App-in-the-Middle is a feature developed for Nfield, in which control of a survey is temporarily handed over to an external application, so that functionalities not offered within Nfield can be incorporated into survey execution. This is useful, for example, if you want to make use of a Sawtooth module for implementing a complex conjoint.

How it works

All you have to do is instruct your Nfield survey when to switch to the external app, and instruct the external app when to switch back to your survey. Making sure, of course, that all the relevant data is handed over during both switching processes.

Within Nfield, this is handled in the sense of pausing and resuming interviews. We’ve accommodated the App-in-the-Middle feature by making it possible to resume paused interviews from an “as is” status, in addition to the usual “as was” status.

“as was” vs “as is”

Under normal circumstances which don’t involve App-in-the-Middle, a paused Nfield Online or CAWI interview simply returns to its last-known state upon resuming, with all variables and routing restored to the point they were at when paused. This is known as resuming from the “as was”.

Incorporating App-in-the-Middle requires Nfield to accept changes made during the external app’s time in control and resume from this updated situation. This means resuming from the “as is” instead of the “as was”. We’ve introduced this option into Nfield, so you can now incorporate extra functionality via App-in-the-Middle whenever you need to.

How to implement App-in-the-Middle in your surveys

This short video shows the basic principles without going into too much detail.

Switching control between Nfield and an external app

Control over conducting the interview is handed over from Nfield to the external app by using the ODIN command: *ENDST 107. This pauses the interview in Nfield and redirects to the relocation link defined for response code 107. The external app hands back to Nfield by redirecting to the survey URL with its corresponding respondent key and other information.

For the interview to continue in Nfield, some special script constructs must be used to prevent the interview executing *ENDST 107 again. This is done by using either the *INIT block or information that is handed over from the external app. You can find out more about using *INIT blocks in our Academy session video on pausing and resuming surveys.

Handing information over to an external app

The relevant information can be handed over to an external app by adding it as query string parameters to the exit link configured for the HandedOver response code (107).

For example:
https://aitm.com?respondent={respondentKey}&extra={extraInfo}

This shows the names of ODIN variables between the {}. The variable names are replaced by the variable values before redirection to the external app’s URL.

Handing information back to the Nfield survey

The external app hands back to Nfield by redirecting to the survey URL. The survey URL must contain the respondentKey. Extra information can be added as query string parameters. The interview will then be resumed (from “as is”) with the ODIN variables having the values now shown in the query string parameters.

For example:
https://{startlink}/{respondentKey}?aitmResult=199

This will resume the interview and if an ODIN variable with the name aitmResult is defined it will get the value 199.

Here to help!

Let us know if you want to integrate a specific app via App-in-the-Middle into your surveys. We’ll be happy to help you set up your project.

The market research industry’s three main channels – Online, CATI and CAPI – each have distinct advantages in different situations. Online is fast, cheap and far-reaching. But better quality responses can sometimes be assured via CATI and CAPI channels, which can be backed up with visual, audio and GPS location evidence.

Circumstantial, cultural, demographic and geographic factors also come into play when it comes to which channel is most effective for delivering response. But these can vary within an individual survey and even within an individual sample.

To offer a comprehensive service that keeps all the options open, market research companies ideally need the ability to deploy all three channels, switching between them as necessary in a mixed-mode project.

Benefits Of multi-mode flexibility

While a particular survey type is often a preferred choice for a specific market research project, there are often circumstances when the ability to switch modes is beneficial.

• CAPI and CATI surveys are sometimes preferential when quality control via a human interviewer is important. But if a target group cannot be reached face-to-face or by phone, then an Online sample can be used to fill in the gaps.
• CAPI interviewers sometimes encounter respondents for whom on-the-spot participation isn’t convenient, but are happy to take the survey online at a time of their choosing.
• Some respondents are not comfortable with completing Online surveys and prefer to provide answers over the phone.
• India, with 370 million internet users, is the second largest online market after China. Yet this figure only represents 30% of the population. So while Online can be a great channel here, it’s not enough on its own.
• Online is the predominant channel in Europe and North America, there’s a preference for face-to-face in Asia and Africa.

Incompatible systems limit market research capability

Because Online, CAPI and CATI channels are driven by different technical systems, switching between them is often not as simple as it needs to be. Making a survey available by all three systems typically requires duplication of set-up work and additional effort to consolidate the results. For some market research companies, this is a burden to far. For the sake of efficiency, they dedicate themselves to a single channel. Thereby cutting off the many opportunities which come with multi-channel capability.

Nfield consolidates Online (CAWI), CATI And CAPI to boost productivity and open up possibilities

Here at NIPO, we are all about empowering market research companies to perform better. So we decided to get a grip on the situation by developing Online, CATI and CAPI systems which are fully compatible with each other.

