Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (174.77 KB, 67 trang )
<span class='text_page_counter'>(1)</span><div class='page_container' data-page=1>
<b>Deputy Director</b>
<b>Deputy Director</b>
<b>Food Consumption and Nutrition Division</b>
<b><sub>Conducting a quantitative evaluation survey is both a </sub><sub>Conducting a quantitative evaluation survey is both a </sub></b>
<b>science and an art.</b>
<b>science and an art.</b>
<b><sub>The science pertains to the construction of a sample that </sub><sub>The science pertains to the construction of a sample that </sub></b>
<b>is representative of the population of interest</b>
<b>is representative of the population of interest</b>
<b><sub>The art pertains to the implementation of a survey </sub><sub>The art pertains to the implementation of a survey </sub></b>
<b>instrument (such as a questionnaire) that generates the </b>
<b>instrument (such as a questionnaire) that generates the </b>
<b>information you need for your evaluation</b>
<b><sub>Deciding on your unit of observation</sub><sub>Deciding on your unit of observation</sub></b>
<b><sub>Describing the universe: The sample frame</sub><sub>Describing the universe: The sample frame</sub></b>
<b><sub>Drawing the sample</sub><sub>Drawing the sample</sub></b>
<b>The Unit of Observation</b>
<b>The Unit of Observation</b>
<b><sub>The unit of observation is simply the unit that is of </sub><sub>The unit of observation is simply the unit that is of </sub></b>
<b>interest given the objectives of your study:</b>
<b>interest given the objectives of your study:</b>
• <b>Students<sub>Students</sub></b>
• <b>Young women<sub>Young women</sub></b>
<b>The Sampling Frame</b>
<b>The Sampling Frame</b>
<b>The frame for a sample is a list of the units in the </b>
<b>The frame for a sample is a list of the units in the </b>
<b>population (or universe) from which the units that will be </b>
<b>population (or universe) from which the units that will be </b>
<b>enumerated in the sample area are selected. It may be an </b>
<b>enumerated in the sample area are selected. It may be an </b>
<b>actual list, a set of index cards, a map, or data stored in a </b>
<b>actual list, a set of index cards, a map, or data stored in a </b>
<b>computer. The frame is a set of physical materials </b>
<b>computer. The frame is a set of physical materials </b>
<b>(census statistics, maps, lists, directories, records) that </b>
<b>(census statistics, maps, lists, directories, records) that </b>
<b>enables us to take hold of the universe piece by piece. </b>
<b>enables us to take hold of the universe piece by piece. </b>
<b>(Casley and Lury, 1987, p. 52)</b>
<b>The Sampling Frame, cont’d</b>
<b>The Sampling Frame, cont’d</b>
<b><sub>It is important that you examine carefully any sampling </sub><sub>It is important that you examine carefully any sampling </sub></b>
<b>frame that is made available to you for:</b>
<b>frame that is made available to you for:</b>
• <b>Duplications<sub>Duplications</sub></b>
• <b>Errors (such as inclusion errors)<sub>Errors (such as inclusion errors)</sub></b>
• <b>OmissionsOmissions</b>
<b><sub>The concern is that errors of inclusion or omission are </sub><sub>The concern is that errors of inclusion or omission are </sub></b>
<b>non-random</b>
<b>non-random</b>
• <b>Lists of students<sub>Lists of students</sub></b>
<b><sub>Probabilistic sampling methods use some mechanism </sub><sub>Probabilistic sampling methods use some mechanism </sub></b>
<b>involving chance to determine which observations appear </b>
<b>involving chance to determine which observations appear </b>
<b>in the sample. These mechanisms include:</b>
<b>in the sample. These mechanisms include:</b>
• <b>Systematic sampling<sub>Systematic sampling</sub></b>
• <b>Systematic random sampling<sub>Systematic random sampling</sub></b>
• <b>Stratified random sampling<sub>Stratified random sampling</sub></b>
<b><sub>Systematic methods involve the selection of every nth </sub><sub>Systematic methods involve the selection of every nth </sub></b>
<b>observation. </b>
<b>observation. </b>
<b><sub>For example, suppose we want a sample of 250 observations </sub><sub>For example, suppose we want a sample of 250 observations </sub></b>
<b>from our population of 1000 students. </b>
<b>from our population of 1000 students. </b>
<b><sub>We could take the first student on our list, the fifth, the ninth </sub><sub>We could take the first student on our list, the fifth, the ninth </sub></b>
<b>and so on. This method is relatively straightforward. </b>
<b>and so on. This method is relatively straightforward. </b>
<b><sub>The drawback is that the ordering of firms from 1 to 1000 </sub><sub>The drawback is that the ordering of firms from 1 to 1000 </sub></b>
<b>must be random. If there is some subtle, difficult-to-observe </b>
<b>must be random. If there is some subtle, difficult-to-observe </b>
<b>ordering of the sample (for example, older children tend to </b>
<b>ordering of the sample (for example, older children tend to </b>
<b>be counted as even numbers) the observations drawn will </b>
<b>be counted as even numbers) the observations drawn will </b>
<b>not be a random sampling of the population.</b>
<b><sub>Simple random sampling is a better alternative.</sub><sub>Simple random sampling is a better alternative.</sub></b>
<b><sub>The simplest way to do this is to use a statistics package </sub><sub>The simplest way to do this is to use a statistics package </sub></b>
<b>like STATA. </b>
<b>like STATA. </b>
<b><sub>Suppose we have a listing of 1000 students and we want </sub><sub>Suppose we have a listing of 1000 students and we want </sub></b>
<b>to randomly select 50 of them for interview. We use the </b>
<b>to randomly select 50 of them for interview. We use the </b>
<b>command:</b>
<b>command:</b>
<b><sub>There is a potential weakness with this approach. Suppose we </sub><sub>There is a potential weakness with this approach. Suppose we </sub></b>
<b>are drawing a sample of 100 students from a population of 1000. </b>
<b>are drawing a sample of 100 students from a population of 1000. </b>
<b>We know that 30% of these have completed grades 5-8 so our </b>
<b>We know that 30% of these have completed grades 5-8 so our </b>
<b>sample should contain 30 such students. However, this is only </b>
<b>sample should contain 30 such students. However, this is only </b>
<b>true on average! Though the likelihood is high that our sample </b>
<b>true on average! Though the likelihood is high that our sample </b>
<b>will contain 30 such students, it is also possible that it contains </b>
