The effects of known source and account labelling on the acceptability of saving by wives in Zambia

This is a survey experiment. The participants will be presented with vignettes in which a wife or husband is deciding to save and will be asked to evaluate the wife's or husband's decision. The following aspects of these vignettes will be exogenously varied: phrases about the origin of the money being saved and the nature of the account it is being saved into; whether the saver is a wife or husband; and whether the saver does or does not tell her/his spouse about the savings.

The participants will also be asked how justifiable it is for a husband to beat his wife having discovered that she has been saving in secret and, in these questions too, phrases about the nature of the account that is being saved into will be exogenously varied.

2023-02-15

2023-02-19

When the focus is wives' and husbands' savings decisions, the outcome of interest is how socially acceptable participants think the decision is.
When the focus is a husband beating his wife, the outcome of interest is whether participants think it is justifiable or not.

Social acceptability will be measured on a 4-point scale: very unacceptable, somewhat unacceptable, somewhat acceptable, very acceptable. In some analyses, this will be treated as a categorical variable. In some analyses it will be treated as a continuous variable ranging from -1 to +1 (very unacceptable -> -1, somewhat unacceptable -> -0.33, somewhat acceptable -> +0.33, very acceptable -> +1). In some analyses, within participant differences in acceptability scores (continuous) across vignette versions will be calculated and analysed as outcome variables of interest, e.g., Acceptability of keeping savings secret from spouse = Acceptability of saving and keeping it secret from spouse - Acceptability of saving and telling spouse.

The beating justifiable outcome variable will equal 1 if a responded indicates that the husband beating his wife would be justified and zero otherwise. (This is consistent with the DHS survey approach.)

None

The participants will be members of savings groups in villages in Luapula Province, Zambia. The aim is to engage 180 participants, 90 male, 90 female, in the survey.

The survey will be conducted via one-to-one interviews by 7 trained field researchers, 3 female, 4 male.

The survey will be conducted in mid-February 2023.

Within each of the three sets of variants, Set A, Set B and Set C, described above, there will be four variants of the vignette. In two of these the saver will decide to tell her/his spouse about the savings and in two she/he will not. Further, in two out of the four variants the saver is a husband and in the other two the saver is a wife. The tell vs don't tell variation and the wife vs husband variation will cross-cut.

Each participant will be randomly assigned to be presented with either Set A or Set B or Set C of the vignette variants. Whether a participant evaluates the decisions to save and tell spouse first and the decisions to save and not tell spouse second or the other way around will also be randomised across participants. The randomizations of Set A vs Set B vs Set C and tell-don't-tell vs don't-tell-tell orderings will cross cut.

Wives will always evaluate decisions by a wife first followed by decisions by a husband. Husbands will always evaluate decisions by a husband first followed by decisions by a wife.

Respondents who have been presented with either Set A or Set B of the wife/husband saving vignettes (described above) will later be asked whether it would be justifiable for a husband to beat his wife if he discovered that she had been saving in secret in a personal savings account and this question will come at the end of the standard series of DHS questions that take a similar form (Do you think a man would be justified in beating his wife if she neglects their children / burns the food / argues with him / visits family or friends without his permission / refuses to have sex with him?) Respondents who have been presented with Set C of the wife/husband saving vignettes (described above) will later be asked whether it would be justifiable for a husband to beat his wife if he discovered that she had been saving in secret in "a savings account for her business" and this question will come at the end of the standard set of DHS questions as described in the previous paragraph.

In total, there will be 12 variants of the questionnaire, 6 for wife/female participants, 6 for husband/male participants. Each variant of the questionnaire will include four savings decision vignettes.

We expect to be able to engage with 180 participants, 90 male, 90 female, by visiting 9 savings groups. However, more savings groups will be visited if required to meet the target sample size.

The male participants will be interviewed by male field researchers. The female participants will be interviewed by female field researchers.
Variants of the survey will be balanced across savings groups and, within the sexes, across field researchers.

Two lists have been drawn up, one for male participants, one for female participants. In each list, survey instrument variants and field researchers have been assigned to participant IDs in a way that assures balance across all relevant dimensions, including the location of a participant in the interview order. In the field, participant IDs will be assigned on a first-come-first-served basis.

The unit of randomization for the Set A vs SET B vs Set C variants is the participant. Each participant will make four evaluations and, across these four, tell vs don't-tell and saver-a-wife vs saver-a-husband are randomized. In the analysis, an observation will be an evaluation or a within-participant difference between two evaluations and these will be clustered by participant.

Yes

180 participants, 90 male, 90 female.

Each participant will evaluate four savings decisions. So, we will have 720 observations (=4*180) for the acceptability-of-savings-decisions analysis.

For the evaluation-of-savings-decision vignettes, the between participant (cluster) treatment arms are Set A vs Set B vs Set C:
- 240 observations from 60 participants (clusters) will relate to the Set A savings decisions (source of savings ambiguous, personal savings account)
- 240 observations from 60 participants (clusters) will relate to the Set B savings decisions (source of savings business, personal savings account)
- 240 observations from 60 participants (clusters) will relate to the Set C savings decisions (source of savings business, business savings account labelled appropriately)
For the evaluation of savings decision vignettes, the within participant treatment arms of interest are saver-tells-spouse vs saver-does-not-tell-spouse:
- 360 observations from 180 participants (clusters) will relate to vignettes in which the saver tells their spouse
- 360 observations from 180 participants (clusters) will relate to vignettes in which the saver does not tell their spouse

For the justification-of-husband-beating-wife questions:
- 120 observations from 120 participants will relate to the husband discovering that the wife had been saving into a personal savings account without telling him (of these, in 60 cases the participant will have earlier been told about a scenario in which the wife's personal savings had come from her business, while in the other 60 cases they will have been told nothing about sources of wives' savings)
- 60 observations from 60 participants will relate to the husband discovering that the wife had been saving into a savings account for her business without telling him.