Design a simple Audio Computer-Assisted Self-Interview (ACASI) using SurveyCTO

This article is accompanied by sample forms and attachments saved in this folder.

1. Introduction    

Audio Computer-Assisted Self-Interviewing (ACASI) empowers respondents to interact with survey questions through a digital interface, using audio to guide them throughout the process. This approach allows respondents to independently navigate and respond to survey questions, ensuring a confidential and private experience. This can have a range of benefits including improved accessibility for respondents across language and literacy barriers. Additionally, more honest answers (because no one is asking the questions or watching the answers) can be recorded which are free of enumerator bias.
You can build user-friendly ACASI workflows using SurveyCTO’s existing features. The sample workflow in this article showcases some of SurveyCTO’s powerful features that can be easily adapted to specific project needs.

This use case covers the following sections:

  1. Deploying the workflow
  2. Overview
  3. Workflow components
  4. Customization and improvements
  5. Bibliography
     
Note: This workflow is optimized for Android. Some features might not work properly with iOS..

2. Deploy the workflow now!

Click below to view this workflow in the Hub and install it on your server.

Install Workflow

Advanced users can find the sample workflow files in this folder. For help with manual deployment, check out our support article Deploying form definitions and server datasets.

3. Overview

In this workflow, you will be able to:

surveycto_icon.png

Practice responding to survey questions using the interface of a digital device

surveycto_icon.png

Receive feedback when a wrong choice is selected

surveycto_icon.png

Repeat each practice question until the correct answer is selected

surveycto_icon.png

Provide your feedback on the usability of the ACASI method

The workflow is composed of two main parts:

surveycto_icon.png

A practice module for respondents to learn to use mobile devices to take the survey

surveycto_icon.png

A self-response module where respondents can answer real survey questions

To be able to effectively test this workflow you can follow these outlined tasks:

  • Follow the audio instructions on using the survey interface components to respond to different types of questions
  • Respond to the self-response module questions using the audio instructions and visual aids

4. Workflow components

Generate Audio Files of Survey Elements

As mentioned earlier, ACASI survey questions and choices are delivered in audio files. So an audio file needs to be generated per field. These audio files should be specified in the media:audio parameter of the corresponding fields. Refer to the "Form field properties" section later in this article for best practices for setting up field properties for ACASI forms.

Note: SurveyCTO supports multiple languages and each can have their own audio. Check out our documentation and support article to learn more.

Another important aspect to consider when generating audio files is selecting the right tool to effectively convey survey elements in audio format is crucial in the successful design process of ACASI survey. 

Recording audio files with a human voice allows the maximum level of customization in ensuring that all the survey elements are effectively delivered to the respondents. Be sure to quality-control the process of recording the audio of each question, minimizing background noise, and ensuring good levels, and the accuracy of each recording. You may need some professional support.

TTS (Text-To-Speech) was used to generate audio for this example. TTS platforms are more cost-effective (even free!) and time-saving tools compared to human voice recording, as you can generate an audio version of your survey questions in real time. There are many open-source TTS platforms that you can use to easily convert field labels into audio. However, most of these platforms only allow non-commercial and personal use of their products under free tier options.

Mozilla's Speech synthesiser demo is an open-source platform that doesn’t have such restrictions and there are several voice options with different accents per language. Users can also personalize the voice speed and pitch. Here’s the Speech synthesiser setup information - Rate: 0.8, Pitch: 1, Voice: Samantha(en-US).

Speech synthesiser.png

 

Audacity is a free and open-source digital audio editor and recording application software. You can use this tool to record real voices or the audio generated using the TTS platform.

Audacity.png

Audacity was used to record and edit audio from Mozilla's TSS platform. 

As all the survey content will be delivered in audio, it’s important to properly name and reference the files throughout the form, and any changes to field names should be seamlessly reflected on the file names. The file names should contain the following information:

  1. File ID in ordinal number format. This can help you organize the files in the right order, independently of the form during production.
  2. Field name. This helps create an obvious link with where the file should be used.
  3. Underscores to connect the file ID and the file name, for example, 1_a_instructions.mp3.