<b>will contain 30 such students, it is also possible that it contains </b>
<b>20, 25 or 40.</b>
<b>20, 25 or 40.</b>
<b><sub>The solution to this problem is random stratified sampling. The </sub><sub>The solution to this problem is random stratified sampling. The </sub></b>
<b>first step is to divide the population into groups or strata. Here, </b>
<b>first step is to divide the population into groups or strata. Here, </b>
<b>the division would be between the 300 students in grades 5-8 </b>
<b>the division would be between the 300 students in grades 5-8 </b>
<b>and 700 other students. Using the random number method, </b>
<b>and 700 other students. Using the random number method, </b>
<b>select 10% of students in each category, so the resultant sample </b>
<b>select 10% of students in each category, so the resultant sample </b>
<b>contains 30 students in grades 5-8 and 70 others. The </b>
<b>contains 30 students in grades 5-8 and 70 others. The </b>
<b>proportions in the sample are identical to those in the </b>
<b>proportions in the sample are identical to those in the </b>
<b>underlying population.</b>
<b><sub>The final approach is to use cluster based sampling.</sub><sub>The final approach is to use cluster based sampling.</sub></b>
<b><sub>Here, you select a unit of observation (eg a school) and </sub><sub>Here, you select a unit of observation (eg a school) and </sub></b>
<b>sample within it.</b>
<b>sample within it.</b>
<b><sub>Cluster based sampling is especially appropriate in the </sub><sub>Cluster based sampling is especially appropriate in the </sub></b>
<b>context of randomized designs where randomization </b>
<b>context of randomized designs where randomization </b>
<b>occurs at the cluster level.</b>
<b>Recall some basic statistical concepts:</b>
<b>Recall some basic statistical concepts:</b>
• <b>Significance level<sub>Significance level</sub>. The probability of rejecting a null . The probability of rejecting a null </b>
<b>hypothesis that is true. – also called Type I errors. This is </b>
<b>hypothesis that is true. – also called Type I errors. This is </b>
<b>often expressed as a percentage so that a test of </b>
<b>often expressed as a percentage so that a test of </b>
<b>significance level, </b>
<b>significance level, </b><i><b>α</b><b>α</b></i><b>, is referred to as a 100, is referred to as a 100</b><i><b>α</b><b>α</b></i><b>% level test% level test</b>
• <b>Power<sub>Power</sub>. The probability, 1- . The probability, 1- </b><i><b>β</b><b>β</b></i><b> of correctly rejecting a null of correctly rejecting a null </b>
<b>hypothesis that is false. </b>
<b>hypothesis that is false. </b>
• <b>For a given sample size, there is a trade off increasing power <sub>For a given sample size, there is a trade off increasing power </sub></b>
<b>and reducing Type I errors</b>
<b>and reducing Type I errors</b>
• <b>We can only increase power and reduce Type I errors We can only increase power and reduce Type I errors </b>
<b>simultaneously by increasing sample size</b>
<b><sub>In addition, in order to calculate sample sizes, we need to </sub><sub>In addition, in order to calculate sample sizes, we need to </sub></b>
<b>know:</b>
<b>know:</b>
• <b>The size of the impact that we would like to detect<sub>The size of the impact that we would like to detect</sub></b>
• <b>Estimates of standard deviations<sub>Estimates of standard deviations</sub></b>
• <b>(If using a cluster design), estimates of intra-cluster <sub>(If using a cluster design), estimates of intra-cluster </sub></b>
<b>correlation, also called the design effect</b>
<b>correlation, also called the design effect</b>
<b><sub>(If using a clustered design), first calculate the intra-</sub><sub>(If using a clustered design), first calculate the </sub></b>
<b>intra-cluster correlation coefficient:</b>
<b>cluster correlation coefficient:</b>
<b><sub>Suppose we have the variable score and our clusters are </sub><sub>Suppose we have the variable score and our clusters are </sub></b>
<b>defined by the variable, school_id</b>
<b>defined by the variable, school_id</b>
<b><sub>The command:</sub><sub>The command:</sub></b>
• <b>loneway score school_id<sub>loneway score school_id</sub></b>
<b><sub>Suppose I want to detect a 20% increase as a result of my </sub><sub>Suppose I want to detect a 20% increase as a result of my </sub></b>
<b>intervention (from 100 for the control group to 120 for the </b>
<b>intervention (from 100 for the control group to 120 for the </b>
<b>treatment group). </b>
<b>treatment group). </b>
<b><sub>I want to have statistical power of 0.80.</sub><sub>I want to have statistical power of 0.80.</sub></b>
<b><sub>Standard deviation is 60.</sub><sub>Standard deviation is 60.</sub></b>
<b><sub>I run the Stata command:</sub><sub>I run the Stata command:</sub></b>
<b>Test Ho: m1 = m2, where m1 is the mean in population 1</b>
<b>Test Ho: m1 = m2, where m1 is the mean in population 1</b>
<b> </b>
<b> and m2 is the mean in population 2and m2 is the mean in population 2</b>
<b>Assumptions:</b>
<b>Assumptions:</b>
<b> </b>
<b> alpha = 0.0500 (two-sided)alpha = 0.0500 (two-sided)</b>
<b> </b>
<b> power = 0.8000power = 0.8000</b>
<b> </b>
<b> m1 = 100m1 = 100</b>
<b> </b>
<b> m2 = 120m2 = 120</b>
<b> </b>
<b> sd1 = 60sd1 = 60</b>
<b> </b>
<b> sd2 = 60sd2 = 60</b>
<b> </b>
<b> n2/n1 = 1.00n2/n1 = 1.00</b>
<b>Estimated required sample sizes:</b>
<b>Estimated required sample sizes:</b>
<b> </b>
<b> n1 = 142n1 = 142</b>
<b> </b>
<b><sub>Suppose I intend to survey 100 schools (50 treatments </sub><sub>Suppose I intend to survey 100 schools (50 treatments </sub></b>
<b>and 50 controls) and the intra-cluster correlation is 0.35</b>
<b>and 50 controls) and the intra-cluster correlation is 0.35</b>
<b><sub>I run</sub><sub>I run</sub></b>
• <b>sampclus, obsclus (25) rho (0.30)<sub>sampclus, obsclus (25) rho (0.30)</sub></b>
<b><sub>Sample Size Adjusted for Cluster Design</sub><sub>Sample Size Adjusted for Cluster Design</sub></b>
<b><sub> </sub><sub> </sub><sub>n1 (uncorrected) = 142</sub><sub>n1 (uncorrected) = 142</sub></b>
<b><sub> </sub><sub> </sub><sub>n2 (uncorrected) = 142</sub><sub>n2 (uncorrected) = 142</sub></b>
<b><sub> </sub><sub> </sub><sub>Intraclass correlation = .3</sub><sub>Intraclass correlation = .3</sub></b>
<b><sub> </sub><sub> </sub><sub>Average obs. per cluster = 25</sub><sub>Average obs. per cluster = 25</sub></b>
<b><sub> </sub><sub> </sub><sub>Minimum number of clusters = 94</sub><sub>Minimum number of clusters = 94</sub></b>
<b><sub> </sub><sub> </sub><sub>Estimated sample size per group:</sub><sub>Estimated sample size per group:</sub></b>
