GENDER Impact
Platform
CGIAR GENDER Impact Platform · Working Paper #016
DECEMBER 2023
How do couples in rural Tanzania 
make decisions? Findings from a 
novel mixed-methods approach 
for understanding intrahousehold 
decision-making
By Steven Cole, Nathaniel Ferguson, Jessica Heckert, Devis 
Mwakanyamale, Greg Seymour, Shiferaw Feleke, Gundula Fischer, 
Innocensia John, Zawadi Lija, Mussa Nyaa, Happy Zacharia
This publication is copyrighted by the International Livestock Research Institute (ILRI). It is 
licensed for use under the Creative Commons Attribution 4.0 International License. To view 
this license, visit https://creativecommons.org/licenses/by/4.0.
BY Unless otherwise noted, you are free to share (copy and redistribute the material in 
any medium or format) and adapt (remix, transform and build upon the material) for any 
purpose, even commercially, under the following conditions:
 ATTRIBUTION. The work must be attributed, but not in any way that suggests endorsement 
by ILRI or the author(s).
ISBN: 92-9146-802-9
Citation: Cole, S., Ferguson, N., Heckert, J., Mwakanyamale, D., Seymour, G., Feleke, S., 
Fischer, G., John, I., Lija, Z., Nyaa, M., and H. Zacharia. 2023. How do couples in rural 
Tanzania make decisions? Findings from a novel mixed-methods approach for 
understanding intrahousehold decision-making. CGIAR GENDER Impact Platform Working 
Paper #016. Nairobi, Kenya: CGIAR GENDER Impact Platform.
ACKNOWLEDGMENTS
The study reported in this working paper was funded by an Innovation Grant awarded by the 
Methods Module of the CGIAR GENDER Impact Platform. The CGIAR GENDER Impact Platform 
is grateful for the support of CGIAR Trust Fund Contributors: www.cgiar.org/funders. 
The development of the working paper was made possible with financial support from the 
CGIAR Harnessing Gender and Social Equality for Resilience in Agrifood Systems (HER+) 
Initiative. The HER+ Initiative is also grateful for the support of CGIAR Trust Fund 
Contributors.
COVER PHOTO CREDIT: A certified cassava seed producer with her husband in her 
cassava field in the Tabora Region, Tanzania. Credit: Gloriana Ndibalema, IITA.
ABOUT CGIAR GENDER IMPACT PLATFORM
Generating Evidence and New Directions for Equitable Results (GENDER) is CGIAR’s 
impact platform designed to put equality and inclusion at the forefront of global agricultural 
research for development. The Platform is transforming the way gender research is done, 
both within and beyond CGIAR, to kick-start a process of genuine change toward greater 
gender equality and better lives for smallholder farmers everywhere. gender.cgiar.org
DISCLAIMER
This working paper has been internally peer-reviewed and the opinions expressed herein 
reflect those of the authors, not necessarily those of the CGIAR GENDER Impact Platform.
CONTACT
Steven Cole
International Institute of Tropical Agriculture (IITA)
Plot No 25 Mikocheni Light Industrial Area, Mwenge Coca-Cola Road, Mikocheni B
Dar es Salaam, Tanzania
s.cole@cgiar.org
CGIAR GENDER Impact Platform 
Working Paper #016
DECEMBER 2023
How do couples in rural Tanzania 
make decisions? Findings from a 
novel mixed-methods approach 
for understanding intrahousehold 
decision-making
Steven Cole Gundula Fischer
International Institute of Tropical IITA
Agriculture (IITA) gundulafischer@gmx.net 
s.cole@cgiar.org
Nathaniel Ferguson Innocensia John 
International Food Policy Research University of Dar es Salaam 
Institute (IFPRI) jinnocensia@gmail.com
N.Ferguson@cgiar.org
Jessica Heckert Zawadi Lija
IFPRI University of Dar es Salaam 
J.Heckert@cgiar.org zawadilija@gmail.com 
Devis Mwakanyamale Mussa Nyaa
IITA IITA
D.Mwakanyamale@cgiar.org m.nyaa@cgiar.org
Greg Seymour Happy Zacharia
IFPRI IITA
G.Seymour@cgiar.org H.Zacharia@cgiar.org
Shiferaw Feleke 
IITA
S.Feleke@cgiar.org
TABLE OF CONTENTS
1. Introduction 1
1.1 Study objectives 2
2. Study context and overview of methods 3
2.1 Study location 3
2.2 Cassava production in Tanzania 4
2.3 An iterative mixed-methods approach 5
2.4 Stakeholder consultations 5
2.5 Ethics approval 5
3. Qualitative dyadic interviews 6
3.1 Study design 6
3.2 Qualitative interview tool 6
3.3 Analysis 7
3.4 Results 9
3.5 Validating the vignettes 10
4. Quantitative household survey 11
4.1 Sampling 11
4.2 Data collection 12
5. Results 13
5.1 Descriptive results of decision-making questions 13
5.2 Regression analysis 16
6. Discussion 19
6.1 Interpreting our findings 19
6.2 Applying this approach to future studies 21
6.3 Conclusion 22
References 23
Annex 26
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making iii
1. Introduction
Despite wide acceptance of women’s empowerment as an important development goal, 
there is less consensus on how best to measure women’s empowerment. Empowerment is 
often defined as women having the ability to make strategic life choices where previously 
these were denied (Kabeer 1999). Women's input into household decision-making is 
considered an important indicator of women’s empowerment (Doss 2013; Laszlo et al. 2020). 
There are several approaches that researchers use to ask study participants about 
intrahousehold decision-making. Demographic and health surveys are used to ask married 
women and men about who usually makes decisions (e.g., about major household purchases) 
and inquire whether these decisions are made mainly alone, jointly with a spouse, or by 
someone else (Kishor and Subaiya 2008). The project-level Women’s Empowerment in 
Agriculture Index (pro-WEAI) asks women and men to indicate which agricultural activities 
they participate in, who normally makes the decisions on a given matter, how much input 
they contributed to the decision, and to what extent they could participate (Malapit et al. 
2019). Nonetheless, there is still uncertainty within the literature on how to best ask decision-
making questions and interpret responses (Acosta et al. 2020; Bernard et al. 2020; Seymour 
and Peterman 2018). These concerns are especially salient for quantitative approaches that 
measure women’s empowerment.
Standard approaches to measuring decision-making have been criticized for different 
reasons. One important criticism is that, in many study instruments, these questions focus 
too much on ascertaining the identity of the decision-maker (or who makes decisions) and 
less on understanding why and how decisions are made within the household (Bernard 
et al. 2020; Seymour and Peterman 2018; Acosta et al. 2020). Bernard et al. (2020) argue 
that determining who makes a particular decision is insufficient when making claims about 
women’s empowerment. For example, neither the demographic and health surveys nor pro-
WEAI questions explore how decision-making processes unfold, or why household members 
make certain decisions alone or jointly with other members. Asking questions beyond who 
makes decisions (and to what extent) is important given that in many rural, low-income 
country contexts (especially in Africa) spouses may not pool resources generated from their 
labor or have the same preferences for how resources are allocated (Duflo 2003; Duflo and 
Udry 2004; Haddad et al. 1997). As such, knowing how and why decisions are made can help 
to better explain the process and rationale used when making joint decisions.
Another criticism is about the heterogeneity in the respondents’ interpretation of joint 
decision-making (Seymour and Peterman 2018) and conflicting reports of decision-making 
between spouses (Ambler et al. 2021). Anderson et al. (2017) found discrepancies in reports 
among spouses about who made various farm and household decisions in rural Tanzania. 
Acosta et  al. (2020) explored differences in reports of intrahousehold decision-making in 
Uganda and found that women reported joint decision-making more often than men about 
the adoption of agricultural practices and consumption expenses. Notably, it has been shown 
that respondents understood joint decision-making as respondents having the final say, while 
most policy and academic literature assumes joint decisions are those in which both spouses 
have an equal say or negotiate the decision. These findings suggest that studies should 
interview both spouses in a marital dyad to better understand intrahousehold decision-
making.
Furthermore, most studies on intrahousehold decision-making, or women’s empowerment 
more broadly, use etic rather than emic perspectives when formulating questions to 
include in their research tools (Elias et  al. 2021). Haapanen and Manninen (2021) explain 
that research that uses an etic perspective takes into account the researcher–analyst’s 
(or outsider) viewpoints, while an emic perspective considers the practitioner–informant’s 
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 1
(or insider) viewpoints. They argue that by combining etic and emic perspectives, a more 
holistic understanding of “complex yet routinized real-life behaviour” is possible (Haapanen 
and Manninen 2021, 1). A transdisciplinary approach entails involving stakeholders from 
diverse educational, work and other backgrounds in the design, implementation and 
dissemination phases of the research, which can strengthen the consideration of both emic 
and etic perspectives. However, the use of transdisciplinary approaches is not common 
when studying intrahousehold decision-making. This means there are missed opportunities 
to strengthen the overall research process and/or increase the understanding of women’s 
involvement in decision-making within the household by different research and development 
actors.
1.1 Study objectives
This paper responds to current concerns about quantitative methods for studying decision-
making by describing the development of and results from an innovative transdisciplinary 
and mixed-methods tool for researching intrahousehold decision-making. The tool focuses 
specifically on decision-making about agricultural- and expenditure-related matters by 
spouses in marital or cohabiting relationships, although it could easily be adapted to 
alternative household compositions (e.g., mother–daughter dyads). The tool explores how 
decisions are made and why they are made alone or jointly. 
We used an iterative approach when designing the tool: each step in the process informing 
the next. We first consulted a diverse group of local research and development people when 
setting up the research. We then conducted in-depth qualitative research with decision-
making dyads to inform the development of vignettes that described local decision-making 
patterns. Once we validated questions both within and outside the places where the 
qualitative research was conducted, they were included in a multitopic household survey. 
We piloted the new tool in rural regions of northwestern Tanzania and contextualized it for 
use with local cassava value chain actors.
While the ultimate objective of the study was to improve the measurement and understanding 
of intrahousehold decision-making on agricultural- and expenditure-related matters, we were 
also able to investigate how decision-making processes are associated with key development 
outcomes, such as crop productivity and sales, food security, and minimum dietary diversity 
for women. Across the study components, we addressed the primary research questions 
presented in Table 1.
Table 1. Primary research questions addressed in this study
1 Who makes cassava production, processing and trading decisions, and those on expenditure-
related matters, within the household?
2 Why does one spouse or the other (or another household member) make certain decisions? 
3 What process(es) do couples or other dyads use to make these agricultural- and expenditure-
related decisions?
4 How are decision-making processes (who decides, how and why) associated with women’s dietary 
diversity, food security and agricultural productivity outcomes at the household level? 
The study components of the tool we developed do not match the research questions “one 
to one”. Instead, most of these questions are addressed across the research components. 
For that reason, we organize the content in this paper as follows. We begin with an overview 
of the context and our methodological approach. We then have a section describing the 
qualitative methods and results, as well as how the results informed the development of 
the vignettes. We then describe the methods and results for the quantitative household 
survey. In the discussion, we return to the research questions and describe how our work 
informed and responded to each question, and consider the overall utility of the tool for 
understanding intrahousehold decision-making.
2 CGIAR GENDER Impact Platform · Working Paper #016
2. Study context and overview of
methods
2.1 Study location
While cassava (Manihot esculenta Crantz) is grown in almost all eight geographical zones 
of Tanzania, it is most widely grown in four zones—Lake, Western, Southern and Eastern. 
We focused this study on the Kigoma and Kagera regions of the Lake and Western zones 
(Figure 1). 
RWANDA
TANZANIA
BURUNDI
Qualitative Study
Quantitative Study
Figure 1. Study locations (qualitative and quantitative studies)
These two regions contributed over one-fifth (22 percent) of the total cassava production in 
the country in the 2018/19 season (Table 2). Kigoma was the top cassava-producing region in 
the country, with over one million [metric] tons produced during the 2018/19 season. Kagera 
was the fifth largest cassava producer, at 0.72 million tons.
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 3
Table 2. Cassava production in Tanzania by region
Region Cassava production (2018/2019)
Dry (ton) Fresh (ton) Fresh (% of total 
for Tanzania)
Kigoma 364,649 1,093,947 13.31
Kagera 240,600 721,800 8.78
Total for Tanzania 2,739,319 8,217,957 100.00
Note: tons are metric tons 
Source: Ministry of Agriculture (2020)
2.2 Cassava production in Tanzania
A large share of the population in Tanzania lives in rural areas (66 percent) (Global Nutrition 
Report 2022). Rural livelihoods rely heavily on agricultural production, which draws on 
the labor of both women and men for production, processing and marketing activities. 
