FAQs

Choosing which WEAI to use depends on the needs of your project. Click here for an interactive tool that will guide you through a series of questions to help you identify the best version for your purpose. Feel free to contact us to discuss.

The pro-WEAI+MI is not yet finalized. The Health and Nutrition add-on module is currently being finalized. We share the draft modules over email upon request – feel free to contact us.

Although we don’t officially provide the WEAI survey modules translated in languages other than English, many of our WEAI users and project partners have translated the WEAI modules into different languages and shared them with our team. Feel free to contact us to check whether we have a language you are looking for. If you plan to use it, we suggest you review the questions for accuracy of translation, especially specific to your project’s context, and compare with the English version used in the same project, as some of the questions may be outdated. If you have suggestions to update and enhance translations of the WEAI surveys, please let our team know.

You can use the pro-WEAI livestock module, or you can use the WELI. The WELI incorporates pro-WEAI but has a more extended set of questions on livestock and decision-making around livestock. For a nutrition project, use the pro-WEAI H&N module. If you are working in India, you may also want to consider the WENI.

The pro-WEAI for market inclusion (pro-WEAI+MI) would be your best option.

Field costs for the WEAI pilots (including enumerator training, translation, and data entry) were $38,000 in Bangladesh (450 households), $56,000 in Guatemala (350 households), and $36,000 in Uganda (350 households). Costs differed across the three pilot countries owing to basic field costs, costs of transportation, as well as translation. Note however that these field costs may not provide an accurate picture; the pilot questionnaires were much longer than the final WEAI module, as various questions were still being tested at that time. The cost information on the pilot surveys is likely to be more helpful for standalone surveys rather than larger multi-purpose household surveys. In the FTF Population-Based Surveys, the WEAI has been collected along with several other modules, making it difficult to isolate the costs for the WEAI alone. However, to give some general parameters, the FTF survey in Rwanda (2000 households) cost $160,000 and collected the WEAI along with two dietary diversity modules and the Household Hunger Scale. The WEAI would likely have accounted for half of the enumeration time in that survey. All other indicators were calculated for FTF using secondary data from the DHS and Rwandan national household expenditure survey. In Tajikistan (2000 households), data collection cost $425,000, but the survey collected many more modules for consumption-expenditure, dietary diversity, and anthropometric measurement, as well as other nutrition/food security information.

We recommend a minimum of 400 households based on our WEAI piloting experience and the OPHI recommendations. However, we highly recommend that you do power calculations for your project and only use the 400 households as an alternative. We have some information on power calculations for WEAI using Stata and can share this information over email upon request.

Use of phone surveys has increased during the pandemic, but pro-WEAI is not validated for phone surveys. Phone surveys are not a good substitute for face-to-face interviews, especially on women’s empowerment, because:

• Women are less likely to have cell phones, especially poorer (and less empowered) women, potentially biasing your sample (check women’s phone ownership rates).
• Phone surveys are much harder to establish rapport, which is important for many pro-WEAI questions.
• It is impossible to verify that the respondent is alone and cannot be overheard. Many women take calls on speakerphone setting, which further complicates privacy and biases responses.
• Phone surveys need to be kept shorter (ideally 15-20 minutes), which is too short for pro-WEAI.

No, we do not recommend systematically excluding non-agricultural households for a number of reasons.

For instance, the LSMS-ISA uses the following screening question: “In the last 12 months, did a member of this household cultivate any land?” When used to screen households for the WEAI module, this means that the survey will capture the WEAI indicators and agriculture activities of only those households that have been “cultivating” land. This is potentially misleading because the survey will not capture livestock activities, small kitchen gardens, access to forest land (gathering), etc. Rural livelihoods are often linked to the agricultural sector in both direct and indirect ways, which is why it is very difficult in practice to come up with a standardized definition of what an agricultural household is, and especially one that would be applicable across countries. For example, in Bangladesh, women typically do postharvest activities and processing but do not consider themselves as working in agriculture, even though they are clearly tied to the agriculture sector. Landless households who do farm wage work are not considered agricultural cultivators and yet their livelihood is directly tied to agriculture. There are potentially many other ways that livelihoods are tied to agriculture and these could vary in different contexts. For this reason, the WEAI Bangladesh survey did not screen for agricultural households.