Using our Nfield Online, CATI and CAPI survey solutions, you only have to set up a survey once for it to automatically translate across all three channels. Each respondent’s answers can be accepted via any channel and results are automatically consolidated.

New to Nfield?

If you’re currently burdened with incompatible systems for multi-mode surveys, switching to Nfield will noticeably boost your productivity.

And if you’ve been limiting yourself to just one or two channels, Nfield provides an easy way to explore the opportunities multi-mode survey capability could bring to your business.

What’s more, because Nfield survey solutions are specifically designed to meet the needs of market research professionals, they’re sully equipped with everything from stunning survey design and advanced scripting options to elaborate sampling, quota handling and much more.

Want to find out more?

Feel free to contact us to discuss your requirements and ask anything you need to know about Nfield Online, CAPI and CATI.

Our article Your Nfield login’s value on the dark web explains why access to your Nfield account is such a tempting prize for hackers. The good news is you can make it almost impossible for them to get in by deploying two-factor authentication. Adding this extra security layer to your username and password login makes your Nfield account more than 99.9% less likely to be compromised, according to research by Alex Weinert, Group Program Manager for Identity Security and Protection at Microsoft¹.

This article explains the concept of two-factor authentication (2FA) and its benefits, as well as giving instructions for setting it up and rolling it out in your organization.

¹ https://techcommunity.microsoft.com/t5/azure-active-directory-identity/your-pa-word-doesn-t-matter/ba-p/731984

How two-factor authentication (2FA) works

You’re probably already familiar with using two-factor authentication to access things such as bank, social media (Facebook, LinkedIn, Instagram) and email accounts. 2FA adds to the first factor – your email address/username and password combination – by asking for a code which can only be obtained via a physical object you have in your possession. This might be a key-like token, an office access card or an SMS message received on your mobile phone. Very highly secured systems may even require a third factor, such as a fingerprint or iris scan.

Nfield accounts secured with two-factor authentication require users to enter a code (a token) generated by a standard authenticator app on a mobile phone. This has the effect of complementing something you know (your username and password) with a code obtained through something you have (your phone). It effectively blocks any hackers who have obtained your username and password from getting into your Nfield account, as they would not be able to retrieve the second factor code from the phone. Your valuable Nfield fieldwork and respondent data is thereby protected from prying eyes.

Data security compliance

Different companies have different policies for protecting different types of data. Even if your organization doesn’t require two-factor authentication, your client’s organization might. Having it set up on your Nfield account means you’ll be compliant with every policy or project requirement.

Two-factor authentication in relation to GDPR, ISO 27001 and ISO 27002

• The General Data Protection Regulation (GDPR) states appropriate data security as being a mandatory requirement. However, it is up to individual organizations to determine how they should achieve this. As an example, the Dutch Data Protection Authority (Dutch DPA) is responsible for a hospital patient portal in which medical data is stored. GDPR² considers this as sensitive personal data. The Dutch DPA has opted to deploy two-factor authentication to satisfy the requirement for appropriate data security.
• ISO 27001 only states that access control should require secret information, such as a password, as a means of authentication³. However, ISO 27002 advises a tighter level of control and recommends using two-factor authentication for a number of scenarios. These scenarios include Nfield, as it falls within the categories A.9.4.2 (secure log-on procedures) and A9.4.4 (use of privileged utility program).

With compliance and IT policies regularly being updated to fend off new security threats, it’s probably only a matter of time before two-factor authentication becomes a standard requirement.

² https://www.zivver.com/blog/which-type-of-2fa-do-i-need-to-use-under-the-gdpr
³ https://advisera.com/27001academy/blog/2017/01/16/how-two-factor-authentication-enables-compliance-with-iso-27001-access-controls/

Enabling two-factor authentication in Nfield

Enabling this feature across an Nfield domain can only be done by domain administrators or local domain managers. The instruction to enable is located in the password policy page in the domain settings.

After enabling two-factor authentication, follow the on-screen instructions for setting up two-factor configuration. You’ll need to start by selecting and setting up an authenticator app, such as Microsoft Authenticator, Google Authenticator or others. Next, use the app to scan the QR code provided by Nfield. Once all is correctly configured, the app will provide a code which needs to be entered into Nfield to complete the two-factor authentication. It is as simple as that! Every time you log in to Nfield, you go through the same process, getting a new code each time.

Two-factor authentication will become effective across your Nfield domain within 30 minutes of being enabled. Any logged-in users will get the same prompt asked them to complete their configuration setup. Other users will get this prompt when they try to log in. Using public API (https://www.nipo.com/api-what-researchers-need-to-know) is excepted from using two-factor authentication.