Undernutrition is a pressing concern. As a staple food, the cassava tuber is an important 
source of staple food, including as processed forms—despite it being considered an energy-
dense and nutrient-poor food (Montagnac et  al. 2009). Additionally, cassava leaves help 
contribute to dietary diversity.
Decades of research have gone into developing new and improved varieties of cassava for 
the east Africa region and associated technologies, and the Government of Tanzania has 
recently moved to prioritize cassava as a cash crop (Ministry of Agriculture 2020). There have 
been and will continue to be large research and development projects on developing and 
upgrading the cassava value chain.
A well-documented phenomenon in the gender and women’s empowerment literature 
is men’s appropriation of certain value chain activities as they move from low-skill/low-
paid status (primarily carried out by women) to higher skill/higher paid status when value 
chain development initiatives take root (Quisumbing et al. 2021; Fischer and Qaim 2012). 
Understanding the who, why and how of decision-making within households is an important 
aspect of informing a gender-responsive and transformative approach to value chain 
development in the future. Overall, the importance of cassava in the region, along with 
the national push to increase cassava productivity and strengthen value chains, makes our 
research on intrahousehold decision-making dynamics with smallholder cassava farmers 
highly relevant for the study area.
2.2.1 Gendered aspects of cassava value chains
Studies in Tanzania illustrate gendered participation in nodes of, or carrying out certain 
activities in, the cassava value chain. Women and men play different roles in the production 
of cassava. Men dominate input supply (e.g., fertilizer) in the cassava supply chain given their 
ability to travel long distances (Masamha et al. 2019). Men tend to clear and till the land for 
cassava production, while women are more involved in preparing cassava cuttings, planting, 
weeding, and disease management (Masamha et al. 2019). Mtunguja et al. (2014) found that 
women and men both play a role in the selection of cassava varieties (landraces) they grow in 
Tanzania. Men own most of the land devoted to agricultural production in the region.
Women and children are highly involved in processing cassava and in cassava food preparation 
(Masamha et al. 2018). Drying cassava leaves is primarily carried out by women (Andersson 
et al. 2016).
Men are more involved in packaging, transporting and marketing cassava using bicycles, 
motorcycles, and sometimes hired vehicles (Masamha et al. 2018). Women’s mobility is often 
restricted by social norms so they are unable to travel like men do (Heckert et al. 2020). 
When women sell locally, they receive lower prices (Masamha et al. 2018).
4 CGIAR GENDER Impact Platform · Working Paper #016
2.3 An iterative mixed-methods approach
We used an eight-step process to develop this research tool to study intrahousehold 
decision-making (Figure 2). We began with a literature review (step 1), the results of which 
are integrated into the background section of this paper. We then conducted stakeholder 
consultations (step 2). The subsequent sections then describe the methods and results for 
the qualitative dyadic interviews (steps 3–6) and the quantitative household survey (steps 7 
and 8) in greater detail.
Figure 2. Process to develop the iterative mixed-methods research tool for 
studying household decision-making
2.4 Stakeholder consultations
In the early stages of developing this tool, we consulted with a diverse set of people, including 
from local research and development organizations. This formed part of the transdisciplinary 
approach. To better understand and measure intrahousehold decision-making for their 
research or development programs, we aimed to:
• learn from people’s experiences with the research topic
• include important lessons in the design of the tool
• gauge their interests in using the proposed tool (once developed)
A brief guide was developed to facilitate the discussions, which was deployed with researchers 
from national programs who are involved in social and gender research, and practitioners 
from development organizations. This step not only increased the relevance of the tool for 
the study context, but also its future usefulness for other people. We recommend that all 
projects intending to use this approach also consult their projects’ relevant stakeholders.
2.5 Ethics approval
The study was approved by the Internal Review Board at the International Institute of 
Tropical Agriculture (reference IRB/008/2021) before the start of the research process. We 
obtained informed consent from all research participants before conducting the qualitative 
and quantitative interviews.
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 5
3. Qualitative dyadic interviews
3.1 Study design
We conducted in-depth interviews in three districts (Muleba, Biharamulo and Kibondo) 
across the two study regions with 40 married/cohabiting couples and other dyads. Married/
cohabiting couples were the predominant household composition in the study regions. 
Potential dyads were identified and purposively selected, with the help of local extension 
officers who used their extensive knowledge of the communities they work in and their 
expertise in agricultural matters. 
The protocol for conducting interviews with couples and other dyads entailed an introduction 
by the research team (comprising one woman and one man) and obtaining the informed 
consent of both study participants. All the interviews were audio recorded. The criteria 
for target couples and dyads for the study were those who (a) lived together and who 
(b) participated in cassava value chain activities and, therefore, (c) together or separately 
within their households made cassava production, processing and/or trading decisions along 
with key expenditure-related decisions. The 40 couples and other dyads who comprised the 
qualitative study sample included:
• married/cohabiting couples (37 total, including monogamous and polygynous couples)—
often, men were heads of households, but not always
• women as heads of households (three total)—not married, but living with their adult 
female or male child(ren)/dependents (>18 years or older)
In polygynous households (four of the 37 married/cohabiting couples), the research team did 
not ask to speak with multiple wives and their husband. Rather, they were asked to organize 
themselves according to who they thought should participate. In all cases, the husband 
decided who among his wives would participate in the interview.
Concerning women heads of household, very few (three) who met this profile were identified 
for interviews during the study. In many cases, a woman who was not married and considered 
herself the sole head of her household was living with children under the age of 18. These 
young members were not eligible to be interviewed according to our study protocol and, 
importantly, were likely not responsible for significant decisions in the household.
3.2 Qualitative interview tool
The first main section of the interview tool explores who does which cassava value chain 
tasks/activities in the household; and the second part explores in depth who makes important 
decisions within the household, how spouses make these decisions jointly or on their own, 
and why. 
Spouses were asked to jointly identify an important decision they made recently (within 
the past agricultural season) on matters related to cassava production, processing and/or 
marketing and on expenditure-related matters. In addition, the interview tool began with 
a section inquiring about basic demographic information, the household’s involvement in 
cassava value chain activities, and their subjective and relative livelihood status.
The semi-structured interviews were designed in a format that allowed the same issues 
to be discussed with multiple types of decision-making dyads. The open-ended questions 
allowed participants to respond in their own words, and the interviewer could frame follow-
up questions as necessary based on responses.
6 CGIAR GENDER Impact Platform · Working Paper #016
To improve study participants’ comprehension and response quality, the interviews were 
conducted with the use of two optional aids. Participants were encouraged to demonstrate 
their answers with beans for questions that required estimates of proportions or percentages. 
Additionally, we developed cards with illustrations that depicted specific tasks within cassava 
production, processing and trading/marketing associated with decision-making questions 
and included them in the qualitative dyadic interviews.
3.3 Analysis
Once the qualitative dyadic interviews were completed, the recordings of the interviews 
were transcribed and translated into English, then added into the ATLAS.ti software for 
analysis. Before the analysis, we developed a coding structure (Figure 3), which included 
inductive and deductive codes, by conducting a preliminary analysis of select transcripts. 
The coding structure represents a relatively full set of possible steps and substeps (or further 
actions dyads might take when making decisions) based on the preliminary analysis. This does 
not infer that the decision-making patterns we identified would necessarily follow all steps or 
contain such details—the coding structure was merely a tool to use to help us highlight any 
and all aspects of each decision of each dyad. The coding structure was based on a process 
framework for decision-making and potential inputs into that process.1 It is also important 
to note that we did not analyze all the qualitative data collected, but only those that could 
help us identify the decision-making patterns presented in the next section. Results from 
analyzing the other qualitative data will be presented in future scientific outputs.
1	 Modified	from	a	step-by-step	decision-making	process	developed	by	University	of	Massachusetts,	Dartmouth	
(see https://www.umassd.edu/media/umassdartmouth/fycm/decision_making_process.pdf,	PDF	89 KB)
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 7
3.4 Results
The analysis resulted in 80 individual decision-making patterns: two each from the 40 sets of 
interviewed couples and other dyads. That is, for each interview, we developed two different 
patterns in terms of visualizations (see Figure 4 for examples) to represent one agriculture- 
and one expenditure-related decision.
(a)
(b)
Figure 4. Illustration of (a) a longer, more complex decision-making pattern 
and (b) a shorter, less complex decision-making pattern
8 CGIAR GENDER Impact Platform · Working Paper #016
Figure 3.  Coding structure for analyzing the qualitative data
3.4 Results
The analysis resulted in 80 individual decision-making patterns: two each from the 40 sets of 
interviewed couples and other dyads. That is, for each interview, we developed two different 
patterns in terms of visualizations (see Figure 4 for examples) to represent one agriculture- 
and one expenditure-related decision.
(a)
(b)
Figure 4. Illustration of (a) a longer, more complex decision-making pattern 
and (b) a shorter, less complex decision-making pattern
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 9
Figure 3.  Coding structure for analyzing the qualitative data
Thereafter, research team members (two women, two men) came together in a workshop to 
analyze the different patterns/visualizations. During the workshop, the visualizations were 
printed and similar decision-making patterns were grouped together. Certain patterns (or 
visualizations) were rather long and complex—depicting more steps (and substeps) used 
by dyads when making decisions—while others were short and relatively simple—depicting 
very few steps that led to making a decision. There were within-group differences, of course, 
but most patterns included in a particular group contained similar elements.
This grouping process resulted in identifying seven common decision-making patterns that 
informed the development of seven vignettes, or short stories, on how married/cohabiting 
couples make agriculture- and expenditure-related decisions—each stemming from one 
of the common decision-making patterns. The vignettes for each of the decision-making 
patterns were developed using the content contained in the visualizations of the individual 
patterns within the seven decision-making pattern groupings. Thereafter, the patterns were 
ready for validation in the study regions.
We note that although we interviewed several dyads who were not married or cohabiting 
(e.g.,  mother–son and mother–daughter dyads), our study did not uncover any additional 
unique patterns based on those interviews. 
3.5 Validating the vignettes
Once the vignettes were developed, we validated them in the two study regions using 
dramatic skits and focus group discussions (FGDs). Doing so allowed us to consult people 
within selected communities about whether the vignettes represented real-life scenarios 
of couples making agriculture- and expenditure-related decisions. First, using the vignettes, 
scripts for the skits were developed by a local writer and reviewed by the research team. 
Then, a drama group was identified and the scripts were provided to the group for rehearsing.
The skits were performed in the three study districts where qualitative data collection was 
conducted. The skits were followed by FGDs. This took place in two study areas (where 
qualitative data were collected) and one in a nonstudy area that was purposively selected 
on the basis of its proximity to the study areas. In total, nine FGDs were conducted, each 
comprising six participants. In each of the three districts, there was one mixed-sex FGD, 
and two separate-sex FGDs (one with women, one with men). The participants of the FGDs 
comprised married/cohabiting women and men, including those in monogamous and 
polygynous relationships. Participants of FGDs were selected based on purposive sampling 
that allowed us to deliberately choose people who could provide rich insights and meaningful 
contributions to our FGDs, increasing the relevance of the discussions for our objective.
In each validation, the drama group acted out all of the different skits while FGD participants 
watched. Then during the FGDs, the research team used a guide to facilitate the discussion 
and took notes using a prepared data-collection sheet.
The validation process indicated that all the seven vignettes represented the decision-
making scenarios in the study area, although certain patterns were found to be more/less 
dominant across the study communities. After the validation process, the seven vignettes 
were ready to be embedded into the household survey.
10 CGIAR GENDER Impact Platform · Working Paper #016
4. Quantitative household survey
4.1 Sampling
We selected a single district from Kagera Region and a single district from Kigoma Region 
based on cassava-production level and population diversity. Muleba was selected from 
Kagera Region, and Kibondo was selected from Kigoma Region. The districts capture a 
diverse population with different ethnic groups, with a population of 540,310 and 261,331, 
respectively (URT Census 2012).
Both Muleba and Kibondo have significant cassava production, processing and trading 
activities in their respective districts. We purposively selected 14 wards (eight from Muleba’s 
43 wards and six from Kibondo’s 19 wards) for this study (Table 3). Once the 14 wards were 
purposively selected based on cassava production and accessibility, we defined the study’s 
target population based on the selection criteria (being married or in informal unions). 
