG	181	232	222	380	152	1,167
experiment participants	508	500	502	501	500	2,511
total survey respondents	1,514	900	1,003	990	1,517	5,924

combination of experiments used: risk - time - trust

variables under analysis:

risk: risk propensity
time: patience propensity
player.trust: trust in others
player.trust_inst: trust in institutions

Figure 1.1: Distribution of variables under analysis

Table 1.2: Average values of variables under analysis by country
variable	KE	MO	TN	TZ	UG	Total
risk	4.68	4.82	5.06	5.26	9.08	5.24
time	6.11	5.56	6.21	7.05	8.69	6.44
player.trust	28.64	40.50	41.99	36.60	40.10	37.96
player.trust_inst	25.00	43.67	54.49	49.82	57.29	46.32

Figure 1.2: Distribution of variables under analysis by country

1.1 Cluster analysis: choice of algorithm and number of groups

Comparison of cluster algorithms and by number of groups; analysis on standardised data; <br>algorithms: kmeans: k-means, pam: partition around medoids, ward.D2: hierarchical clustering by Ward's method; <br>indices: CH=Calinski-Harabasz, S=Silhouette, C=Hubert & Levin C index, DB=Davies-Bouldin

Figure 1.3: Comparison of cluster algorithms and by number of groups; analysis on standardised data;
algorithms: kmeans: k-means, pam: partition around medoids, ward.D2: hierarchical clustering by Ward’s method;
indices: CH=Calinski-Harabasz, S=Silhouette, C=Hubert & Levin C index, DB=Davies-Bouldin

La scelta dell’algoritmo di cluster e del numero ottimale di gruppi è stata fatta:
1. standardizzando le variabili in analisi;
2. calcolando la matrice di distanze euclidee;
3. confrontando i valori degli indici di Calinski-Harabasz, Silhouette, Hubert & Levin C e Davies-Bouldin ottenuti per le possibili soluzioni da 2 a 10 gruppi, forniti dagli algoritmi k-means, partition around medoids e cluster gerarchica con metodo di Ward

Escludendo la soluzione a due gruppi, ritenuta troppo generica, osserviamo che gli indici di Calinski-Harabasz e Silhouette presentano il valore massimo in corrispondenza del raggruppamento in quattro cluster per l’algoritmo di k-means; l’indice C di Hubert & Levin, per cui l’ottimo corrisponde al valore minimo, tende a decrescere fino ai dieci cluster, tuttavia si può notare un gomito in corrispondenza del partizionamento in quattro gruppi, dopodichè la decrescita rallenta.

choice: k-means with 4 clusters

1.2 Cluster analysis results

Table 1.3: number of consumers by cluster
levels	N	%
1	259	30.7
2	178	21.1
3	210	24.9
4	196	23.3
Total	843	100.0

Table 1.4: Average value of variables under analysis by cluster
variable	1	2	3	4	Total
risk	4.00	8.22	4.94	4.47	5.24
time	7.88	8.80	1.60	7.59	6.44
player.trust	23.07	35.96	38.81	58.55	37.96
player.trust_inst	24.90	44.94	47.62	74.49	46.32

Figure 1.4: Cluster profiling: average value of variables under analysis by cluster

Table 1.5: Pairwise comparisons of Experiment clusters by cluster variables
variable	comparison	p.value	signif
risk	1 - 2	<0.001	***
	1 - 3	<0.001	***
	1 - 4	0.0205	*
	2 - 3	<0.001	***
	2 - 4	<0.001	***
	3 - 4	0.5437
time	1 - 2	<0.001	***
	1 - 3	<0.001	***
	1 - 4	1
	2 - 3	<0.001	***
	2 - 4	<0.001	***
	3 - 4	<0.001	***
player.trust	1 - 2	<0.001	***
	1 - 3	<0.001	***
	1 - 4	<0.001	***
	2 - 3	1
	2 - 4	<0.001	***
	3 - 4	<0.001	***
player.trust_inst	1 - 2	<0.001	***
	1 - 3	<0.001	***
	1 - 4	<0.001	***
	2 - 3	1
	2 - 4	<0.001	***
	3 - 4	<0.001	***
Pairwise comparisons using Wilcoxon rank sum test with continuity correction P value adjustment method: bonferroni

Clusters summary:

1: patient but untrusting (orange)
2: high risk-takers (red)
3: impatient (green)
4: trusting (blue)

Figure 1.5: clusters distribution

Figure 1.6: Clusters by socioeconomic variables

Table 1.6: Clusters by socioeconomic variables - row %
variable	levels	1	2	3	4	Total
Total	Total	30.8	21.2	24.9	23.1	100
country	KE	47.3	22.7	20.0	10.0	100
	MO	34.1	12.2	35.4	18.3	100
	TN	24.1	16.7	21.3	37.9	100
	TZ	30.9	20.1	23.7	25.2	100
	UG	0.0	83.3	6.2	10.4	100
gender	Male	32.0	19.5	26.5	22.0	100
	Female	29.4	23.1	23.1	24.4	100
cl.age	≤25	40.9	13.0	22.1	24.0	100
	26-35	28.2	29.1	25.4	17.4	100
	36-45	25.7	19.0	25.1	30.2	100
	46-55	30.4	24.4	23.0	22.2	100
	>55	30.4	18.4	28.5	22.8	100
educ	No qualification literate	30.2	18.9	37.7	13.2	100
	Primary	30.2	24.6	23.4	21.8	100
	Secondary	35.3	20.0	24.0	20.7	100
	More than secondary	29.7	20.3	23.1	26.9	100
	Other	12.8	17.0	31.9	38.3	100
income_avg	lower than average	30.3	21.5	29.9	18.4	100
	somewhat lower than average	29.8	21.5	24.9	23.9	100
	about average	30.5	18.8	25.8	25.0	100
	somewhat higher than average	28.9	23.3	15.6	32.2	100
	higher than average	39.4	21.1	16.9	22.5	100

Table 1.7: Clusters by socioeconomic variables - col %
variable	livelli	1	2	3	4	Total
Total		100.0	100.0	100.0	100.0	100.0
country	KE	20.2	14.0	10.5	5.7	13.1
	MO	30.2	15.7	38.8	21.6	27.3
	TN	16.3	16.3	17.7	34.0	20.7
	TZ	33.3	31.5	31.6	36.1	33.1
	UG	0.0	22.5	1.4	2.6	5.7
gender	Male	54.3	47.8	55.5	49.5	52.1
	Female	45.7	52.2	44.5	50.5	47.9
cl.age	≤25	24.4	11.2	16.3	19.1	18.4
	26-35	23.3	34.8	25.8	19.1	25.4
	36-45	17.8	19.1	21.5	27.8	21.3
	46-55	15.9	18.5	14.8	15.5	16.1
	>55	18.6	16.3	21.5	18.6	18.8
educ	No qualification literate	6.2	5.6	9.6	3.6	6.3
	Primary	29.5	34.8	28.2	28.4	30.0
	Secondary	37.6	30.9	31.6	29.4	32.8
	More than secondary	24.4	24.2	23.4	29.4	25.3
	Other	2.3	4.5	7.2	9.3	5.6
income_avg	lower than average	30.6	31.5	37.3	24.7	31.1
	somewhat lower than average	33.3	34.8	34.4	35.6	34.4
	about average	15.1	13.5	15.8	16.5	15.3
	somewhat higher than average	10.1	11.8	6.7	14.9	10.7
	higher than average	10.9	8.4	5.7	8.2	8.5