Timing your 2FA roll-out

To minimize disruption to your team, please plan this carefully. We recommend you consider the following:

• read and share this article.
• communicate with your team about why 2FA is necessary and what to expect.
• best practice is to use the authenticator via a phone, not via a PC. The second factor is far more secure when coming from a different source. A two-factor authenticator installed on a PC browser is vulnerable to hackers, as they can get hold of it when hacking into your PC.
• find a good time to enable your 2FA. Doing this at the start or end of the day may be best, so setting up the two-factor configuration doesn’t disrupt other tasks. Also, try to avoid days on which your users are likely to be especially busy. It’s a very good idea to plan it for during our Amsterdam helpdesk office hour 9 AM – 6 PM (GMT+1), so we can help you out of necessary. If necessary, our team can also disable the two-factor authentication on your domain.
• Enabling 2FA in Nfield won’t apply it to Single-Sign-On login. You should consult your IT department about adding 2FA to Single-Sign-On accounts.
• Set the password of the system default login “DA” to very strong and keep it in a secured place.

Whether or not you feel you need it right now, we highly recommend enabling two-factor authentication for your Nfield domain, to enjoy better security protection and gain more benefits from Nfield. If you have any questions, please contact our helpdesk. And, of course, we’re always curious to get your feedback via your account manager.

Invisible to standard internet browsers and search engines, the Dark Web is a place where users anonymously access content which is either illegal or related to illegal activities. This includes the lucrative marketplace for stolen login details, which are bought by criminals who are able to monetize the data and permissions they unlock. Among these, your Nfield login is a desirable prize, as it opens the door to a large pool of personal information.

How desirable? This can be worked out by taking a look at trading prices for the items your Nfield data helps unlock.

Dark Web Price Index

The illustration below shows the process a hacker would go through to harvest your survey participants’ login details from Nfield and make them saleable. If we work backwards from the end prize, we get an idea of the financial temptation.

According to Dark Web Price Index 2020, a stolen Gmail login (email address/username + password) sells for $155.73. A hacker who obtains your Nfield Manager login name and password gains access to all the sample information in all your surveys. That’s likely to include a high percentage of valid Gmail addresses. To make these saleable on the dark web, all the hacker now has to do is get the corresponding passwords.

Stealing passwords

Hackers commonly obtain passwords via phishing emails and database matching. Both of which are automated processes that easily deliver the goods.

Phishing Emails

Because login to Google accounts is protected by limitation on incorrect password attempts, there’s no point in trying to obtain passwords by endless guessing. Phishing, whereby a hacker approaches you via fake emails pretending to be from a trusted source and asks you to sign in to a related account – thereby revealing your password – is a far more successful method.

Every successful attempt can reap multiple payouts for the hacker. According to LastPass, 65% of people re-use passwords. A lucky hacker could net themselves an additional $74.50 if the same login details give access to a Facebook account and $50 each for Twitter and Instagram accounts¹.

¹ Dark Web Price Index 2020

Database matching

Another popular commodity traded on the dark web is credit card details, which sell for between $12 and $20 each.

Credit card details are composed of two parts. The first part is the info displayed on the card itself – credit card number, expiry date, CVV number (on the back) and holder name. These are plentifully available in a nicely formatted database, obtained through skimming devices planted in ATM machines, train ticket machines and are even sold by unscrupulous retailers who’ve been paid via card swipe machines.

The second part is the card owner zip code (derived from address), city, address, email and phone. Details which are commonly held in survey sample data! Armed with this information, a hacker can easily run a database matching tool between a purchased credit card database and your sample data to obtain complete sets of credit card details, ready to sell.

If you think this seems a lot of work for relatively modest returns, think again. Phishing and database matching are automated processes which take no more effort than a few clicks. Easy money.

How your Nfield account could get hacked

Hackers have all the tools they need to crack your password, take over your browser session and use malware to get hold of your Nfield login.

Cracking passwords using software is a basic skill in the hacking world. It’s simply a matter of time. Did you know 90% of passwords can be cracked in less than six hours².

Using shared computers in public spaces, such as cafés, airport lounges or even client meeting rooms, can lay out the welcome mat for hackers. You never know who’s hanging around waiting to inspect your activity in the event of you rushing off without logging out, or through having installed password capture software.

At the time of writing, a great many of us are working from home due to the COVID-19 pandemic. In this situation, the companies we work for usually have very limited control over the equipment we’re using for both work and recreation. It’s a vulnerability that makes it easier to implant malware that records your key strokes and can hack into your Nfield account.

² https://www.servalsystems.co.uk/6-facts-about-passwords-that-will-make-you-think/

Protect your gold mine

Although the sales values for individual login details and credit cards aren’t particularly high, they easily add up. How much do you think all the sample data held in your Nfield account is potentially worth to a criminal? Selling just one Gmail login per day would net a hacker over $56,000 a year. That’s too tempting to ignore. Which is why Nfield is now equipped with two-factor authentication, to keep hackers out of your account, even if they have cracked your password.

Learn how to implement two-factor authentication in Nfield.