The target population included households with married couples or informal unions, who 
were engaged in cassava production and processing (but not necessarily trading as some 
households only cultivate cassava as a food crop) in 35 villages of the eight wards in Muleba 
and 19 villages of the six wards in Kibondo.
Table 3. Population, number of wards and villages in the study districts
Description Muleba Kibondo
District population 540,310 261,331
Total number of wards 43 19
Total number of villages 166 50
Number of wards with significant cassava production 8 6
Number of selected villages with significant cassava production 35 19
Sample size (households) 816 536
The sample size was determined using Cochran’s formula (1991), given 
as where N is the total sample size, e is the margin of error, z is the z-score, p is the 
proportion of joint decision-making. This sample size formula for a single proportion was 
applied because measuring empowerment in joint spousal decision-making has traditionally 
followed a dichotomous approach in which decisions are either joint or sole, with the former 
assumed to reflect more empowerment than the latter (Acosta et al. 2020). Since we had 
no information on the proportion of joint spousal decision-making, we estimated p at 0.5. 
Considering a z-score of 2.575 (99 percent confidence level) and e of 3.5 percent, we settled 
on a combined sample size for the two districts of 1,352. Strata (district) sample sizes were 
determined using  where Ni is the sample size of strata or district i, wi and is the
proportion of total cassava production in strata or district i (60.3 percent for Muleba and 
39.7 percent for Kibondo).
Once we determined the sample size of each district, we created a sampling frame of villages 
arranged in alphabetical order, assigned the respective population size to each village, 
and determined the number of sample households to be drawn from each village based 
on probability proportional to population size. Village leaders and local extension agents 
provided lists of households in the selected villages to the research team. Households with 
formally or informally married couples in monogamous or polygynous relationships were 
sampled. 
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 11
Finally, we randomly selected individual households from each village based on the sampling 
frame. We opted to oversample in each village to ensure the availability of replacement 
households in case couples from selected households were away during the survey period 
or had left the area. In total, 1,352 households were selected from 54 villages (i.e., 816 from 
35 villages in Muleba and 536 from 19 villages in Kibondo).
4.2 Data collection
4.2.1 Interview format
We used one of two interview formats:
1. joint (husband–wife pair) interviews that covered the 12 modules in the questionnaire 
—except the final module (l), which collected data on minimum dietary diversity for 
women (MDD-W) that only the wife responded to
2. joint interviews that covered the first nine modules (a to i), individual interviews for the 
remaining three modules (j and k) of the questionnaire, and only women responded to 
the last module (l)
The two different interview formats will, in future work, allow us to compare responses across 
interview formats. The respondents were randomly assigned to their interview format.
In total, 2,028 respondents completed the survey: (i) 676 couples participated in interview 
format one; and (ii) 676 couples participated in interview format two, doubling the number 
of respondents from this interview format to 1,352. 
All households grew and processed cassava, but not all households traded cassava. As a 
result, the functional sample for questions related to cassava trading is smaller—451 couples 
from jointly conducted interviews and 498  couples individually interviewed—a total of 
1,447 respondents.
4.2.2 Survey questionnaire
The data were collected by trained enumerators using tablets equipped with SurveyCTO 
software. The household survey questionnaire included modules on (a) household 
identification; (b) household membership; (c) dwelling characteristics; (d) productive assets; 
(e) group membership; (f) credit and finance; (g) food security; (h) cassava production, 
processing and sales; (i) extension; (j) intrahousehold decision-making; (k) participation in 
activities linked to the decision-making topics; and (l) MDD-W.
For the decision-making module, respondents were questioned about the process used in 
their household to make decisions about several topics. In total, we asked about eight topics 
across four domains: three about cassava production, one about cassava processing, two 
about cassava trading, and two about expenditure. These eight were identified as important 
topics discussed by couples in the qualitative interviews. 
For each topic, respondents were asked three questions about the decision-making process 
in their household. The first question asked who in the household generally makes the 
final decision about the topic. Based on their response, respondents were read up to three 
vignettes describing different decision-making processes involving a husband and wife, 
and asked which vignette is most like how they and their spouse make decisions about the 
topic. The final question asked respondents about the most important reason for why final 
decisions in their household are made by a specific individual or jointly.
12 CGIAR GENDER Impact Platform · Working Paper #016
Questions similar to our first question about the final decision-maker have often been used by 
large-scale, international household survey programs to measure respondents’ participation 
in decision-making within a particular domain (The DHS Program 2020). This single-question 
approach contrasts with the multiquestion approach we used to measure intrahousehold 
decision-making. By comparing responses to the final decision-maker question with 
responses to the vignettes (on how decisions are made) and the why question, we were 
able to highlight the additional information these questions provide about intrahousehold 
decision-making processes that is missed by single-question approaches.
Annex table 1 presents the seven vignettes used to collect data on intrahousehold decision-
making variables in the four domains: (i) cassava production, (ii) cassava processing, 
(iii) cassava trading and (iv) expenditure.
The vignettes in Annex table 1 were read to respondents during the interview as they were 
shown corresponding cards with illustrations of the vignettes. The illustration cards were 
designed to be self-explanatory and without the need for literacy, so that respondents could 
understand the different components of each vignette (e.g., who comes up with the idea 
to do X, who shares the idea, whether there is any discussion and/or suggestions for an 
alternative option, and who makes the final decision).
The number of cards presented to the respondents during the interview differed depending 
on the answer to the first question in the module, “Who in the household generally makes the 
final decisions about a given issue?” The question was first posed to the respondents before 
reading aloud the vignettes and presenting the graphic cards. If, for example, respondents 
answered, “husband only” to this question, then they were read vignettes A, B, and G in 
Annex table 1, presented with the accompanying illustration cards, then asked to choose the 
one vignette that most represented the decision-making pattern they use. The same process 
was used if respondents answered, “wife only” (using vignettes E and F) or, “jointly” (using 
vignettes C and D).
5. Results
5.1 Descriptive results of decision-making 
questions
This section presents descriptive statistics for the intrahousehold decision-making questions 
and vignettes. Throughout this section, we pool data from the two types of interviews 
(i.e., one-on-one and joint) conducted using the survey questionnaire. Hence, the proportions 
shown in Annex tables 2–17 represent the share of the interviews conducted during the 
survey (combined: joint, individual women, and individual men). 
Wives’ and husbands’ responses to decision-making questions were not always in concord, 
nor should they be expected to be (Ambler et al. 2021; Seymour and Peterman 2018). For 
this reason, we believe considering wives’ and husbands’ responses separately, as we do 
below, provides a more accurate characterization of the data than simply relying on the 
responses for a single spouse per household.
5.1.1 Decision 1: Which variety to use
The first decision concerns the variety of cassava to be planted in the main plot. 
Overwhelmingly, the final say for this decision is had by the husband (39.6 percent) and the 
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 13
husband and wife jointly (49.3 percent); see Annex table 2. In 11.1 percent of cases (225) the 
wife has the final say for this decision. Among the 800 respondents who reported that the 
husband is the final decision-maker, 79.3 percent (635) reported that the decision originates 
with him, and he brings it to his wife for a discussion before making the decision himself 
(vignette A). The husband’s final say may alternatively be preceded by the wife sharing 
an idea (vignette B) or by no discussion at all (vignette G), but these scenarios were less 
common. Among the 635 households where the husband is the final decision-maker, these 
two scenarios were only reported by 10.2 percent (82) and 10.4 percent (83) of respondents, 
respectively. Among the 225 respondents who reported that the wife is the final decision-
maker, 83.6 percent (188) reported that she shared the idea in the first place (vignette E). 
Among the 995 households where the decision was made jointly, decisions usually resulted 
from the husband initiating the idea (vignette C)—78.2 percent (778) couples selected this 
scenario—compared to 21.8 percent (217) who identified with the scenario where the wife 
initially shares the idea (vignette D). Only 1.8 percent (37) reported that the wife had the 
final say after the husband shared the idea (vignette F).
As shown in Annex table 3, after disaggregating the data by final say, 26  percent of the 
jointly deciding dyads said the most important reason for the joint decision about selecting 
a variety is because both spouses agreed/supported the decision. The next most common 
reason (21.5 percent; 435) was that the husband made the decision because he is the head 
of the household, followed by the couple making the decision jointly because they make all 
decisions together (15.9 percent; 321).
5.1.2 Decision 2: Which area of land to plant
The second decision concerns which area of land to be devoted to the main cassava production. 
This decision was also primarily made by the husband or jointly, with only 7.8 percent (158) 
out of 2,025 total respondents reporting that the wife has the final say (Annex table 4). For 
decisions made solely by either the wife or husband, the final decision-maker was almost 
always reported to be the one who initiated the idea. Among the 884 households where 
the decision is made jointly, 74.9 percent (662) reported that the husband initiated the idea 
(vignette C) compared to 25.1 percent (222) for the wife (vignette D).
The most common reason for the decision (on which land to use) where the husband had the 
final say is because he is the head of the household (32.6 percent; 661) (Annex table 5). Next, 
21 percent (426) reported that the decision is made jointly because the couple agreed/both 
supported the decision. Just over one-eighth (14.3 percent; 290) reported that the couple 
makes the decision jointly because they make all decisions about this matter together.
5.1.3 Decision 3: When to harvest cassava from main plot
The third decision refers to when to harvest cassava from the main plot. This decision was 
mostly made jointly (51.3  percent; 1,038). Wives were reported to have the final say by 
23.3 percent (472) of respondents, compared to 25.3 percent (512) who reported that the 
husband made the final decision (Annex table 6). Notably, among the 1,038 respondents 
who reported that the final decision is made jointly, there is a much smaller difference in the 
number of respondents who reported that the husband or the wife initiated than for the 
other decisions in the production domain—61.5  percent (638) respondents reported the 
husband to have initiated the idea (vignette C), compared to 38.5 percent (400) reporting 
that the wife initiated the idea (vignette D).
The most common reason for making the decision of when to harvest is that the couple 
agreed/supported the decision, leading them to jointly have the final say (23 percent; 466) 
(Annex table 7). Next, 16.1 percent (325) reported that the decision is made jointly because 
the couple makes all decisions about this matter together, and 14.2 percent (288) reported 
that the husband makes the decision due to him being the head of the household.
14 CGIAR GENDER Impact Platform · Working Paper #016
5.1.4 Decision 4: How to process the cassava from main plot
The fourth decision concerns how to process the cassava from the main plot. This decision 
is the only one for which more respondents reported the wife having the sole final say than 
the husband. Overall, 46.2 percent (930) reported the final say on the decision was made 
jointly, 34 percent (684) reported the wife as having the final say, and 19.7 percent (396) 
reported the husband had the final say (Annex table 8). Among the 684 who reported that 
the wife has the final say, 89.6 percent (613) reported that she also initially shared the idea 
(vignette F). In the 930 cases of a joint decision, the wife shared the initial idea less frequently 
(42.6 percent; 396) (vignette D) compared to the husband.
The most commonly reported reason for processing decisions with the wife having the 
final say was because she is the most knowledgeable about the matter (21.9 percent; 440) 
(Annex table 9). Next, 20 percent (401) reported that the decision was made jointly because 
the couple agreed/supported the decision. Only 14 percent (281) reported that the couple 
makes the decision jointly because they make all decisions about this matter together.
5.1.5 Decision 5: Where to sell cassava from your main plot
The fifth decision is about where to sell cassava from the main plot. Fifty-one percent of 
(734) respondents reported that the husband had the final say, while 43.4  percent (625) 
reported the decision was made jointly and 5.6 percent (81) reported the wife had the final 
say (Annex table 10). Trends among individual decision-makers are similar to those presented 
in the first four decision-making scenarios—the final decision-maker was usually the person 
who introduced the idea.
The most common reason for the decision on where to sell cassava where the husband had 
the final say was because he is the head of the household (27.5 percent; 396) (Annex table 11). 
Next, 21  percent (302) reported that the decision was made jointly because the couple 
agreed/supported the decision. Just over 17 percent (251) reported that the couple makes 
the decision jointly because they make all decisions about this matter together.
5.1.6 Decision 6: How much cassava to sell from your main 
plot
The sixth decision concerns the amount of cassava from the main plot to sell. For this decision, 
50.7  percent (730) respondents reported that the decision is made jointly, 35.4  percent 
(510) reported that the husband had the final say, and 13.9  percent (200) reported that 
the wife had the final say (Annex table 12). Among the 730 respondents who reported that 
the final decision is made jointly, 65.2 percent (476) reported that the idea originated with 
the husband (vignette C). Among the 510 respondents who reported that the husband had 
the final say, 90.2  percent (460) reported that the husband initially brought up the idea 
(vignettes A and G)—12.4 percent (57) reported that the husband has the final say without a 
discussion with their wife (vignette G). Among the 200 respondents who reported the wife 
as having the final say, 88.5 percent (177) reported that she was also the one to initially share 
the idea (vignette E).
