9.4 ASHA-Mother Vignettes Results

First, analyses are conducted for each question, generalizing across all vignettes. Then, further analyses divide the vignettes into two broad categories. Lastly, we describe patterns specific to each vignette.

9.4.1 Across-vignette analyses

For each question, we determined a series of response questions that applied across the eight vignettes. Complete coding rubrics are provided in the appendix. For most questions, the response categories are not mutually exclusive because respondents were free to give multiple reasons for any given decisions. So, for every question, we analyze each kind of response independently as a binary dependent variable (i.e., either the response contains information relevant to this response category or it does not). Thus, for every dependent variable in every question, we conducted a mixed-effects logistic regression (also referred to as Generalized Linear Mixed Model, GLMM), analyzing what factors make that specific kind of response more likely.

The basic form of the GLMM that generalize across vignettes are:

Fixed-effects: 1. Respondent (Mother; ASHA) and 2. Condition (Consistent; Inconsistent) and their interaction Random intercepts for cluster variables: Subjects; Vignette

When we want to compare specific questions (e.g., 1 vs. 3), the model structure will change to include Question as a fixed effect. Because there are many analyses, the tables below will just present the odds ratios of statistically significant fixed-effects.

Table 9.5: Codes categories for vignettes

Code	Meaning	Example
Health-Biology	Health or the body is mentioned	Feeding colostrum is good for the baby’s health
Specific Illness	A subset of health-biology responses where a specific illness is mentioned	Feeding colostrum prevents anemia
Social Dynamics	General social norm or a specific social context is mentioned	Everyone in my family feeds colostrum to their baby
Other Benefits-Costs	Reasons beyond social or health implications	Financial costs, logistical issues, preference
Knowledge-Ignorance	An individual’s knoledge or lack-there-of was mentioned	Ignorance of what, specifically, was not always mentioned

9.4.2 Question 1

Question 1 asked participants why the mother in the vignette made the decision that she did. Before presenting the results of the mixed-effects models, we first present the proportions of responses of each kind, broken down by respondent and condition. This table shows that mothers and ASHAs have many similarities, but there are still key places where they diverge. ASHAs are more likely to refer to social dynamics in their explanations than mothers. Further, the ASHAs are more likely to cite specific illnesses and symptoms than mothers, reflecting deeper knowledge.

Table 9.5: Question 1: Proportion of results in each category

Respondent	Condition	Health - Biology	Specific illness	Social Dynamics	Other Benefits - Costs	Knowledge - Ignorance
ASHA	Consistent	0.84	0.47	0.12	0.21	0.05
Mother	Consistent	0.86	0.32	0.05	0.19	0.00
ASHA	Inconsistent	0.59	.25	0.36	0.18	0.37
Mother	Inconsistent	0.64	.25	0.21	0.20	0.26

Table 9.6: Question 1: Mixed-effects model results across all vignettes

Effect	Health- Biology	Specific Illness	Social Dynamics	Other Benefits - Costs	Knowledge/ Ignorance
Consistent - Inconsistent	9.671***	2.207***	0.312***		0.032***
ASHA - Mother		1.796***	2.244***		3.474***
Condition X Respondent		2.013**	1.986*	2.629*	6.706**

9.4.3 Question 3

Due to the related content between Q’s 1 and 3, these questions are discussed consecutively in this report. Q3 asked what the ASHA could do to convince the mother given (or if) she was not convinced. Here, there is an even greater focus on health than in question 1. This suggests an opening here to help ASHAs be more effective. They note in the inconsistent condition for question 1, that social dynamics are a barrier to optimal decision-making. However, they cite that re-explaining the health benefits is the best strategy for compliance of the mother. Certainly, making sure mothers understand the health benefits are important, but these results suggest equipping ASHAs with tools to navigate the social barriers is an area for improvement.

Table 9.7: Question 3: Proportion of results in each category

Respondent	Condition	Health - Biology	Social Dynamics	Other Benefits - Costs	Knowledge - Ignorance
ASHA	Consistent	0.93	0.18	0.17	0.01
Mother	Consistent	0.90	0.12	0.21	0.00
ASHA	Inconsistent	0.84	0.10	0.24	0.01
Mother	Inconsistent	0.86	0.05	0.21	0.01

Because Q3 had so few Knowledge-Ignorance responses, they are not included in GLMMs. Looking at the effects just within Q3 before comparing to Q1 shows that in the Consistent condition, Health and Social responses are more likely than in the Inconsistent Condition. This suggests that they generally recognize the important role of the social dynamics in compliance (the consistent condition), but then this is not cited as much as what to focus on to be more convincing in the Inconsistent condition.