At the very end of 2020, NIPO further enhanced its status within the Microsoft Partner Network by adding a third Gold Competency. Complementing already-held Gold competencies for Application Development and Cloud Platform, the addition of Gold status for Application Integration means NIPO’s team has now received the highest possible recognition from Microsoft for its proficiency in three separate areas.

This achievement underscores NIPO’s ability to provide its customers with a cutting-edge SaaS Azure Cloud based platform, leveraging the latest Microsoft technologies and fully meeting Microsoft’s standards. NIPO has earned its Gold competencies by demonstrating “best-in-class” ability and commitment to meeting Microsoft customers’ evolving needs in today’s mobile-first, cloud-first world.

Jeroen Noordman, Managing Director of NIPO: “The Application Integration competency is highly relevant to NIPO. Our Nfield platform is open by nature, so we have been very eager to demonstrate our comprehensive understanding of all the opportunities and challenges this topic presents. Attaining this third Gold Competency is also proof of our ongoing commitment to continuous investment in keeping our NIPO team members’ knowledge fully up-to-date. This recognition from the Microsoft Partner Network ecosystem benefits both NIPO and our customers. We are very proud to announce this latest achievement and look forward to having more exciting news to share relating to our MPN involvement during 2021”.

Gavriella Schuster, corporate vice president at One Commercial Partner (OCP) at Microsoft Corp.: “Achieving Gold Competency confirms that partners have demonstrated the highest, most consistent capability and commitment to the latest Microsoft technology. These partners have deep expertise that positions them at the top of our partner ecosystem, with a proficiency which can help customers drive innovative solutions”.

Surveys contain valuable, and sometimes sensitive, information. It’s therefore essential to restrict access to certain parts of surveys to those who really need it, in order to do their jobs. This works by assigning users with specific roles which only allow access to designated areas and functionality.

Roles and rights

Nfield surveys incorporate a number of default roles: Domain Administrator, Power User, Regular User, Scripter, Supervisor, Limited User and Quota Manager. Each of which allows an appropriate scope of access.

For example, Scripters can create questionnaires and publish surveys for testing purposes. But cannot publish surveys to go live, send out email invitations or access survey data.

By default, each role is granted their access rights for every survey within a domain. But there might be reasons to also restrict which surveys each user can access.

For example:

• If you have any surveys classified as highly confidential.
• If you are working with competing clients (Cola Cola vs Pepsi).
• If you want to allow a client to access real time quota progress.

How to specify individual access rights

Nfield survey access rights can be customized to fit your needs. This requires API implementation and is an add-on to your Nfield subscription.

To explain the set-up, we need to look at how Survey Groups work.

A Survey Group is a container for surveys in which users are assigned their access. By default, existing and newly created surveys are put under the Default Survey Group, so your existing and newly created users are given the access specified in this Default Survey Group.

To demonstrate, let’s say you have a survey about Cola, which is highly confidential and access must therefore be restricted to an individual Scripter. This would need to be set up as shown below, with a new container called “Survey Group Cola” that contains this Cola Survey and only specifies access to your designated Scripter.

In terms of API calls, this breaks down into the following steps:

Step	API call
1) Create the group “Survey Group Cola”.	POST v1/SurveyGroups
2) Move the Cola survey from “Default Survey Group” to “Survey Group Cola”.	PUT v1/Surveys/{surveyId}/SurveyGroup
3) Assign the Scripter to “Survey Group Cola”.	PUT v1/SurveyGroups/{surveyGroupId}/AssignLocal
4) Unassign the Scripter from “Default Assignment”.*	PUT v1/SurveyGroups/{surveyGroupId}/UnassignLocal

*remarks: One user can be assigned to multiple survey groups. So you may skip this step to keep the default survey group access according to your wish.

The following illustration demonstrates this using Postman for executing the API calls.

Putting you in control

As described in ISO 27001, “Put simply, access control is about who needs to know, who needs to use and how much they get access to.” It’s something people are always talking about on our own work floor. This Nfield feature gives you the control you need to restrict access to survey rights as appropriate, to safeguard data security while enabling people to do their jobs.

In an ideal world, every survey you do will begin with a perfectly set up quota frame structure. Reality, however, can be a different story. Even after the most considered planning, a need for changes sometimes only becomes apparent once survey fieldwork has got underway. Nfield gives you the ability to edit quota frames even after fieldwork has begun, while retaining valid responses already collected.