The most common reason for making decisions on how much cassava to sell from the 
main plot was the husband having the final say because he is the head of the household 
(26 percent; 375) (Annex table 13). Next, 22.8 percent (329) reported that the decision is 
made jointly because the couple agreed/both supported the decision. Just over 20 percent 
(20.6 percent; 296) reported that the couple makes the decision jointly because they make 
all decisions about this matter together.
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 15
5.1.7 Decision 7: Building/refurbishing a house
The seventh decision refers to building/refurbishing a house. For this decision, 
64.7  percent (1,183) reported that the decision is made solely by the husband, which is 
the highest proportion for decisions made solely by the husband across all eight decisions 
(Annex table 14). Just over one-third (612) reported that the decision is made jointly, and 
1.9 percent (34) reported that the wife had the final say. Among the 612 who reported that 
the decision is made jointly, 87.7 percent (537) reported that the idea originated with the 
husband (vignette C). Among the 1,183 who reported that the husband has the final say, 
92.5 percent (1,094) reported that the husband initially brought up the idea (vignettes A 
and G); 12.5  percent of those respondents (137) reported that the husband makes the 
decision without a discussion with his wife (vignette G). Among the 34 respondents who 
reported the wife as having the final say, 82.4 percent (28) reported that she was also the 
one to initially share the idea (vignette E).
The most common reason for the husband having the final say on building/refurbishing a 
house is because he is the head of the household (38.8 percent; 709) (Annex table 15). Next, 
15.5 percent (283) reported that the decision is made jointly because the couple agreed/both 
supported the decision. About one-seventh (14.1 percent; 258) reported that the husband 
had the final say because he contributed the most resources, income or labor.
5.1.8 Decision 8: Paying school fees
The eighth decision refers to paying school fees. For this decision, 51.1 percent (823) reported 
that the decision is made jointly, 44 percent (710) reported that the husband had the final 
say, and 4.9 percent (79) reported that the wife had the final say (Annex table 16). Among 
the 823 who reported that the decision is made jointly, 448 (54.4 percent) reported that the 
idea originated with the husband (vignette C). Among the 710 respondents who reported 
that the husband has the final say, 74.2 percent (527) reported that the husband initially 
brought up the idea (vignettes A and G); 13.5 percent (71) reported that the husband makes 
the decision without a discussion with his wife (vignette G). Among the 79 who reported 
the wife having the final say, 83.5 percent (66) reported that she initially shared the idea 
(vignette E).
The most common reason that school fee decisions are finally made by the husband is because 
he is the head of the household, which was cited by 38.8 percent (446) (Annex table 17). 
Next, 19.4 percent (312) respondents reported that the decision is made jointly because the 
couple agreed/both supported the decision. Nearly one-fifth (18.3 percent; 295) reported 
that the couple makes the decision jointly because they make all decisions about this matter 
together.
5.2 Regression analysis
In this section, we investigate whether information on how and why couples make decisions 
can help to explain differences in economic, food security and dietary outcomes across 
households beyond simply who in the household makes the final decision. We consider the 
following outcomes:
• Cassava productivity (kg/acres) of the main plot: defined as the total cassava in kg 
produced from the main cassava plot, divided by the size of that plot measured in acres. 
The main plot was the plot determined by respondents to be the most important for their 
household’s cassava production.
• Processed cassava sales productivity (sales/acre) of the main plot: constructed as the 
total sales of processed cassava from the main plot in units of 1,000 Tanzanian Shillings 
(TZS), divided by the size of the plot in acres.
16 CGIAR GENDER Impact Platform · Working Paper #016
• Dietary diversity of women (measured using MDD-W food groups) (FAO 2021): a 
numeric variable which measures the number of food groups that the respondent (the 
wife) consumed the previous day. Respondents were asked whether they had eaten 14 
different food groups at any point during the previous day, and these responses were 
aggregated into the 10 food groups defined in the MDD-W.2
• Food insecurity (measured using the Food Insecurity Experiences Scale) (Cafiero et  al. 
2016): a continuous variable that represents the total number of affirmative responses 
given by a respondent to eight questions about food security in their household over the 
past 30 days.3
Table 4 presents descriptive statistics for each of four outcomes considered in our analysis.
Table 4. Descriptive statistics for main outcome variables
Observations Mean Standard Deviation
Cassava productivity (kg/acre) 1,352 889.83 1,208.18
Processed cassava sales 912* 372.75 500.33
productivity (TZS1,000/acre)
Minimum diet diversity for women 1,352 3.07 1.17
Food Insecurity Experience Scale 1,349 3.15 2.36
* Per our selection criteria, all households who participated in the study grew and processed cassava; but it was not required 
that all households traded cassava (given that some regard cassava as only a food crop not a cash crop). Therefore, the 
number of observations is smaller for processed-cassava sales productivity.
Using Ordinary Least Squares (OLS) regression, we assess how respondents’ participation 
in different stages of the decision-making process correlates with the outcome of interest, 
using data on each of the eight decisions included in the survey. 
As a benchmark, we start by analyzing the correlation between who makes the final decision 
and the outcome variables, as is usually reported in empirical analyses of decision-making, 
based on the following specification:
.            (1) 
 is outcome d for individual i from household h.  is a vector of dummy variables 
indicating whether the wife made the final decision without her husband, the husband made 
the final decision without his wife, or the wife and husband made the final decision together 
for decision d. The omitted category is the husband making the final decision without his wife, 
such that the  coefficients measure the difference in means of each category compared to 
the husband making the final decision alone. 
 is a vector of control variables that includes age, highest level of education attained, 
and primary occupation for the wife and husband, respectively; household size; youth-
dependency ratio; dummy variables indicating whether the wife or husband owns the main 
plot; household-wealth quintile; dummy variables indicating whether an improved variety of 
cassava was grown and whether agrochemical inputs were used on the main plot; dummy 
variables indicating whether the dyad is in a formal union, monogamous nonformal union 
or polygynous nonformal union; dummy variables indicating, respectively, whether anyone 
in the household received cassava-related extension advice or training during the past 
12 months, borrowed money from someone outside the household or from an institution 
during the past 12 months, and is an active member of a formal or informal group of any type; 
2	 This	variable	is	sometimes	expressed	by	a	binary	variable	in	which	consumption	of	food	from	five	food	groups	
is	considered	to	be	an	adequately	diverse	diet,	but	we	left	the	variable	as	the	food	group	sum	to	express	the	
full range of dietary diversity in the sample (FAO 2021).
3	 While	this	score	is	normally	expressed	as	a	population	statistic,	on	the	individual	level	it	is	appropriate	to	use	
the	raw	score,	with	a	higher	score	indicating	a	higher	probability	that	the	household	is	food	insecure	(Cafiero	
et al. 2016).
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 17
and dummy variables indicating whether the wife and husband, respectively, performed 
work related to the four domains (cassava production, processing or trading; expenditure) 
during the 2020/21 agricultural season.
We then add variables from other questions following each vignette that the respondent 
identified with, about whether the wife or husband shared her/his idea with the other 
member of the dyad (origin) and whether discussion on the topic occurred before the final 
decision (discussion occurred), holding constant whether the wife or husband makes the final 
decision, as reflected in the following specification:
.            (2) 
 is a dummy variable indicating whether the wife initially recognized a need for a decision 
on the topic and shared her idea with the other member of the dyad.  is a dummy variable 
indicating that discussion on the topic occurred before the final decision.
We estimate equations  1 and 2 for each of our specified outcomes: cassava productivity, 
processed cassava sales productivity, MDD-W score (continuous), and Food Insecurity 
Experiences Scale (continuous). In each case, we limit our investigation to only those 
decision-making topics which are related to each outcome. The results of these regressions 
are shown in Annex tables 18–20.
We begin, in Annex table 18, by assessing the relationship between cassava productivity and 
decision-making on production-related topics. In Annex table 19, we focus on the relationship 
between processed-cassava sales productivity and decision-making on processing- and sales-
related topics. 
The overall trend, broadly apparent in both regressions, suggests a negative relationship 
between productivity and wives’ involvement in decision-making on these topics. In 
Annex table 18, for instance, the coefficient associated with the wife having the final say 
in the decision is negative in all cases and statistically significant in all cases except one 
specification (involving the variable based on the vignette). 
In Annex table 19, for decisions involving how to process the cassava from the main plot and 
where to sell cassava from the main plot, the coefficients associated with the wife either 
solely or jointly having the final say in the decisions are negative and statistically significant 
in all but two specifications. 
Across both Annex tables 18 and 19, the coefficients associated with the wife initiating 
the decision-making process and discussion between the couple are negative, although 
they are statistically insignificant for three decisions. Overall, there are no systematic 
differences between the specifications involving variables based on the traditional final-say 
survey questions and those based on the vignettes.
In Annex table 20, we assess the relationships between MDD-W (Panel A) and food security 
(Panel B) and decision-making on production-, processing- and sales-related topics. The 
coefficients associated with the wife solely or jointly having the final say are frequently 
positive and statistically significant. Because higher values of the respective variables are 
associated with more diverse diets, and less food-secure households, greater involvement 
by wives in decision-making is therefore associated with improved MDD-W, but 
with decreased food security for households. While there are instances in which these 
coefficients are negative and instances in which they are statistically insignificant, it is 
revealing that there are no specifications in which the wife either solely or jointly having 
the final say is statistically significant and negatively related to MDD-W or food security. 
As above, there is generally little difference in the results between the specifications 
involving variables based on the traditional final-say survey questions and those based on 
the vignettes.
18 CGIAR GENDER Impact Platform · Working Paper #016
6. Discussion
6.1 Interpreting our findings
In this section, we return to the four key research questions motivating this work:
1. Who makes cassava production, processing and trading decisions, and those on
expenditure-related matters, within the household?
2. Why does one spouse or the other (or another household member) make certain
decisions?
3. What process(es) do couples or other dyads use to make these agricultural- and
expenditure-related decisions?
4. How are decision-making processes (who decides, how and why) associated with
women’s dietary diversity, food security and agricultural productivity outcomes at the
household level?
We reflect on how this new tool enables us to answer these questions to a depth that would 
not have been possible using traditional (i.e.,  single-question) approaches to measuring 
intrahousehold decision-making, and how our findings converge or diverge from the broader 
literature on intrahousehold decision-making.
In general, our findings reveal variability and complexity in couples’ decision-making 
processes, beyond simply who in the household makes the final decision on a given topic. 
Decision-making processes are complex, yet this complexity is missed in research using 
traditional approaches to capture or measure decision-making. The variations in how married/
cohabiting couples make decisions are too great to be captured by a single question, such as 
who generally makes decisions (or the final decision) on a given topic. In particular, the tool 
provides a more nuanced portraal of women’s multiple and varied roles in decision-making 
processes and why they make sole or joint decisions with their spouses or are left out of 
those processes.
Turning first to our research questions about who makes decisions and what processes are 
used, we see across the eight topics considered that final decisions are most often either 
made jointly or made solely by the husband. With the exception of cassava-processing 
decisions—a topic often thought to be women’s domain and where they are perceived to 
be most knowledgeable—wives are less likely than husbands to be reported as the sole final 
decision-maker. Broadly, these findings align with other research on women’s empowerment 
from sub-Saharan Africa. Patriarchal gender norms remain strong throughout the region: 
women typically have less decision-making powers than men (Van Aelst 2014; Anderson 
et al. 2017). This evidence, however, relies mostly on studies using single-question 
approaches to measure decision-making. Our findings inform the relatively less-studied 
area of women’s participation in specific aspects of decision-making processes.
Going deeper, our results show that wives have some level of involvement in nearly all 
decisions, including those in which husbands are reported as the final decision-maker. 
Thus, even in these decisions, wives are not necessarily excluded from the decision-making 
process, as could be concluded if based only on a single question about the final decision-
maker. Overall, only a small share of decisions (4.4 percent across all topics) are reported as 
being made solely by the husband and not involving any discussions with his wife. 
