Menu
Home
Log in / Register
 
Home arrow Environment arrow Research Methods in Anthropology: Qualitative and Quantitative Approaches
Source

REPRESENTING COMPLICATED MODELS WITH TABLES AND IF-THEN CHARTS Young and Garro's EDM: Decision Tables

James Young and Linda Garro studied how Tarascan people in Pichataro, Mexico, choose one of four ways to treat an illness: Use a home remedy, go to a native curer, see a practicante (a local, nonphysician practitioner of modern medicine), or go to a physician (see Garro 1986; Young 1980; Young and Garro 1982, 1994 [1981]). From their ethnographic work, Young and Garro believed that the decision to use one or another of these treatments depended on four factors:

  • 1. how serious an illness was perceived to be (gravity);
  • 2. whether a home remedy for the illness was known;
  • 3. whether the informant had confidence in the general efficacy of a mode of treatment for a particular illness; and
  • 4. accessibility (in terms of cost and transportation) of a particular mode of treatment.

The choice situations emerged from structured interviews with eight men and seven women who were asked:

If you or another person in your household were ill, when—for what reasons—would

you [consult] [use] _instead of [consulting] [using] _? (Young

and Garro 1994 [1981]:132)

Young and Garro used this question frame to elicit responses about all six possible pairs of treatment alternatives: home remedy vs. a physician, curer vs. home remedy, and so on. To check the validity of the statements made in the interviews, Young and Garro collected case histories of actual illnesses and their treatments from each of the 15 informants.

Next, the researchers completed interviews with 20 informants using a series of‘‘What if . . .’’ questions to generate decisions, under various combinations of circumstances, regarding the selection of treatments for illnesses. For example, informants were asked:

FIGURE 17.4.

Ethnographic decision model for recycling cans.

SOURCE: G.W. Ryan and H. R. Bernard, ''Testing an Ethnographic Decision Tree Model on a National Sample: Recycling Beverage Cans,'' Human Organization, Vol. 65, pp. 103-14, 2006. Reprinted with permission of the Society for Applied Anthropology.

FIGURE 17.5.

National test of an ethnographic decision model.

SOURCE: G. W. Ryan and H. R. Bernard, ''Testing an Ethnographic Decision Tree Model on a National Sample: Recycling Beverage Cans,'' Human Organization, Vol. 65, pp. 103-14, 2006. Reprinted with permission of the Society for Applied Anthropology.

Let’s say there is a person who has a very grave illness. In this family, money is scarce— sure, they’re eating, but there is just not anything left over. They have had this illness in the family before, and they now know of the remedy that benefited the illness on the previous occasion. What do you think they are going to do? (Young and Garro 1994 [1981]:137)

This vignette combines the condition of a serious illness (level 3 on gravity in tables 17.1 and 17.2), with lack of accessibility (no money), and a known remedy that can be applied at home. Young and Garro used the three levels of gravity, two possible conditions of knowing a remedy (yes and no), and two possible conditions of accessibility (yes and no) in making up the vignettes, which meant that they had to make up eight of them. Each vignette was presented to each informant for a response. Tables 17.1 and 17.2 show the decision tables for Young and Garro’s data.

Table 17.1 Young and Garro's Decision Table Showing How Pichatarenos Choose an Initial Method of Treating an Illness

Rules:

1

2

3

4

5

6

7

8

9

Conditions

gravity0

1

1

1

2

2

2

3

3

3

known home remedy10

Y

N

N

Y

N

faithc

F

M

(F)

F

M

F

M

(M)

accessibility11

N

Y

Choices

self-treatment

X

X

curer

X

X

X

practicante

physician

X

X

X

X

a. 1 = nonserious, 2 = moderately serious, 3 = grave

b. Y = yes, N = no

c. F = favors folk treatment, M = favors medical treatment

d. Y = money and transportation available, N = either money or transportation not available

SOURCE: Reprinted by permission of Waveland Press, Inc., from J. C. Young and L. C. Garro, Medical Choice in a Mexican Village, 1981 (reissued 1994), p. 154. All rights reserved.

From these qualitative data, collected in structured interviews, Young and Garro developed their decision model, for the initial choice of treatment. The model, containing nine decision rules, is shown in table 17.1. Rule number 1, for example, says that if the illness is not serious and there is a known home remedy, then treat the illness yourself. Rule number 9 says that for grave illnesses there is an implicit understanding that physicians are better (hence the M in parentheses), so if there is money, then go to a physician.

Rule number 9 also says that for the few cases of very grave illnesses where physicians are commonly thought not to be effective, apply rule number 7 and go to a curer. The blank spaces in the top part of table 17.1 indicate irrelevant conditions. In rule number 1, for example, there is no question about accessibility for home remedies because they cost little or nothing and everyone has access to them.