Table 9.8: Question 3: Mixed-effects model results

Effect	Health - Biology	Social Dynamics	Other Benefits- Costs
Consistent - Inconsistent	2.400***	2.598***
ASHA - Mother		2.030***	2.043*
Condition X Respondent

For direct comparison of Q3 to Q1, a new GLMM added the additional variable of Question. This makes 3 fixed factors: Condition, Respondent, Question, with random intercepts for participant and vignette. The model does not include all possible interactions because the critical analyses are potential interactions between Question and the other two variables. Thus, the model included three main effects of the fixed variables, and those two interactions. The DVs of interest are Health and Social Dynamics.

These analyses reveal multiple reliable effects of interest. The main effect of Question was significant for both Health (Q3 more) and Social (Q1 more). Critically, Question significantly interacted with Condition for both Health and Social. For the Consistent condition, Health responses are quite similar for Q1 and Q3, but for Inconsistent, Health responses are more frequent for Q3. The tradeoff is that for Q3, Social responses are much less frequent specifically in the Inconsistent Condition. They show that, as discussed before; respondents recognize that Social Dynamics are the barrier to compliance as shown in Q1, but they rely heavily of Health as the persuasive tool to overcome that barrier, as shown by Q3.

Table 9.9: Questions 1 and 3 Compared: Mixed-effects model results

Health	Social Dynamics
Consistent - Inconsistent	4.718***
ASHA - Mother		1.867***
Question: 3-1	3.375***	0.682**
Condition X Question	0.242***	7.935***
Respondent X Question	1.764*

9.4.4 Question 2

Question 2 asked “who typically makes this decision and why?” Here we break the responses down between the two questions. For “who” the only answers that gained more than 3% of responses were the mother herself, her husband, and her mother in-law (MIL). The descriptive statistics table also show, “ASHA” and “Other,” which covers all other people mentioned (such as a sister) for comparison. However, for the GLMM only the three frequent responses were analyzed.

Question 2 was framed to be about this kind of decision in general, and so was not hypothesized to show an effect of condition (Consistent vs. Inconsistent). The GLMM confirmed there were no reliable effects of Condition for the three frequent response types. For clarity, we present the respondent differences generalizing across condition.

Table 9.10: Question 2: Response proportions for ‘who?’

Respondent	Mother	MIL	Husband	ASHA	Other
ASHA	0.56	0.20	0.20	0.02	0.02
Mother	0.46	0.37	0.16	0.02	0.03

The results show that ASHAs attribute more decision-making power to mothers than Mothers do! In addition, ASHAs attribute more decision-making power to husbands than Mothers do. The tradeoff is that Mothers attribute more decision-making power to MILs than ASHAs do. These patterns are all statistically significant.

Table 9.11: Question 2: Mixed-effects model results

Effect	Mother	MIL	Husband
ASHA - Mother	2.31***	.338***	1.729**

For the “Why” question, the pattern revealed by the GLMM is a bit more complicated. There were two primary kinds of responses. The first kind is called “Health Knowledge” and these responses attributed the reason for decision-making to be because that individual had knowledge of what is most healthy. The other kind of reason is called “Role” and that means that person made the decision because it is their role in the household to make that decision. This including responses that refer to duty, responsibility, and authority. Here another wrinkle is added. Whether or not they responded that the mother herself makes the decision is a factor in the “Why” model to see if reasons differ depending on who is attributed decision-making power. Another complicating factor is that Condition consistently interacts with Respondent in the GLMM.

The descriptive table below shows that ASHAs are more likely to give Health Knowledge explanations in the Consistent condition, and Mothers are more likely to give these kinds of explanations in the Inconsistent condition. This pattern is reversed for the Role explanations. The explanation for this pattern is unclear.

Table 9.12: Question 2: Response proportions for ‘why?’