Reasons for editing quota frames

Just because you can edit active quota frames, that doesn’t mean you shouldn’t strive to get them right from the start. This will create the smoothest ride. However, mistakes and unforeseen circumstances still happen. The most common scenarios which require quota frames to be edited are:

Spellings
Quota variable and items names (e.g. Leisure Activity Type – Dance / Football / Volleyball / Tennis) need to be spelled exactly the same as they are in the survey script. It sounds simple enough, but mismatches can still happen if:

• UK and US spellings and terms are switched out. E.g. Football (UK) vs Soccer (US) / Colour (UK) vs Color (US)
• Brand names change. E.g. Apple Computers becomes Apple Inc. / NIPO Software becomes NIPO
• Additional space breaks have been created when copy-pasting
• Scripting errors have resulted in incorrectly entered item values

Additional brand names
In surveys which gather information about competing brands, the number of brands specified is usually kept to a minimum to keep costs down. However, as the survey progresses, the selection may prove insufficient. E.g. Netflix is surveying rivals Videoland and Amazon Prime Video. When Disney Plus starts emerging as a strong competitor, this needs to be added.

Inappropriate structure
After a survey has got underway, it may become apparent that the quota frame is too complex, requiring responses from specific groups which are too difficult to achieve. E.g. region-by-region targets may have to be abandoned to a broader geographic grouping.

How to edit a quota frame in an active survey

You can access quota in two pages – Quota tab under Setup Survey; and Quota tab under Monitoring Survey. Let’s see how it works.

Quota tab under Setup Survey

Setting up quota frame is now available after fieldwork. You can go back to previous steps to define quota variables and order and nest quota variables, same way as in the first time setup.

Quota frame editing works with a concept of version. When the quota frame is changed, it stores as a new version. It needs publishing to make it effective. Successful interviews are now added to its corresponding quota frame version. It implies that succesful interviews before this change is counted in version 1.

Version 1:

Video Streaming Providers	Target	Successful Interviews
Netflix	40	30
Videoland	40	30
Amazon Prime Video	40	30

After changing the quota frame (e.g. adding Disney Plus), this quota frame is now saved as a new version (thus version 2) to start with. Depending on your scenario and the purpose of changes, you may want to adjust the target. In this case of adding a new brand to catch up with the other existing brands, target of each brand in the new version is adjusted as Target minus Successful Interviews of version 1.

Version 2:

Video Streaming Providers	Target	Successful Interviews
Netflix	10	0
Videoland	10	0
Amazon Prime Video	10	0
Disney Plus	40	0

As a general rule of thumb, check all the quota targets to make sure they are updated as you wish and remember to update the ssurvey script before publishing.

Quota tab under Monitor fieldwork

This page allows you to edit quota targets and monitor quota progress. The addition is a timestamp of quota frame version. By default, it shows the latest quota frame version and its successful interviews added in since its last publish. You can also select other versions to show its quota progress.

Conclusion

Editing quota frame during fieldwork is a very powerful feature to adapt changes easier. The change should be carefully managed. We hope it benefits you. If you have any feedback on this feature or any challenges you have, please feel free to contact us.

To uphold the quality of your CAPI and Online research, you need to ensure survey results aren’t contaminated by ‘bad’ interviews. This calls for identifying and disqualifying responses which show evidence of being falsified.

Keeping it real

Simple pointers to unreal answers and/or respondents include short interview duration, short time spent on each question, short or meaningless open-ended answers and lack of cohesion between answers to related questions. Another giveaway in CAPI fieldwork could be when a particular interviewer’s work is completed via the shortest-possible route.

Methods for identifying these pointers in Nfield CAPI interviews are explained in CAPI Quality Control: Audit Trail helps you eliminate falsified fieldwork and Nfield’s Quality Control Options for Rock-Solid CAPI Data.

In addition to identifying suspicious interviews from interview completion data, you can also adopt a practice of contacting certain respondents to verify their input.

Setting your threshold

Every company or survey project has its own criteria for whether an interview should be classed as ‘bad’ and therefore disqualified from results.

Samples from panel providers are generally trustworthy, as panel providers deploy their own mechanisms for maintaining an honest and active respondent pool. But if your survey is conducted outside the confines of a panel, there is a higher chance of contribution from respondents who are just making mischief, possibly to get promised rewards, whose answers are not genuine.

To determine where to set your tolerance, you need to get to know your own typical response patterns for the criteria described above (under “Keeping it Real). When answers fall outside of these, it’s time to raise a red flag.

Throwing out bad apples to purify results

Just as bad apples need to be disposed of to avoid the whole crop becoming unusable, ‘bad’ interviews need to be disqualified from survey results in order to come away with genuinely useful insights.

This is easily done in via the Nfield Manager for Nfield Online and Nfield CAPI.

Once you reject an interview, it is automatically removed from all counts related to successful completes and corresponding quota cells. So disqualification of ‘bad’ interviews may mean continuing your survey a bit longer to reach all the required targets.

Disqualification process

In Nfield Manager, go to the Quality Control tab. Here you’ll find a Qualification Control overview, as well as each individual interview’s validation status.

To disqualify an untrusted interview from survey results, simply select the interview ID and click Reject.