Sometimes the treatment selected for an illness doesn’t work and another decision has to be made. Table 17.2, with 11 decision rules, shows Young and Garro’s analysis of this second stage of decision making. Their entire two-stage model is based on their sense of emerging patterns in the data they collected about decision-making. The question, of course, is: Does it work?

Young and Garro tested their model against 444 treatment choices gathered from 62

Table 17.2 Young and Garro's Decision Table Showing How Pichatarenos Choose a Method of Treating an Illness When Their First Choice Doesn't Work

Rules:

1

2

3

4

5

6

7

8

9

10

11

Conditions

preceding choice"

ST

ST

ST

ST

C-P

C-P

C

P

Dr

Dr

Dr

current gravityb

1-2

3

3

1

2-3

2-3

2-3

faithc

F

M

M

(M)

M

accessibility[1]

N

Y

Y

N

N

N

Y

Choices

self-treatment

X

curer

X

X

X

X

practicante

physician

X

X

X

X

X

X

a. ST = self-treatment, C = curer, P = practicante, Dr = physician

b. 1 = nonserious, 2 = moderately serious, 3 = grave

c. F = favors folk treatment, M = favors medical treatment

d. Y = money and transportation available, N = either money or transportation not currently available SOURCE: Reprinted by permission of Waveland Press, Inc., from J. C. Young and L. C. Garro, Medical Choice in a Mexican Village, 1981 (reissued 1994), p. 156. All rights reserved.

Table 17.3 Test Results of Young and Garro's Decision Model of How Pichatarenos Choose a Treatment Method When They Are Ill

Table

Rule

Treatment Method Chosen

Totals

Percentage

correct

Self

treatment

Curer

Practicante

Physician

18.1

1

157

157

2

4

4

3

5

5

4

67

(1)

68

5

8

8

6

(2)

20

(7)

29

7

8

8

8

(2)

4

(2)

8

9

(2)

11

13

Subtotal

300

94.7%

18.2

1

19

19

2

(1)

28

(6)

35

3

(3)

6

9

4

(2)

22

24

5

3

(1)

4

6

(2)

(2)

(1)

24

29

7

(1)

3

(2)

6

8

2

(1)

3

9

(1)

7

8

10

0

11

7

7

Subtotal

144

84.0%

Total

444

91.2%

SOURCE: Reprinted by permission of Waveland Press, Inc., from J. C. Young and L. C. Garro, Medical Choice in a Mexican Village, 1981 (reissued 1994), p. 165. All rights reserved.

Treatment Method Chosen

duration of the episode

perceived cause (from worms, from empacho, from food, etc.)

whether there was mucous in the stool

whether there was blood in the stool

whether the stools smelled bad

whether the stools were frequent or not

whether the stools were loose or not

whether the child had fever

color of the stool

whether the child had a dry mouth whether the child had dry eyes whether the child was vomiting whether the child had swollen glands

Table 17.4 shows the data from the 17 women in Ryan and Martinez’s original sample and the decision to take the child to the doctor. Read the table like this: Mother #1 said that her child’s last episode of diarrhea lasted 2 days and was caused by bad food. The stools contained mucous, but did not contain blood. The stools smelled bad, were frequent, and were loose. The child had fever, the stools were yellow. The child had dry

Table 17.4 Decisions to Take a Child to the Doctor in San Jose, Mexico

mother doctor days

cause

muc.

blood

smell

freq

loose

fever

color

mouth

eyes

vomit gland

1

N

2

C

Y

N

Y

Y

Y

Y

A

Y

Y

N

N

2

N

20

E

Y

N

Y

Y

Y

N

A

Y

Y

N

N

3

Y

8

T

N

N

Y

Y

Y

Y

N

Y

Y

N

N

4

Y

8

C

Y

N

Y

N

Y

Y

V

Y

Y

5

N

3

P

Y

N

Y

Y

Y

Y

A

Y

Y

N

Y

6

N

3

L

N

N

Y

Y

Y

N

B

Y

Y

N

N

7

Y

8

D

Y

N

Y

Y

Y

N

A

Y

Y

N

N

8

N

1

D

N

N

Y

Y

Y

N

A

N

N

9

N

C

Y

N

N

Y

Y

N

B

Y

Y

N

N

10

N

3

O

N

N

Y

Y

Y

N

A

Y

Y

N

N

11

N

2

C

N

N

N

N

N

N

A

N

N

12

N

C

N

N

Y

Y

Y

N

A

Y

Y

N

N

13

N

4

C

N

N

Y

N

N

N

A

Y

Y

N

N

14

Y

4

E

N

N

Y

Y

Y

Y

V

Y

N

N

15

Y

3

I

Y

Y

Y

Y

Y

Y

A

Y

Y

Y

N

16

N

2

C

Y

N

N

Y

Y

N

V

Y

Y

N

N

17

N

7

E

N

N

N

Y

Y

N

A

N

N

N

N

Cause Color

C = food D = teething A = yellow

L = worms T = dirt V = green

E = empacho P = parasites B = white

I = indigestion O = other N = black

SOURCE: G. W Ryan and H. Martinez, ''Can We Predict What Mothers Do? Modeling Childhood Diarrhea in Rural Mexico,'' Human Organization, Vol. 55, pp. 47-57, 1996. Reprinted with permission of the Society for Applied Anthropology.