Respondent	Condition	Health-Knowledge	Role
ASHA	Consistent	0.55	0.53
Mother	Consistent	0.46	0.58
ASHA	Inconsistent	0.42	0.63
Mother	Inconsistent	0.46	0.57

This next table shows that when respondents indicate mothers as the health decision maker, it is more likely that they do so because of their knowledge of health. However, when someone else makes the decision (e.g., the husband or MIL), it is because it is their role in the household to decide on these health behaviors. This again speaks to Q1 responses to the social barriers to compliance. Mothers are seen as knowledgeable, and when they do not make the decisions, it is not just because another has more knowledge, but because it is their social role to make the decision.

Table 9.13: Question 2: Response proportions for ‘why?’ broken down by ‘who?’

Respondent	Decision-Maker	Health-Knowledge	Role
ASHA	Mother	0.52	0.51
Mother	Mother	0.50	0.49
ASHA	Not Mother	0.44	0.68
Mother	Not Mother	0.42	0.66

Table 9.14: Question 2: Mixed-effects model results for ‘why?’

Health Knowledge	Role
Consistent - Inconsistent	1.361*
ASHA - Mother
Mother makes the decision	1.549**	0.341***
Condition X Question	1.762*	0.551*

9.4.5 Question 4

This question asks whether there was a conflict eiither between the mother and her family member (consistent condition) when the mother did what the ASHA prescribed and conflicted with the wishes of the family member or themother and her ASHA (inconsistent condition), when the Mother did not follow ASHAs recommendation. Unlike prior questions, here the options were mutually exclusive. So the GLMM just analyzed “yes” responses. The majority of responses were “yes” or “no” though we note below that between 5% and 10% of responses indicated that people were upset, but did not mention any conflict specifically.

The analyses show that there were many more conflicts between Mothers and family members when the mother complied with the ASHA than between Mothers and ASHAs when Mothers did not comply. This pattern was the same for ASHA respondents and Mother respondents.

Table 9.15: Question 4: Response frequencies

Respondent	Condition	Yes	No	Upset but no mentioned conflict
ASHA	Consistent	0.84	0.11	0.04
Mother	Consistent	0.88	0.08	0.04
ASHA	Inconsistent	0.71	0.18	0.09
Mother	Inconsistent	0.77	0.13	0.09

Table 9.16: Question 4: Mixed-effects model results

Effect	Yes there was a conflict
Consistent - Inconsistent	2.975***
ASHA - Mother
Condition X Respondent

9.4.6 Question 5

This question asked how the conflict should be resolved either between the Mother and her family member (Consistent condition) or between the Mother and ASHA (Inconsistent condition). Respondents often broke these responses into two parts, and these parts were analyzed separately. One part examines what to communicate to help resolve the conflict- whether the benefits of the choices, or the social dynamics of making the choice. The other part focuses on who initiates the resolution.

For what to communicate, analyses show that explaining the benefits of the choice to the family member in the Consistent condition is much more common than to the ASHA in the Inconsistent condition. This makes sense, as respondents assume family members’ objections to ASHA recommendations are seen to be rooted in a lack of understanding, and everyone assumes ASHAs are knowledgeable. For social dynamic communication, the pattern is more complex. There are effects of Condition, Respondent, and an interaction between the factors. ASHAs are more likely to discuss social dynamics than Mothers, but this difference is much larger in the Consistent condition. This suggests that ASHAs think mothers will discuss social dynamics (between the ASHA and the mother) with their family members as a way to resolve a conflict with them more than Mothers think mothers will use this conflict resolution strategy. ASHAs and Mothers roughly think social dynamics (between mothers and their families) will be discussed as part of the resolution between Mothers and ASHAs equally. Overall, this contrasting pattern can be compared with Q3 above where ASHAs don’t utilize social information well when trying to convince mothers to comply but seem to think they are a key part to resolving conflicts with mother’s family after they do comply.

Not offering any kind of information to communicate is much more common to resolve the conflict between mother and ASHA in the Inconsistent condition. In contrast it was rare to not have anything specific to add to resolve a conflict between a mother and her family in the Consistent condition. There was a significant interaction where the difference between conditions was stronger for ASHA respondents. It was very rare for them to not offer a way for mothers to resolve conflicts with their families. Furthermore, over 1/3 of the time, they did not offer a way to resolve a conflict with the mother. The respondents still suggested conflicts were resolved, but just did not specific what kind of information was key to resolving it.