As an alternative to rejection, you can carry out positive selection by approving individual interviews. And if you aren’t sure whether to approve or reject an interview, you can un-verify it, indicating a requirement to conduct further checking.

Clicking “Reset” will restore all validation states to “Not checked”. This will include any interviews which had been set to “Rejected”, thereby re-including them in corresponding quota counts as successful completes.

Automated disqualification

If you want to automate identification and disqualification of untrusted interviews, this can be achieved by integrating appropriate tools with Nfield via an API. See API – Developer’s guide to learn more about Nfield API.

Downloading Rejected Answers

You can download a record of all rejected interviews in the same way as you do for other interview records.

If you have any questions or comments, please do not hesitate to contact us.

It is estimated that half of the world’s population uses two or more languages in daily life. This is no surprise to us, as 91.6% companies of our customers live in countries where the primary language is not their native one. With so many people having multilingual backgrounds, it’s a good idea to offer survey respondents a choice of languages, with the option to switch part way through.

This is because respondents sometimes choose to begin answering a survey in the local language, even when it’s not their native one, but start to struggle if terminology gets more specific. This could happen if questioning goes into detail about things such as medical issues.

Respondents can often deal with this using Google Translate, but that’s not ideal. Enabling language switching within the survey itself provides a far easier and more reliable experience, as you can control terminology and phrasing by embedding approved translations.

Switchable multi-language surveys In Nfield

Nfield supports multiple languages with ease, allowing respondents to choose their preferred language at the start of the questionnaire. In the case of known individual preferences, these are automatically offered.
Thanks to a recently introduced feature, Nfield also now lets respondents switch language at any point during the survey. It’s as easy as turning a dial!

To create a multi-language survey in Nfield, you simply have to append the translations with the main script. The language switching feature is enabled by specifying the language codes. So, if the questionnaire is offered in English (“eng”) and Chinese (“chi”), the command should be added like this.

*UIOPTIONS "languages=eng,chi"

When you have more translations, their language names should be added in the above with a comma separator.

The Nfield survey will automatically add a drop-down box on each page, allowing respondents to switch at any time without disrupting their completion of the survey.

If you have any questions about Nfield’s support of multiple languages within a survey, please feel free to drop us a line.

In our documentation there is information about the NIPO CATI Client configuration, but in this news item you will find a quick overview on how to set up your CATI system in such a way that you can have your interviewers connecting from any location. Very useful if you want to give your interviewers the flexibility to work from home, or to give a partnering firm access to your CATI system so you can do projects together while working on the same server.

So, what are the steps to set up CATI@Home? To start with the obvious, your CATI@Home interviewers need an internet connection to be able to connect to the NIPO CATI Master. Since they will connect through the internet, the connection protocol for the NIPO CATI Client needs to be set to PureTcpIp. We advice you to place your CATI Master behind a firewall in which you open the specific port of the NIPO CATI Master. In a default configuration this port is 8001. The port the NIPO CATI Master is listening to can be changed from the CATI Manager › File › Configure › Master.

Settings on the CATI Master machine
Remote CATI Clients cannot simply connect to the CATI Master, but require additional verification in the CATI Master. The CATI Master will keep a list of so called StationKeys for which it will allow incoming remote connections. Connections are recognized as remote connections based on the IP addresses of NIPO CATI Master and NIPO CATI Client. The StationKey of the incoming connection needs to match a known StationKey value on the CATI Master.

In order to arrange this additional verification you will need to add some registry settings to the CATI Master. These registry settings are all stringvalues to be added under \HKLM\SOFTWARE\NIPO\CatiMaster or in case of a 64 bit operating system under \HKLM\SOFTWARE\Wow6432Node\NIPO\CatiMaster.

Add the following registry settings:

• StationKeyDatabase=NipoFieldworkSystem
  This is the database in which you store the StationKey values.
• StationKeyTable=Interviewer
  This is the table in the database where the StationKey values are best stored. By using the Interviewer table you can easily use the NIPO FMS Client as an interface to configure the values per connection.
• StationKeyTableKey=AuthenticationKey
  This is the fieldname in the interviewer table where the actual StationKey per connection is stored. If you choose the fieldname ‘AuthenticationKey’ be aware that you cannot assign values to this field manually, but the StationKey value will then be derived from the interviewer’s SocialSecurityNumber. If you do not use the SocialSecurityNumber field, you may find it easier to manually add a new nvarchar field to the interviewer table in your database and use that field instead as the StationKeyTableKey.
• StationKeyTableStation=AuthenticationStation
  This field is used to store the computer name of the remote NIPO CATI Client machine. Initially this field is empty, but after the first successful connection through CATI@Home setup, this field will be filled with the computer name of the interviewer machine. Consecutive connections will not only check if the StationKey of the interviewer matches the one known to the NIPO CATI Master, but also the computer name must match. This way you can make sure the interviewer always connects from the same machine. In case the interviewer would need to use another computer at home, you can simply remove the computer name for the interviewer in the NIPO FMS Client.
• StationKeyTableStatus=AuthenticationStatus
  This is an optional field. When used it allows you to easily toggle the activation / deactivation of the specific StationKey. Once configured you need to set this field to 1 to allow an incoming connection using this StationKey. Setting it to 0 would disable that StationKey. You could use this for instance in cases where an interviewer will not be working for you for a certain period, but might return at a later stage.
• StationKeyTableRemoteTelnr =RemoteTelnr
  This field is only relevant in case your using a centralized Dialer system with the NIPO CATI Master Dialer API. This allows you to specify the home telephone number of the interviewer which the dialer should use to call the interviewer. Note there are three ways to specify the telephone extension of the interviewer and they are handled in the following order: an argument on the command line of the NIPO CATI Client with -TELNR, in the RemoteTelnr field in the database and configured as extension in the NIPO CATI/Web manager.

