Table 1.

Definition of variables used in the regressions, and summary statistics for the latest survey (2010)

Name	Definition	N	Mean	SD
[A] Outcome variables:
[i] Perceived stress and perceived morbidity:
Stress	“During the last 7 days how often did you feel……?” Negative emotions included: sad, fear, angry, and unhappy Mean of total episodes of negative emotions used.	607	4.35	1.33
Beda	Person reported being bed-ridden in last 2 weeks=1; zero otherwise.	612	0.21	0.41
Illa	Person reported having had an illness or symptom of illness in last 2 weeks=1; zero otherwise.	612	0.51	0.50
[ii] Addiction:
Addictionb	Principal component factor analysis used to create a Z score for consumption in last seven days of commercial cigarettes and alcoholic beverages.	657	0	1
[iii] Anthropometric indicators of nutritional status:
Arm (muscle)c	Age and sex-Z score of arm muscle area based on Frisancho (2008) norms	611	−0.62	0.87
BMId	Body-mass index (weight [kg]/standing height [m2])	454	23.86	2.93
[B] Gini coefficients of village inequality (H, M, and L = high, medium, and low visibility):
Wildlife	Kg of wildlife (wild meat, fish, and birds) eaten by person in last 7 days/# of household residents. Per capita indices computed from household data/household size (H)	13	0.43	0.09
Meat	Kg of meat from chickens, pigs, cattle, and ducks eaten by person in last 7 days/# of household residents. Per capita indices computed from household data/household size (H)	13	0.57	0.14
Durablese	Monetary expenditures in physical durable assets in last 7 days (e.g., metal tools, plastic buckets). Per capita indices computed from individual data (M)	13	0.88	0.27
Luxuries	Monetary expenditures in luxuries in last seven days (e.g., watches, perfume). Per capita indices computed from individual data (M)	13	0.83	0.06
Plantings	Hectares (ha) planted with main crops last year: manioc, rice, maize, plantains. Per capita indices computed from household data (L)	13	0.36	0.06
Forest	Hectares (ha) of old-growth and fallow forest cleared for farming last year. Per capita indices computed from household data (L)	13	0.38	0.07
[C] Control variables:f
Male	Subject’s sex (male=1; female=0)	612	0.49	0.5
Education	Maximum school grade attained by subject	612	3.0	2.9
Age	Subject’s age in years	612	36	16
Year	Survey year
Count	# of times person was surveyed in panel	612	7.2	2.6

Notes: We limited the recall period for our measures of expenditures in the panel surveys to seven days for two reasons. First, we piloted several different ways of measuring socioeconomic variables. As is true in other native Amazonian societies (Gordon, 2004; Pica et al., 2004) among the Tsimane’ one finds limited cultural significance for numbers (Piantadosi et al., 2014; Undurraga et al., 2013). Because of Tsimane’ limited use of numbers and thus faulty recall we have found it hard to obtain reliable measures of expenditures, particularly in ordinary goods compared with salient goods. Second, because asking about expenditures and consumption over longer recall periods takes more time, we limited recall to one week in the 9-year panel study.

^aThe variables bed and ill capture different points in the continuum of perceived morbidity, with Bed capturing morbidity that is perceived as so severe that it restricts the person to bed and limits ordinary activities.

^bWe used addiction as a proxy for stress and negative emotions. Previous research suggest that economic inequalities affect mental health, including addiction (Pickett and Wilkinson, 2015), although some have argued that poorer health is rather a consequence of differing lifestyle and habits by socioeconomic status (Costa-Font et al., 2014). We also collected data on consumption of coca leaves and of a traditional home-fermented beverage (chicha), but did not include them as addictive substances. Chewing coca leaves is a highland trait that Tsimane’ have adopted to relieve fatigue, hunger, and thirst when doing strenuous physical work. The consumption of chicha is a traditional form of displaying sociability, takes place at home with guests, and is associated with a festive group spirit and not with drunkenness, as is the consumption of commercial alcoholic beverages, which is often done alone. By construction, addiction has mean=0 and SD=1.

^cWe used calipers to measure mid-arm muscle area following the norms of Frisancho (2008).

^dWe followed Lohman’s protocol to measure standing height and body weight (Lohman et al., 1988), and excluded pregnant and lactating women (Zeng et al., 2013). Unlike other native Amazonian populations, the Tsimane’ show little evidence of increased obesity (Zeng et al., 2013). Previous studies have found a significant association between BMI and wealth inequality (Carson, 2013).

^eThe durable category includes the value of industrial goods acquired in the market, including knives, fishhooks, cutlasses, axes, and cooking pots, rifles, shotguns, fishnets, mosquito nets, radios watches, bicycles, and grinders. In contrast, locally-produced assets included the value of canoes, bows and arrows, cotton bags, grinding platforms, and wooden mortars.

^fControls variables also include village-to-town travel time, annual village median of the resources whose inequality we measured, subject’s level of resource k in section [B] (e.g., if regression includes Gini of luxuries, resource=person’s expenditure in luxuries), total number of households in the village each year, and number of annual measures taken from the subject during the nine-years of the panel.