Table 1.8: Testing the relationship between clusters and the following variables at project level
variable	test	statistic	df	p.value	signif
country	Chi.square	175.54	12	3.5e-31	***
gender	Chi.square	3.33	3	0.3431
cl.age	Chi.square	28.52	12	0.0046	**
educ	Chi.square	23.17	12	0.0263	*
income_avg	Chi.square	16.18	12	0.1831

Figure 1.7: Correspondence Analysis - Clusters Experiments

Figure 1.8: Clusters by gender, age, education level and income at country level

Table 1.9: Clusters by gender, age, education level and income at country level - row %
fattore	variable	levels	1	2	3	4	Total
KE	Total	Total	47.3	22.7	20.0	10.0	100
	gender	Male	48.2	26.8	14.3	10.7	100
		Female	46.3	18.5	25.9	9.3	100
	cl.age	≤25	59.4	12.5	21.9	6.2	100
		26-35	46.2	30.8	17.9	5.1	100
		36-45	33.3	14.3	28.6	23.8	100
		46-55	30.0	40.0	10.0	20.0	100
		>55	62.5	25.0	12.5	0.0	100
	educ	No qualification literate	66.7	33.3	0.0	0.0	100
		Primary	45.5	21.2	15.2	18.2	100
		Secondary	61.1	19.4	16.7	2.8	100
		More than secondary	38.2	23.5	29.4	8.8	100
		Other	0.0	50.0	25.0	25.0	100
	income_avg	lower than average	50.0	0.0	50.0	0.0	100
		somewhat lower than average	77.8	11.1	0.0	11.1	100
		about average	50.0	11.1	33.3	5.6	100
		somewhat higher than average	42.4	27.3	15.2	15.2	100
		higher than average	43.8	27.1	20.8	8.3	100
MO	Total	Total	34.1	12.2	35.4	18.3	100
	gender	Male	38.1	11.9	35.1	14.9	100
		Female	23.0	13.1	36.1	27.9	100
	cl.age	≤25	14.3	0.0	28.6	57.1	100
		26-35	37.5	12.5	31.2	18.8	100
		36-45	29.5	13.6	34.1	22.7	100
		46-55	41.4	13.8	32.8	12.1	100
		>55	31.8	11.4	39.8	17.0	100
	educ	No qualification literate	35.9	5.1	48.7	10.3	100
		Primary	31.3	19.4	31.3	17.9	100
		Secondary	36.6	13.4	29.3	20.7	100
		More than secondary	33.3	5.6	38.9	22.2	100
		Other	20.0	0.0	60.0	20.0	100
	income_avg	lower than average	34.6	15.0	32.7	17.8	100
		somewhat lower than average	32.1	10.3	42.3	15.4	100
		about average	43.3	6.7	30.0	20.0	100
		somewhat higher than average	11.1	22.2	33.3	33.3	100
		higher than average	40.0	0.0	20.0	40.0	100
TN	Total	Total	24.1	16.7	21.3	37.9	100
	gender	Male	17.8	20.5	23.3	38.4	100
		Female	28.7	13.9	19.8	37.6	100
	cl.age	≤25	33.3	2.8	27.8	36.1	100
		26-35	20.0	23.3	20.0	36.7	100
		36-45	23.3	20.9	23.3	32.6	100
		46-55	16.7	19.4	19.4	44.4	100
		>55	27.6	17.2	13.8	41.4	100
	educ	No qualification literate	0.0	60.0	0.0	40.0	100
		Primary	25.0	20.0	20.0	35.0	100
		Secondary	26.3	13.2	31.6	28.9	100
		More than secondary	26.8	17.1	17.1	39.0	100
		Other	17.2	10.3	24.1	48.3	100
	income_avg	lower than average	12.5	18.8	31.2	37.5	100
		somewhat lower than average	27.1	18.6	20.0	34.3	100
		about average	24.5	16.3	24.5	34.7	100
		somewhat higher than average	20.0	16.7	16.7	46.7	100
		higher than average	33.3	0.0	11.1	55.6	100
TZ	Total	Total	30.9	20.1	23.7	25.2	100
	gender	Male	29.5	15.6	26.2	28.7	100
		Female	32.1	23.7	21.8	22.4	100
	cl.age	≤25	43.1	15.3	19.4	22.2	100
		26-35	26.7	21.1	33.3	18.9	100
		36-45	26.2	13.1	21.3	39.3	100
		46-55	33.3	33.3	16.7	16.7	100
		>55	22.6	32.3	16.1	29.0	100
	educ	No qualification literate	0.0	0.0	100.0	0.0	100
		Primary	29.2	23.3	22.5	25.0	100
		Secondary	34.3	18.6	22.5	24.5	100
		More than secondary	33.3	16.7	22.9	27.1	100
		Other	0.0	14.3	57.1	28.6	100
	income_avg	lower than average	32.2	19.0	30.6	18.2	100
		somewhat lower than average	33.3	19.0	21.0	26.7	100
		about average	18.5	29.6	22.2	29.6	100
		somewhat higher than average	31.2	18.8	6.2	43.8	100
		higher than average	22.2	22.2	0.0	55.6	100
UG	Total	Total	0.0	83.3	6.2	10.4	100
	gender	Male	0.0	88.9	0.0	11.1	100
		Female	0.0	80.0	10.0	10.0	100
	cl.age	≤25	0.0	57.1	14.3	28.6	100
		26-35	0.0	90.9	4.5	4.5	100
		36-45	0.0	80.0	10.0	10.0	100
		46-55	0.0	85.7	0.0	14.3	100
		>55	0.0	100.0	0.0	0.0	100
	educ	No qualification literate	0.0	80.0	0.0	20.0	100
		Primary	0.0	83.3	16.7	0.0	100
		Secondary	0.0	76.5	5.9	17.6	100
		More than secondary	0.0	91.7	0.0	8.3	100
		Other	0.0	100.0	0.0	0.0	100
	income_avg	lower than average	0.0	93.3	0.0	6.7	100
		somewhat lower than average	0.0	74.1	11.1	14.8	100
		about average	0.0	100.0	0.0	0.0	100
		somewhat higher than average	0.0	100.0	0.0	0.0	100
		higher than average	0.0	0.0	0.0	0.0	0
Total	Total	Total	30.8	21.2	24.9	23.1	100
	gender	Male	32.0	19.5	26.5	22.0	100
		Female	29.4	23.1	23.1	24.4	100
	cl.age	≤25	40.9	13.0	22.1	24.0	100
		26-35	28.2	29.1	25.4	17.4	100
		36-45	25.7	19.0	25.1	30.2	100
		46-55	30.4	24.4	23.0	22.2	100
		>55	30.4	18.4	28.5	22.8	100
	educ	No qualification literate	30.2	18.9	37.7	13.2	100
		Primary	30.2	24.6	23.4	21.8	100
		Secondary	35.3	20.0	24.0	20.7	100
		More than secondary	29.7	20.3	23.1	26.9	100
		Other	12.8	17.0	31.9	38.3	100
	income_avg	lower than average	30.3	21.5	29.9	18.4	100
		somewhat lower than average	29.8	21.5	24.9	23.9	100
		about average	30.5	18.8	25.8	25.0	100
		somewhat higher than average	28.9	23.3	15.6	32.2	100
		higher than average	39.4	21.1	16.9	22.5	100