Restart the NIPO CATI/Web Master
After adding the registry settings and checking the port number of the NIPO CATI Master, you will need to restart the NIPO CATI/Web Master service to activate the new settings.

Prepare a remote CATI connection in NIPO FMS Client
Open the NIPO FMS Client and go the interviewer you want to enable to connect remotely. In the ‘Details’ view you may need to change the view to actually show the Authentication fields. Right-click on the page and from the popup dialog select ‘Fields…’ This brings up a dialog in which all fields of the interviewer table are shown, but only the ones ticked are shown on the interviewer’s ‘Details’ page. Make sure to tick the fields for AuthenticationKey (or the fieldname of the field you added to the Interviewer table), AuthenticationStation and AuthenticationStatus. And when a dialer is used tick RemoteTelnr as well. Then click ‘OK’ and the fields are shown on the interviewer’s Detail page.

• If you make use of the SocialSecurityNumber, you can click the button ‘Create Key’ next to the AuthenticationKey field. This creates a unique key of 256 characters for that interviewer.
• When you defined another field as the AuthenticationKey field, you can type in a unique Key yourself in that specific field.
• set the AuthenticationStatus to 1.
• leave AuthenticationStation empty.
• when using a dialer system: specify the remote telephone number of the interviewer.

Configure the remote CATI Client application
The remote CATI Client machine uses the normal CATI Client software with in addition the value of the StationKey. You have to define yourself what is the best way to get the software on the interviewer’s home computer; you could for instance prepare a client installation including the shortcut and make a self-extracting zip file and email/ftp that to your interviewers.

There are 2 ways you can specify the key on the NIPO CATI Client; either by adding a parameter to the shortcut: -KEY xxxxxx, or by adding an entry Settings=xxxxxx to the [OdQes] section of the niposys.ini file (where xxxxxx represents the actual AuthenticationKey as specified for the interviewer). Since this Key value can be a long string, it is preferred to add the settings=xxxxxx to the niposys.ini file. Note when the key is used on the command line it should be present as the first argument.

The shortcut to the CATI Client then looks like this:
"C:\your path here\OdQesu.exe" /N 100.100.100.100:8001 /P PureTcpIp
Of course, 100.100.100.100 should be the IP address of your NIPO CATI Master, and 8001 is the port number as defined above. And you may also use the other normal parameters like ? /SIZE depending on your preference.

Test the connection
By itself, the above should enable access for the remote interviewer, so go ahead and try to connect. After successful contact to the CATI Master, check the interviewer details in the NIPO FMS Client, and you will see the computer name in the AuthenticationStation field. The workstation is now successfully configured for CATI@Home use!

Remark: Even though we use the interviewer table to store the StationKey values, there is no specific link between the interviewer number and the StationKey itself. Using the StationKey is only a control mechanism to determine if a certain remote NIPO CATI Client is allowed to connect or not. Once the connection is established, the interviewer could theoretically still logon with any interviewer number. It is however good practice to use StationKeys as described above, because each StationKey will then only be used by 1 interviewer which makes management of the StationKeys much easier.

Good quality CAPI research relies on trustworthy interviews. You need to be sure fieldwork integrity isn’t compromised by interviewers taking shortcuts, submitting false responses. Nfield CAPI now builds on standard verification checks with an Audit Trail / Logging feature that highlights suspicious input patterns. By analyzing statistical details, this enables you to identify when interviewers are likely to be cheating the system, and take appropriate action.

Building on the basics

Traditional quality control components such as field validation, logic checking, ordering and randomization are, of course, all supported in Nfield. For face-to-face CAPI interviews, additional features such as GPS locations, photographs and audio recordings can also be activated to provide evidence of interviews taking place. (More information about these can be found at Nfield’s quality control options for rock-solid CAPI data.)

