Category Archives: Analysis

21 Mar

7 Tips to Matching DHS Final Report Tables

Can't match DHS Final Report tables?
Feeling frustrated because you can’t match DHS Final Report tables in your statistical software?

 

Our new four-part video series shows the Top 7 Tips & Tricks for Matching DHS Final Report Tables.

In this four-part video series, we will be covering the top 7 tips and tricks to matching The DHS Program Final Reports using a statistical software program.

The videos will guide you through the following questions:

  1. Are you using the correct data file?
  2. Are you using the correct denominator of cases?
  3. Are you using the correct variable(s)?
  4. Are you properly recoding?
  5. Are you applying the correct weights?
  6. Are you selecting the correct software specific code?
  7. Are you properly coding the tabulation commands in your statistical program?

Watch the four videos in the series below on Matching DHS Final Report tables to get all the details on the top 7 tips and tricks.


Additional help can be found on our website and the User Forum.
16 Nov

From National to Local: A New Way to Leverage DHS Data

In DHS survey final reports, data are presented on a national or first-level administrative sub-national level. However, this is usually not the level at which program planning and decision making are truly happening. To support more decentralized decision making at lower administrative levels, data need to be presented on a more disaggregated level.

The DHS Program is producing a standard set of spatially modeled map surfaces for each population-based survey for a select list of indicators that provide smaller area estimates of data. Geostatistics are used to predict (interpolate) the indicator value for unsampled areas based on data from sampled data locations. DHS creates standardized modeled map surfaces using DHS survey data along with global covariate datasets. Currently, sets of standard surfaces are available for 16 surveys. Spatial data packages and stand-alone maps are available for download through The DHS Program’s Spatial Data Repository.

How can modeled map surfaces be used?

These new spatially modeled surfaces can help in several ways to improve decision making for many development sectors that include health, population, nutrition, and water and sanitation programs on multiple levels. Users can combine the maps with other resources to support:

  1. Monitoring and evaluation: analysis and evaluation of past initiatives (impact analysis) or understanding existing situations
  2. Program planning: future planning of appropriate programs and policies

Data in the modeled surfaces can be used to evaluate past programs or to better understand existing situations. Such evaluations can help to understand deviations from the norm, attribute cause, or to contribute to impact evaluations, which analyze what would have happened to the population of an area if a program had not been implemented.

Program managers can also use modeled surfaces to plan, target, and develop interventions and programs that aim to improve situations in targeted geographic areas. Interventions can be targeted more precisely, saving money, time, and human resources in the search for the most effective outcomes.

The matrix below shows potential approaches for monitoring and evaluating past and planning future programs using modeled surfaces.

This matrix is by no means comprehensive, and it is expected that map users will come up with many more potential uses after analyzing their particular situation and maps for their country.

To read more, please see the Spatial Analysis Report 14, “Guidance for Use of The DHS Program Modeled Map Surfaces.” The report delivers more in-depth information on what modeled surfaces The DHS Program is creating, as well as an explanation of their creation process. In addition, the report provides guidance on limitations and assumptions.

The DHS Program is looking forward to seeing how groups will use this new data product to enhance their activities. There is enormous potential for innovative uses of these modeled surfaces beyond those discussed in the report. Users are encouraged to submit ideas and case studies to The DHS Program (spatialdata@dhsprogram.com) as only a large community of users who share their experiences will fully expose the maps’ potential.


Aileen Marshall is the Knowledge Management/Monitoring & Evaluation Specialist at The DHS Program. She is responsible for planning, development, implementation and evaluation of the KM strategy, KM activities as well as the project-wide SharePoint site. Additionally, she is involved in measuring and evaluating capacity strengthening activities at DHS and works closely with all teams to ensure knowledge at DHS is captured, stored and shared efficiently among staff. Aileen holds an MA in English Linguistics from the Westfaelische Wilhelms-University in Muenster, Germany, and an MLIS from the University of South Carolina.