 

Generate Images for Choices

In ACASI surveys, images are used along with audio instructions to help respondents respond to questions. When designing the images as choices in ACASI, it’s important to take into account the following:

  1. The main purpose of using images is to facilitate respondents selecting the choices that they hear in the audio. In other words, the content of the survey question and choices are delivered via audio and the images work as “buttons” to help respondents select the choices that represent their response the best.
  2. Due to the sensitive nature of these questions and options, the images used for choices shouldn’t be representative of the content of the audio.
  3. Instead, the images should be something that the respondents can easily differentiate, for example, easily identifiable shapes, daily used items or objects that can be easily found in the surroundings of respondents' locations, etc. This point is important, especially for those with low literacy levels. It’s important to test different images with target populations to select the most appropriate images to use.
  4. Use accessible colors (for example, yellow and blue) and avoid certain colors (red and green) that can be difficult for color-blind people to identify. Check out online resources such as this one for more information.
  5. There is a wide range of options available when it comes to creating images for an ACASI survey. Unsplash is a good source for royalty-free images. You can also draw your own image with professional drawing software such as Photoshop or with the help of free AI-powered platforms such as AutoDraw. Images used in the sample workflow are made using AutoDraw. 
When you’re using form attachments such as images and audio, make sure that they don’t exceed the maximum attachment size (100MB per file and 300MB for total attachment size). Check out our support article to learn more about SurveyCTO limitations.

 

Practice Module

The purpose of this module is to help respondents with low digital literacy learn to use digital devices to take the survey on their own.

Take into account the following when designing this module:

  • Respondents' previous experience and knowledge with using digital devices
  • Numeracy level of the respondents to design any questions that require them to insert numeric values, such as number of items, age, periods, etc.
  • Also, make sure to test your workflow with people with similar demographic characteristics before going to the field.

The sample practice module has the following features:

  1. Select one option among different items
  2. Likert question to demonstrate different grades of respondent’s opinion on a subject
  3. Insert a numeric value
  4. Repeat the question until the respondent selects the correct answer
  5. Provide feedback in the case that the selected option is incorrect

 

Form Field Properties

In the sample workflow, the following field properties are used for enhanced user experience in the ACASI survey:

  • appearance options for ACASI sample form fields
    “quick” appearance can be used to automatically advance to the next question as soon as the respondent selects a choice. This makes the user experience seamless as the respondents just need to select the choice that best represents their response without having to worry about additional steps to take to operate the survey.
    In addition to the “quick” appearance, “quickcompact-#” appearance can be used to customize the number of choices appearing in one row. For example, when “quickcompact-4” is used as the appearance, up to 4 choices will appear on each row. In case there are 6 choices with this appearance, the first 4 choices will appear on the first row and the last 2 choices will appear on the second row. And when the respondent selects a choice, it will automatically advance to the next question.
  • media:audio
    In ACASI survey, each survey question and its choices are delivered in a single audio file. In order to make the audio file play during the survey, specify the full file name in the media:audio column of the field(i.e. field1.mp3), on your spreadsheet form definition.
    In case  you’re using the online form designer,
    1. Go to the “Other options” section of the field
    2. Click on Edit next to media:audio
    3. Upload the audio file for this field.
    The most commonly used formats are .mp3 and .wav. Make sure to test the audio files on your mobile device.
  • autoplay
    Add the “autoplay” column to the survey sheet of the spreadsheet form design template and specify “audio” in this column to automatically play the audio file when that field is accessed for the first time (if you're working in the online form designer, you can export the spreadsheet form definition and make this change and then re-deploy your form to do this). Please note that the audio won’t be automatically played again when the field is accessed more than once. For example, the audio won’t automatically play again when the respondent decides to go back to this field after they already responded to this field and moved to the next field. In this case, the respondent needs to press the audio icon to listen to the audio instructions again.
  • minimum_seconds(only on SurveyCTO Collect)
    Use the minimum_seconds column to specify a minimum number of seconds that enumerators should spend on any given field. To calculate appropriate time that the respondent needs to spend on each field, add around 5 to 10 seconds to the duration of the audio file.
    On SurveyCTO Collect, you can enforce the minimum seconds requirement by enabling the Enforce minimum times for fields option within Collect's Admin Settings (from the main Collect menu, open the three-dot menu, then choose Admin Settings). That way, the respondents will need to spend the minimum required time on each field before moving on to the next field. Please note that this requirement doesn’t apply when the respondents decide to go back to the previous fields to change their responses. Check out our documentation to learn more about quality assurance features.
  • image column on the choices tab:
    In the ACASI survey, all the choices will be delivered in audio, accompanied by images to guide respondents to select the choices of their preference. To include an image to display for a choice, its filename should be specified in the images column of the choices tab of the spreadsheet form definition and the file should be uploaded with the form definition. All devices support common image formats like .jpg or .png. For example, bluecircle.png.
    If you're working in the online form designer, you can upload an image per choice in a choice list.

