A Mixed Methods Exploration of the Role of Participation in a Nutrition-Sensitive Agroecology Intervention in Rural Tanzania

Abstract

Background: Participation is key to the successful implementation of nutrition-related interventions, but it has been relatively overlooked.

Objective: We sought to describe participation intensity among smallholder farmers in a randomized nutrition-sensitive agroecology study in rural Tanzania. We explored the association between baseline characteristics and overall participation intensity (quantitatively at the individual level and qualitatively at the group level), the association of participation intensity with 2 process indicators, and the association between participation intensity and key study outcomes.

Methods: Data came from 7 rounds of surveys with 295 women and 267 men across 29 months and 2 rounds of semi-structured interviews with the 20 "mentor farmers" who delivered the intervention. Participation intensity was based on the number of months of attendance at village-level project meetings or household visits (range: 0-29). Multivariable models of participation were built.

Results: Women and men participated for 17.5 ± 7.2 and 13.6 ± 8.3 months, respectively. Participation intensity followed 1 latent trajectory: initially low, with a sharp increase after month 7, and plateaued after the first year. At baseline, higher participation intensity was associated with older age, higher education, level of women's empowerment, being in the middle quintile of wealth, and qualitatively, village residence. Higher participation intensity was associated with 2 process indicators - better recall of topics discussed during meetings and greater knowledge about key agroecological methods. High participation intensity was positively associated with increased use of sustainable agricultural practices among all participants, and among women, with husband's involvement in household tasks and child's dietary diversity score.

Conclusions: Participation intensity covaried with key study outcomes, suggesting the value of increased attention to implementation in nutrition-related programs for providing insights into drivers of impact. We hope that investigations of participation, including participation intensity, will become more widespread so that intervention impacts, or lack thereof, can be better understood.

Keywords: agroecology; implementation science; nutrition-sensitive agriculture; participation; participation intensity; process evaluation; smallholder farmer.

FIGURE 1

Timeline of Singida Nutrition and Agroecology Project detailing when mentor farmer election, training, seed distribution, mentor farmer quarterly meetings (M), and surveys (S) took place. The timing of data collection of target variables and outcomes is shown; attendance, lessons learned, and knowledge change according to themes of agriculture, gender equity, or nutrition. ∗In July of 2016, only women’s responses for nutrition knowledge change were collected.

FIGURE 2

(A) The mean and 95% CI of mean program participation intensity over the course of the Singida Nutrition and Agroecology Project, disaggregated by gender, as indicated by attendance at village-level project meetings and household visits by mentor farmers each month. (B) Average participation intensity (number of months of reported attendance at village-level project meetings and household visits over 29 months) by the intervention village. Villages are grouped based on participation intensity. Villages 5, 1, and 17 had the highest participation intensity, villages 2, 3, and 12 had a moderate participation intensity, and villages 7, 11, and 13 had the lowest.

FIGURE 3

Percentage of the number of mentions of themes that would explain higher or lower program participation intensity identified in mentor farmer interviews according to the attendance tier of participating villages. Villages are grouped based on participation intensity. Villages 5, 1, and 17 had the highest participation intensity, villages 2, 3, and 12 had a moderate participation intensity, and villages 7, 11, and 13 had the lowest (Table 1).

FIGURE 4

Percentage of men (M, n = 259), women (W, n = 286), or men and women answering together (T, n = 256) who recall discussion of sustainable agriculture (A), gender equity (B), and dietary diversity when asked about topics discussed during village-level project meetings and who answered according to the message intended by intervention about sustainable agriculture (D), gender equity (E), and dietary diversity (F). G indicates data collected during the Growing Season (January and February), whereas H indicates data collected during the harvest season (May, June, and July).

FIGURE 5

Odds ratio of high participation intensity (attending 4 or more months in any 6 months) on the recall of meeting topics (“topic recall”) and knowledge change (“knowledge”) in Singida Nutrition and Agroecology Project at the end of those 6 months, by gender. W indicates women (n = 286), M indicates men (n = 259), and T indicates men and women interviewed together (n = 259). Models were analyzed separately for each sample. Estimates controlled for social desirability bias and baseline covariates that were correlated with participation intensity: household wealth quintile, individual’s age, education, 5 domains of empowerment score, and attitudes toward intimate partner violence. For models on knowledge change, we also controlled for baseline knowledge on the topic.

FIGURE 6

Beta coefficients of the association between participation intensity on various study outcomes of the Singida Nutrition and Agroecology Project by gender. W indicates women (n = 286) and M indicates men (n = 259). We controlled for social desirability bias, outcome measures at baseline, and baseline covariates that were correlated with participation intensity: household wealth quintile, individual’s age, education, 5 domains of empowerment score, and attitudes toward intimate partner violence. For models for child’s dietary diversity score, we also controlled for the child’s age.