<b><sub>The final point to note here is that having insufficient </sub><sub>The final point to note here is that having insufficient </sub></b>
<b>sample sizes dooms the evaluation survey even before </b>
<b>sample sizes dooms the evaluation survey even before </b>
<b>you leave the office</b>
<b>you leave the office</b>
<b><sub>You need to make these calculations on the basis of </sub><sub>You need to make these calculations on the basis of </sub></b>
<b>conservative assumptions regarding:</b>
<b>conservative assumptions regarding:</b>
• <b>Effect size<sub>Effect size</sub></b>
• <b>Questionnaire design and length; pilot testing<sub>Questionnaire design and length; pilot testing</sub></b>
• <b>Use of PDAs<sub>Use of PDAs</sub></b>
• <b>Issues in data entry, checking and cleaning<sub>Issues in data entry, checking and cleaning</sub></b>
• <b>BudgetsBudgets</b>
<b><sub>The “Soft Arts”</sub><sub>The “Soft Arts”</sub></b>
• <b>Selecting enumeratorsSelecting enumerators</b>
• <b>Selecting Supervisors<sub>Selecting Supervisors</sub></b>
• <b>Interacting with respondents<sub>Interacting with respondents</sub></b>
– <b>Informed consent<sub>Informed consent</sub></b>
– <b>Payments<sub>Payments</sub></b>
• <b>Questionnaire design and length; pilot testing<sub>Questionnaire design and length; pilot testing</sub></b>
• <b>Use of PDAsUse of PDAs</b>
• <b>Issues in data entry, checking and cleaning<sub>Issues in data entry, checking and cleaning</sub></b>
<b><sub>Getting this right is critical to the success of your project.</sub><sub>Getting this right is critical to the success of your project.</sub></b>
<b><sub>The best way to design the questionnaire is to ‘work </sub><sub>The best way to design the questionnaire is to ‘work </sub></b>
<b>backwards’ . That is, start by thinking about what your </b>
<b>backwards’ . That is, start by thinking about what your </b>
<b>report will look like:</b>
<b>report will look like:</b>
• <b>What are the outcomes that you want to measure?<sub>What are the outcomes that you want to measure?</sub></b>
• <b>(In the case of non-randomized designs), what variables <sub>(In the case of non-randomized designs), what variables </sub></b>
<b>determine participation. What covariates would you put in </b>
<b>determine participation. What covariates would you put in </b>
• <b>Do treatment observations actually receive the treatment? <sub>Do treatment observations actually receive the treatment? </sub></b>
<b>Do they receive only partial treatment? Are there problems </b>
<b>Do they receive only partial treatment? Are there problems </b>
<b>with quality? What constraints/problems did they face in </b>
<b>with quality? What constraints/problems did they face in </b>
<b>accessing intervention</b>
<b>accessing intervention</b>
– <b>Helps explain why you might not find significant impactHelps explain why you might not find significant impact</b>
– <b>Allows you to set up a “treatment on the treated” model as an Allows you to set up a “treatment on the treated” model as an </b>
<b>alternative to your “intent to treat”</b>
<b>alternative to your “intent to treat”</b>
– <b>Operational details are of considerable interest to program Operational details are of considerable interest to program </b>
<b>managers</b>
<b>managers</b>
• <b>What are the characteristics of your sample? Are there <sub>What are the characteristics of your sample? Are there </sub></b>
<b>particular sub-groups that you want to identify?</b>
• <b>Develop a logical sequence of questions – think of this as a <sub>Develop a logical sequence of questions – think of this as a </sub></b>
<b>conversation rather than an interview</b>
<b>conversation rather than an interview</b>
• <b>Start with easy/gently questions of the “tell me about Start with easy/gently questions of the “tell me about </b>
<b>yourself” type</b>
<b>yourself” type</b>
• <b>Consider which questions should be pre-coded and which <sub>Consider which questions should be pre-coded and which </sub></b>
<b>should be open-ended. For example, a question on marital </b>
<b>should be open-ended. For example, a question on marital </b>
<b>status could be</b>
<b>status could be</b>
– <b>Precoded: <sub>Precoded: </sub>1 if single; 2 if married; 3 if widowed; 4 if 1 if single; 2 if married; 3 if widowed; 4 if </b>
<b>divorced; 5 if other. </b>
<b>divorced; 5 if other. </b>
– <b>Open ended: How would you describe your marital Open ended: How would you describe your marital </b>
<b>status?</b>
• <b>Use skip patterns:<sub>Use skip patterns:</sub></b>
– <b>“If yes, go to Q10”<sub>“</sub>If yes, go to Q10”</b>
• <b>Think about recall periods:<sub>Think about recall periods:</sub></b>
– <b>Telescoping<sub>Telescoping</sub></b>
– <b>Important but infrequent events<sub>Important but infrequent events</sub></b>
• <b>Think carefully about how to approach sensitive topics<sub>Think carefully about how to approach sensitive topics</sub></b>
• <b>Break down complex questions into parts<sub>Break down complex questions into parts</sub></b>
– <b>Not. How much did you earn from wage work.Not. How much did you earn from wage work.</b>
– <b>Better. How many hours per day did you work in the last seven <sub>Better. How many hours per day did you work in the last seven </sub></b>
<b>days? How many days per week did you work in the last seven </b>
<b>days? How many days per week did you work in the last seven </b>
<b>days? How much are you paid for this wage work, allowing </b>
<b>days? How much are you paid for this wage work, allowing </b>
<b>respondent to answer in terms of hourly, daily or weekly wage</b>
<b>respondent to answer in terms of hourly, daily or weekly wage</b>
<b><sub>Before starting your survey, you need to make sure that </sub><sub>Before starting your survey, you need to make sure that </sub></b>
<b>your questionnaire works – this is called pilot testing.</b>
<b>your questionnaire works – this is called pilot testing.</b>
<b><sub>You should try the questionnaire on 10-20 respondents, </sub><sub>You should try the questionnaire on 10-20 respondents, </sub></b>
<b>who represent a variety of respondent ‘types’</b>
<b>who represent a variety of respondent ‘types’</b>
<b><sub>Pilot testing should reveal the following:</sub><sub>Pilot testing should reveal the following:</sub></b>
<b>a)</b>
<b>a)</b> <b>Are definitions used in the questionnaire appropriate. This Are definitions used in the questionnaire appropriate. This </b>
<b>applies both to definitions of units of observation (does the </b>
<b>applies both to definitions of units of observation (does the </b>
<b>definition of a household correspond with the definition </b>