Most decisions are either made jointly by the husband and wife (46.1 percent across all topics) 
or are at least discussed between partners (95.6  percent across all topics). Whether the 
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 19
decision is made jointly with equal weight by both spouses or the decision is only discussed 
briefly (see discussion on this by Acosta et al. 2020) cannot be determined based on our data, 
which warrants further investigation. Nonetheless, our results show that wives are still part 
of most decision-making processes. 
Given that proxy measures for empowerment are often based on the outcome of the 
decision-making process, the nuance introduced by incorporating process-oriented 
questions about decision-making into these proxy measures are likely important for use in 
designing or evaluating programs, especially when the program being evaluated aims to 
increase communication between spouses as they make important decisions. One possible 
use of our approach is, thus, to generate richer, more complete metrics of intrahousehold 
decision-making and empowerment than traditional measurement approaches.
While this information is valuable for better understanding gendered power relations and 
intrahousehold decision-making dynamics, we recognize that this incentive alone may not 
be enough to justify the added costs of implementing this new approach. Hence, our final 
research question focuses on how decision-making processes—in particular, the process-
level information gleaned by this in-depth approach—relate to specific outcomes about 
food and nutrition security and agricultural productivity.
For this discussion, we turn to the regression results from the previous section. Two general 
trends are apparent in these results. Greater involvement by wives in decision-making is 
associated with:
• lower productivity and household food security
• higher MDD-W
We consider potential explanations for why these outcomes are differentially associated 
with wives’ decision-making input. Evidence from sub-Saharan African countries on gender 
gaps in smallholder agricultural productivity find that women farmers achieve lower levels 
of productivity than men because of varying combinations of unequal access to inputs, 
differences in individual personal characteristics, and lower returns for their inputs and 
characteristics (Aguilar et al. 2015; Slavchevska 2015; Oseni et al. 2015). Women farmers, 
for instance, may employ different techniques than men do for cultivating the same crop, 
use lower quality inputs (including land) compared to men, or face different estimated 
(shadow) prices for inputs and credit than men—all of which might lead women to appear to 
be less productive than men (Peterman et al. 2011). Moreover, women’s increased input into 
decisions may necessarily free them from these constraints and increase their productivity 
or profits.
Although we control for whether the wife and husband solely or jointly own the main plot, 
the negative association we observe between wives’ involvement in decision-making and 
cassava productivity and sales (i.e.,  households where women have more involvement in 
decision-making have lower cassava production and sales) may be driven by these same 
factors of unequal access to inputs, individual characteristics, and lower returns to inputs 
that commonly explain gender productivity gaps. Cassava production and sales in households 
in which wives are more involved in decision-making may be more exposed to underlying 
gender inequities inherent in carrying out these tasks, compared to households in which 
women are less involved in cassava-related decision-making. Thus, production and sales in 
these households might be lower relative to production and sales in households in which 
wives are less involved in decision-making. Wives who are more involved in decision-making 
may also be more involved in the actual day-to-day management and/or labor associated 
with cassava production and sales activities, although importantly this does not necessarily 
have to be the case. Our data, for instance, show a wide variety of reasons—other than 
resources, income and labor contributions—for why specific household members make 
production- and sales-related decisions. Specifically, the choice of decision-maker is often 
motivated by who is the most knowledgeable or by default reasons (e.g., that is the person 
who always makes the decision, or they are the head of household).
20 CGIAR GENDER Impact Platform · Working Paper #016
The negative association between wives’ participation in decision-making and household food 
security, alongside a positive association with MDD-W may initially appear counterintuitive, 
given that these are both nutrition-related outcomes. However, they measure very different 
aspects of nutrition. MDD-W measures dietary diversity in terms of the number of different 
food groups that a woman consumed the previous day. Nearly all individuals are consuming 
staples, such as cassava. However, to have higher dietary diversity scores, individuals need 
to consume a range of food groups; for example, iron- or vitamin A–rich foods. Intense 
cassava farming may compete for the space and energy needed to produce or acquire these 
products. The household food security measure, on the other hand, taps into food quantity 
and the extent to which household members were able to consume enough food to avoid 
being hungry in the past month. Staple foods, such as cassava, often have an important 
role in meeting basic food security needs. Given the importance of quantity in ensuring 
household food security, it is not surprising that food security and productivity follow similar 
associations with women’s decision-making.
To explain the positive association between women’s increased input into decision-making 
and MDD-W, we see similar examples in an emerging body of literature, based mostly on 
evidence from the WEAI and pro-WEAI, that establishes a consistent positive relationship 
between women’s empowerment and their health and nutrition outcomes (Amugsi et al. 
2016; Malapit and Quisumbing 2015; Malapit et al. 2015; Sinharoy et al. 2018). Given the 
central role of decision-making power in the WEAI and pro-WEAI, it is unsurprising that 
we find a similar result for the influence of women’s involvement in decision-making on 
improved MDD-W.
6.2 Applying this approach to future studies
The step-by-step process we took to develop the mixed-methods research tool was initially 
rather generic in nature, so we feel that research or development projects focusing on other 
agricultural value chains and geographies could easily modify their focus within each step to 
fit their needs and context. 
Different qualitative data analyses could be carried out in addition to the ones we did to 
inform the development of additional vignettes. For example, we could have followed 
Bernard et  al. (2020) to focus our analysis and the vignettes on why decisions are made 
between spouses in a marital dyad. We instead chose to focus on how decisions are made 
because we believe that decision-making processes used by spouses are an important and 
understudied theme that can help us better understand decision-making. 
Moreover, understanding how decisions are made is more conducive to short vignettes 
in quantitative survey instruments than describing why a decision is made. We included 
questions on why spouses make decisions alone or jointly in follow-up questions. 
Also, the main decision-making module we included in the household survey could instead be 
integrated into an existing survey instrument for studies that have an interest in (but not a 
central focus on) measuring intrahousehold decision-making and the topics we concentrated 
on in the other modules.
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 21
6.3 Conclusion
Intrahousehold decision-making is an important aspect of women’s empowerment, yet 
questions remain on how to ask about decision-making and how to interpret the responses. 
We designed a multistep, mixed-methods, transdisciplinary approach to understand 
intrahousehold decision-making and applied it to households active in the cassava value 
chain in rural northwestern Tanzania. 
This approach drew on past literature and stakeholder consultations to design and 
conduct qualitative dyadic interviews on how marital dyads make decisions. The results 
of these interviews informed the development and validation of vignettes that were later 
incorporated into a household survey to better understand who makes important decisions, 
as well as how and why these decisions are made.
The qualitative part of the study revealed seven different decision-making patterns in which 
one of the spouses shares the idea, the two discuss, and the final decision is made either by 
the wife, husband, or jointly. In our quantitative results, we find somewhat superficially—
yet consistent with the literature—that women have less input than men into the types of 
decisions that we asked about. However, women do have some input in most decisions—
either because the decision is discussed by the dyad or because the decision is made jointly. 
How then does input into decisions matter for important outcomes? We find that greater 
involvement of wives in decision-making is linked to lower agricultural productivity and lower 
food security, but higher dietary diversity of women. These findings may be attributable to 
the fact that these are associational analyses of cross-sectional data and could be driven, for 
example, by the fact that women may only have control over lower quality cropping land. It is 
also worth considering that dietary diversity is related to the quality of diets (variety of foods 
consumed), not the amount of food. Meanwhile, individuals may respond to food-security 
questions with a primary focus on the amount of food available. Women’s activities and 
decision-making input, however, are often dedicated to activities that enhance the diversity 
of available foods (such as managing kitchen gardens), not necessarily the staple crops that 
drive the quantity of food available.
Overall, the novel approach we developed to studying decision-making has the potential to be 
applied across a wide variety of geographic contexts, as well as to different types of decisions, 
including those outside of agriculture. Further application of this approach has the potential 
to help us better understand decision-making processes, women’s (dis)empowerment, and 
the implications for important development outcomes.
22 CGIAR GENDER Impact Platform · Working Paper #016
References
Acosta, Mariola, Margit van Wessel, Severine van Bommel, Edidah L. Ampaire, Jennifer Twyman, Laurence 
Jassogne, and Peter H. Feindt. 2020. “What Does It Mean to Make a ‘Joint’ Decision? Unpacking 
Intra-Household Decision-Making in Agriculture: Implications for Policy and Practice.” The Journal of 
Development Studies 56 (6): 1210–29. https://doi.org/10.1080/00220388.2019.1650169.
Aguilar, Arturo, Eliana Carranza, Markus Goldstein, Talip Kilic, and Gbemisola Oseni. 2015. “Decomposition 
of Gender Differentials in Agricultural Productivity in Ethiopia.” Agricultural Economics 46 (3): 311–
34. https://doi.org/10.1111/agec.12167.
Ambler, Kate, Cheryl Doss, Caitlin Kieran, and Simone Passarelli. 2021. “He Says, She Says: Spousal 
Disagreement in Survey Measures of Bargaining Power.” Economic Development and Cultural Change 
69 (2): 765–88. https://doi.org/10.1086/703082.
Amugsi, Dickson A., Anna Lartey, Elizabeth Kimani-Murage, and Blessing U. Mberu. 2016. “Women’s 
Participation in Household Decision-Making and Higher Dietary Diversity: Findings from Nationally 
Representative Data from Ghana.” Journal of Health, Population and Nutrition 35 (1): 16. https://doi.
org/10.1186/s41043-016-0053-1.
Anderson, C. Leigh, Travis W. Reynolds, and Mary Kay Gugerty. 2017. “Husband and Wife Perspectives 
on Farm Household Decision-Making Authority and Evidence on Intra-Household Accord 
in Rural Tanzania.” World Development 90 (February): 169–83. https://doi.org/10.1016/j.
worlddev.2016.09.005.
Andersson, Karolin, Johanna Bergman Lodin, and Linley Chiwona-Karltun. 2016. “Gender Dynamics in 
Cassava Leaves Value Chains: The Case of Tanzania.” Journal of Gender, Agriculture and Food Security 
(Agri-Gender) 1 (302-2016-4753): 84–109. http://dx.doi.org/10.22004/ag.econ.246031.
Bernard, Tanguy, Cheryl Doss, Melissa Hidrobo, Jessica Hoel, and Caitlin Kieran. 2020. “Ask Me Why: 
Patterns of Intrahousehold Decision-Making.” World Development 125. https://doi.org/10.1016/j.
worlddev.2019.104671.
Cafiero, Carlo, Mark Nord, Sara Viviani, M. Eduardo Del Grossi, Terri Ballard, Anne Kepple, and C. Nwosu. 
2016. “Methods for Estimating Comparable Prevalence Rates of Food Insecurity Experienced 
by Adults throughout the World.” Rome: Food and Agriculture Organization of the United Nations. 
https://iopscience.iop.org/article/10.1088/1742-6596/772/1/012060/pdf.
Cochran, William G. 1991. “Sampling Techniques.” In Wiley.Com, 3rd ed. Wiley. https://www.wiley.com/en-
us/Sampling+Techniques%2C+3rd+Edition-p-9780471162407.
Doss, Cheryl R. 2013. “Intrahousehold Bargaining and Resource Allocation in Developing Countries.” World 
Bank Research Observer 28 (1): 52–78. https://doi.org/10.1093/wbro/lkt001.
Duflo, Esther. 2003. “Grandmothers and Granddaughters: Old-Age Pensions and Intrahousehold Allocation 
in South Africa.” The World Bank Economic Review 17 (1): 1–25. https://doi.org/10.1093/wber/
lhg013.
Duflo, Esther, and Christopher R. Udry. 2004. Intrahousehold Resource Allocation in Cote d’Ivoire: Social 
Norms, Separate Accounts and Consumption Choices. Mass., USA: National Bureau of Economic 
Research Cambridge. https://www.nber.org/papers/w10498.
Elias, Marlène, Steven Michael Cole, Agnes Quisumbing, A. P. Paez Valencia, Ruth Meinzen-Dick, and 
Jennifer Twyman. 2021. “Assessing Women’s Empowerment in Agricultural Research.” Advancing 
Gender Equality through Agricultural and Environmental Research: Past, Present, and Future, 329–64. 
https://doi.org/10.2499/9780896293915_09.
FAO. 2021. Minimum dietary diversity for women. Rome. https://doi.org/10.4060/cb3434en.
Fischer, Elisabeth, and Matin Qaim. 2012. “Gender, Agricultural Commercialization, and Collective Action in 
Kenya.” Food Security 4 (3): 441–53. https://doi.org/10.1007/s12571-012-0199-7.
Global Nutrition Report. 2022. Global Nutrition Report: Stronger Commitments for Greater Action. Bristol, 
UK: Development Initiatives. https://globalnutritionreport.org/.