Another important implication of screening is that the survey will not be able to capture movements in and out of agriculture. If FTF is providing agricultural and other support (e.g. credit) services, then these movements are among the key issues that the survey is trying to capture with the WEAI (and presumably other key indicators) – so this would be a significant loss. Lastly, surveys that screen for agricultural households will not be comparable to surveys in other countries that do not screen. This will limit the potential for analyzing the WEAI across countries.

If you DO screen for agricultural households, we suggest you make sure to establish individuals’ “involvement” in agriculture very carefully, and irrespective of whether they own and/or operate land themselves and irrespective of the scale of farming that they might practice. Landless individuals who work in agriculture post-harvest activities, wage workers, and those working in homestead gardens/subsistence farming are all legitimate participants in agriculture and therefore should be captured in the WEAI surveys.

With regard to the Feed the Future (FTF) initiative, yes, in most cases it would not be necessary to include urban areas since the WEAI was designed to monitor FTF agriculture programs. Even if there is some FTF programming that creeps into urban areas, it is usually a different type of programming such as health or nutrition interventions, rather than programs that are likely to “move” the WEAI indicators. In many countries, small urban centers may just be living/trading areas for people who are still doing agricultural work, but the distinction should be made at some higher FTF strategy level for what is/is not included in the sampling and how these classifications are made. This decision should be made taking a lot more into account than impacts for the WEAI. The most recent iteration of the WEAI, called the WEAI for Value Chains or WEAI4VC, measures women's empowerment across the agricultural value chain. Unlike the original WEAI which focuses heavily on agricultural production, the WEAI4VC aims to also capture activities in later nodes of the value chain such as post-harvest processing, trading, and marketing in addition to production. The WEAI4VC can be relevant to an urban setting as long as the survey participants are agricultural value chain actors (producers, entrepreneurs, or wage workers) and working in production, processing, and other post-harvest activities across the value chain.

If you only interview women, you can measure their empowerment using 5DE (WEAI, A-WEAI) or 3DE (pro-WEAI). However, you will not be able to compute men’s 5DE, 3DE, nor the GPI. You won’t know about gender equality. This might be important if you’re interested in potential backlash of programs targeted to women, or if you have a gender-transformative project. To know whether gender norms are changing, you need to look at women AND men. You can’t judge whether a project is gender-transformative by looking at women alone. You also won’t know whether a gap exists because of gender, or whether there are other reasons the gap exists. For example: is it that only women lack access to credit, or do men also lack credit? Knowing this is very important for programming purposes.

Unfortunately, no. To calculate the WEAI, which is a score, you would need data from men and women in the same household (for the Gender Parity Index (GPI)). We recommend that you collect data on both men and women in the household so that you can assess gender disparities. However, if you collect data only on women, you can calculate the Five Domains of Empowerment (5DE) sub-index for male- and female-headed households.

We have reduced the number of indicators from 12 to 10, and number of questions required per indicator as well. Check the questionnaire carefully for the optional questions. Next, check whether some of your questionnaire covers things that are similar to pro-WEAI. There may be opportunities to reconcile them (as we have done with IFAD I-WEAI). For example, if you are asking land ownership questions, you can tie that to the household roster to identify who owns land. Or, you could reduce the number of questions (items) that make up an indicator. This is the approach taken in the IFAD R-WEAI (reduced WEAI), which keeps the same number of indicators, but fewer questions for each indicator. Garbero and Perge (2017) reduced the number of questions using multiple correspondence analysis. If you need to cut further, consider which indicators are least related to your theory of change. If you drop an indicator, you will have to figure out adequacy and pro-WEAI scores based on the reduced number of indicators. Keep in mind that all of these changes require significant adaptations and are no longer comparable to pro-WEAI.