<b>definition of a household correspond with the definition </b>
<b>used by the people being studied) and to particular </b>
<b>used by the people being studied) and to particular </b>
<b>questions (eg "holdings"; "assets"; "income")</b>
<b>questions (eg "holdings"; "assets"; "income")</b>
<b>b)</b>
<b>b)</b> <b>Do respondents understand the questionsDo respondents understand the questions</b>
<b>c)</b>
<b>d)</b>
<b>d)</b> <b>Are questions being asked that cause respondents unease Are questions being asked that cause respondents unease </b>
<b>or do they refuse to answer</b>
<b>or do they refuse to answer</b>
<b>e)</b>
<b>e)</b> <b>Are there problems associated with translating particular Are there problems associated with translating particular </b>
<b>concepts</b>
<b>concepts</b>
<b>f)</b>
<b>f)</b> <b>Is the layout and sequencing of questions sensibleIs the layout and sequencing of questions sensible</b>
<b>g)</b>
<b>g)</b> <b>Can greater use be made of pre-coding Can greater use be made of pre-coding </b>
<b>h)</b>
<b><sub>After pilot testing, review results with enumerators</sub><sub>After pilot testing, review results with enumerators</sub></b>
<b><sub>Depending on results, re-do with a smaller number of </sub><sub>Depending on results, re-do with a smaller number of </sub></b>
<b>respondents (5-10)</b>
<b><sub>When doing an evaluation survey with a longitudinal </sub><sub>When doing an evaluation survey with a longitudinal </sub></b>
<b>design, it is extremely important that at the time of the </b>
<b>design, it is extremely important that at the time of the </b>
<b>baseline, you obtain information on how to contact </b>
<b>baseline, you obtain information on how to contact </b>
<b>respondents in the future:</b>
<b>respondents in the future:</b>
• <b>GPS is ideal<sub>GPS is ideal</sub></b>
• <b>Cellphone numbers<sub>Cellphone numbers</sub></b>
• <b>Information on people who could aid in a follow up contactInformation on people who could aid in a follow up contact</b>
<b><sub>Not only is this information helpful, it can also be used as </sub><sub>Not only is this information helpful, it can also be used as </sub></b>
<b>additional regressors in attrition probits</b>
<b>Assistants (PDAs) such as PalmPilots or handheld </b>
<b>Assistants (PDAs) such as PalmPilots or handheld </b>
<b>computers for data collection.</b>
<b>computers for data collection.</b>
<b><sub>Using these requires:</sub><sub>Using these requires:</sub></b>
• <b>Purchasing the hardware ($200-$500 each)<sub>Purchasing the hardware ($200-$500 each)</sub></b>
• <b>Purchasing the software, such as Pendragon Forms or PC <sub>Purchasing the software, such as Pendragon Forms or PC </sub></b>
<b>Pocket Creations</b>
<b>Pocket Creations</b>
• <b>Some one to write the data entry program using this <sub>Some one to write the data entry program using this </sub></b>
<b>software</b>
<b><sub>Speed: Data are available for use immediately</sub><sub>Speed: Data are available for use immediately</sub></b>
<b><sub>Filters and skips can ensure that unnecessary questions </sub><sub>Filters and skips can ensure that unnecessary questions </sub></b>
<b>are not asked</b>
<b>are not asked</b>
• <b>Not in our experience.<sub>Not in our experience.</sub></b>
<b><sub>Battery life is a problem</sub><sub>Battery life is a problem</sub></b>
• <b>Buy extra batteries<sub>Buy extra batteries</sub></b>
<b><sub>Transferring data to computers is hard</sub><sub>Transferring data to computers is hard</sub></b>
<b><sub>In our experience, data entry errors are more frequent </sub><sub>In our experience, data entry errors are more frequent </sub></b>
<b>when using PDAs compared to paper questionnaires and </b>
<b>when using PDAs compared to paper questionnaires and </b>
<b>back office data entry, although these may converge with </b>
<b>back office data entry, although these may converge with </b>
<b>further experience</b>
<b>further experience</b>
<b><sub>In our experience, surveys take slightly longer when using </sub><sub>In our experience, surveys take slightly longer when using </sub></b>
<b>handhelds for numeric information and much longer for </b>
<b>handhelds for numeric information and much longer for </b>
<b>text. Consequently, enumerators are much less likely to </b>
<b><sub>If a response does not seem quite right, it is relatively </sub><sub>If a response does not seem quite right, it is relatively </sub></b>
<b>straightforward to take the questionnaire back into the </b>
<b>straightforward to take the questionnaire back into the </b>
<b>field. Note too that checking one response often leads to </b>
<b>field. Note too that checking one response often leads to </b>
<b>revisions to other responses.</b>
<b>revisions to other responses.</b>
<b><sub>If a PDA is lost, then all data are lost. In some cases (such </sub><sub>If a PDA is lost, then all data are lost. In some cases (such </sub></b>
<b>as Pocket PCs, this is also true if the battery dies)</b>
<b>as Pocket PCs, this is also true if the battery dies)</b>
<b><sub>Safety/security of enumerators – carrying an expensive </sub><sub>Safety/security of enumerators – carrying an expensive </sub></b>
<b>electronic device can make them a target for criminals</b>
<b><sub>Researchers typically spend a lot of time thinking about </sub><sub>Researchers typically spend a lot of time thinking about </sub></b>
<b>critical issues at the outset of their project (sample size </b>
<b>critical issues at the outset of their project (sample size </b>
<b>calculations, questionnaires) and at the end of the project </b>
<b>calculations, questionnaires) and at the end of the project </b>
<b>(data analysis, report writing).</b>
<b>(data analysis, report writing).</b>
<b><sub>They typically spend less time worrying about the </sub><sub>They typically spend less time worrying about the </sub></b>
<b>intermediate phase: data entry, cleaning and </b>
<b>intermediate phase: data entry, cleaning and </b>
<b>management.</b>
<b>management.</b>
<b><sub>This is a mistake. Many evaluation studies fall apart, or </sub><sub>This is a mistake. Many evaluation studies fall apart, or </sub></b>
<b>fall badly behind schedule, because insufficient attention </b>
<b>fall badly behind schedule, because insufficient attention </b>
<b>is paid to data management</b>
• <b>Paying attention to the design and implementation of the <sub>Paying attention to the design and implementation of the </sub></b>
<b>data entry software early</b>
<b>data entry software early</b>
• <b>CS PRO (<sub>CS PRO (</sub>www.cspro.orgwww.cspro.org))</b>
• <b>Microsoft Access (often bundled with Microsoft Office)<sub>Microsoft Access (often bundled with Microsoft Office)</sub></b>
