1 Module 5 Gender Issues in Data Collection, Sampling and Analysis.

1

Module 5

Gender Issues in Data Collection, Sampling and Analysis

2

Exercise Review Working with the questions you

developed earlier, which ones would you now use to monitor and evaluate this project?

How would you define and operationalize the key terms?

How would you measure your key terms?

What steps would you take to ensure that your data are gender sensitive?

What information did you enter into the design matrix?

3

Learning Objectives

At the end of this session participants will understand:data collection optionssampling optionsdata analysis optionsgender issues related to data collection and data analysis

4

Data Collection StrategyThe strategy depends upon:What you want to know

•“Numbers” or “stories”Where the data resides

•Environment, files, peopleResources available

•Money•Time•Expertise

Gender issues

5

Multiple Approaches Quantitative

• When you want to do statistical analysis, to be precise, to know exactly what you want to measure and/or want to cover a large group

• Hard to develop, easy to analyze

Qualitative• When you want anecdotes or in-depth

information, when you cannot measure what you want to measure, want to know reasons for achievements and problems and/or there is no need to quantify

• Easy to develop, hard to analyze

6

Common Data Collection Approaches

In-person interviews• Structured or unstructured

Self-administered questionnaires

Focus group discussions

Diaries, self-administered time-use reports and income and expenditure reporting

7

Common Data Collection Approaches (Continued)

Observation • Participant observation, unobtrusive

observation, obtrusive observation

Secondary data • Prior studies, existing reports• Media• Existing data

8

Why Data Collection Methods Often Are Not Gender Sensitive Managers, researchers, and technical staff

are not aware of gender issues in the projects or lack experience with gender issues and methods.

Surveys frequently interview only the (male) “household head.”

Formal interviews are not an adequate way to capture information on sensitive topics.

Women may not be able to speak freely in interviews or to attend or speak in community meetings.

9

Gender Sensitive Approaches to Data Collection

Collect data in ways that allow men and women to participate and speak freely

• Consider time of day, child care arrangements, safe settings

Combine quantitative and qualitative data collection methods

Collect information on priorities, constraints and opportunities of individual household members

Ask questions specific to men or women but maintain common core questions so responses can be compared

10

Checklist for Assessing Gender Sensitivity of Data Collection

Situations/Issues to avoid

Actions to ensure methods adequately address gender issues

Sex disaggregated data is available but not used

Information is not collected from the right people

Household surveys are not the appropriate data collection method

Inadequate analysis of gender differences in control of resources within the household

•Assess the availability of gender-responsive data before considering the need to collect new data.

•Include additional questions on gender-specific topics

•Use special methods to analyze gender differences in household decision-making and control of resources.

•Use special methods to study domestic and public violence

•Budget time and resources for follow-up field visits to interpret and further explore statistical findings.

11

Case Discussion

What are some data collection strategies that could be used in your Family Health Project Case?

How can you ensure the data collection is gender-sensitive?

Refer to Your Family Health Project Design Matrix

12

Sampling Samples are required because it is often

not feasible or necessary to collect data on all subjects in the universe being studied.

Random samples ensure that the findings are representative and can be generalized to all families, communities, etc., covered by the study.

Always report sampling procedures used for selection, number of participants, and participation rate (response rate), even in qualitative monitoring and evaluation.

13

Importance of Sampling with Qualitative Methods

Often necessary to generalize from qualitative studies.

Focus groups, PRA techniques etc often do not pay sufficient attention to sampling issues.

Conclusions and recommendations can be very misleading.

14

Types of Sampling

Random: each has an equal probability of being selected (statistical sample).• Results are generalizable

Non-Random• Accidental• Judgmental/Purposive • Convenience• Results are not generalizable

Random

15

Random Samples Level of precision required. Level of disaggregation required

By region By economic group By sex of household head

Type of estimates to be made Point estimate (average income) Difference between means Impacts of different project components

16

1. Sample too large Waste time and money

2. Sample too small Cannot do required analysis

3. Sample covers wrong population Wrong conclusions

4. Parts of the population not covered Wrong conclusions

Four Dangers in Random Samples

17

Random Sampling Problems in Field No sampling frame [list] Difficult to define unit of analysis Important groups missing [illegal

residents, squatters/renters/ landless]

Some groups not accessible Some groups remote and expensive

to reach

18

Non-Random Sampling

The results of non-probability samples cannot be generalized. Data is reported in terms “Of the

respondents….”