Table 1.10: Clusters by gender, age, education level and income at country level - col %
fattore	variable	livelli	1	2	3	4	Total
KE	Total		100.0	100.0	100.0	100.0	100.0
	gender	Male	51.9	60.0	36.4	54.5	50.9
		Female	48.1	40.0	63.6	45.5	49.1
	cl.age	≤25	36.5	16.0	31.8	18.2	29.1
		26-35	34.6	48.0	31.8	18.2	35.5
		36-45	13.5	12.0	27.3	45.5	19.1
		46-55	5.8	16.0	4.5	18.2	9.1
		>55	9.6	8.0	4.5	0.0	7.3
	educ	No qualification literate	3.8	4.0	0.0	0.0	2.7
		Primary	28.8	28.0	22.7	54.5	30.0
		Secondary	42.3	28.0	27.3	9.1	32.7
		More than secondary	25.0	32.0	45.5	27.3	30.9
		Other	0.0	8.0	4.5	9.1	3.6
	income_avg	lower than average	1.9	0.0	4.5	0.0	1.8
		somewhat lower than average	13.5	4.0	0.0	9.1	8.2
		about average	17.3	8.0	27.3	9.1	16.4
		somewhat higher than average	26.9	36.0	22.7	45.5	30.0
		higher than average	40.4	52.0	45.5	36.4	43.6
MO	Total		100.0	100.0	100.0	100.0	100.0
	gender	Male	82.1	71.4	72.8	59.5	73.4
		Female	17.9	28.6	27.2	40.5	26.6
	cl.age	≤25	1.3	0.0	2.5	9.5	3.1
		26-35	15.4	14.3	12.3	14.3	14.0
		36-45	16.7	21.4	18.5	23.8	19.2
		46-55	30.8	28.6	23.5	16.7	25.3
		>55	35.9	35.7	43.2	35.7	38.4
	educ	No qualification literate	17.9	7.1	23.5	9.5	17.0
		Primary	26.9	46.4	25.9	28.6	29.3
		Secondary	38.5	39.3	29.6	40.5	35.8
		More than secondary	15.4	7.1	17.3	19.0	15.7
		Other	1.3	0.0	3.7	2.4	2.2
	income_avg	lower than average	47.4	57.1	43.2	45.2	46.7
		somewhat lower than average	32.1	28.6	40.7	28.6	34.1
		about average	16.7	7.1	11.1	14.3	13.1
		somewhat higher than average	1.3	7.1	3.7	7.1	3.9
		higher than average	2.6	0.0	1.2	4.8	2.2
TN	Total		100.0	100.0	100.0	100.0	100.0
	gender	Male	31.0	51.7	45.9	42.4	42.0
		Female	69.0	48.3	54.1	57.6	58.0
	cl.age	≤25	28.6	3.4	27.0	19.7	20.7
		26-35	14.3	24.1	16.2	16.7	17.2
		36-45	23.8	31.0	27.0	21.2	24.7
		46-55	14.3	24.1	18.9	24.2	20.7
		>55	19.0	17.2	10.8	18.2	16.7
	educ	No qualification literate	0.0	10.3	0.0	3.0	2.9
		Primary	11.9	13.8	10.8	10.6	11.5
		Secondary	23.8	17.2	32.4	16.7	21.8
		More than secondary	52.4	48.3	37.8	48.5	47.1
		Other	11.9	10.3	18.9	21.2	16.7
	income_avg	lower than average	4.8	10.3	13.5	9.1	9.2
		somewhat lower than average	45.2	44.8	37.8	36.4	40.2
		about average	28.6	27.6	32.4	25.8	28.2
		somewhat higher than average	14.3	17.2	13.5	21.2	17.2
		higher than average	7.1	0.0	2.7	7.6	5.2
TZ	Total		100.0	100.0	100.0	100.0	100.0
	gender	Male	41.9	33.9	48.5	50.0	43.9
		Female	58.1	66.1	51.5	50.0	56.1
	cl.age	≤25	36.0	19.6	21.2	22.9	25.9
		26-35	27.9	33.9	45.5	24.3	32.4
		36-45	18.6	14.3	19.7	34.3	21.9
		46-55	9.3	14.3	6.1	5.7	8.6
		>55	8.1	17.9	7.6	12.9	11.2
	educ	No qualification literate	0.0	0.0	1.5	0.0	0.4
		Primary	40.7	50.0	40.9	42.9	43.2
		Secondary	40.7	33.9	34.8	35.7	36.7
		More than secondary	18.6	14.3	16.7	18.6	17.3
		Other	0.0	1.8	6.1	2.9	2.5
	income_avg	lower than average	45.3	41.1	56.1	31.4	43.5
		somewhat lower than average	40.7	35.7	33.3	40.0	37.8
		about average	5.8	14.3	9.1	11.4	9.7
		somewhat higher than average	5.8	5.4	1.5	10.0	5.8
		higher than average	2.3	3.6	0.0	7.1	3.2
UG	Total			100.0	100.0	100.0	100.0
	gender	Male		40.0	0.0	40.0	37.5
		Female		60.0	100.0	60.0	62.5
	cl.age	≤25		10.0	33.3	40.0	14.6
		26-35		50.0	33.3	20.0	45.8
		36-45		20.0	33.3	20.0	20.8
		46-55		15.0	0.0	20.0	14.6
		>55		5.0	0.0	0.0	4.2
	educ	No qualification literate		10.0	0.0	20.0	10.4
		Primary		25.0	66.7	0.0	25.0
		Secondary		32.5	33.3	60.0	35.4
		More than secondary		27.5	0.0	20.0	25.0
		Other		5.0	0.0	0.0	4.2
	income_avg	lower than average		35.0	0.0	20.0	31.2
		somewhat lower than average		50.0	100.0	80.0	56.2
		about average		10.0	0.0	0.0	8.3
		somewhat higher than average		5.0	0.0	0.0	4.2
		higher than average		0.0	0.0	0.0	0.0

Table 1.11: Testing the relationship between clusters and the following variables at country-level (Chi-squared test)
variable	country	statistic	df	p.value	Signif
gender	KE	2.77	3	0.4287
	MO	7.19	3	0.0660	.
	TN	3.47	3	0.3244
	TZ	4.03	3	0.2585
	UG	1.92	2	0.3829
	Total	3.33	3	0.3431
age	KE	16.28	12	0.1789
	MO	11.71	12	0.4689
	TN	10.57	12	0.5657
	TZ	24.71	12	0.0163	*
	UG	5.68	8	0.6831
	Total	28.52	12	0.0046	**
education	KE	13.95	12	0.3041
	MO	13.15	12	0.3581
	TN	14.33	12	0.2802
	TZ	10.66	12	0.5585
	UG	6.35	8	0.6085
	Total	23.17	12	0.0263	*
income	KE	11.13	12	0.5177
	MO	9.99	12	0.6170
	TN	6.89	12	0.8650
	TZ	18.31	12	0.1067
	UG	4.32	6	0.6335
	Total	16.18	12	0.1831

2 DQQ

Table 2.1: variable under analysis in DQQ cluster
Variabile	Tipo	Levels
vegetable_fruit	factor	No; Yes
animalSource	factor	No; Yes
sweetBeverage	factor	No; Yes
sugarSweetDrink	factor	No; Yes
sweetFood	factor	No; Yes
saltyFriedSnack	factor	No; Yes
WholeGrain	factor	No; Yes
Pulse	factor	No; Yes
NutsSeeds	factor	No; Yes
ProcessedMeat	factor	No; Yes
All.5	factor	No; Yes
NCD_P	numeric	0:9
NCD_R	numeric	0:9
GDR	numeric	0:18
FGDS	numeric	0:10

2.1 Cluster analysis: choice of algorithm and number of groups

The choice of the agorithm and the optimal number of clusters was made using a distance matrix obtained with the Gower metric.