Trinadh Dontamsetti is the Health Geographic Analyst for The DHS Program. He contributes to geospatial analysis, mapmaking, and geographic data processing activities. His research interests include geospatial interpolation, tuberculosis, and vector arthropod-borne diseases.

 

Clara R. Burgert is the GIS Coordinator for The DHS Program. She oversees all  geographic data, mapping, and geospatial analysis activities at The DHS Program.  Additionally, she facilitates workshops in partner countries on using maps for better decision making using open source GIS software.

Save

Save

Save

Save

Save

18 May

Building Awareness of the Link between Fistula and Gender-Based Violence

Genital fistula, an abnormal hole between the vagina and rectum or bladder that causes urinary or fecal incontinence, is a devastating, preventable condition that no woman should have to endure. It usually results from inadequately managed, prolonged or obstructed labor, surgical error, or trauma [1, 2].  Although rare, it can be completely debilitating—physically, socially, and economically—particularly to women who live in remote areas without access to treatment; women with fistula are often shunned from the household or society, which can cause immense suffering  [3].

While sexual violence can cause traumatic fistula, the vulnerable state of women with fistula gives reason to suspect that the risk of violence could also increase after the onset of fistula [4, 5, 6], though no studies have attempted to evaluate this to date.  Moreover, because it is so rare, it is difficult to capture statistically significant associations with the condition.

The DHS Program provides an opportunity to study such rare events because of the inclusion of standardized questions in numerous, nationally-representative  surveys with large sample sizes. In a study conducted to further examine the relationship between fistula and violence, data were pooled from 12 DHS surveys, 11 conducted in Sub-Saharan African countries and one in Haiti, where standardized modules (sets of questions) on the two topics were included.

In total, 90,276 women were included in the analysis.  Among these women, the prevalence of self-reported symptoms of fistula ranged from 0.4% to 2.0%.  Regression analyses confirmed an association with sexual violence: women who have experienced sexual violence, both ever as well as within the 12 months preceding the survey, have almost twice the odds of reporting symptoms of fistula. Although there are no questions posed on timing of onset of symptoms of fistula in the DHS, the association with lifetime as well as recent experience of sexual violence suggests that violence could occur both before as well as after fistula’s onset.

One other finding of interest was that women whose first experience of sexual violence was committed by a non-partner had over four times the odds of reporting symptoms of fistula than women who did not report sexual violence.  Although inferences from these findings can only be made with caution, the temporality relationship between fistula and sexual violence deserves further investigation.

In light of International Day to End Fistula on May 23, it is imperative to continue to work towards minimizing occurrence of fistula by building awareness around conditions that contribute to and result from this morbidity. This study shows yet another disheartening correlation between gender-based violence and poor health outcomes for women. It provides even more impetus for training and sensitivity for women’s health care providers in this area.

full studyA poster presentation of the study was exhibited at the 2016 Annual Meeting of the Population Association of America (PAA) in Washington DC. More information can be found in this poster.

 

 

 


  1. Longombe AO, Claude KM, Ruminjo J. Fistula and traumatic genital injury from sexual violence in a conflict setting in Eastern Congo: case studies. Reprod Health Matters. 2008 May;16(31):132-41
  2. Raassen TJ, Ngongo CJ, Mahendeka MM. Iatrogenic genitourinary fistula: an 18-year retrospective review of 805 injuries. Int Urogynecol J. 2014 Dec;25(12):1699-706.
  3. Baloch, B.A., A. Salam, D. ZaibUnnisa, and H. Nawaz. 2014. Vesico-Vaginal Fistulae. The Professional Medical Journal, 21(5), 851-855.
  4. ACQUIRE. 2006. Traumatic gynecologic fistula: A Consequence of Sexual Violence in Conflict Settings. A report of a meeting held in Addis Ababa, Ethiopia, September 6-8, 2005. New York, The ACQUIRE Project/EngenderHealth.
  5. Peterman A, Johnson K. Incontinence and trauma: sexual violence, female genital cutting and proxy measures of gynecological fistula. Soc Sci Med. 2009 Mar;68(5):971-9.
  6. Naved RT, Blum LS, Chowdhury S, Khan R, Bilkis S, Koblinsky M. Violence against women with chronic maternal disabilities in rural Bangladesh. J Health Popul Nutr. 2012 Jun;30(2):181-92.