Sample size not that important Enough so it seems reasonable Purposeful selection

19

Gender Issues in Sampling

Sample should be representative of men and women in the community or population of interest.

Sample should include sufficient amount of variation for making comparisons.• Different age groups, different marital status,

different villages

20

Gender Issues (continued)

Sample designs to reach women: Separate module for women Snowball sampling

Cultural problems to interview women Intersection theory: need to compare

gender, race, class. Large enough sample to stratify

21

Case Discussion Micro-Credit Studies

Economic Study: Why did they use a statistical sample?• What are the advantages and disadvantages?

Social Study: Why did they use a purposive sample?• What are the advantages and disadvantages?

22

Case Discussion: Family Health Project

Would you use a sample for some data collection? Why or why not?

What kind of sample would you use? What are the advantages and

disadvantages?

23

Data Analysis

Two Basic Types:

Quantitative Data Analysis

Qualitative Data Analysis

24

Quantitative Data Analysis Frequencies, percentage distributions

Rates of change

Cross tabulations

Measures of central tendency• Means, medians and modes

Measures of dispersion.• Standard deviation

Analysis of relationships between variables

25

Social Micro-credit Study

Full control 18%Significant control 19%Partial control 24%Very limited control 17%No control 22%

The study found that less than 40% had significant control.

26

Measures of Association

How strong is the association between two variables? (e.g., income and education)

Several different measures of association• Some measures of association range from 0 to 1• Others range from -1 to +1

Perfect Relationship = 1 or –1

Closer to 0: no relationship

27

Deterministic Statistics (estimates/predictions)

For every change in one variable, we expect an estimated amount of change in another variable.

For example, simple linear regression:• Change in education (x) results in a

change () in income (y)

28

Deterministic Statistics: Economic Micro-credit Study

For every 10% increase in male borrowing in the Grameen Bank, per capita spending by men increases .18.• So if male borrowing increases 20%, we would

predict a .36 increase in per capita spending.

For every 10% increase in female borrowing in the Grameen Bank, per capita spending by women increases .43.

Turn to the Micro-credit Design Matrix

29

Case Discussion Micro-credit Results

Are these results what you expected?

What surprised you?

What might explain these results?

What might you want to ask in the next study?

Looking at the results of this study, what conclusions would you draw about the impact of micro-credit programs?

30

Qualitative Data Analysis

Data from narrative documents, open-ended interviews, focus groups, unstructured observations

Conduct content analysis: Identify common words, ideas, themes

Write on cards Keep track of where they are located Have a second person do the analysis

• Compare results• Work out differences

Identify “quotable quotes” Greatest Risk: Bias

• Hard to recognize things you don’t expect

31

Challenges in Qualitative Data Analysis

Maintaining uniqueness while seeking uniformities and patterns.

Is the purpose of the analysis:• Exploratory and hypothesis generation?• Generalization and testing of hypotheses?

Avoid the trap of selecting extreme or dramatic cases and implying they are typical.

!

32

Integrating Quantitative/Qualitative Analysis

Exploratory studies followed by surveys. Defining key concepts

Quantitative and qualitative research in parallel. Understanding the setting

Follow-up qualitative research to interpret survey findings. Discuss survey results to understand why

people responded as they did; context.

33

Gender Issues in Data Analysis

1. Ensure sex-disaggregated data analysis is conducted for all key variables.

2. Avoid exclusive focus on household level data: go into the household.

3. Do not rely solely on comparison of male and female headed households for analysis of gender differences.

!

34

Gender Issues in Data Analysis

4. Break down female-headed households into voluntary and involuntary households.

5. Study cultural traditions and other factors limiting women’s control over productive assets and ability to take advantage of economic development projects.

6. Consider cultural traditions that would impact acceptance of family planning, education and employment.

!

35

Group Exercise

Each group should complete the Design Matrix for the Family Health Project Case.

Select someone to present elements of the Design Matrix.

Finish Design Matrix