Comparison between algorithms and by number of clusters; analysis on standardised data; <br>algorithms: pam: partition around medoids, ward.D2: hierarchical clustering by Ward's method; <br>indices: S=Silhouette, C=Hubert & Levin C index

Figure 2.1: Comparison between algorithms and by number of clusters; analysis on standardised data;
algorithms: pam: partition around medoids, ward.D2: hierarchical clustering by Ward’s method;
indices: S=Silhouette, C=Hubert & Levin C index

In questo caso, avendo un mix di variabili quantitative e qualitative, si sono potuti mettere a confronto solo gli algoritmi di Partition Around Medoids e di cluster gerarchica, sempre ottenuta con il metodo di Ward;
in particolare si è calcolata la matrice di distanze usando la metrica di Gower e sono stati valutati i valori degli indici di Silhouette e C di Hubert & Levin (data la struttura dei dati non è possibile usare Calinski-Harabasz e Davies-Bouldin) per tutti i possibili raggruppamenti da 2 a 10.
In generale l’algoritmo PAM fornisce risultati migliori per entrambi gli indici; per quanto riguarda invece la scelta del numero di gruppi, non essendoci differenze significative nei valori di Silhouette tra i diversi raggruppamenti, si è optato per una soluzione che non portasse ad un overfitting dei gruppi.

choice: partition around medoids with 4 clusters

2.2 Cluster analysis results

Table 2.2: number of consumers by cluster
levels	N	%
1	1,556	29.8
2	757	14.5
3	1,445	27.7
4	1,468	28.1
Total	5,226	100.0

Table 2.3: clusters profiling by the variables in analysis
variable	1	2	3	4	Total
vegetable_fruit	94.30	97.50	100.00	100.00	98.00
animalSource	79.90	95.50	96.20	98.00	91.80
sweetBeverage	78.50	94.70	85.70	87.00	85.20
sugarSweetDrink	6.60	72.50	19.40	33.10	27.10
sweetFood	19.60	81.00	23.90	63.50	42.00
saltyFriedSnack	20.40	80.60	30.30	76.20	47.50
WholeGrain	27.40	26.40	28.90	75.10	41.10
Pulse	21.30	19.30	89.00	76.30	55.20
NutsSeeds	12.50	21.70	28.40	79.30	37.00
ProcessedMeat	5.80	22.20	9.10	19.70	13.00
All.5	2.40	19.20	90.20	95.20	55.20
NCD_P	3.09	3.79	5.03	6.56	4.70
NCD_R	1.16	4.08	1.76	3.73	2.47
GDR	10.93	8.71	12.27	11.84	11.23
FGDS	4.91	6.15	7.14	8.33	6.67
variable from vegetable_fruit to All.5: % values variable from NCD_P to FGDS: average values

Figure 2.2: Cluster profiling by the variables in analysis

Table 2.4: Pairwise comparisons of DQQ clusters by cluster variables
variable	comparison	p.value	signif
NCD_P	1 - 2	<0.001	***
	1 - 3	<0.001	***
	1 - 4	<0.001	***
	2 - 3	<0.001	***
	2 - 4	<0.001	***
	3 - 4	<0.001	***
NCD_R	1 - 2	<0.001	***
	1 - 3	<0.001	***
	1 - 4	<0.001	***
	2 - 3	<0.001	***
	2 - 4	<0.001	***
	3 - 4	<0.001	***
GDR	1 - 2	<0.001	***
	1 - 3	<0.001	***
	1 - 4	<0.001	***
	2 - 3	<0.001	***
	2 - 4	<0.001	***
	3 - 4	<0.001	***
FGDS	1 - 2	<0.001	***
	1 - 3	<0.001	***
	1 - 4	<0.001	***
	2 - 3	<0.001	***
	2 - 4	<0.001	***
	3 - 4	<0.001	***
Pairwise comparisons using Wilcoxon rank sum test with continuity correction P value adjustment method: bonferroni

Figure 2.3: Mosaic plot DQQ cluster and qualitative cluster variables

Clusters summary:

1: lack (only 2.4% consume all five recommended food groups, have the lowest FGDS value);
2: little and badly (less than 20 per cent consume all five recommended food groups but at the same time more than 80 per cent consume sweetFood, saltyFriedSnack, sweetBeverage, and are those who consume processedMeat the most);
3: balanced (90% consume the five recommended food groups with an above-average FGDS);
4: much but also badly (almost all consume the five recommended food groups (95%) but, at the same time have a high NCD-Risk score, with consumption of sweetFood, saltyFriedSnack, sweetBeverage and processedMeat).

Figure 2.4: clusters distribution

Figure 2.5: Clusters DQQ by socioeconomic variables

Table 2.5: Clusters DQQ by socioeconomic variables - row %
variable	levels	1	2	3	4	Total
Total	Total	29.8	14.5	27.7	28.1	100
country	KE	22.4	13.7	28.8	35.1	100
	MO	51.0	14.5	19.4	15.0	100
	TN	23.2	20.5	26.2	30.1	100
	TZ	22.3	11.7	36.3	29.7	100
	UG	35.7	13.4	25.9	24.9	100
gender	Male	30.5	13.9	28.3	27.4	100
	Female	29.2	15.0	27.1	28.7	100
cl.age	≤25	27.6	19.3	26.6	26.5	100
	26-35	27.0	15.9	28.2	28.9	100
	36-45	27.8	12.5	28.3	31.4	100
	46-55	32.0	12.3	25.7	30.0	100
	>55	40.3	9.6	29.2	20.8	100
educ	No qualification literate	55.4	7.8	20.0	16.8	100
	Primary	29.7	12.5	32.2	25.7	100
	Secondary	27.7	15.3	26.7	30.3	100
	More than secondary	25.1	17.2	26.8	30.9	100
	Other	24.3	17.9	26.0	31.8	100
income_avg	lower than average	40.5	12.9	24.0	22.7	100
	somewhat lower than average	30.2	14.9	26.5	28.5	100
	about average	22.9	17.1	26.4	33.7	100
	somewhat higher than average	26.6	15.1	28.1	30.2	100
	higher than average	23.5	12.3	37.5	26.7	100

Table 2.6: Clusters DQQ by socioeconomic variables - col %
variable	livelli	1	2	3	4	Total
Total		100.0	100.0	100.0	100.0	100.0
country	KE	21.3	26.7	29.4	35.4	28.3
	MO	24.0	14.0	9.8	7.5	14.0
	TN	12.4	22.6	15.1	17.1	15.9
	TZ	13.0	14.0	22.8	18.3	17.3
	UG	29.4	22.7	22.9	21.7	24.5
gender	Male	45.9	42.9	45.9	43.7	44.9
	Female	54.1	57.1	54.1	56.3	55.1
cl.age	≤25	19.4	27.9	20.1	19.8	20.9
	26-35	26.3	31.7	29.6	29.8	29.0
	36-45	20.9	19.3	22.8	25.0	22.3
	46-55	15.9	12.5	13.8	15.8	14.8
	>55	17.5	8.6	13.7	9.6	13.0
educ	No qualification literate	15.9	4.6	6.2	5.1	8.5
	Primary	27.6	23.8	32.2	25.3	27.7
	Secondary	30.7	34.7	31.9	35.6	33.0
	More than secondary	23.2	32.8	26.6	30.2	27.5
	Other	2.7	4.1	3.1	3.7	3.3
income_avg	lower than average	28.7	18.8	18.3	17.1	21.1
	somewhat lower than average	35.8	36.3	33.8	35.8	35.3
	about average	12.2	18.8	15.2	19.1	15.9
	somewhat higher than average	12.6	14.7	14.3	15.1	14.1
	higher than average	10.7	11.5	18.3	12.9	13.5