You can certainly measure particular domains (or even indicators) by themselves, but please note that doing so does not result in the WEAI. The WEAI and A-WEAI are obtained by taking a weighted average of two sub-indices, the 5DE and GPI, and both are obtained by taking the weighted average of the 10 indicators representing the 5 domains. These 10 indicators can each be interpreted on their own, so if you do not have time to administer the entire module but wish to collect some gender-relevant indicators, you can try to see which domains/indicators are most relevant to you. There is a discussion paper available that describes some of the validity testing that was done for the indicators.

Based on IFPRI’s experience implementing the pilots through paper-based surveys, the original WEAI is estimated to take 30-40 minutes, and A-WEAI took 20% less time than the original WEAI. If the surveys are done concurrently with men and women, then the additional time per dual-adult household is also 30-40 minutes. Based on recent estimates from computer-assisted interviews (which tend to take longer than equivalent paper and pencil surveys), A-WEAI requires 40-60 minutes. The main pro-WEAI module takes approximately 45-75 minutes to administer; the add-on health/nutrition add-on module takes approximately 20 minutes. All these estimates were reported by a team of extremely well-trained enumerators who have been using various versions of the WEAI since 2011.

One option is to administer the WEAI at a lag. For example, the WEAI team could follow the Baseline survey team and go into a cluster that has been completed. Since the WEAI survey will be collected at a different time, this will help minimize interview fatigue. One advantage of doing this might be that the interviewer would have already built rapport with the household, and a follow-up interview (particularly with sensitive questions about decision making) would not be viewed as a great imposition. Another recommendation is to split enumeration of other non-WEAI modules between members of the household (primary male and female decision-makers) based on who is best suited to know about the subject matter and administer them concurrently. For example, modules on dietary diversity are typically administered to a woman respondent while the household roster, dwelling characteristics, and expenditure modules might be administered to a man respondent.

Try not to put the pro-WEAI questions at the end of the survey or following especially tedious modules (e.g., plot-level agricultural production). This should help to reduce respondent fatigue on pro-WEAI. Check whether the respondent can answer questions without someone listening. If possible, have respondents interviewed by interviewers of the same sex. Try randomizing the order of questions in the item set.

The question on whether the day was typical was originally asked in the pilot survey but was later excluded in the FTF PBS to reduce its length since this question is not used to construct the Index. However, the WEAI team highly recommends including this question if time and budget permits (recommended questions are highlighted in blue in the Uganda pilot questionnaire). It would be optional extra information, but as the example in this question points out, it can be very useful for interpreting the data. With such information you can recompute the 5DE/GPI/WEAI for the sample with and without the atypical cases, so you can see if this makes a difference. It is recommended that enumeration schedules be planned to not collect data the day after a cultural religious day or Sabbath to minimize the effects on this indicator. Please contact us if you have any more questions regarding this indicator.

The key issue with the administration of the time use module is that enumerators did not ask respondents to recount activities or assign 15 min intervals to them. Rather, respondents were asked to narrate their days and they themselves allocated time periods. It is very true that many respondents do not have time in minutes and hours “in their heads” as we do where our days are structured around a 24-hour period. In this way, time spans allocated to activities will be more of an approximation, especially because there is rounding, than a strict 15-minute interval. In these calculations, it is imperative that the enumerators understand the local culture and context where the respondents live – i.e., knowing at what time the sun rises, at what time it sets, how long it takes to travel to the nearest water point or market, what the prayer times are in Islamic societies, etc. In the pilots, the 15-minute intervals were actually more useful in portioning out secondary activities. For example, if someone is eating for 30 minutes and they say they were watching TV for part of the time, “watching TV” could be 15 minutes. However, if the intervals are longer, say 30-minute segments, we would simply lose the watching TV activity. That said, several WEAI collection efforts (the Nepal Suaahara IFPRI survey and the Cambodia FTF PBS) have changed the time intervals to 30 minutes. That change will essentially reduce the diversity in activities, and in the case of the FTF surveys that adopt this modification, their WEAI will no longer be strictly comparable to other countries. The secondary activities are collected only for the pro-WEAI and are not required for the original WEAI and A-WEAI.

