Difference between revisions of "Computer-Assisted Personal Interviews (CAPI)"

Jump to: navigation, search
Line 1: Line 1:
<onlyinclude>
<onlyinclude>
Computer-Assisted Personal Interviews(CAPI) is a face-to-face interview method where the interviewer uses a tablet, mobile phone or a computer to record answers given during the interview. There is a wide variety of CAPI software in the market which makes it easy even for someone with no programming experience to program a questionnaire for computer assisted interviews.  
Computer-Assisted Personal Interviews (CAPI) is a face-to-face data collection method in which the interviewer uses a tablet, mobile phone or a computer to record answers given during the interview. This page outlines the advantages and disadvantages of CAPI and provides points to consider when deciding whether or not CAPI fits well into your questionnaire [[Questionnaire Programming | programming]] and [[Questionnaire Design | design]] plan. </onlyinclude>
</onlyinclude>
== Read First ==
== Read First ==
* When programming a questionnaire in CAPI software, careful attention must be paid to a few things which is covered in the topic [[Questionnaire Programming]].
* The wide variety of CAPI software makes it easy even for someone with no programming experience to program a CAPI questionnaire.
* When programming a questionnaire in CAPI software, pay careful attention to [[Questionnaire Programming | questionnaire programming]].


== When to use CAPI surveys ==
== Considerations ==
 
CAPI method has advantages and disadvantages when compared with the [[Pen-and-Paper_Personal_Interviews_(PAPI) | PAPI ]] method of data collection. The advantages of the CAPI method are listed as follows:
===Advantages===
 
* CAPI facilitates logic checks, skip patterns, and validations during the interview. This makes the survey more efficient and helps [[Data Quality Assurance Plan | assure]] higher quality data. It also saves later efforts on [[Data Cleaning | data cleaning]].  
*The data can be validated during the interview leading to better quality data. Along with data validation, the data is also in electronic format which cuts down on the time and effort needed to digitize paper survey data and removes mistakes made when copying the data from paper to digital format.  
* CAPI is a good tool to [[Monitoring Data Quality | monitor]] enumerators. It can automatically record each interview’s start time, end time and GPS location, making it easy for supervisors to check whether an enumerator indeed conducted a given interview or not by comparing its time and GPS data to that of other interviews during which a supervisor was present.
*Complex skips in the questionnaire can be programmed directly into the software, making the survey more efficient and accurate.
* Given strong connection to wifi or data, data collected by CAPI surveys is immediately uploaded in an electronic format. This allows the research team to run high frequency checks.  
*It is also a good tool to monitor enumerators. Since, the software can record start and end time and gps location of an interview automatically, supervisors can check if the interview team did actually conduct the interview or not by comparing that data to other interviews where a supervisor was present.
 
Even though CAPI surveys have benefits in terms of the data quality and management, PAPI methods are very useful in certain situations :
 
*PAPI surveys are especially useful in interviews where a lot of comments/qualitative information needs to be taken. For example - qualitative surveys, [[Survey Pilot | survey pilot]], etc
*PAPI surveys provide a hard copy backup. CAPI data gets corrupted before it can be uploaded to a server, it is lost for ever.
*PAPI surveys are useful in areas without good electricity or data connectivity. While this can be worked using battery packs and offline data storage during CAPI surveys, in some remote parts of the world, it is easier just to do a PAPI survey.
*PAPI surveys are safer to conduct in crime-ridden areas as tablets/phone can be an easy target for thieves.  
*Some respondents are not comfortable during CAPI surveys due to concerns due to spying.
*CAPI surveys require technologically competent enumerators or longer training periods for enumerators in cases which might be a problem in surveys done in emergency situations.
 
All these points should be carefully considered before deciding on the method of interview done for the survey.


== CAPI Software ==  
=== Disadvantages ===


DIME most commonly uses [https://www.surveycto.com/index.html SurveyCTO] (based on Open Data Kit) for CAPI surveys, and a support team in place for technical assistance with SurveyCTO.  
*CAPI surveys may be difficult to conduct in crime-ridden areas, since tablets or phones are easy targets for thieves. In these cases, [[Pen-and-Paper Personal Interviews (PAPI) | PAPI]] or [[Computer-Assisted Field Entry (CAFE) | CAFE]] surveys may have a comparative advantage.
*CAPI relies largely on electricity or data connectivity, which may not be feasible in some remote parts of the world. If CAPI data gets corrupted before it is uploaded to a server, it is lost forever. Battery packs and offline data storage can certainly facilitate CAPI surveys in regions without good electricity or data connectivity; however, PAPI may be the best option in these cases.  
*Relative to PAPI, CAPI less effectively collects data in interviews that entail many comments or qualitative information (i.e. qualitative surveys, [[Survey Pilot | survey pilots]], etc).
*Some respondents are not comfortable during CAPI surveys due to concerns due to spying. This may result in either no data from such respondents or low-quality data due to a lack of trust.
*CAPI surveys require technologically competent enumerators or longer training periods for enumerators. This may pose a problem for surveys conducted in emergency or last-minute situations.