Table 2.7: Test on differences in clusters DQQ by the following variables at project level
variable	test	statistic	df	p.value	signif
country	Chi.square	331.95	12	9.0e-64	***
gender	Chi.square	3.16	3	0.3669
cl.age	Chi.square	92.00	12	2.0e-14	***
educ	Chi.square	192.89	12	9.5e-35	***
income_avg	Chi.square	130.15	12	5.8e-22	***

Figure 2.6: Correspondence Analysis - Clusters DQQ

Figure 2.7: Clusters DQQ by gender, age, education level and income at country level

Table 2.8: Clusters DQQ by gender, age, education level and income at country level - row %
fattore	variable	levels	1	2	3	4	Total
KE	Total	Total	22.4	13.7	28.8	35.1	100
	gender	Male	23.2	12.5	30.9	33.4	100
		Female	21.8	14.6	27.2	36.4	100
	cl.age	≤25	23.4	18.5	31.8	26.3	100
		26-35	22.2	12.7	27.8	37.3	100
		36-45	18.3	13.2	28.1	40.4	100
		46-55	23.4	10.8	26.8	39.0	100
		>55	27.8	12.2	29.4	30.6	100
	educ	No qualification literate	32.5	14.3	20.8	32.5	100
		Primary	22.2	13.0	35.5	29.2	100
		Secondary	21.7	13.5	27.7	37.0	100
		More than secondary	21.4	14.8	24.2	39.5	100
		Other	30.0	0.0	50.0	20.0	100
	income_avg	lower than average	24.9	17.1	15.1	42.9	100
		somewhat lower than average	15.1	13.2	17.1	54.6	100
		about average	19.9	12.9	23.7	43.6	100
		somewhat higher than average	25.4	14.1	29.9	30.7	100
		higher than average	22.5	12.5	39.9	25.2	100
MO	Total	Total	51.0	14.5	19.4	15.0	100
	gender	Male	52.6	13.5	18.3	15.5	100
		Female	47.6	16.6	21.8	14.0	100
	cl.age	≤25	36.4	20.5	19.3	23.9	100
		26-35	47.6	17.3	20.2	14.9	100
		36-45	48.2	15.8	18.7	17.3	100
		46-55	53.8	12.7	17.7	15.8	100
		>55	61.2	9.6	20.8	8.4	100
	educ	No qualification literate	76.5	5.1	16.2	2.2	100
		Primary	52.9	13.4	21.0	12.7	100
		Secondary	52.1	18.0	14.4	15.5	100
		More than secondary	33.6	16.1	27.0	23.2	100
		Other	42.4	27.3	6.1	24.2	100
	income_avg	lower than average	63.9	11.7	17.4	7.0	100
		somewhat lower than average	49.8	15.8	19.6	14.8	100
		about average	29.5	20.5	19.3	30.7	100
		somewhat higher than average	40.8	17.1	23.7	18.4	100
		higher than average	35.6	11.9	23.7	28.8	100
TN	Total	Total	23.2	20.5	26.2	30.1	100
	gender	Male	21.4	20.3	28.5	29.9	100
		Female	24.6	20.7	24.4	30.3	100
	cl.age	≤25	22.9	32.4	17.6	27.1	100
		26-35	16.9	24.9	27.7	30.5	100
		36-45	21.8	16.8	25.1	36.3	100
		46-55	24.7	15.6	28.6	31.2	100
		>55	30.7	11.8	32.7	24.8	100
	educ	No qualification literate	47.1	23.5	11.8	17.6	100
		Primary	27.0	15.1	32.5	25.4	100
		Secondary	22.9	22.0	26.2	29.0	100
		More than secondary	21.9	21.6	25.2	31.4	100
		Other	19.5	19.5	24.1	36.8	100
	income_avg	lower than average	37.5	23.4	20.3	18.8	100
		somewhat lower than average	23.5	21.8	27.0	27.6	100
		about average	20.5	22.4	26.3	30.9	100
		somewhat higher than average	24.3	15.5	23.6	36.5	100
		higher than average	15.9	15.9	33.3	34.8	100
TZ	Total	Total	22.3	11.7	36.3	29.7	100
	gender	Male	22.0	9.6	37.3	31.1	100
		Female	22.5	13.5	35.5	28.5	100
	cl.age	≤25	20.1	13.2	32.3	34.4	100
		26-35	20.8	13.8	34.0	31.4	100
		36-45	23.3	9.3	39.4	28.0	100
		46-55	21.0	13.0	35.0	31.0	100
		>55	33.3	4.3	49.3	13.0	100
	educ	No qualification literate	8.3	8.3	58.3	25.0	100
		Primary	26.4	9.1	38.7	25.8	100
		Secondary	17.9	13.0	32.7	36.4	100
		More than secondary	23.2	15.3	35.6	26.0	100
		Other	20.7	10.3	41.4	27.6	100
	income_avg	lower than average	27.3	11.0	38.3	23.4	100
		somewhat lower than average	17.9	12.6	35.6	33.9	100
		about average	19.4	13.6	35.0	32.0	100
		somewhat higher than average	21.3	14.8	31.1	32.8	100
		higher than average	27.3	5.2	37.7	29.9	100
UG	Total	Total	35.7	13.4	25.9	24.9	100
	gender	Male	31.3	15.1	27.1	26.6	100
		Female	37.9	12.6	25.3	24.1	100
	cl.age	≤25	35.7	16.5	25.1	22.7	100
		26-35	32.3	16.2	27.7	23.8	100
		36-45	36.8	10.0	25.7	27.5	100
		46-55	37.7	10.0	23.1	29.2	100
		>55	45.4	5.2	24.7	24.7	100
	educ	No qualification literate	53.4	5.9	20.6	20.1	100
		Primary	32.3	13.8	26.8	27.1	100
		Secondary	34.8	14.7	26.7	23.8	100
		More than secondary	29.7	16.7	26.8	26.8	100
		Other	14.3	14.3	35.7	35.7	100
	income_avg	lower than average	42.4	10.0	22.3	25.3	100
		somewhat lower than average	35.5	13.5	25.8	25.1	100
		about average	30.7	15.0	29.3	25.0	100
		somewhat higher than average	28.3	18.9	30.2	22.6	100
		higher than average	23.8	23.8	33.3	19.0	100
Total	Total	Total	29.8	14.5	27.7	28.1	100
	gender	Male	30.5	13.9	28.3	27.4	100
		Female	29.2	15.0	27.1	28.7	100
	cl.age	≤25	27.6	19.3	26.6	26.5	100
		26-35	27.0	15.9	28.2	28.9	100
		36-45	27.8	12.5	28.3	31.4	100
		46-55	32.0	12.3	25.7	30.0	100
		>55	40.3	9.6	29.2	20.8	100
	educ	No qualification literate	55.4	7.8	20.0	16.8	100
		Primary	29.7	12.5	32.2	25.7	100
		Secondary	27.7	15.3	26.7	30.3	100
		More than secondary	25.1	17.2	26.8	30.9	100
		Other	24.3	17.9	26.0	31.8	100
	income_avg	lower than average	40.5	12.9	24.0	22.7	100
		somewhat lower than average	30.2	14.9	26.5	28.5	100
		about average	22.9	17.1	26.4	33.7	100
		somewhat higher than average	26.6	15.1	28.1	30.2	100
		higher than average	23.5	12.3	37.5	26.7	100