Many of the FTF PBS surveys will be using CAPI, so these surveys will not be able to use a time grid to "draw" the responses as in the pilots. As far as capturing the time information, it should be the same so long as enumerators follow the same procedure of asking respondents to narrate their activities throughout the 24-hour period. Respondents themselves assign the time periods, and the enumerators log the information at 15-minute intervals. Whether paper-based or computerized, the most important issue is to verify that the data entry instrument is set up in a way that can differentiate and capture secondary versus primary activities—especially when one activity is marked as a primary activity in some time periods, and a secondary activity in other time periods. This issue came up in the Uganda and Guatemala pilots, and resulted in re-examining individual questionnaires. In CAPI, entering information in smaller chunks of time may take longer. Time grids are usually easier for enumerators to "map" activities and see them visually, which may also lead to less error in marking end/beginning points. However, CAPI software can also be pre-programmed to flag any issues.

Although we don’t officially provide the WEAI CAPI modules, many WEAI projects have shared their forms with us. Please contact us with information on the CAPI platform and version of the WEAI and we can check if we have what you are looking for. We suggest closely reviewing the questions and modules as they may be outdated.

For the original WEAI and A-WEAI, the domains were assigned equal weights because each domain is equally important for women’s empowerment, and there is not enough evidence to support that one is more important than another across contexts. Although the weights can be modified, we highly recommend leaving them as is to ensure comparability to the WEAI – changing the weights will make them incomparable to other studies. Weights can be modified in the sensitivity analysis. If you decide to modify the weights or other parts of the WEAI, please make these changes clear in any reporting/analysis and call your measure a modified WEAI in any publications.

The 80% cutoff for pro-WEAI was initially determined based on sensitivity analysis conducted that compared all possible cutoffs using all WEAI data available at the time. Similar analysis was done prior to the launch of pro-WEAI, which confirmed that this cutoff remained reasonable.

What is the theory of change of your project? Are you affecting some domains of empowerment more than others, for example, instrumental agency? Look at the indicator-by-indicator results. That will tell you which domains/indicators are most affected by your project. Use qualitative work to help you unpack the results.

Did you collect a household roster with the age and education of others in the household? Did you collect data on livestock, and who within the household owns livestock? It is very important to collect not only pro-WEAI, but also basic information about the household and its members, and the outcomes that your project wants to affect (see theory of change). Without collecting this information, you won’t know the relationship between empowerment and other outcomes.

Empowerment diagnostics from projects are likely to be different from numbers from nationally representative samples. If your project requires having livestock or a plot of land to participate, it’s likely that women participants will be empowered with respect to assets. Similarly, in group-based projects, participants will be empowered with respect to group membership. Look at project eligibility criteria and see if some will automatically classify women as empowered.

• Solution 1: Adjust the empowerment threshold. Keep in mind this won’t be comparable to other countries in the portfolio.
• Solution 2: Do qualitative work and triangulate! Try to understand what are the aspects of disempowerment that are important.
• Solution 3: Consider intersectionality. Women may be faring well as a whole, but are there areas of disadvantage? Caste, ethnicity, race?

We strongly recommend completing the WEAI on the same households sampled for the rest of the survey. If the WEAI is administered to households that are different from the rest of the Feed the Future PBS or another similar household survey, it will still be possible to compute the overall Index, but it will not be possible to link the Index with any other individual or household level outcomes collected in the other Feed the Future modules, such as nutrition or poverty. This very much limits the usefulness of the WEAI.