Other CAPI software commonly used in data collection is listed in the table below:
== Software ==
=== Options ===
DIME most commonly uses [https://www.surveycto.com/index.html SurveyCTO],  an Open Data Kit platform, for CAPI surveys. Other CAPI software commonly used in data collection is listed in the table below.


{| class="wikitable"  style="margin-left: 0px; margin-right: auto; border: none;"
{| class="wikitable"  style="margin-left: 0px; margin-right: auto; border: none;"
|+ List of other CAPI software in alphabetical order
|-
|-
! Name
! Name
! Developer
! Developer
! Website
|-
|-
|Blaise
|[http://blaise.com/ Blaise]
|Statistics Netherlands
|Statistics Netherlands
|http://blaise.com/
|-
|-
|CSPro
|[https://www.census.gov/population/international/software/cspro/ CSPro]
|United States Census Bureau
|United States Census Bureau
|https://www.census.gov/population/international/software/cspro/
|-
|-
|Dooblo
|[http://www.dooblo.net/ Dooblo]
|Dooblo Ltd., Israel
|Dooblo Ltd., Israel
|http://www.dooblo.net/
|-
|-
|Open Data Kit(ODK)
|[https://opendatakit.org/ Open Data Kit(ODK)]
|University of Washington's Department of Computer Science and Engineering
|University of Washington's Department of Computer Science and Engineering
|https://opendatakit.org/
|-
|-
|SurveyBe
|[http://surveybe.com/ SurveyBe]
|Economic Development Initiatives Limited, UK
|Economic Development Initiatives Limited, UK
|http://surveybe.com/
|-
|-
|SurveySolutions
|[https://solutions.worldbank.org/ SurveySolutions]
|The World Bank
|The World Bank
|https://solutions.worldbank.org/
|}
|}


=== Choosing a CAPI software for your needs ===
=== Considerations ===


When choosing a CAPI software, you should keep a few things in mind to make sure that you choose the best one for your project:  
When choosing a CAPI software, consider the following factors:


* Data Capture = Does the CAPI Software you're going to choose capture all forms of data that you need i.e. text, numbers, pictures, audio, etc? Does it have the language support that you need to conduct multilingual surveys or record multilingual answers ?  
* Data capture: does the software capture all forms of data that you need (i.e. text, numbers, pictures, audio, etc.)? Does it have the language support that you need to conduct multilingual surveys or record multilingual answers ?  
* Questionnaire Navigation = Is it easy to navigate through the questionnaire in the software?  Can you perform the skips and loops that you need to do in your survey in the software ?
* Questionnaire navigation: is it easy to navigate through the questionnaire in the software?  Can you perform the skips and loops that you need to do in your survey in the software ?
* Data Quality Control =  Does the CAPI software provide ways of controlling the quality of data(values within certain range, values with a required character, values without a certain character, etc)?
* Data quality control: does the CAPI software provide ways of controlling data quality (values within certain range, values with a required character, values without a certain character, etc.)?
* Data Management = Is the data output file from the CAPI Software compatible with the statistical tool you use to analyze datasets?
* Data management: is the data output file from the CAPI Software compatible with the statistical tool you use to analyze datasets?
* Case Management = How easy does the software make for management of tasks for various people in the hierarchy(team leaders, enumerators, etc) during a survey ?
* Case management: does the software facilitate management of tasks for various people within the [[Impact Evaluation Team | impact evaluation team]] during a survey?  
 