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 23
Haapanen, Lauri, and Ville JE Manninen. 2021. “Etic and Emic Data Production Methods in the Study of 
Journalistic Work Practices: A Systematic Literature Review.” Journalism May, 14648849211016997. 
https://doi.org/10.1177/14648849211016997.
Haddad, Lawrence, John Hoddinott, and Harold Alderman. 1997. Intrahousehold Resource Allocation in 
Developing Countries: Models, Methods and Policies. International Food Policy Research Institute, 
Johns Hopkins University Press. https://www.ifpri.org/publication/intrahousehold-resource-
allocation-developing-countries-0.
Heckert, Jessica, Emily Myers, and Hazel J. Malapit. 2020. "Developing Survey-Based Measures of 
Gendered Freedom of Movement for Use in Studies of Agricultural Value Chains." Intl Food Policy 
Res Inst. Vol. 1966. https://doi.org/10.2499/p15738coll2.134048.
Kabeer, Naila. 1999. “Resources, Agency, Achievements: Reflections on the Measurement of Women’s 
Empowerment.” Development and Change 30(3): 435–464. https://doi.org/10.1111/1467-
7660.00125.
Kishor, Sunita, and Lekha Subaiya. 2008. “Understanding Women’s Empowerment: A Comparative Analysis 
of Demographic and Health Surveys (DHS) Data.” DHS Comparative Reports 20. Calverton, MD: 
Macro International. https://dhsprogram.com/publications/publication-cr20-comparative-reports.
cfm.
Laszlo, Sonia, Kate Grantham, Ecem Oskay, and Tingting Zhang. 2020. “Grappling with the Challenges of 
Measuring Women’s Economic Empowerment in Intrahousehold Settings.” World Development 
132: 104959. https://doi.org/10.1016/j.worlddev.2020.104959.
Malapit, Hazel Jean L., Suneetha Kadiyala, Agnes R. Quisumbing, Kenda Cunningham, and Parul Tyagi. 
2015. “Women’s Empowerment Mitigates the Negative Effects of Low Production Diversity on 
Maternal and Child Nutrition in Nepal.” The Journal of Development Studies 51 (8): 1097–1123. 
https://doi.org/10.1080/00220388.2015.1018904.
Malapit, Hazel Jean L., Agnes Quisumbing, Ruth Meinzen-Dick, Greg Seymour, Elena M. Martinez, Jessica 
Heckert, Deborah Rubin, Ana Vaz, Kathryn M. Yount, and Gender Agriculture Assets Project Phase. 
2019. “Development of the Project-Level Women’s Empowerment in Agriculture Index (pro-WEAI).” 
World Development 122: 675–92. https://doi.org/10.1016/j.worlddev.2019.06.018.
Malapit, Hazel Jean L., and Agnes R. Quisumbing. 2015. “What Dimensions of Women’s Empowerment in 
Agriculture Matter for Nutrition in Ghana?” Food Policy 52 (April): 54–63. https://doi.org/10.1016/j.
foodpol.2015.02.003.
Masamha, Blessing, Vusilizwe Thebe, and Veronica N. E. Uzokwe. 2018. “Mapping Cassava Food 
Value Chains in Tanzania’s Smallholder Farming Sector: The Implications of Intra-Household 
Gender Dynamics.” Journal of Rural Studies 58 (February): 82–92. https://doi.org/10.1016/j.
jrurstud.2017.12.011.
Masamha, Blessing, Veronica Ne Uzokwe, Frederick Evans Ntagwabira, Damien Gabagambi, and Peter 
Mamiro. 2019. “Gender influence on participation in cassava value chains in smallholder farming 
sectors: evidence from Kigoma Region, Tanzania.” Experimental Agriculture 55 (1): 57–72. https://
doi.org/10.1017/S0014479717000552.
Ministry of Agriculture. 2020. National Cassava Development Strategy (NCDS) 2020-2030. Dodoma, 
Tanzania: The United Republic of Tanzania Ministry of Agriculture. https://www.kilimo.go.tz/
uploads/CASSAVA_DEVELOPMENT_STRATEGY.pdf.
Montagnac, Julie A., Christopher R. Davis, and Sherry A. Tanumihardjo. 2009. “Nutritional Value of Cassava 
for Use as a Staple Food and Recent Advances for Improvement.” Comprehensive Reviews in Food 
Science and Food Safety 8 (3): 181–94. https://doi.org/10.1111/j.1541-4337.2009.00077.x.
Mtunguja, M. K., H. S. Laswai, Y. C. Muzanila, and J. Ndunguru. 2014. “Farmer’s Knowledge on Selection 
and Conservation of Cassava (Manihot Esculanta) Genetic Resources in Tanzania.” Journal of Biology, 
Agriculture and Healthcare 4 (10): 120–29. https://www.iiste.org/Journals/index.php/JBAH/article/
viewFile/12738/24272.
Oseni, Gbemisola, Paul Corral, Markus Goldstein, and Paul Winters. 2015. “Explaining Gender Differentials 
in Agricultural Production in Nigeria.” Agricultural Economics 46 (3): 285–310. https://doi.
org/10.1111/agec.12166.
24 CGIAR GENDER Impact Platform · Working Paper #016
Peterman, Amber, Agnes R. Quisumbing, Julia A. Behrman, and Ephraim Nkonya. 2011. “Understanding the 
Complexities Surrounding Gender Differences in Agricultural Productivity in Nigeria and Uganda.” 
Journal of Development Studies 47 (10): 1482–1509. https://doi.org/10.1080/00220388.2010.53622
2.
Quisumbing, Agnes, Jessica Heckert, Simone Faas, Gayathri Ramani, Kalyani Raghunathan, Hazel Malapit, 
and The pro-WEAI for Market Inclusion Study Team. 2021. “Women’s Empowerment and Gender 
Equality in Agricultural Value Chains: Evidence from Four Countries in Asia and Africa.” Food Security 
13 (5): 1101–24. https://doi.org/10.1007/s12571-021-01193-5.
Seymour, Greg, and Amber Peterman. 2018. “Context and Measurement: An Analysis of the Relationship 
between Intrahousehold Decision-Making and Autonomy.” World Development 111 (November): 97–
112. https://doi.org/10.1016/j.worlddev.2018.06.027.
Sinharoy, Sheela S., Jillian L. Waid, Regine Haardörfer, Amanda Wendt, Sabine Gabrysch, and Kathryn 
M. Yount. 2018. “Women’s Dietary Diversity in Rural Bangladesh: Pathways through Women’s 
Empowerment.” Maternal & Child Nutrition 14 (1): e12489. https://doi.org/10.1111/mcn.12489.
Slavchevska, Vanya. 2015. “Gender Differences in Agricultural Productivity: The Case of Tanzania.” 
Agricultural Economics 46 (3): 335–55. https://doi.org/10.1111/agec.12168.
The DHS Program. 2020. DHS Model Questionnaire—Phase 8. https://dhsprogram.com/publications/
publication-DHSQ8-DHS-Questionnaires-and-Manuals.cfm. 
URT (United Republic of Tanzania). 2013. 2012 POPULATION AND HOUSING CENSUS: Population 
Distribution by Administrative Area. National Bureau of Statistics Ministry of Finance Dar es Salaam 
and Office of Chief Government Statistician President’s Office, Finance, Economy and Development 
Planning Zanzibar, March 2013. https://www.nbs.go.tz/index.php/en/census-surveys/population-
and-housing-census/162-2012-phcpopulation-distribution-by-administrative-areas 
Van Aelst, Katrien. 2014. “Household Decision-Making and Gender Relations in Tanzania: Literature and 
Theory Review.” IOB Working Paper/2014.02. Antwerp, Belgium: Institute of Development Policy 
and Management. https://ideas.repec.org/p/iob/wpaper/201402.html.
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 25
Annex
Annex table 1. Vignettes applied to collect data on intrahousehold decision-making 
variables
  Description Vignette
A. Husband shares The husband realizes something needs to happen or a decision needs to be made 
idea, discusses with regarding X. The husband shares his idea with his wife and they discuss jointly about the 
wife, and husband idea. After discussing, the wife confirms she is supportive of the idea and/or suggests an 
makes the final alternative option that the husband considers before he makes the final decision. 
decision
B. Wife shares idea, The wife realizes something needs to happen or a decision needs to be made regarding 
discusses with X. The wife shares her idea with her husband and they discuss jointly about the idea. After 
husband, and the discussing, the husband confirms he is supportive of the idea and/or alternative options 
husband makes the from either the husband or wife are considered before the husband makes the final 
final decision decision.
C. Husband shares The husband realizes something needs to happen or a decision needs to be made 
idea, discusses with regarding X. The husband shares his idea with his wife and they discuss jointly about the 
wife, and they make idea. After discussing, the wife confirms she is supportive of the idea and/or suggests an 
a joint final decision alternative option that the husband considers before the husband and wife make a joint 
final decision.
D. Wife shares idea, The wife realizes something needs to happen or a decision needs to be made regarding 
discusses with X. The wife shares her idea with her husband and they discuss jointly about the idea. 
husband, and they After discussing, the husband confirms he is supportive of the idea and/or suggests an 
make a joint final alternative option that the wife considers before the wife and husband make a joint final 
decision decision.
E. Wife shares idea, The wife realizes something needs to happen or a decision needs to be made regarding 
discusses with X. The wife shares her idea with her husband and they discuss jointly about the idea. After 
husband, and wife discussing, the husband confirms he is supportive of the idea and/or the wife considers an 
makes the final alternative option from her husband before she makes the final decision.
decision
F. Husband shares The husband realizes something needs to happen or a decision needs to be made 
idea, discusses regarding X. The husband shares his idea with his wife and they discuss jointly about the 
with wife, and wife idea. After discussing, the wife confirms she is supportive of the idea and/or suggests an 
makes the final alternative option that the husband considers before the wife makes the final decision.
decision
G. Husband shares The husband realizes something needs to happen or a decision needs to be made 
idea with his wife regarding X. After sharing his idea with his wife, he makes the final decision.
before he makes 
the final decision
Note: X represents eight decisions: three production decisions—(1) which cassava variety to use; (2) which plot to use for cassava production; 
(3) when to harvest the cassava; one processing decision—(4) how to process cassava; two trading decisions—(5) where to sell cassava; (6) 
how much cassava to sell; and two expenditure decisions—(7) building/refurbishing a house; (8) paying school fees. 
26 CGIAR GENDER Impact Platform · Working Paper #016
Annex table 2. Descriptive analysis of decisions on which cassava variety to use, 
according to who makes the decision and the type of couple resembled
Among the couples that you resemble, Who in your household generally makes the final decision 
which is the MOST similar to your couple? about which cassava variety to use?
Husband only Wife only Jointly
Vignette A. Husband shares idea, discusses with 635 
wife, and husband makes the final decision (229/227/179)
Vignette B. Wife shares idea, discusses with 82 
husband, and the husband makes the final (23/36/23)
decision
Vignette C. Husband shares idea, discusses with 778 
wife, and they make a joint final decision (254/222/302)
Vignette D. Wife shares idea, discusses with 217 
husband, and they make a joint final decision (67/73/77)
Vignette E. Wife shares idea, discusses with 188 
husband, and wife makes the final decision (56/70/62)
Vignette F. Husband shares idea, discusses with 37 
wife, and wife makes the final decision (14/16/7)
Vignette G. Husband shares idea with his wife 83 
before he makes the final decision (29/29/25)
Total observations 800 225 995 
(281/292/227) (70/86/69) (321/295/379)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
Annex table 3. Descriptive analysis of decisions on which cassava variety to use, 
according to who makes the decision and why
Most important reason why the final Who in your household generally makes the final decision 
decisions about which variety of cassava about which cassava variety to use?
to use are made by a sole member or Husband only Wife only Jointly
jointly.
Couple agreed/supported the decision 525 
(176/161/188)
Person X contributes the most resources, 52 26 
income, or labor (15/20/17) (6/10/10)
Couple contributes resources, income, or labor 149 
(43/46/60)
Person X is head of household 435 1 
(126/195/114) (0/1/0)
Person X introduced the idea 81 25 
(43/19/19) (2/14/9)
Person X makes decisions about this and Person 44 58 
Y makes decisions about other things (17/12/15) (22/19/17)
Couple makes all decisions about this activity 321 
together (102/88/131)
Person X is the most knowledgeable about this 185 113 
activity (80/46/59) (39/41/33)
Other 3 2 0 
(0/0/3) (1/1/0) (0/0/0)
Total observations 800 225 995 
(281/292/227) (70/86/69) (321/295/379)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 27
Annex table 4. Descriptive analysis of decisions on which land to use, according to 
who makes the decision and type of couple resembled
Among the couples that you resemble, Who in your household generally makes the final decision 
which is the MOST similar to your couple? about which area of land to use?