mouth and dry eyes, but was not vomiting and did not have swollen glands. In the end, Mother #1 did not take her child to the doctor. (The codes for cause and color are from Spanish; see the legend just below the table.)

Table 17.4 makes it clear that mothers took their children to the doctor if the child had blood in the stool, had swollen glands, or was vomiting, or if the diarrhea had lasted more than 7 days. None of the other factors played a part in the final decision to take the child to the doctor.

But remember: There were seven different treatments, and mothers often try several treatments in any given episode. Ryan and Martinez looked at the pattern of circumstances for all seven treatments and built a model that accounted for the treatment decisions made by the 17 mothers. Their model had just six rules and three constraints.

Figure 17.6 shows the model as a series of IF-THEN statements. Notice the constraints: For a woman to choose a modern medication, she has to know about it and it has to be easy to get and cheap. The constraints to the rules are derived from ethnographic interviews. So was the observation that mothers distinguished between curative and palliative treatments—treatments that stop diarrhea and treatments that simply make the child feel better until the episode is over. The model postdicted (accounted for) 89% of the treatments that the 17 mothers had reported.

Next, Ryan and Martinez tested their model. They interviewed 20 more mothers, but this time they asked each woman every question in the model. In other words, they asked each woman: ‘‘In your child’s last episode of diarrhea, did the stools have blood in them? Did the child have swollen glands? What caused the diarrhea?’’ and so on. The IF-THEN model in figure 17.6 accounted for 84% of the second group’s treatment decisions (Further Reading: ethnographic decision models).

FIGURE 17.6.

Ryan and Martinez's decision model as a series of IF-THEN rules.

SOURCE: G. W. Ryan and H. Martinez, ''Can We Predict What Mothers Do? Modeling Childhood Diarrhea in Rural Mexico,'' Human Organization, Vol. 55, pp. 47-57, 1996. Reprinted with permission of the Society for Applied Anthropology.

  • [1] households over a 6-month period. To make the test fair, none of the informants in thetest were among those whose data were used in developing the model. Table 17.3 showsthe results of the test. There were 157 cases covered by rule number 1 from table 17.1(first-stage decision), and in every single case informants did what the rule predicted. Intable 17.3, errors (informants’ choices that are not predicted by the model) are in parentheses, so informants did what rule number 6 predicted 20 out of 29 times. Overall, for the first stage, Young and Garro’s decision rules predict about 94% ofinformants’ reported behavior. After removing the cases covered by rules 1 and 4 (whichaccount for half the cases in the data, but which could be dismissed as common-sense,routine decisions and not in need of any pretentious ‘‘analysis’’), their model still predictsalmost 83% of reported behavior. Even for the second stage, after first-stage decisions failto result in a cure, and decisions get more complex and tougher to predict, the modelpredicts an impressive 84% of reported behavior. Ryan and Martinez's EDM: IF-THEN Charts Gery Ryan and Homero Martinez (1996) built an EDM for how mothers in San Jose,Mexico, treated children who have diarrhea. Ryan and Martinez knew, from living in thevillage, that mothers there use seven different treatments in treating their children’s diarrhea. Five of the treatments consist of giving the child one or more of the following: (1)tea, (2) homemade rice water, (3) medication from the pharmacy (their informants toldthem ‘‘If you can say it, you can buy it’’), (4) a carbonated beverage, or (5) a commerciallyproduced oral rehydration solution. The other two treatments are: (6) manipulating thechild’s body (massaging the child’s body, pinching the child’s back) or (7) taking the childto the doctor. Ryan and Martinez asked 17 mothers in San Jose who had children under age 5 whatthey did the last time their children had diarrhea. Then they went systematically throughthe treatments, asking each mother why she had used X instead of A, X instead of B, Xinstead of C, and so on down through the list. Mothers in San Jose listed the following factors for choosing one treatment overanother:
 
Source
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
 
Subjects
Accounting
Business & Finance
Communication
Computer Science
Economics
Education
Engineering
Environment
Geography
Health
History
Language & Literature
Law
Management
Marketing
Mathematics
Political science
Philosophy
Psychology
Religion
Sociology
Travel