Source material

Survey questions used in the sample form are from the sexual health module from the UNAIDS report(title: “Knowledge, Attitude and Practices in Sexual Behaviour and HIV/AIDS in Belize”, page 104 - 106).

Workflow Component Files
Click on the section to access the files.
Forms
Name ACASI sample form
File(s) Definition: Form definition - ACASI sample form 
Description This sample form is optimized for a field-ready experience.
Name ACASI sample form with labels
File(s) Definition: Form definition - ACASI sample form with labels 
Description This sample form is optimized for quickly understanding how ACASI works. However, it is is not best practice to include labels in fields or choice lists without images.
Form Attachments
Name Audios
Files(s) File: Audios 
Description Audio files in .zip format
Name Images
Files(s) Definition: Images 
Description Image files in .zip format

5. Customization and improvements

SurveyCTO has endless possibilities, so this workflow is only the beginning. Feel free to modify and expand on this workflow to fit your own needs. Here are some tips on what you can add:

  • The survey tab of the sample form shouldn’t have labels in the text format except for the guidelines for the enumerators. This is because the survey contents are provided only through audio and images, so that third parties cannot tell what’s going on with the device when the respondent is taking a survey. However, it can be harder to build a form where the fields do not have self-explanatory labels. A tip to effectively manage this is just working on the survey tab as usual with the questions in the labels. After it’s completed, duplicate the label column and name the duplicated column “labels_hidden”. Then delete the texts of the labels column on the survey tab (do not remove the labels header though). That way, you’ll always have a copy of the original label texts with full content that you can refer to whenever you need to make changes.
  • The above method is not necessary when it comes to the text labels for the choices on the choices tab. For select_one questions with “quickcompact-#” appearance that has more than one choice, text labels of the choices get replaced with the images on SurveyCTO Collect, which makes the images ‘buttons’.
  • Currently, group labels and repeat group labels cannot be hidden on Collect. On the web form, you can hide them by selecting “OFF” on the Show location within groups option on the form settings section on the Collect tab on the server. In order to ensure the maximum privacy, use generic names that are not representative of the groups and repeat groups on the survey tab,
  • This sample doesn’t technically cater to a low literacy population, given the use of integer fields. There’d be ways of using visual choice lists to get a rough sense of quantity and number using visual symbols and clear audio descriptions to better cater to low-literacy populations.
  • For sensitive topics, there should be a preamble that checks whether the respondent and interviewer have enough privacy and what to do if privacy is compromised.
  • Think about providing respondents with headphones during the survey to ensure their privacy.
  • On topics that are a risk for the respondent (such as intimate partner violence), far more stringent protocols should be established to protect respondents.
  • In order to guarantee privacy and ensure honest answers, you can make the questions disappear right after the respondents answer the questions. You can learn how to implement this workflow and also its pros and cons to this approach in our support article Is it possible to hide answers in a form?.
  • Encryption should be used to at least hide the respondents' identity in the data (if it is being collected) in the case of sensitive questions. Check out our support article How to encrypt a form to learn more.
  • As mentioned above, respondents’ context should be taken into consideration in designing ACASI surveys in order to make the survey accessible. Make sure to test ACASI features throughout the design process. Especially for the practice module, test images, audio, and mobile device interface with the target population during the design phase and make adjustments to reflect the test results. And then test again. Repeat this process until the desired results are achieved.

6. Bibliography

Do you have thoughts on this guide? We'd love to hear them! Feel free to fill out this feedback form.

0 Comments

Please sign in to leave a comment.