• <b>SPSS/Data Entry Module (but this can be expensive)<sub>SPSS/Data Entry Module (but this can be expensive)</sub></b>
<b><sub>Start work on data entry programs as soon as </sub><sub>Start work on data entry programs as soon as </sub></b>
<b>questionnaire design (or designs of certain modules) is </b>
<b>questionnaire design (or designs of certain modules) is </b>
<b>finalized</b>
<b>finalized</b>
<b>1.</b>
<b>1.</b> <b>Enumerator checks questionnaire for completeness before Enumerator checks questionnaire for completeness before </b>
<b>giving it to supervisor</b>
<b>giving it to supervisor</b>
<b>2.</b>
<b>2.</b> <b>Supervisor does quick check to make sure form is complete Supervisor does quick check to make sure form is complete </b>
<b>and that critical topics are correct. She passes form to:</b>
<b>and that critical topics are correct. She passes form to:</b>
<b>3.</b>
<b>3.</b> <b>Data checkers/verifiers who go through form in detail. Data checkers/verifiers who go through form in detail. </b>
<b>Forms are sent back to supervisors/enumerators for </b>
<b>Forms are sent back to supervisors/enumerators for </b>
<b>correction. Otherwise, they are given to data entry team.</b>
<b>correction. Otherwise, they are given to data entry team.</b>
<b>4.</b>
<b>4.</b> <b>Data are entered:Data are entered:</b>
– <b>Question as to whether to use single or double data entry<sub>Question as to whether to use single or double data entry</sub></b>
– <b>Compromise; enter 10-20% of data twice to do quality <sub>Compromise; enter 10-20% of data twice to do quality </sub></b>
<b>check on entry</b>
<b>check on entry</b>
<b>5.</b>
<b>5.</b> <b>Violations of range/value are sent back to field for checkingViolations of range/value are sent back to field for checking</b>
<b>6.</b>
<b>6.</b> <b>Additional checks are made when variable aggregates are Additional checks are made when variable aggregates are </b>
<b>constructed</b>
• <b>Minimum once a week<sub>Minimum once a week</sub></b>
• <b>This means that you have more than one copy held in <sub>This means that you have more than one copy held in </sub></b>
<b>different locations</b>
<b><sub>Budgeting for quantitative surveys has three </sub><sub>Budgeting for quantitative surveys has three </sub></b>
<b>components:</b>
<b>components:</b>
• <b>What do you need?<sub>What do you need?</sub></b>
• <b>How many ___ do you need?How many ___ do you need?</b>
<b><sub>Typical line items in a survey budget:</sub><sub>Typical line items in a survey budget:</sub></b>
• <b>Pilot testing of questionnairesPilot testing of questionnaires</b>
• <b>Training of survey staff<sub>Training of survey staff</sub></b>
• <b>Enumerators (salary, payroll taxes and possibly per diem)<sub>Enumerators (salary, payroll taxes and possibly per diem)</sub></b>
• <b>Supervisors (salary payroll taxes and possibly per diem)Supervisors (salary payroll taxes and possibly per diem)</b>
• <b>Drivers and Transport<sub>Drivers and Transport</sub></b>
• <b>Printing of questionnaires<sub>Printing of questionnaires</sub></b>
• <b>Supplies (clipboards, pens, first aid kits etc)<sub>Supplies (clipboards, pens, first aid kits etc)</sub></b>
• <b>Payments to respondents(?)<sub>Payments to respondents(?)</sub></b>
• <b>Verifiers<sub>Verifiers</sub></b>
• <b>Training of data entry staff<sub>Training of data entry staff</sub></b>
• <b>The size of the sample you are collecting<sub>The size of the sample you are collecting</sub></b>
• <b>Whether it is concentrated in a few places or widely <sub>Whether it is concentrated in a few places or widely </sub></b>
<b>disbursed</b>
<b>disbursed</b>
• <b>How quickly you want to complete the survey<sub>How quickly you want to complete the survey</sub></b>
<b><sub>Example.</sub><sub>Example.</sub></b>
• <b>My survey is working in 40 spatially disbursed locations.<sub>My survey is working in 40 spatially disbursed locations.</sub></b>
• <b>I am interviewing 30 households in each location. I am interviewing 30 households in each location. </b>
• <b>Each interview lasts approximately two hours and so I Each interview lasts approximately two hours and so I </b>
<b>assume that an interviewer can complete two interviews per </b>
<b>assume that an interviewer can complete two interviews per </b>
<b>day</b>
<b>day</b>
• <b>I assume that enumerators work five days, then have one I assume that enumerators work five days, then have one </b>
<b>day off</b>
<b><sub>If I hire three enumerators, they will complete the survey </sub><sub>If I hire three enumerators, they will complete the survey </sub></b>
<b>in one locality in five days</b>
<b>in one locality in five days</b>
• <b>Each day, 6 interviews are completed (3 enumerators x 2 interviews <sub>Each day, 6 interviews are completed (3 enumerators x 2 interviews </sub></b>
<b>per day)</b>
<b>per day)</b>
• <b>In five days, 30 interviews are completed (6 per day x 5 days)In five days, 30 interviews are completed (6 per day x 5 days)</b>
• <b>One day to move to new site, one day off<sub>One day to move to new site, one day off</sub></b>
• <b>Repeat three timesRepeat three times</b>
<b><sub>This implies that one team of three enumerators will cover </sub><sub>This implies that one team of three enumerators will cover </sub></b>
<b>four localities in one month</b>
<b>four localities in one month</b>
<b><sub>So, ten teams of enumerators (30 enumerators in total) </sub><sub>So, ten teams of enumerators (30 enumerators in total) </sub></b>
<b>will complete survey over a four week period</b>
<b>will complete survey over a four week period</b>
<b><sub>Suppose you allocate one supervisor to two teams.</sub><sub>Suppose you allocate one supervisor to two teams.</sub></b>
<b><sub>Selecting Supervisors</sub><sub>Selecting Supervisors</sub></b>
<b><sub>Interacting with respondents</sub><sub>Interacting with respondents</sub></b>
• <b>Preparing the ground <sub>Preparing the ground </sub></b>
• <b>Interview setting<sub>Interview setting</sub></b>
• <b>Interacting with respondents, including:Interacting with respondents, including:</b>
– <b>Informed consentInformed consent</b>
– <b>PaymentsPayments</b>
<b><sub>The job description for the ideal enumerator would </sub><sub>The job description for the ideal enumerator would </sub></b>
<b>include:</b>
<b>include:</b>
• <b>Communications skills<sub>Communications skills</sub></b>
• <b>Good knowledge of English (or French, or Portuguese or <sub>Good knowledge of English (or French, or Portuguese or </sub></b>
<b>Spanish) as well as the local language(s)</b>
<b>Spanish) as well as the local language(s)</b>