== Hardware ==
   
   
=== Programming ===
=== Considerations ===


For programming a survey for CAPI interviews, please read the page on [[Questionnaire_Programming | questionnaire programming]].
When choosing a tablet to use with CAPI software, aim to meet the following specifications:
 
== CAPI Hardware ==
 
Tablets to be used for CAPI interviews should meet the following specifications:  
* Wifi AND cellular connectivity (preferably to a network with widespread 4G coverage in country)
* Front and back cameras (cheapest models have only front cameras; if you need pictures of anything this won’t work well)
* Up-to-date android operating system (current fully-supported versions are marshmallow and nougat)
* Size – depends on the complexity of the instrument. 7 inches is typically fine. However, if lots of tables, may make sense to go up to 10”. If very simple instrument, smaller screen (e.g. 3.5 inch will be fine)
* 1GB memory. More if survey is particularly complex or multiple instruments.
* Internal storage: only a concern if you plan to capture lots of photos, video or audio. If so, tablets with the capacity to use SD cards to expand available memory can help.
* Screen: if you need to show videos or photos, may consider a HD screen. Otherwise no special consideration needed here.
* GPS – should have GPS capacity. However, GPS is typically pretty poor (long time to lock on to a signal, accuracy of 10-15m). Bluetooth receivers (e.g. GlobalSat BT-821 Bluetooth GPS Receiver) can be a good way to enhance GPS capacity
* Font and language requirements – if you’re working in a place with a non-Roman alphabet, make sure that’s supported


* Wifi and cellular connectivity: preferably the tablet can connect to a network with widespread 4G coverage throughout the survey territory.
* Front and back cameras: the cheapest tablet models have only front cameras, but if you need pictures of anything, these won’t work well.
* Up-to-date Android operating system: the current fully-supported versions are marshmallow and nougat.
* Size: the ideal size of your tablet depends on your [[Questionnaire Design | questionnaire design]] and complexity. 7 inches is typically fine, though if your questionnaire includes a lot of tables, you may want to use 10 inch tablets. On the other hand, if your questionnaire is very simple, a 3.5 inch tablet may be sufficient.
* Internal storage: aim for at least 1 GB memory. If the survey is particularly complex or you are implementing multiple questionnaires simultaneously, you may need more than 1 GB.  You may also consider using tablets with SD card capacity to expand available memory.
* Screen: if you need to show videos or photos to respondents, consider a HD screen. It not, no special screen considerations are needed.
* GPS: your tablet should have GPS capacity in order to assure high quality data and monitor enumerators. However, GPS is typically pretty poor: it may take a long time to lock on to a signal and only offer an accuracy of 10-15 meters. Bluetooth receivers (i.e. GlobalSat BT-821 Bluetooth GPS Receiver) can help to enhance GPS capacity.
* Font and language requirement: if you’re working in a place with a non-Roman alphabet, make sure that’s supported.
* Battery life: how long can your tablet last in the field before it needs to be recharged?


== Back to Parent ==
== Back to Parent ==
Line 94: Line 78:


== Additional Resources ==
== Additional Resources ==
*Randomized Survey Experiment comparing CAPI and PAPI survey methods:  [http://siteresources.worldbank.org/INTLSMS/Resources/3358986-1199367264546/Caeyers_Chalmers_DeWeerdt_CAPIvsPAPI.pdf Caeyers, B., N. Chalmers, and J. De Weerdt. "A comparison of CAPI and PAPI through a randomized field experiment (2010)."]
 
*Comparative Assessment Study of CAPI Software done by the World Bank : http://siteresources.worldbank.org/INTSURAGRI/Resources/7420178-1294259038276/CAPI.Software.Assessment.Main.Report.pdf
*The World Bank's [http://siteresources.worldbank.org/INTSURAGRI/Resources/7420178-1294259038276/CAPI.Software.Assessment.Main.Report.pdf Comparative Assessment Study of CAPI Software]
*World Bank slides on [http://web.worldbank.org/archive/website01542/WEB/IMAGES/1_ELECTR.PDF CAPI]
*Cayers, Chalmers and Weerdt's [http://siteresources.worldbank.org/INTLSMS/Resources/3358986-1199367264546/Caeyers_Chalmers_DeWeerdt_CAPIvsPAPI.pdf Randomized Survey Experiment] comparing CAPI and PAPI survey methods


[[Category: Preparing for Data Collection]]
[[Category: Preparing for Data Collection]]

Revision as of 15:08, 12 April 2019

Computer-Assisted Personal Interviews (CAPI) is a face-to-face data collection method in which the interviewer uses a tablet, mobile phone or a computer to record answers given during the interview. This page outlines the advantages and disadvantages of CAPI and provides points to consider when deciding whether or not CAPI fits well into your questionnaire programming and design plan.

Read First

  • The wide variety of CAPI software makes it easy even for someone with no programming experience to program a CAPI questionnaire.
  • When programming a questionnaire in CAPI software, pay careful attention to questionnaire programming.