Table 2.9: Clusters DQQ by gender, age, education level and income at country level - col %
fattore	variable	livelli	1	2	3	4	Total
KE	Total		100.0	100.0	100.0	100.0	100.0
	gender	Male	44.4	39.1	46.1	40.8	42.9
		Female	55.6	60.9	53.9	59.2	57.1
	cl.age	≤25	21.8	28.2	23.1	15.6	20.9
		26-35	28.4	26.7	27.8	30.4	28.7
		36-45	18.4	21.8	22.1	26.0	22.6
		46-55	16.3	12.4	14.6	17.3	15.6
		>55	15.1	10.9	12.5	10.6	12.2
	educ	No qualification literate	7.6	5.4	3.8	4.8	5.2
		Primary	27.8	26.7	34.6	23.3	28.0
		Secondary	38.4	39.1	38.1	41.6	39.5
		More than secondary	25.4	28.7	22.4	29.9	26.5
		Other	0.9	0.0	1.2	0.4	0.7
	income_avg	lower than average	15.4	17.3	7.3	17.0	13.9
		somewhat lower than average	6.9	9.9	6.1	16.0	10.3
		about average	14.5	15.3	13.4	20.2	16.3
		somewhat higher than average	30.5	27.7	28.0	23.5	26.9
		higher than average	32.6	29.7	45.2	23.3	32.6
MO	Total		100.0	100.0	100.0	100.0	100.0
	gender	Male	70.8	64.2	64.8	70.9	68.7
		Female	29.2	35.8	35.2	29.1	31.3
	cl.age	≤25	8.6	17.0	12.0	19.1	12.0
		26-35	21.4	27.4	23.9	22.7	23.0
		36-45	18.0	20.8	18.3	21.8	19.0
		46-55	22.8	18.9	19.7	22.7	21.6
		>55	29.2	16.0	26.1	13.6	24.4
	educ	No qualification literate	27.9	6.6	15.5	2.7	18.6
		Primary	22.3	19.8	23.2	18.2	21.5
		Secondary	27.1	33.0	19.7	27.3	26.5
		More than secondary	19.0	32.1	40.1	44.5	28.9
		Other	3.8	8.5	1.4	7.3	4.5
	income_avg	lower than average	51.2	33.0	36.6	19.1	40.9
		somewhat lower than average	27.9	31.1	28.9	28.2	28.6
		about average	7.0	17.0	12.0	24.5	12.0
		somewhat higher than average	8.3	12.3	12.7	12.7	10.4
		higher than average	5.6	6.6	9.9	15.5	8.1
TN	Total		100.0	100.0	100.0	100.0	100.0
	gender	Male	40.4	43.3	47.7	43.4	43.8
		Female	59.6	56.7	52.3	56.6	56.2
	cl.age	≤25	20.2	32.2	13.8	18.3	20.4
		26-35	15.5	25.7	22.5	21.5	21.2
		36-45	20.2	17.5	20.6	25.9	21.5
		46-55	19.7	14.0	20.2	19.1	18.5
		>55	24.4	10.5	22.9	15.1	18.4
	educ	No qualification literate	4.1	2.3	0.9	1.2	2.0
		Primary	17.6	11.1	18.8	12.7	15.1
		Secondary	25.4	27.5	25.7	24.7	25.7
		More than secondary	44.0	49.1	45.0	48.6	46.7
		Other	8.8	9.9	9.6	12.7	10.4
	income_avg	lower than average	12.4	8.8	6.0	4.8	7.7
		somewhat lower than average	35.8	37.4	36.2	32.3	35.2
		about average	27.5	33.9	31.2	31.9	31.1
		somewhat higher than average	18.7	13.5	16.1	21.5	17.8
		higher than average	5.7	6.4	10.6	9.6	8.3
TZ	Total		100.0	100.0	100.0	100.0	100.0
	gender	Male	45.5	37.7	47.4	48.3	46.1
		Female	54.5	62.3	52.6	51.7	53.9
	cl.age	≤25	18.8	23.6	18.5	24.2	20.9
		26-35	32.2	40.6	32.2	36.4	34.4
		36-45	27.2	20.8	28.3	24.5	26.0
		46-55	10.4	12.3	10.6	11.5	11.0
		>55	11.4	2.8	10.3	3.3	7.6
	educ	No qualification literate	0.5	0.9	2.1	1.1	1.3
		Primary	47.5	31.1	42.9	34.9	40.2
		Secondary	28.7	39.6	32.2	43.9	35.8
		More than secondary	20.3	25.5	19.1	17.1	19.5
		Other	3.0	2.8	3.6	3.0	3.2
	income_avg	lower than average	41.6	32.1	35.9	26.8	34.0
		somewhat lower than average	31.7	42.5	38.6	45.0	39.4
		about average	9.9	13.2	10.9	12.3	11.4
		somewhat higher than average	6.4	8.5	5.8	7.4	6.7
		higher than average	10.4	3.8	8.8	8.6	8.5
UG	Total		100.0	100.0	100.0	100.0	100.0
	gender	Male	29.1	37.2	34.7	35.4	33.2
		Female	70.9	62.8	65.3	64.6	66.8
	cl.age	≤25	26.5	32.6	25.7	24.1	26.5
		26-35	30.6	40.7	36.3	32.3	33.9
		36-45	22.5	16.3	21.8	24.1	21.9
		46-55	10.7	7.6	9.1	11.9	10.2
		>55	9.6	2.9	7.3	7.5	7.6
	educ	No qualification literate	23.9	7.0	12.7	12.9	15.9
		Primary	27.1	30.8	31.1	32.6	30.0
		Secondary	31.1	34.9	32.9	30.4	31.9
		More than secondary	17.5	26.2	21.8	22.6	21.0
		Other	0.4	1.2	1.5	1.6	1.1
	income_avg	lower than average	21.2	13.4	15.4	18.2	17.9
		somewhat lower than average	65.0	65.7	65.3	65.8	65.4
		about average	9.4	12.2	12.4	11.0	10.9
		somewhat higher than average	3.3	5.8	4.8	3.8	4.1
		higher than average	1.1	2.9	2.1	1.3	1.6

Table 2.10: Test on differences in clusters DQQ by the following variables at country level (Chi-squared test)
variable	country	statistic	df	p.value	Signif
gender	KE	4.18	3	0.2425
	MO	3.03	3	0.3875
	TN	2.28	3	0.5157
	TZ	3.78	3	0.2868
	UG	5.77	3	0.1234
	Total	3.16	3	0.3669
age	KE	25.94	12	0.0110	*
	MO	25.51	12	0.0126	*
	TN	41.07	12	4.8e-05	***
	TZ	24.30	12	0.0185	*
	UG	21.18	12	0.0478	*
	Total	92.00	12	2.0e-14	***
education	KE	26.38	12	0.0095	**
	MO	83.91	12	7.4e-13	***
	TN	15.11	12	0.2357
	TZ	22.64	12	0.0310	*
	UG	42.03	12	3.3e-05	***
	Total	192.89	12	9.5e-35	***
income	KE	93.14	12	1.2e-14	***
	MO	63.67	12	4.8e-09	***
	TN	20.10	12	0.0651	.
	TZ	19.27	12	0.0821	.
	UG	13.57	12	0.3292
	Total	130.15	12	5.8e-22	***

3 comparison between cluster Experiments - DQQ

Table 3.1: cluster DQQ - cluster Expriment
stat	clusterDQQ	1-untrusting	2-risk-takers	3-impatient	4-trusting	Total
count	1-lack	65	37	57	43	202
	2-little and badly	34	14	21	23	92
	3-balanced	73	48	63	44	228
	4-much but badly	65	50	40	67	222
	Total	237	149	181	177	744
row.%	1-lack	32.2	18.3	28.2	21.3	100.0
	2-little and badly	37.0	15.2	22.8	25.0	100.0
	3-balanced	32.0	21.1	27.6	19.3	100.0
	4-much but badly	29.3	22.5	18.0	30.2	100.0
	Total	31.9	20.0	24.3	23.8	100.0
col.%	1-lack	27.4	24.8	31.5	24.3	27.2
	2-little and badly	14.3	9.4	11.6	13.0	12.4
	3-balanced	30.8	32.2	34.8	24.9	30.6
	4-much but badly	27.4	33.6	22.1	37.9	29.8
	Total	100.0	100.0	100.0	100.0	100.0
overall.%	1-lack	8.7	5.0	7.7	5.8	27.2
	2-little and badly	4.6	1.9	2.8	3.1	12.4
	3-balanced	9.8	6.5	8.5	5.9	30.6
	4-much but badly	8.7	6.7	5.4	9.0	29.8
	Total	31.9	20.0	24.3	23.8	100.0

Table 3.2: Chi-squared test between clusterDQQ and clusterExp
Test	Statistic	df	p.value	Signif
Pearson's Chi-squared test	15.72	9	0.0730	.