Husband only Wife only Jointly
Vignette A. Husband shares idea, discusses with 795 
wife, and husband makes the final decision (237/276/282)
Vignette B. Wife shares idea, discusses with 98 
husband, and the husband makes the final (24/35/39)
decision
Vignette C. Husband shares idea, discusses with 662 
wife, and they make a joint final decision (263/221/178)
Vignette D. Wife shares idea, discusses with 222 
husband, and they make a joint final decision (70/67/85)
Vignette E. Wife shares idea, discusses with 136 
husband, and wife makes the final decision (53/40/43)
Vignette F. Husband shares idea, discusses with 22 
wife, and wife makes the final decision (6/8/8)
Vignette G. Husband shares idea with his wife 90 
before he makes the final decision (21/29/40)
Total observations 983 158 884 
(340/361/282) (48/51/59) (288/263/333)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
Annex table 5. Descriptive analysis of decisions on which land to use, according to 
who makes the decision and why
Most important reason why the final Who in your household generally makes the final decision 
decisions about which area of land to use about which area of land to use?
are made by a sole member or jointly. Husband only Wife only Jointly
Couple agreed/supported the decision 426 
(136/126/164)
Person X contributes the most resources, 86 31 
income, or labor (32/15/39) (8/10/13)
Couple contributes resources, income, or labor 165 
(46/56/63)
Person X is head of household 661 0 
(207/278/176) (0/0/0)
Person X introduced the idea 57 20 
(25/20/12) (4/6/10)
Person X makes decisions about this and Person 47 47 
Y makes decisions about other things (20/12/15) (15/14/18)
Couple makes all decisions about this activity 290 
together (105/80/105)
Person X is the most knowledgeable about this 130 59 
activity (56/35/39) (21/20/18)
Other 2 1 3 
(0/1/1) (0/1/0) (1/1/1)
Total observations 983 158 884 
(340/361/282) (48/51/59) (288/263/333)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
28 CGIAR GENDER Impact Platform · Working Paper #016
Annex table 6. Descriptive analysis of decisions on when to harvest cassava, 
according to who makes the decision and type of couple resembled
Among the couples that you resemble, Who in your household generally makes the final decision 
which is the MOST similar to your couple? about when to harvest cassava?
Husband only Wife only Jointly
Vignette A. Husband shares idea, discusses with 366 
wife, and husband makes the final decision (138/127/101)
Vignette B. Wife shares idea, discusses with 84 
husband, and the husband makes the final (28/30/26)
decision
Vignette C. Husband shares idea, discusses with 638 
wife, and they make a joint final decision (212/192/234)
Vignette D. Wife shares idea, discusses with 400 
husband, and they make a joint final decision (122/141/137)
Vignette E. Wife shares idea, discusses with 425 
husband, and wife makes the final decision (137/143/145)
Vignette F. Husband shares idea, discusses with 47 
wife, and wife makes the final decision (13/16/18)
Vignette G. Husband shares idea with his wife 62 
before he makes the final decision (24/24/14)
Total observations 512 472 1,038 
(190/181/141) (150/159/163) (334/333/371)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
Annex table 7. Descriptive analysis of decisions on when to harvest cassava, 
according to who makes the decision and why
Most important reason why the final Who in your household generally makes the final decision 
decisions about when to harvest cassava about when to harvest cassava?
are made by a sole member or jointly. Husband only Wife only Jointly
Couple agreed/supported the decision 466 
(157/156/153)
Person X contributes the most resources, 66 35 
income, or labor (24/23/19) (12/11/12)
Couple contributes resources, income, or labor 247 
(76/78/93)
Person X is head of household 288 5 
(94/121/73) (1/3/1)
Person X introduced the idea 33 33 
(15/8/10) (15/8/10)
Person X makes decisions about this and Person 47 128 
Y makes decisions about other things (19/12/16) (38/51/39)
Couple makes all decisions about this activity 325 
together (101/99/125)
Person X is the most knowledgeable about this 78 270 
activity (38/17/23) (82/88/100)
Other 0 1 0 
(0/0/0) (0/0/1) (0/0/0)
Total observations 512 472 1,038 
(190/181/141) (150/159/163) (334/333/371)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 29
Annex table 8. Descriptive analysis of decisions on how to process cassava, according 
to who makes the decision and type of couple resembled
Among the couples that you resemble, Who in your household generally makes the final decision 
which is the MOST similar to your couple? about how to process cassava?
Husband only Wife only Jointly
Vignette A. Husband shares idea, discusses with 294 
wife, and husband makes the final decision (114/89/91)
Vignette B. Wife shares idea, discusses with 63 
husband, and the husband makes the final (19/20/24)
decision
Vignette C. Husband shares idea, discusses with 534 
wife, and they make a joint final decision (180/173/181)
Vignette D. Wife shares idea, discusses with 396 
husband, and they make a joint final decision (130/129/137)
Vignette E. Wife shares idea, discusses with 613 
husband, and wife makes the final decision (185/221/207)
Vignette F. Husband shares idea, discusses with 71 
wife, and wife makes the final decision (33/17/21)
Vignette G. Husband shares idea with his wife 39 
before he makes the final decision (10/19/10)
Total observations 396 684 930 
(143/128/125) (218/238/228) (310/302/318)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
Annex table 9. Descriptive analysis of decisions on how to process cassava, according 
to who makes the decision and why
Most important reason why the final Who in your household generally makes the final decision 
decisions about how to process cassava about how to process cassava?
are made by a sole member or jointly. Husband only Wife only Jointly
Couple agreed/supported the decision 401 
(141/130/130)
Person X contributes the most resources, 48 48 
income, or labor (23/10/15) (20/16/12)
Couple contributes resources, income, or labor 248 
(70/85/93)
Person X is head of household 240 6 
(79/85/76) (3/1/2)
Person X introduced the idea 15 23 
(9/3/3) (9/6/8)
Person X makes decisions about this and Person 34 165 
Y makes decisions about other things (7/14/13) (49/57/59)
Couple makes all decisions about this activity 281 
together (99/87/95)
Person X is the most knowledgeable about this 59 440 
activity (25/16/18) (137/157/146)
Other 0 2 0 
(0/0/0) (1/0/1) (0/0/0)
Total observations 396 684 930 
(143/128/125) (218/238/228) (310/302/318)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
30 CGIAR GENDER Impact Platform · Working Paper #016
Annex table 10. Descriptive analysis of decisions on where to sell cassava, according 
to who makes the decision and type of couple resembled
Among the couples that you resemble, Who in your household generally makes the final decision 
which is the MOST similar to your couple? about where to sell cassava?
Husband only Wife only Jointly
Vignette A. Husband shares idea, discusses with 598 
wife, and husband makes the final decision (197/251/150)
Vignette B. Wife shares idea, discusses with 46 
husband, and the husband makes the final (19/13/14)
decision
Vignette C. Husband shares idea, discusses with 506 
wife, and they make a joint final decision (178/139/189)
Vignette D. Wife shares idea, discusses with 119 
husband, and they make a joint final decision (46/34/39)
Vignette E. Wife shares idea, discusses with 70 
husband, and wife makes the final decision (27/16/27)
Vignette F. Husband shares idea, discusses with 11 
wife, and wife makes the final decision (1/3/7)
Vignette G. Husband shares idea with his wife 90 
before he makes the final decision (27/39/24)
Total observations 734 81 625 
(243/303/188) (28/19/34) (224/173/228)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
Annex table 11. Descriptive analysis of decisions on where to sell cassava, according 
to who makes the decision and why
Most important reason why the final Who in your household generally makes the final decision 
decisions about where to sell cassava are about where to sell cassava?
made by a sole member or jointly. Husband only Wife only Jointly
Couple agreed/supported the decision 302 
(103/82/117)
Person X contributes the most resources, 46 11 
income, or labor (17/13/16) (3/2/6)
Couple contributes resources, income, or labor 72 
(26/18/28)
Person X is head of household 396 0 
(119/183/94) (0/0/0)
Person X introduced the idea 64 8 
(23/27/14) (1/4/3)
Person X makes decisions about this and Person 37 30 
Y makes decisions about other things (13/12/12) (10/6/14)
Couple makes all decisions about this activity 251 
together (95/73/83)
Person X is the most knowledgeable about this 190 32 
activity (70/68/52) (14/7/11)
Other 1 0 0 
(0/1/0) (0/0/0) (0/0/0)
Total observations 734 81 625 
(243/303/188) (28/19/34) (224/173/228)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 31
Annex table 12. Descriptive analysis of decisions on how much cassava to sell, 
according to who makes the decision and type of couple resembled
Among the couples that you resemble, Who in your household generally makes the final decision 
which is the MOST similar to your couple? about how much cassava to sell?
Husband only Wife only Jointly
Vignette A. Husband shares idea, discusses with 403 
wife, and husband makes the final decision (137/158/108)
Vignette B. Wife shares idea, discusses with 50 
husband, and the husband makes the final (15/21/14)
decision
Vignette C. Husband shares idea, discusses with 476 
wife, and they make a joint final decision (180/148/148)
Vignette D. Wife shares idea, discusses with 254 
husband, and they make a joint final decision (75/86/93)
Vignette E. Wife shares idea, discusses with 177 
husband, and wife makes the final decision (63/46/68)
Vignette F. Husband shares idea, discusses with 23 
wife, and wife makes the final decision (8/6/9)
Vignette G. Husband shares idea with his wife 57 
before he makes the final decision (17/30/10)
Total observations 510 200 730 
(169/209/132) (71/52/77) (255/234/241)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
Annex table 13. Descriptive analysis of decisions on how much cassava to sell, 
according to who makes the decision and why
Most important reason why the final Who in your household generally makes the final decision 
decisions about how much to sell are about how much cassava to sell?
made by a sole member or jointly. Husband only Wife only Jointly
Couple agreed/supported the decision 329 
(113/111/105)
Person X contributes the most resources, 30 11 
income, or labor (10/11/9) (5/3/3)
Couple contributes resources, income, or labor 104 
(40/37/27)
Person X is head of household 375 5 
(117/166/92) (2/1/2)
Person X introduced the idea 21 7 
(7/9/5) (3/3/1)
Person X makes decisions about this and Person 33 55 
Y makes decisions about other things (12/9/12) (20/12/23)
Couple makes all decisions about this activity 296 
together (101/86/109)
Person X is the most knowledgeable about this 50 122 
activity (23/13/14) (41/33/48)
Other 1 0 1 
(0/1/0) (0/0/0) (0/1/0)
Total observations 510 200 730 
(169/209/132) (71/52/77) (255/234/241)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
32 CGIAR GENDER Impact Platform · Working Paper #016
Annex table 14. Descriptive analysis of decisions on building/refurbishing a house, 
according to who makes the decision and type of couple resembled
Among the couples that you resemble, Who in your household generally makes the final decision 
which is the MOST similar to your couple? about building/refurbishing a house?
Husband only Wife only Jointly
Vignette A. Husband shares idea, discusses with 957 
wife, and husband makes the final decision (318/338/301)
Vignette B. Wife shares idea, discusses with 89 
husband, and the husband makes the final (28/27/34)
decision
Vignette C. Husband shares idea, discusses with 537 
wife, and they make a joint final decision (175/146/216)
Vignette D. Wife shares idea, discusses with 75 
husband, and they make a joint final decision (19/29/27)
Vignette E. Wife shares idea, discusses with 28 
husband, and wife makes the final decision (11/12/5)
Vignette F. Husband shares idea, discusses with 6 
wife, and wife makes the final decision (3/2/1)
Vignette G. Husband shares idea with his wife 137 
before he makes the final decision (48/48/41)
Total observations 1,183 34 612 
(394/413/376) (14/14/6) (194/175/243)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
Annex table 15. Descriptive analysis of decisions on building/refurbishing a house, 
according to who makes the decision and why
Most important reason why the final Who in your household generally makes the final decision 
decisions about building/refurbishing about building/refurbishing a house?
a house are made by a sole member or Husband only Wife only Jointly
jointly.