Considerations

Advantages

  • CAPI facilitates logic checks, skip patterns, and validations during the interview. This makes the survey more efficient and helps assure higher quality data. It also saves later efforts on data cleaning.
  • CAPI is a good tool to monitor enumerators. It can automatically record each interview’s start time, end time and GPS location, making it easy for supervisors to check whether an enumerator indeed conducted a given interview or not by comparing its time and GPS data to that of other interviews during which a supervisor was present.
  • Given strong connection to wifi or data, data collected by CAPI surveys is immediately uploaded in an electronic format. This allows the research team to run high frequency checks.

Disadvantages

  • CAPI surveys may be difficult to conduct in crime-ridden areas, since tablets or phones are easy targets for thieves. In these cases, PAPI or CAFE surveys may have a comparative advantage.
  • CAPI relies largely on electricity or data connectivity, which may not be feasible in some remote parts of the world. If CAPI data gets corrupted before it is uploaded to a server, it is lost forever. Battery packs and offline data storage can certainly facilitate CAPI surveys in regions without good electricity or data connectivity; however, PAPI may be the best option in these cases.
  • Relative to PAPI, CAPI less effectively collects data in interviews that entail many comments or qualitative information (i.e. qualitative surveys, survey pilots, etc).
  • Some respondents are not comfortable during CAPI surveys due to concerns due to spying. This may result in either no data from such respondents or low-quality data due to a lack of trust.
  • CAPI surveys require technologically competent enumerators or longer training periods for enumerators. This may pose a problem for surveys conducted in emergency or last-minute situations.

Software

Options

DIME most commonly uses SurveyCTO, an Open Data Kit platform, for CAPI surveys. Other CAPI software commonly used in data collection is listed in the table below.

Name Developer
Blaise Statistics Netherlands
CSPro United States Census Bureau
Dooblo Dooblo Ltd., Israel
Open Data Kit(ODK) University of Washington's Department of Computer Science and Engineering
SurveyBe Economic Development Initiatives Limited, UK
SurveySolutions The World Bank

Considerations

When choosing a CAPI software, consider the following factors:

  • Data capture: does the software capture all forms of data that you need (i.e. text, numbers, pictures, audio, etc.)? Does it have the language support that you need to conduct multilingual surveys or record multilingual answers ?
  • Questionnaire navigation: is it easy to navigate through the questionnaire in the software? Can you perform the skips and loops that you need to do in your survey in the software ?
  • Data quality control: does the CAPI software provide ways of controlling data quality (values within certain range, values with a required character, values without a certain character, etc.)?
  • Data management: is the data output file from the CAPI Software compatible with the statistical tool you use to analyze datasets?
  • Case management: does the software facilitate management of tasks for various people within the impact evaluation team during a survey?

Hardware

Considerations

When choosing a tablet to use with CAPI software, aim to meet the following specifications:

  • Wifi and cellular connectivity: preferably the tablet can connect to a network with widespread 4G coverage throughout the survey territory.
  • Front and back cameras: the cheapest tablet models have only front cameras, but if you need pictures of anything, these won’t work well.
  • Up-to-date Android operating system: the current fully-supported versions are marshmallow and nougat.
  • Size: the ideal size of your tablet depends on your questionnaire design and complexity. 7 inches is typically fine, though if your questionnaire includes a lot of tables, you may want to use 10 inch tablets. On the other hand, if your questionnaire is very simple, a 3.5 inch tablet may be sufficient.
  • Internal storage: aim for at least 1 GB memory. If the survey is particularly complex or you are implementing multiple questionnaires simultaneously, you may need more than 1 GB. You may also consider using tablets with SD card capacity to expand available memory.
  • Screen: if you need to show videos or photos to respondents, consider a HD screen. It not, no special screen considerations are needed.
  • GPS: your tablet should have GPS capacity in order to assure high quality data and monitor enumerators. However, GPS is typically pretty poor: it may take a long time to lock on to a signal and only offer an accuracy of 10-15 meters. Bluetooth receivers (i.e. GlobalSat BT-821 Bluetooth GPS Receiver) can help to enhance GPS capacity.
  • Font and language requirement: if you’re working in a place with a non-Roman alphabet, make sure that’s supported.
  • Battery life: how long can your tablet last in the field before it needs to be recharged?

Back to Parent

This article is part of the topic Preparing for Data Collection

Additional Resources