Figure 3.1: clusterDQQ - clusterExp

Table 3.3: Test on relationship between clusterDQQ and clusterExp at country level (Chi-squared test)
country	statistic	df	p.value	Signif
KE	6.91	9	0.6465
MO	8.84	9	0.4521
TN	5.63	9	0.7768
TZ	8.58	9	0.4773
UG	12.61	6	0.0496	*
Total	15.72	9	0.0729	.

3.1 DQQ variables by Experiment clusters

Table 3.4: Testing the relationship between Experiment clusters and the DQQ variables at project level
variable	valid.N	test	statistic	df	p.value	signif
MDD.W	366	Chi.square	2.83	3	0.4187
All.5	748	Chi.square	2.64	3	0.4512
HDDS	747	Kruskal-Wallis	0.55	3	0.9075
FGDS	748	Kruskal-Wallis	11.52	3	0.0092	**
NCD_P	748	Kruskal-Wallis	10.27	3	0.0164	*
NCD_R	748	Kruskal-Wallis	0.59	3	0.8981
GDR	748	Kruskal-Wallis	4.80	3	0.1871

Table 3.5: Average values of quantitative DQQ variables by Experiment clusters
variable	1	2	3	4	Total
HDDS	8.34	8.21	8.25	8.27	8.28
FGDS	6.96	7.05	6.65	7.24	6.97
NCD_P	4.88	5.22	4.66	5.19	4.97
NCD_R	2.46	2.62	2.52	2.50	2.52
GDR	11.42	11.60	11.14	11.69	11.45

Figure 3.2: Average values of quantitative DQQ variables by Experiment clusters

Table 3.6: FGDS-Experiment cluster: Pairwise comparisons using Wilcoxon rank sum test with continuity correction
cluster	1	2	3
2	1.0000
3	0.3992	0.2494
4	0.3896	1.0000	0.0059
correction method: bonferroni

Table 3.7: NCD_P-Experiment cluster: Pairwise comparisons using Wilcoxon rank sum test with continuity correction
cluster	1	2	3
2	0.5332
3	1.0000	0.0312
4	0.6971	1.0000	0.0652
correction method: bonferroni

Table 3.8: qualitative DQQ variables by Experiment clusters - row %
variable	cluster	No	Yes	Total
MDD.W	1	7.0	93.0	100
	2	9.2	90.8	100
	3	14.0	86.0	100
	4	8.9	91.1	100
All.5	1	39.5	60.5	100
	2	37.6	62.4	100
	3	45.6	54.4	100
	4	39.7	60.3	100

Figure 3.3: qualitative DQQ variables by Experiment clusters

3.2 Experiment variables by DQQ clusters

3.2.1 Risk

Table 3.9: Risk propensity by DQQ clusters - summary statistics
clusterDQQ	n	%	Mean	st.Dev	min	Q1	Median	Q3	max
1	375	27.96	5.08	2.25	1	4	5	6.00	10
2	192	14.32	5.36	2.45	1	4	5	6.00	10
3	394	29.38	5.07	2.54	1	3	5	6.00	10
4	380	28.34	5.50	2.41	1	4	5	6.25	10
Total	1,341	100.00	5.24	2.42	1	4	5	6.00	10

Figure 3.4: Risk propensity distribution by DQQ clusters and country

Table 3.10: Average risk propensity by DQQ clusters and country
country	1	2	3	4	Total
KE	5.54	5.04	4.20	4.90	4.79
MO	4.73	4.50	5.05	4.94	4.80
TN	4.79	5.09	4.99	5.32	5.09
TZ	4.81	5.85	5.31	5.26	5.25
UG	9.25	8.33	8.40	8.86	8.72
Total	5.08	5.36	5.07	5.50	5.24

Table 3.11: Test to verify the relationship between risk propensity and DQQ cluster at country level (Kruskal-Wallis test)
country	statistic	df	p.value	Signif
KE	8.27	3	0.0408	*
MO	4.44	3	0.2181
TN	3.06	3	0.3820
TZ	4.78	3	0.1885
UG	2.51	3	0.4729
Total	9.36	3	0.0249	*

3.2.2 Time

Table 3.12: Time Preferences by DQQ clusters - summary statistics
clusterDQQ	n	%	Mean	st.Dev	Q1	Median	Q3	max
1	478	27.69	6.05	3.52	4	6	10	10
2	277	16.05	5.63	3.68	3	6	9	10
3	485	28.10	5.80	3.65	3	6	10	10
4	486	28.16	6.49	3.48	4	7	10	10
Total	1,726	100.00	6.04	3.58	4	6	10	10

Figure 3.5: Time Preferences distribution by DQQ clusters and country

Table 3.13: Average Time Preferences by DQQ clusters and country
country	1	2	3	4	Total
KE	5.77	5.97	5.62	5.80	5.75
MO	5.64	5.25	5.05	5.89	5.51
TN	6.14	5.87	5.52	6.08	5.91
TZ	6.31	6.00	6.47	7.78	6.80
UG	7.19	5.10	6.23	6.54	6.30
Total	6.05	5.63	5.80	6.49	6.04

Table 3.14: Test to verify the relationship between Time Preferences and DQQ cluster at country level (Kruskal-Wallis test)
country	statistic	df	p.value	Signif
KE	0.17	3	0.9824
MO	3.50	3	0.3212
TN	1.83	3	0.6083
TZ	9.23	3	0.0264	*
UG	8.70	3	0.0336	*
Total	13.64	3	0.0034	**

3.2.3 Trust game

Table 3.15: Trust by DQQ clusters - summary statistics
variable	clusterDQQ	n	%	Mean	st.Dev	Q1	Median	Q3	max
player.trust	1	326	25.00	39.72	22.74	25	25	50	100
	2	188	14.42	43.48	26.63	25	50	50	100
	3	392	30.06	36.80	23.84	25	25	50	100
	4	398	30.52	40.70	23.97	25	25	50	100
	Total	1,304	100.00	39.69	24.11	25	25	50	100
player.trust_inst	1	326	25.00	46.47	26.07	25	50	50	100
	2	188	14.42	50.80	28.19	25	50	75	100
	3	392	30.06	45.34	26.86	25	50	75	100
	4	398	30.52	51.13	26.47	25	50	75	100
	Total	1,304	100.00	48.18	26.84	25	50	75	100