• <b>A perceptive intelligence,<sub>A perceptive intelligence,</sub></b>
• <b>inexhaustible patience<sub>inexhaustible patience</sub></b>
• <b>unfailing dependability<sub>unfailing dependability</sub></b>
• <b>Wonderful people skills<sub>Wonderful people skills</sub></b>
• <b>Willingness to work long hours<sub>Willingness to work long hours</sub></b>
• <b>An ability to get along with all elements of the local <sub>An ability to get along with all elements of the local </sub></b>
<b>population. </b>
<b>population. </b>
• <b>Level of schoolingLevel of schooling</b>
• <b>Gender<sub>Gender</sub></b>
<b><sub>Supervisors should be able to do the following</sub><sub>Supervisors should be able to do the following</sub></b>
<b>a)</b>
<b>a)</b> <b>Maintaining time schedules and co-ordinating workMaintaining time schedules and co-ordinating work</b>
<b>b)</b>
<b>b)</b> <b>Checking the enumerator's work by sitting in on interviewsChecking the enumerator's work by sitting in on interviews</b>
<b>c)</b>
<b>c)</b> <b>Checking completed questionnaires for misunderstandings Checking completed questionnaires for misunderstandings </b>
<b>and recording errors</b>
<b>and recording errors</b>
<b>d)</b>
<b>d)</b> <b>Liaising and trouble-shooting with respondents and local Liaising and trouble-shooting with respondents and local </b>
<b>community leaders</b>
<b>community leaders</b>
<b>e)</b>
<b>e)</b> <b>Acting as courier - transferring equipment and documents Acting as courier - transferring equipment and documents </b>
<b>from the office to the field and vice versa</b>
<b>from the office to the field and vice versa</b>
<b>f)</b>
<b>f)</b> <b>Covering in emergencies (eg arranging interviews if an Covering in emergencies (eg arranging interviews if an </b>
<b>enumerator falls sick)</b>
<b>enumerator falls sick)</b>
<b>g)</b>
<b><sub>In both rural and urban areas, even if you have research </sub><sub>In both rural and urban areas, even if you have research </sub></b>
<b>clearance, it is a good idea to contact local government </b>
<b>clearance, it is a good idea to contact local government </b>
<b>officials to inform them about the survey work you intend </b>
<b>officials to inform them about the survey work you intend </b>
<b>to carry out.</b>
<b>to carry out.</b>
<b><sub>It is also helpful to have a standard ‘story’ or explanation </sub><sub>It is also helpful to have a standard ‘story’ or explanation </sub></b>
<b>that you and your research team can provide to any one </b>
<b>that you and your research team can provide to any one </b>
<b>curious about your work.</b>
<b>curious about your work.</b>
<b>“informed consent” from your respondents.</b>
<b>“informed consent” from your respondents.</b>
<b><sub>Specifically, you need to explain to </sub><sub>Specifically, you need to explain to </sub><sub>those being </sub><sub>those being </sub></b>
<b>interviewed:</b>
<b>interviewed:</b>
• <b>The purpose of their study<sub>The purpose of their study</sub></b>
• <b>That participation is voluntary<sub>That participation is voluntary</sub></b>
• <b>Who has funded and supported itWho has funded and supported it</b>
• <b>How the data will be collectedHow the data will be collected</b>
• <b>Who they can contact if they have questions<sub>Who they can contact if they have questions</sub></b>
• <b>How the results may be used.How the results may be used.</b>
• <b>How researchers will maintain anonymity and/or confidentiality.How researchers will maintain anonymity and/or confidentiality.</b>
<b><sub>Consent must be derived from the actual people involved, </sub><sub>Consent must be derived from the actual people involved, </sub></b>
<b>not just officials or leaders. </b>
<b><sub>The best approach to informed consent is to prepare two copies </sub><sub>The best approach to informed consent is to prepare two copies </sub></b>
<b>of a document (or letter) which contains the information </b>
<b>of a document (or letter) which contains the information </b>
<b>described on the previous slide.</b>
<b>described on the previous slide.</b>
<b><sub>At the time of first contact, enumerators review the letter with </sub><sub>At the time of first contact, enumerators review the letter with </sub></b>
<b>respondents. The respondent keeps one copy and signs the </b>
<b>respondents. The respondent keeps one copy and signs the </b>
<b>other which the survey team retains.</b>
<b>other which the survey team retains.</b>
<b><sub>There should be a log kept indicating who has given voluntary </sub><sub>There should be a log kept indicating who has given voluntary </sub></b>
<b>consent.</b>
<b>consent.</b>
<b><sub>In the case of school children (or minors), there is an issue as to </sub><sub>In the case of school children (or minors), there is an issue as to </sub></b>
<b><sub>The decision regarding their setting has two components:</sub><sub>The decision regarding their setting has two components:</sub></b>
• <b>Where to conduct an interview, and Where to conduct an interview, and </b>
• <b>When. When. </b>
<b><sub>The principal decision with respect to interview location is </sub><sub>The principal decision with respect to interview location is </sub></b>
<b>the relative merits of a private versus public encounter. </b>
<b>the relative merits of a private versus public encounter. </b>
<b><sub>A public encounter may encourage a more general </sub><sub>A public encounter may encourage a more general </sub></b>
<b>discussion of the issues; and where information is in the </b>
<b>discussion of the issues; and where information is in the </b>
<b>public domain, onlookers may assist or correct respondents </b>
<b>public domain, onlookers may assist or correct respondents </b>
<b>with particular questions (although there is no guarantee </b>
<b>with particular questions (although there is no guarantee </b>
<b>that they will). </b>
<b>that they will). </b>
<b><sub>When the research deals with matters that are personally </sub><sub>When the research deals with matters that are personally </sub></b>
<b>sensitive, the presence of outsiders or even other family </b>
<b>sensitive, the presence of outsiders or even other family </b>
<b>members may inhibit respondents, embarrassing them into </b>
<b>members may inhibit respondents, embarrassing them into </b>
<b>evasion or silence.</b>
<b><sub>Onlookers may even encourage respondents to answer </sub><sub>Onlookers may even encourage respondents to answer </sub></b>
<b>untruthfully. For example, working during food crises is </b>
<b>untruthfully. For example, working during food crises is </b>
<b>problematic (apart from posing moral dilemmas) because </b>