Table 9.17: Question 5: Response frequencies

Respondent	Condition	Explain benefits	Explain social dynamics	No explanation given
ASHA	Consistent	0.82	0.40	0.03
Mother	Consistent	0.81	0.23	0.11
ASHA	Inconsistent	0.37	0.38	0.36
Mother	Inconsistent	0.38	0.35	0.32

Table 9.18: Question 5: Mixed-effects model results

Effect	Explain Benefits	Explain Social Dynamics	No explanation
Consistent - Inconsistent	15.03***	.769*	.0816***
ASHA - Mother		1.709***	.5407*
Condition X Respondent		2.045**	.2860**

Please see Appendix B.0.0 to review more analyses on conflict resolution data.

Question 6: We wait for analyses for Q6 until we break down specific vignettes. This question was about when ASHAs should give the relevant information for this behavior, and there were not clear cross-vignette generalizations to analyze. Every behavior had their own specific preferred timeline.

9.4.7 Effects of Vignette

The first analysis of the effects of vignette breaks vignettes into two categories. Looking at the descriptive statistics of Question 1 “why was the decision made?” two categories of vignettes pop out- those with 0 or very rare “Other Benefits-Costs” explanations (and instead extremely high “Health” explanations, particularly in the Consistent condition), and those with frequent “Other Benefits - Costs” explanations. “Other Benefits-Costs” refer to financial, or quality of life benefits outside of direct health benefits. The vignettes that cluster along this distinction are quite sensible. See the table below. Below are analyses of Questions 1 – 5 that focus on the distinction between these two categories of Vignettes. We refer to the two vignette categories as “Health” and “Other Benefits.”

Vignette	Vignette Category	Respondent	Condition	Health	Specific illness	Social	Other Benefits - Costs	Knowledge
Colostrum	Health	ASHA	Consistent	1.00	0.31	0.19	0.00	0.27
	Mother	Consistent	0.92	0.17	0.14	0.00	0.01
	ASHA	Inconsistent	0.70	0.04	0.30	0.00	0.41
	Mother	Inconsistent	0.81	0.28	0.13	0.00	0.20
IFA	Health	ASHA	Consistent	0.98	0.81	0.14	0.00	0.10
	Mother	Consistent	0.99	0.51	0.04	0.00	0.01
	ASHA	Inconsistent	0.76	0.62	0.21	0.00	0.36
	Mother	Inconsistent	0.90	0.55	0.14	0.00	0.19
Exclusive Breastfeed	Health	ASHA	Consistent	0.86	0.40	0.19	0.00	0.02
	Mother	Consistent	0.99	0.39	0.03	0.00	0.01
	ASHA	Inconsistent	0.80	0.12	0.15	0.00	0.34
	Mother	Inconsistent	0.67	0.13	0.13	0.00	0.24
Vaccines - Pregnancy	Health	ASHA	Consistent	0.93	0.98	0.05	0.00	0.00
	Mother	Consistent	1.00	0.63	0.00	0.00	0.00
	ASHA	Inconsistent	0.33	0.21	0.31	0.00	0.64
	Mother	Inconsistent	0.52	0.19	0.30	0.00	0.63
Vaccines - Infancy	Health	ASHA	Consistent	1.00	0.73	0.00	0.00	0.00
	Mother	Consistent	1.00	0.62	0.09	0.00	0.00
	ASHA	Inconsistent	0.85	0.77	0.15	0.04	0.31
	Mother	Inconsistent	0.85	0.61	0.04	0.00	0.19
Institutional Delivery	Other Benefits	ASHA	Consistent	0.96	0.15	0.00	0.58	0.00
	Mother	Consistent	0.97	0.04	0.00	0.48	0.00
	ASHA	Inconsistent	0.46	0.00	0.42	0.73	0.46
	Mother	Inconsistent	0.42	0.02	0.18	0.82	0.52
Family Planning 1 - no children	Other Benefits	ASHA	Consistent	0.77	0.25	0.27	0.18	0.00
	Mother	Consistent	0.85	0.09	0.11	0.11	0.00
	ASHA	Inconsistent	0.52	0.09	0.64	0.00	0.23
	Mother	Inconsistent	0.41	0.15	0.63	0.44	0.11
Family Planning 2 - multiple children	Other Benefits	ASHA	Consistent	0.23	0.08	0.15	0.92	0.00
	Mother	Consistent	0.20	0.03	0.00	0.90	0.00
	ASHA	Inconsistent	0.25	0.15	0.71	0.68	0.21
	Mother	Inconsistent	0.57	0.07	0.14	0.36	0.04