 

13 Apr

How Many Demographers Does It Take to Make a Great Visualization?

How much time do you budget to create a data visualization?  The best visualizations, though they appear to be simple and clear, are often the result of dozens of attempts.

Demographers spend countless hours crunching data and preparing journal submissions, but not all take full advantage of data visualization, either in their exploratory analysis, or in communication of their findings.  Last month, data visualization enthusiasts met at the Population Reference Bureau for a hands-on workshop as part of the Population Association of America (PAA) Conference.  The 4 hour interactive workshop featured presentations from DC-based data viz expert Jon Schwabish, Dr. Tim Riffe, demographer at the Max Planck Institute for Demographic Research (MPIDR), Jonas Schoeley (MPIDR), and Dr. Audrey Dorelien of the Minnesota Population Center.  While each presenter had a unique focus, a common theme was clear:  your first draft visualization should never be your final visualization.  This lesson was put into practice as participants shared works-in-progress, received constructive feedback, and prepared “makeovers”.


Clara Burgert and I have been working on a visualization project for over a year.  The original was published last summer but we’ve been reworking it for a journal submission. Our colleagues at the data viz workshop provided constructive feedback, and we have made yet another round of changes. Some of the many stages of our chart “makeover” are presented below.

SAR12

Clara’s recently published analysis looks at 27 countries and 6 child health indicators. The goals of our visualization were to compare countries across these 6 indicators and to illustrate the inequity within countries, by highlighting the worst and best performing sub-national regions. While some countries have a very high measles vaccination prevalence, such as Tanzania, there are regions in Tanzania that are performing very poorly. Meanwhile, other countries have moderately good vaccination rates with very little variation among regions (like Rwanda). Our first real attempt at a publishable graphic looked like this:

indicators for journal

One of the challenges with this first graphic was that it didn’t use color very well. Clara needed to use color to distinguish between the 6 indicators in other places in the report, so we wanted to integrate that color scheme here for consistency. Simultaneously we realized that we could also simplify our use of color in this first draft: while we had originally plotted the red circle as the lowest region, the reader doesn’t need that color to know that that plot is the lowest- it’s obvious based on the axis and the left-to-right understanding of a numerical timeline. So we tried this:

indicators for journal with color coding

This color scheme worked better to unify the other graphics in the report, and we were feeling pretty good about it. But we still had a few concerns and questions:

  • Was it okay to have the axis for the stunting indicator and under-five mortality the same size as the others even though they aren’t at the same scale?
  • Was it okay that we were sorting lowest to highest, instead of ordering countries in a consistent way?
  • How should we handle ordering of the data when for 4 of our indicators, a high data value is “good”, like vaccination coverage, while for 2 of our indicators, a high data value is bad, like mortality?
  • Were there any formatting tweaks we could make to improve readability?

It was this version that was shared at the PAA data visualization workshop. Through the feedback of experts and colleagues, we made some final decisions:

  1. Change the axis of the stunting indicator to go to 100% so that it is consistent with the other percentages in the graphic. Some suggested that we move stunting and under-five mortality to a separate page to visually remind readers that the interpretation of these indicators is different (i.e., high values are bad). Ultimately, we decided that the layout of the 6 indicators was better for us in terms of publication, but agree that this is a trade-off and may confuse some less technical audiences.
  2. We decided to keep our sorting from low to high, as the main audience for this paper is looking at general trends, not for data for a specific country. However, reports by The DHS Program often have many audiences, and with that in mind, we created an additional graphic (not shown) that summarizes each of the indicators by country so that a stakeholder in Ghana can see his or her relevant data in one view, without searching for Ghana in each of the above graphics.
  3. Jon Schwabish had some quick and practical suggestions for making this graphic easier to read. His critique that it felt “heavy” resonated with us as the creators. He suggested thinning out the lines and substituting the big “X” marking the national average with a smaller circle.