Couple agreed/supported the decision 283 
(91/85/107)
Person X contributes the most resources, 258 3 
income, or labor (91/90/77) (2/0/1)
Couple contributes resources, income, or labor 150 
(50/41/59)
Person X is head of household 709 1 
(220/267/222) (0/1/0)
Person X introduced the idea 36 24 
(12/8/16) (10/12/2)
Person X makes decisions about this and Person 31 5 
Y makes decisions about other things (13/8/10) (2/1/2)
Couple makes all decisions about this activity 179 
together (53/49/77)
Person X is the most knowledgeable about this 149 1 
activity (58/40/51) (0/0/1)
Other 0 0 0 
(0/0/0) (0/0/0) (0/0/0)
Total observations 1,183 34 612 
(394/413/376) (14/14/6) (194/175/243)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 33
Annex table 16. Descriptive analysis of decisions on paying school fees, according to 
who makes the decision and type of couple resembled
Among the couples that you resemble, Who in your household generally makes the final decision 
which is the MOST similar to your couple? about paying school fees?
Husband only Wife only Jointly
Vignette A. Husband shares idea, discusses with 456 
wife, and husband makes the final decision (164/155/137)
Vignette B. Wife shares idea, discusses with 183 
husband, and the husband makes the final (60/61/62)
decision
Vignette C. Husband shares idea, discusses with 448 
wife, and they make a joint final decision (158/143/147)
Vignette D. Wife shares idea, discusses with 375 
husband, and they make a joint final decision (113/129/133)
Vignette E. Wife shares idea, discusses with 66 
husband, and wife makes the final decision (16/19/31)
Vignette F. Husband shares idea, discusses with 13 
wife, and wife makes the final decision (5/5/3)
Vignette G. Husband shares idea with his wife 71 
before he makes the final decision (21/26/24)
Total observations 710 79 823 
(245/242/223) (21/24/34) (271/272/280)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
Annex table 17. Descriptive analysis of decisions on paying school fees, according to 
who makes the decision and why
Most important reason why the final Who in your household generally makes the final decision 
decisions about paying school fees are about paying school fees?
made by a sole member or jointly. Husband only Wife only Jointly
Couple agreed/supported the decision 312 
(96/108/108)
Person X contributes the most resources, 221 14 
income, or labor (71/74/76) (4/4/6)
Couple contributes resources, income, or labor 216 
(75/68/73)
Person X is head of household 446 2 
(155/155/136) (1/0/1)
Person X introduced the idea 10 7 
(2/6/2) (0/0/7)
Person X makes decisions about this and Person 16 33 
Y makes decisions about other things (9/2/5) (11/11/11)
Couple makes all decisions about this activity 295 
together (96/100/99)
Person X is the most knowledgeable about this 17 23 
activity (5/8/4) (9/5/9)
Other 0 0 0 
(0/0/0) (0/0/0) (0/0/0)
Total observations 710 79 823 
(245/242/223) (21/24/34) (271/272/280)
Note: Respondent identity tally for each response is given as (# husband/# wife/# joint).
34 CGIAR GENDER Impact Platform · Working Paper #016
Annex table 18. Relationship between cassava productivity and decision-making on 
production-related topics
Dependent variable: cassava productivity (kg/acre) in main plot
Which variety to plant in your What area of land should be When to harvest cassava from 
main cassava plot? devoted to the main cassava your main plot?
production?
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 
(Final say) (Final say (Final say) (Final say (Final say) (Final say 
+ origin + + origin + + origin + 
discussion) discussion) discussion)
Final say: wife −241.372*** −183.312* −150.171* −38.732 −382.319*** −244.175**
(80.225) (102.006) (82.779) (84.725) (99.207) (116.856)
Final say: joint 38.612 54.779 −41.175 5.314 −96.408 −32.428
(55.227) (58.562) (56.683) (58.936) (73.660) (86.719)
Origin: wife −73.676 −118.025 −155.267**
(79.188) (79.284) (76.361)
Discussion −82.231 −333.064 −263.848
occurred
(171.045) (208.941) (262.797)
Adjusted 0.132 0.132 0.131 0.135 0.140 0.144
R-squared
Joint sig. test: 6.174 3.953 1.784 2.725 8.353 5.302
F stat.
Joint sig. test: 0.004 0.007 0.178 0.039 0.001 0.001
P value
N 2,017 2,017 2,022 2,022 2,019 2,019
Notes: Standard errors in parentheses. Significance denoted as: (* ~ p<0.10, ** ~ p<0.05, *** ~ p<0.01). Final say responses refer to answers 
to the question “Who in your household generally makes the decisions about X?” Origin and discussion variables are drawn from elements of 
the vignette that respondents selected as being most reflective of their household decision-making process—specifically, whether or not the 
couple discussed it together, and who had the final say. Base level: final say: husband, no discussion, decider: husband. Control variables: age of 
husband/wife, household size, youth dependency ratio, husband’s/wife’s ownership of main plot, household wealth quintile, improved variety 
on main plot, agrochemical inputs on main plot, marital status, educational status of husband/wife, husband’s/wife’s occupation, extension 
services, access to credit, group membership, husband’s/wife’s participation in activities (1–3). Hidden control variables are not included in 
tests for joint significance.
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 35
Annex table 19. Relationship between processed cassava sales productivity and 
decision-making on processing- and sales-related topics
Dependent variable: processed cassava sales productivity from main plot (TSZ1,000/acre)
How to process the cassava Where to sell cassava from How much cassava to sell from 
from your main plot? your main plot? your main plot?
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 
(Final say) (Final say (Final say) (Final say (Final say) (Final say 
+ origin + + origin + + origin + 
discussion) discussion) discussion)
Final say: wife −183.856*** −138.391*** −50.403 18.056 91.469 102.625
(50.146) (40.758) (78.307) (74.104) (127.125) (122.801)
Final say: joint −157.898*** −131.508*** −116.401** −97.189* −32.370 −28.111
(51.914) (48.918) (56.126) (55.027) (19.857) (22.748)
Origin: wife −41.313*** −76.429** −14.126
(14.221) (31.686) (19.899)
Discussion −158.884 −85.466 −7.308
occurred
(221.483) (100.578) (54.158)
Adjusted 0.078 0.080 0.071 0.074 0.065 0.064
R-squared
Joint sig. test: 6.723 4.500 2.444 3.019 1.403 0.951
F stat.
Joint sig. test: 0.003 0.003 0.097 0.026 0.255 0.442
P value
N 1,383 1,383 1,382 1,382 1,382 1,382
Notes: Standard errors in parentheses. Significance denoted as: (* ~ p<0.10, ** ~ p<0.05, *** ~ p<0.01). Final say responses refer to answers 
to the question “Who in your household generally makes the decisions about X?” Origin and discussion variables are drawn from elements of 
the vignette that respondents selected as being most reflective of their household decision-making process—specifically, whether or not the 
couple discussed it together, and who had the final say. Base level: final say: husband, no discussion, decider: husband. Control variables: age of 
husband/wife, household size, youth dependency ratio, husband’s/wife’s ownership of main plot, household wealth quintile, improved variety 
on main plot, agrochemical inputs on main plot, marital status, educational status of husband/wife, husband’s/wife’s occupation, extension 
services, access to credit, group membership, husband’s/wife’s participation in activities (1–3). Hidden control variables are not included in 
tests for joint significance.
36 CGIAR GENDER Impact Platform · Working Paper #016
How do couples in rural Tanzania make decisions? A novel approach for understanding intrahousehold decision-making 37
Annex table 20. Relationship between minimum dietary diversity, food security and decision-making on production-, processing- 
and sales-related topics
Panel A. Dependent variable: minimum diet diversity for women
Which variety to plant What area of land When to harvest How to process the Where to sell cassava How much cassava to 
in your main cassava should be devoted cassava from your cassava from your from your main plot? sell from your main 
plot? to the main cassava main plot? main plot? plot?
production?
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11 Model 12
Final say: wife 0.018 −0.072 −0.099 −0.139 0.346** 0.402** 0.258* 0.126 0.262 0.164 −0.071 −0.065
(0.152) (0.168) (0.156) (0.175) (0.148) (0.170) (0.131) (0.132) (0.283) (0.288) (0.164) (0.185)
Final say: joint 0.018 −0.018 0.169** 0.169** 0.142** 0.160** 0.059 0.005 0.035 −0.000 0.020 0.022
(0.069) (0.072) (0.075) (0.075) (0.073) (0.074) (0.079) (0.082) (0.080) (0.079) (0.083) (0.097)
Origin: wife 0.089 0.065 −0.078 0.173* 0.117 −0.005
(0.111) (0.108) (0.106) (0.091) (0.085) (0.097)
Discussion occurred 0.265 −0.128 −0.034 0.131 0.157 −0.013
(0.210) (0.182) (0.175) (0.164) (0.138) (0.131)
Adjusted  0.113 0.114 0.117 0.116 0.121 0.120 0.119 0.121 0.115 0.115 0.114 0.112
R-squared
Joint sig. test:  0.036 0.602 2.830 1.435 3.481 1.910 1.955 1.697 0.482 1.086 0.251 0.128
F stat.
Joint sig. test:  0.965 0.663 0.068 0.235 0.038 0.122 0.152 0.164 0.620 0.373 0.779 0.972
P value
N 1,021 1,021 1,023 1,023 1,021 1,021 1,021 1,021 1,021 1,021 1,021 1,021
38 CGIAR GENDER Impact Platform · Working Paper #016
Panel B. Dependent variable: Food Insecurity Experience Scale
Which variety to plant What area of land When to harvest How to process the Where to sell cassava How much cassava to 
in your main cassava should be devoted cassava from your cassava from your from your main plot? sell from your main 
plot? to the main cassava main plot? main plot? plot?
production?
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11 Model 12
Final say: wife 0.000 0.146 0.243 0.052 0.118 0.177 0.009 −0.006 0.594* 0.369 0.369* 0.208
(0.251) (0.291) (0.292) (0.342) (0.191) (0.223) (0.217) (0.199) (0.307) (0.361) (0.205) (0.220)
Final say: joint −0.016 0.068 0.337** 0.334** 0.010 0.028 0.194 0.208 0.348** 0.336** 0.031 −0.016
(0.148) (0.147) (0.139) (0.137) (0.141) (0.147) (0.159) (0.166) (0.142) (0.140) (0.170) (0.163)
Origin: wife −0.089 0.310* −0.085 0.068 0.378** 0.238
(0.201) (0.180) (0.150) (0.201) (0.183) (0.148)
Discussion occurred −0.778** −0.581 −0.022 −0.347 −0.353 −0.125
(0.375) (0.403) (0.320) (0.495) (0.256) (0.279)
Adjusted  0.169 0.172 0.174 0.177 0.169 0.168 0.171 0.170 0.176 0.179 0.172 0.172
R-squared
Joint sig. test:  0.006 1.264 2.925 3.103 0.199 0.168 0.796 0.571 4.493 4.048 1.863 1.648
F stat
Joint sig. test:  0.994 0.296 0.062 0.023 0.820 0.954 0.456 0.685 0.016 0.006 0.165 0.176
P value
N 1,020 1,020 1,022 1,022 1,020 1,020 1,020 1,020 1,020 1,020 1,020 1,020
Notes: Standard errors in parentheses. Significance denoted as: (* ~ p<0.10, ** ~ p<0.05, *** ~ p<0.01). Final say responses refer to answers to the question “Who in your household generally makes the decisions 
about X?” Origin and discussion variables are drawn from elements of the vignette that respondents selected as being most reflective of their household decision-making process; specifically, whether or not the 
couple discussed it together, and who had the final say. Base level: final say: husband, no discussion, decider: husband. Control variables: age of husband/wife, household size, youth dependency ratio, husband’s/
wife’s ownership of main plot, household wealth quintile, improved variety on main plot, agrochemical inputs on main plot, marital status, educational status of husband/wife, husband’s/wife’s occupation, extension 
services, access to credit, group membership, husband’s/wife’s participation in activities (1–7). Hidden control variables are not included in tests for joint significance.
GENDER Impact
Platform
Generating Evidence and New Directions for Equitable 
Results (GENDER) is CGIAR’s impact platform designed 
to put equality and inclusion at the forefront of global 
agricultural research for development. The Platform 
is transforming the way gender research is done, both 
within and beyond CGIAR, to kick-start a process of 
genuine change toward greater gender equality and 
better lives for smallholder farmers everywhere.
gender.cgiar.org
CGIAR is a global research partnership for a food-secure 
future dedicated to reducing poverty, enhancing food 
and nutrition security, and improving natural resources. 
cgiar.org