Figure 3.6: player.trust distribution by DQQ clusters and country

Figure 3.7: player.trust_inst distribution by DQQ clusters and country

Table 3.16: Average trust by DQQ clusters and country
variable	country	1	2	3	4	Total
player.trust	KE	35.8	34.1	31.1	31.2	32.8
	MO	39.0	38.3	39.6	41.7	39.6
	TN	42.3	42.2	44.6	45.0	43.9
	TZ	40.2	41.0	34.5	39.7	38.1
	UG	43.8	58.3	48.7	45.8	48.7
	Total	39.7	43.5	36.8	40.7	39.7
player.trust_inst	KE	30.6	31.8	30.6	35.3	31.8
	MO	42.3	41.7	41.0	49.2	43.3
	TN	53.2	50.5	55.9	55.8	54.3
	TZ	51.1	50.0	48.0	52.3	50.3
	UG	63.1	71.8	62.8	57.4	63.2
	Total	46.5	50.8	45.3	51.1	48.2

Table 3.17: Test to verify the relationship between trust and DQQ cluster at country level (Kruskal-Wallis test)
variable	country	statistic	df	p.value	Signif
player.trust	KE	2.19	3	0.5337
	MO	0.84	3	0.8395
	TN	0.74	3	0.8640
	TZ	6.80	3	0.0784	.
	UG	6.51	3	0.0893	.
	Total	9.83	3	0.0200	*
player.trust_inst	KE	2.19	3	0.5339
	MO	1.79	3	0.6178
	TN	1.40	3	0.7053
	TZ	2.44	3	0.4870
	UG	6.92	3	0.0744	.
	Total	11.31	3	0.0101	*

3.2.4 Public Goods Game

Table 3.18: PGG by DQQ clusters - summary statistics
variable	clusterDQQ	n	%	Mean	st.Dev	Q1	Median	Q3	max
PGG.1	1	240	27.65	71.22	39.88	50	50	100	150
	2	161	18.55	70.61	38.87	50	50	100	150
	3	235	27.07	69.57	41.03	50	50	100	150
	4	232	26.73	73.84	43.23	50	70	100	150
	Total	868	100.00	71.36	40.89	50	50	100	150
PGG.2	1	240	27.65	71.49	42.03	50	50	100	150
	2	161	18.55	72.26	41.83	50	60	100	150
	3	235	27.07	69.38	40.15	50	50	100	150
	4	232	26.73	74.34	46.03	50	70	100	150
	Total	868	100.00	71.82	42.58	50	60	100	150

Figure 3.8: PGG.1 distribution by DQQ clusters and country

Figure 3.9: PGG.2 distribution by DQQ clusters and country

Table 3.19: Average PGG by DQQ clusters and country
variable	country	1	2	3	4	Total
PGG.1	KE	66.0	60.6	63.7	60.4	63.0
	MO	68.5	75.6	67.4	88.8	72.8
	TN	78.9	74.5	82.4	73.2	77.0
	TZ
	UG	76.1	70.4	69.4	79.1	74.2
	Total	71.2	70.6	69.6	73.8	71.4
PGG.2	KE	64.4	58.0	64.5	62.0	63.0
	MO	67.5	81.4	67.5	70.9	70.5
	TN	78.9	77.8	74.2	79.7	77.7
	TZ
	UG	81.0	70.6	74.9	81.5	77.4
	Total	71.5	72.3	69.4	74.3	71.8

Table 3.20: Test to verify the relationship between PGG and DQQ cluster at country level (Kruskal-Wallis test)
variable	country	statistic	df	p.value	Signif
PGG.1	KE	0.49	3	0.9216
	MO	9.74	3	0.0209	*
	TN	1.71	3	0.6347
	UG	2.50	3	0.4762
	Total	1.61	3	0.6570
PGG.2	KE	0.50	3	0.9196
	MO	4.73	3	0.1929
	TN	0.38	3	0.9437
	UG	2.23	3	0.5257
	Total	1.22	3	0.7476

4 Innovation by clusters

4.1 DQQ cluster by interest in innovation and in local products

4.1.1 Q12. New Food Product

Table 4.1: Choice on a new food product by DQQ cluster
stat	tech_avsb	1	2	3	4	Total
count	A;C	262	122	187	355	926
	B;D	818	473	901	825	3017
	Total	1080	595	1088	1180	3943
row.%	A;C	28.3	13.2	20.2	38.3	100.0
	B;D	27.1	15.7	29.9	27.3	100.0
	Total	27.4	15.1	27.6	29.9	100.0
col.%	A;C	24.3	20.5	17.2	30.1	23.5
	B;D	75.7	79.5	82.8	69.9	76.5
	Total	100.0	100.0	100.0	100.0	100.0

Table 4.2: Chi-squared test on relationship between Choice on a new food product by DQQ cluster
Test	Statistic	df	p.value	Signif
Pearson's Chi-squared test	55.92	3	4.4e-12	***

Figure 4.1: Mosaic plot Choice on a new food product by DQQ cluster

4.1.2 Q13. Interest in a new nutrient dense Food Product

Table 4.3: Interest in a new Food Product by DQQ cluster - summary statistics
clusterDQQ	n	%	Mean	st.Dev	min	Q1	Median	Q3	max	NAs
1	1,556	29.77	3.96	1.20	1	3	4	5	5	5
2	757	14.49	4.14	1.05	1	4	4	5	5	2
3	1,445	27.65	4.17	0.99	1	4	4	5	5	4
4	1,468	28.09	4.34	0.96	1	4	5	5	5	4
Total	5,226	100.00	4.15	1.07	1	4	4	5	5	15

Table 4.4: Kruskal-Wallis test on relationship between Interest in a new Food Product and DQQ cluster
Test	Statistic	df	p.value	Signif
Kruskal-Wallis rank sum test	86.81	3	1.1e-18	***

Table 4.5: multiple pairwise comparisons (with Bonferroni correction)
cluster	1	2	3
2	0.0283	-	-
3	0.0022	1	-
4	<0.001	<0.001	<0.001

Figure 4.2: multiple pairwise comparisons (with Bonferroni correction)

4.1.3 Q6. Interest in a new local food product

Table 4.6: Interest in a new local food Product by DQQ cluster - summary statistics
clusterDQQ	n	%	Mean	st.Dev	min	Q1	Median	Q3	max	NAs
1	1,556	29.77	3.88	1.15	1	3	4	5	5	0
2	757	14.49	4.07	1.06	1	4	4	5	5	1
3	1,445	27.65	3.99	1.11	1	3	4	5	5	1
4	1,468	28.09	4.07	1.01	1	3	4	5	5	1
Total	5,226	100.00	3.99	1.09	1	3	4	5	5	3

Table 4.7: Kruskal-Wallis test on relationship between Interest in a new local Food Product and DQQ cluster
Test	Statistic	df	p.value	Signif
Kruskal-Wallis rank sum test	22.56	3	5e-05	***

Table 4.8: multiple pairwise comparisons (with Bonferroni correction)
cluster	1	2	3
2	<0.001	-	-
3	0.0416	0.71	-
4	<0.001	1	1

Figure 4.3: multiple pairwise comparisons (with Bonferroni correction)

4.2 Experiment cluster by interest in innovation and in local products

4.2.1 Q12. New Food Product

Table 4.9: Choice on a new food product by Experiment cluster
stat	tech_avsb	1	2	3	4	Total
count	A;C	53	23	41	43	160
	B;D	206	155	169	153	683
	Total	259	178	210	196	843
row.%	A;C	33.1	14.4	25.6	26.9	100.0
	B;D	30.2	22.7	24.7	22.4	100.0
	Total	30.7	21.1	24.9	23.3	100.0
col.%	A;C	20.5	12.9	19.5	21.9	19.0
	B;D	79.5	87.1	80.5	78.1	81.0
	Total	100.0	100.0	100.0	100.0	100.0

Figure 4.4: Mosaic plot Choice on a new food product by Experiment cluster

Urban consumers - clustering DQQ and experiments

2023-03-24