<b>problematic (apart from posing moral dilemmas) because </b>
<b>respondents have an incentive to understate their stocks </b>
<b>respondents have an incentive to understate their stocks </b>
<b>of grain, and their general wealth, in the expectation that </b>
<b>of grain, and their general wealth, in the expectation that </b>
<b>food aid will be brought into the community. This </b>
<b>food aid will be brought into the community. This </b>
<b>tendency will be exacerbated in a public interview, since </b>
<b>tendency will be exacerbated in a public interview, since </b>
<b>the respondent who admits to being wealthy may face </b>
<b>the respondent who admits to being wealthy may face </b>
<b>demands for help from poorer neighbours.</b>
<b>demands for help from poorer neighbours.</b>
<b><sub>In general, therefore, the more sensitive the topic, the </sub><sub>In general, therefore, the more sensitive the topic, the </sub></b>
<b>stronger the case for conducting the interview in private.</b>
<b><sub>However, often there is no choice. It may be difficult, for </sub><sub>However, often there is no choice. It may be difficult, for </sub></b>
<b>example, to interview workers in firms without their </b>
<b>example, to interview workers in firms without their </b>
<b>supervisors, or fellow employees being present. In some </b>
<b>supervisors, or fellow employees being present. In some </b>
<b>rural areas, it may be impossible to keep neighbours and </b>
<b>rural areas, it may be impossible to keep neighbours and </b>
<b>visitors away. </b>
<b>visitors away. </b>
<b><sub>But ways can be found around this problem.</sub><sub>But ways can be found around this problem.</sub></b>
• <b>Changed the subject of the interview to avoid sensitive Changed the subject of the interview to avoid sensitive </b>
<b>topics while other people were around. </b>
<b>topics while other people were around. </b>
• <b>Another solution is to catch people alone - on coffee breaks, <sub>Another solution is to catch people alone - on coffee breaks, </sub></b>
<b>in the fields working, or on their way to market.</b>
<b><sub>Interviews can be a real imposition on respondents' time. </sub><sub>Interviews can be a real imposition on respondents' time. </sub></b>
<b><sub>It is important to meet on days and at times convenient to </sub><sub>It is important to meet on days and at times convenient to </sub></b>
<b>respondents. </b>
<b>respondents. </b>
<b><sub>One way of ensuring that interviews are not an imposition </sub><sub>One way of ensuring that interviews are not an imposition </sub></b>
<b>is to make appointments to see people.</b>
<b>is to make appointments to see people.</b>
• <b>For example, by making initial contact (and obtaining <sub>For example, by making initial contact (and obtaining </sub></b>
<b>informed consent) and leaving 'calling cards' and/or </b>
<b>informed consent) and leaving 'calling cards' and/or </b>
<b>calenders.</b>
<b>calenders.</b>
<b><sub>A related issue is deciding when in the year to conduct the </sub><sub>A related issue is deciding when in the year to conduct the </sub></b>
<b>interview. </b>
<b>interview. </b>
<b><sub>This is not usually an issue for traders or manufacturing </sub><sub>This is not usually an issue for traders or manufacturing </sub></b>
<b>firms (though food traders are likely to be very busy </b>
<b>firms (though food traders are likely to be very busy </b>
<b>around the harvest period) but can be a matter of concern </b>
<b>around the harvest period) but can be a matter of concern </b>
<b>in rural localities.</b>
<b>in rural localities.</b>
<b><sub>In semi-arid areas, for instance, it may be easier to work </sub><sub>In semi-arid areas, for instance, it may be easier to work </sub></b>
<b>mainly during the dry season when people who are </b>
<b>mainly during the dry season when people who are </b>
<b>actively engaged in farming have time to spare. </b>
<b>actively engaged in farming have time to spare. </b>
<b><sub>At a practical level, travelling around a community may be </sub><sub>At a practical level, travelling around a community may be </sub></b>
<b>debilitating during the heat of the dry season, but can </b>
<b>debilitating during the heat of the dry season, but can </b>
<b>become virtually impossible during the rains. </b>
<b><sub>On the other hand, farmers in semi-arid areas do not work </sub><sub>On the other hand, farmers in semi-arid areas do not work </sub></b>
<b>on their farms for half the year and sit around under trees </b>
<b>on their farms for half the year and sit around under trees </b>
<b>waiting to be interviewed by researchers for the other </b>
<b>waiting to be interviewed by researchers for the other </b>
<b>half! Seasonal out-migration is common in such </b>
<b>half! Seasonal out-migration is common in such </b>
<b>communities, notably among young men, and it would be </b>
<b>communities, notably among young men, and it would be </b>
<b>myopic to defer interviewing until after the harvest if a </b>
<b>myopic to defer interviewing until after the harvest if a </b>
<b>large number of respondents then disappears from the </b>
<b>large number of respondents then disappears from the </b>
<b>village. </b>
<b>village. </b>
<b><sub>Interviewing schedules should be built around the </sub><sub>Interviewing schedules should be built around the </sub></b>
<b>seasonal (and daily and weekly) pattern of domestic and </b>
<b>seasonal (and daily and weekly) pattern of domestic and </b>
<b>economic activities.</b>
<b><sub>In general, researchers do not like to pay respondents in cash for </sub><sub>In general, researchers do not like to pay respondents in cash for </sub></b>
<b>interviews. But interviewing does take time, and can cost </b>
<b>interviews. But interviewing does take time, and can cost </b>
<b>respondents money in terms of lost income. </b>
<b>respondents money in terms of lost income. </b>
<b><sub>When interviews imply a financial sacrifice, some form of </sub><sub>When interviews imply a financial sacrifice, some form of </sub></b>
<b>compensation may be appropriate. </b>
<b>compensation may be appropriate. </b>
<b><sub>In the context of longitudinal or evaluation surveys, payments are </sub><sub>In the context of longitudinal or evaluation surveys, payments are </sub></b>
<b>a means of reducing attrition.</b>
<b>a means of reducing attrition.</b>
<b><sub>One approach is to pay respondents the equivalent of a daily </sub><sub>One approach is to pay respondents the equivalent of a daily </sub></b>
<b>wage, in recognition of the fact that they were giving up a day's </b>