6 indicators for journal April 4

There is a science to data visualization, but there is also a lot of subjectivity. Many solutions can be found only through trial and error. Often it takes time, several new sets of eyes, and dozens of drafts to settle on the best possible visualization for your data. While this is a big investment, there is growing evidence that it’s worth it. We are competing for just 1 or 2 minutes of our audience’s attention in a world filled with data and information. We hope to create a few visualizations that are worth stopping to explore.

 

22 Mar

A Closer Look at Unmet Need in Ghana

View from Elmina Castle in Cape Coast Ghana. © Cameron Taylor/ICF International

For over 30 years, data from DHS surveys have been widely used to assess use of family planning, and monitor family planning programs around the world. DHS data are the gold standard for quality, but nuanced information related to fertility intentions and family planning use is often challenging to collect in a large-scale quantitative survey. Information from in-depth interviews with DHS respondents can help us understand and interpret survey results.

QRS20DHS recently published a follow-up study to the 2014 Ghana Demographic and Health Survey (GDHS). The study reflects an evolving model of qualitative and mixed-methods research within The DHS Program: projects linked to the DHS survey process itself, rather than fielded separately. At the heart of the study was the opportunity to speak with a sub-sample of DHS respondents a few weeks after their DHS interview, which gives us some insight on data quality and reliability when we re-ask a few of the same questions.

But the real purpose of the study was to help us make sense of quantitative data. What does it mean when women say that they want to delay or space their births but that they are not using family planning? Programmatically, there is an important distinction between women who may be classified as having an unmet need for family planning versus women who are willing and ready to contracept. The reasons why a country with a relatively strong family planning program such as Ghana would have one of the continent’s highest levels of unmet need are not something we can always understand very well through the existing questions asked in large-scale surveys. A small number of systematically planned in-depth interviews can help us understand the individual narratives behind survey answers that give rise to the classification of unmet need.

Approximate locations of the final 13 clusters selected for the follow up study. Cluster locations have been randomly displaced to ensure respondent confidentiality.

Approximate locations of the final 13 clusters selected for the follow up study. Cluster locations have been randomly displaced to ensure respondent confidentiality.

Following up with DHS survey respondents was ethically and logistically complicated. We had to first get women’s consent during the initial interview for re-contact, use a computer program to select eligible women, and then try to re-identify women using an approximate address, structure number, name of head of household, and relationship to head of household. Once we approached original respondents we then had to start the process of obtaining consent and scheduling an interview anew. Fieldwork was conducted by the Institute of Statistical, Social, and Economic Research (ISSER) at the University of Ghana, Legon. Ghana Statistical Services, which implemented the GDHS, helped ISSER interviewers re-locate original households and randomly audited the follow-up interviews for data quality.

We re-asked some of the same questions posed by the GDHS and then inquired about any discrepancies. Did the respondent think that there was an error in transcription, or had her circumstances changed in the interim period between surveys?

Perceived cost and access barriers to contraception among follow-up respondents who were not using a modern method of family planning.

Perceived cost and access barriers to contraception among follow-up respondents who were not using a modern method of family planning.

Key findings from the study include: women seem to underreport traditional method use, intentional abstinence as a method of family planning is not well-captured by our surveys, husbands and partners have both a positive and a negative influence on use, women are most concerned about menstrual irregularities caused by hormonal methods, and opposition to modern methods among non-users is stronger than apparent from survey data.

You can download the full study, “Understanding Unmet Need in Ghana: Results from a Follow-up Study to the 2014 Ghana Demographic and Health Survey” from The DHS Program website.

The information provided on this Web site is not official U.S. Government information and does not represent the views or positions of the U.S. Agency for International Development or the U.S. Government.

The DHS Program, ICF
530 Gaither Road, Suite 500, Rockville, MD 20850
Tel: +1 (301) 407-6500 • Fax: +1 (301) 407-6501
dhsprogram.com