<b>wage, in recognition of the fact that they were giving up a day's </b>
<b>labour. </b>
<b>labour. </b>
<b><sub>An alternative to cash payment is to offer small gifts as a tangible </sub><sub>An alternative to cash payment is to offer small gifts as a tangible </sub></b>
<b>expression of thanks (for education related interventions, pens, </b>
<b>expression of thanks (for education related interventions, pens, </b>
<b>notebooks or inexpensive calculators are an option). </b>
<b><sub>But compensation might create friction between </sub><sub>But compensation might create friction between </sub></b>
<b>respondents as well. </b>
<b>respondents as well. </b>
• <b>If significant inequalities exist within the community, equal If significant inequalities exist within the community, equal </b>
<b>payments are inappropriate - a sizeable sum to a landless labourer </b>
<b>payments are inappropriate - a sizeable sum to a landless labourer </b>
<b>will be an insulting pittance to a wealthy merchant. </b>
<b>will be an insulting pittance to a wealthy merchant. </b>
• <b>Paying wealthy respondents more than the poor only exacerbates Paying wealthy respondents more than the poor only exacerbates </b>
<b>existing inequalities, whereas offering payment or gifts only to the </b>
<b>existing inequalities, whereas offering payment or gifts only to the </b>
<b>poor appears to favour some people at the expense of others. </b>
<b>poor appears to favour some people at the expense of others. </b>
• <b>Further, cash compensation can interfere with the data collection <sub>Further, cash compensation can interfere with the data collection </sub></b>
<b>exercise, particularly if it is more than nominal, and if it relates </b>
<b>exercise, particularly if it is more than nominal, and if it relates </b>
<b>(directly or indirectly) to the research issues.</b>
<b><sub>Further, as soon as fieldworkers give gifts, pay informants in </sub><sub>Further, as soon as fieldworkers give gifts, pay informants in </sub></b>
<b>cash, or lend money, a precedent is set which inevitably leads to </b>
<b>cash, or lend money, a precedent is set which inevitably leads to </b>
<b>escalating expectations. </b>
<b>escalating expectations. </b>
<b><sub>Failing to meet these may compromise both the current </sub><sub>Failing to meet these may compromise both the current </sub></b>
<b>fieldwork and subsequent research projects. </b>
<b>fieldwork and subsequent research projects. </b>
<b><sub>Conversely, if no payment is offered, non respondents may </sub><sub>Conversely, if no payment is offered, non respondents may </sub><sub>‑</sub><sub>‑</sub></b>
<b>ridicule participants for being 'used' for nothing. </b>
<b>ridicule participants for being 'used' for nothing. </b>
<b><sub>Others who are excluded from the sampling frame may want to </sub><sub>Others who are excluded from the sampling frame may want to </sub></b>
<b>be interviewed (and may resent being excluded, believing that </b>
<b>be interviewed (and may resent being excluded, believing that </b>
<b>those selected for interviews will benefit somehow), while some </b>
<b>those selected for interviews will benefit somehow), while some </b>
<b>who are selected may complain that they have been picked on to </b>
<b>who are selected may complain that they have been picked on to </b>
<b>answer interminable boring questions.</b>
<b><sub>In some cases, it may be more appropriate to offer a </sub><sub>In some cases, it may be more appropriate to offer a </sub></b>
<b>single, sizeable contribution to the wider community as a </b>
<b>single, sizeable contribution to the wider community as a </b>
<b>whole. </b>
<b>whole. </b>
<b><sub>This, too, can take a number of forms (depending on what </sub><sub>This, too, can take a number of forms (depending on what </sub></b>
<b>seems appropriate), such as a social event, contributions </b>
<b>seems appropriate), such as a social event, contributions </b>
<b>to community fund raising, a health fair. </b>
<b>to community fund raising, a health fair. </b>
<b><sub>It is important to donate something to which all members </sub><sub>It is important to donate something to which all members </sub></b>
<b>of the community will have equal access and to ensure </b>
<b>of the community will have equal access and to ensure </b>
<b>that the gift does not distort research-related outcomes, </b>
<b>that the gift does not distort research-related outcomes, </b>
<b>unless its presentation is deferred until departure.</b>
<b><sub>There is no simple answer to the question of payment to </sub><sub>There is no simple answer to the question of payment to </sub></b>
<b>respondents. </b>
<b>respondents. </b>
<b><sub>There are strong reasons for providing some form of </sub><sub>There are strong reasons for providing some form of </sub></b>
<b>recompense for the time and trouble taken by </b>
<b>recompense for the time and trouble taken by </b>
<b>respondents</b>
<b>respondents</b>
<b><sub>But there are many dangers in doing so. </sub><sub>But there are many dangers in doing so. </sub></b>
<b><sub>Perhaps the best advice is to provide some tangible </sub><sub>Perhaps the best advice is to provide some tangible </sub></b>
<b>means of appreciation, but do so in a manner sensitive to </b>
<b>means of appreciation, but do so in a manner sensitive to </b>
<b>the circumstances of the research.</b>
<b>completion of the last questionnaire. Researchers have </b>
<b>completion of the last questionnaire. Researchers have </b>
<b>responsibilities to their respondents. </b>
<b>responsibilities to their respondents. </b>
<b><sub>A primary obligation is to tell respondents' story </sub><sub>A primary obligation is to tell respondents' story </sub></b>
<b>truthfully. </b>
<b>truthfully. </b>
<b><sub>An equally important obligation is to maintain the </sub><sub>An equally important obligation is to maintain the </sub></b>
<b>anonymity of respondents. This extends beyond not </b>
<b>anonymity of respondents. This extends beyond not </b>
<b>using names in the context of writing research reports; </b>
<b>using names in the context of writing research reports; </b>
<b>there is a strong case for not entering individual names </b>
<b>there is a strong case for not entering individual names </b>
<b>onto computer files (and relying on identification numbers </b>
<b><sub>Finally, respondents are often interested in the findings of </sub><sub>Finally, respondents are often interested in the findings of </sub></b>
<b>your research. Some means of disseminating results </b>
<b>your research. Some means of disseminating results </b>
<b>should be built into the project design. </b>