All this gets you off to a great start, but still leaves room for cheating. To further close off the possibility of falsified responses, you need to look more closely at response input patterns. Nfield CAPI’s Audit Trail function enables you to spot the tell-tale signs of dishonest interviewer behavior, such as:

Within an interview:

• Short overall interview time¹
• Long answer time for a simple question¹
• Short answer time for a difficult question¹
• Interview happened in an incorrect location²

Within an interviewer’s individual behavior:

• Gap between two interviews is shorter than the expected travel time between locations²
• Completing an overly high number of interviews in a day for the interviewer’s known ability²

Within an interviewer’s comparable performance:

• Much shorter interview times than average¹
• Higher-than-average percentage of inconsistency between GPS locations and respondent addresses²
• Higher-than-average percentage of unentered respondent tracing information (failure to enter phone number, personal details, etc.)³
• Lower-than-average number of visits²
• Higher-than-average number of interviews/visits per day²
• Significantly higher or lower numbers for project-specific tasks. E.g. collecting samples of hair for laboratory analysis as part of a haircare survey.

The data for annotated items 1, 2 and 3 can be found in the Nfield downloads:

• ¹ the new audit log file ending with “-auditlog.csv”
• ² paradata csv file ending with “.csv”
• ³ survey result data file ending with “.dat”

Comparing performance via Audit Trail

Nfield CAPI’s new Audit Trail feature records how long an interviewer stayed on a page and which button they clicked to move on. The results are provided in a csv file. The example below shows how this might look for a single interview.

You can compare each result line against your reasonable expectations and set benchmarks to flag up interviews that fall outside these parameters.

The best method for reviewing this is to build a report (in Microsoft PowerBI or other reporting system) to show suspicious interviews. Incorporating a drill-down function will give you an easy way to view more information.

The two illustrations below show how your reports might look.

Here, the results are sorted to show interview durations in descending order.

Intelligence at your fingertips

Armed with these insights, you can better fortify the quality of your CAPI surveys by eliminating falsified responses. Not only can you challenge interviewers whose input appears suspicious, and remove fake responses from your results, you can also use this feature to deter dishonest interviewer behavior from taking place at all.

And in case you are wondering, this Audit Trail feature will also be introduced for Nfield Online in the near future.

Have you ever concluded a survey to discover it’s generated many more completes than specified in your quota targets? This is called Quota Overshoot, and it happens when a lot of respondents fill in a survey at the same time. Quota Overshoot doesn’t reduce a survey’s quality, but it does result in higher-than-necessary expenditure on reward, sample and interview costs.

The good news is Nfield’s Max Overshoot feature enables you to control the number of excess completes, and so avoid unnecessary costs.

Why Quota Overshoot happens

To understand how Max Overshoot works, you first need to understand why Quota Overshoot happens. The one-word technical answer to this is ‘concurrency’.

Concurrency is when two or more respondents are filling in a survey at the same time. For the most part, this is exactly what you need to get your survey completed in the shortest possible time. But if multiple responders remain active when just one more complete is needed, you’ll end up with more completes than the quota. This is because all the interviews already underway at the moment of the last required completion will also continue to completion

How Nfield solves this

Nfield’s Max Overshoot feature gives you control over excess completes by limiting concurrency as your survey nears completion. You can set it to reflect your preferred balance between speed and excess respondent costs

The Plane Game

To explain how this balance shifts, let’s play the Plane Game. Or in these COVID-19 times, imagine playing it!

THE GOAL:
Get 20 paper planes made, each by a different person. Of these, 10 have to be made by males and the other 10 by females. (Want to know how to make a “world record paper plane”? This video shows you!)

THE RULES:
Each plane-maker has to sit at a separate desk while performing the task. When finished, they vacate the desk for a new plane-maker to take their place. This all happens inside a room that players can only enter when a free desk is available. Players who’ve completed their planes stay in the room. Players who decide to give up have to leave the room. When you’ve reached the target number of planes, no more people can come in.

Everyone in the room at the moment the last required plane is completed gets a voucher worth $8 as a reward, even if they’re still finishing their plane.

SCENARIO 1:
You invite a lot of people (let’s say 100) to participate and provide the same number of desks. This way, everyone is making planes at the same time, with nobody having to wait their turn.

This will produce a really fast result. But because you had 100 people in the room, and none of them had given up at the point the 10+10 target was reached, you have to pay them all. That’s a bill of $800.

SCENARIO 2:
You have two rooms, each with just 3 desks. One room is only for the males, the other only for the females. Both rooms operate at the same time. Once 10 male planes have been completed, the male game ends. The same applies for the female game.

This limits each category’s concurrency and, as a result, the number of participants in both rooms combined when the targets are reached will only be 24. An excess of two males and two females. The total reward payout will be just $192. But the game will have taken much longer to complete.

Finding the right balance

Of course this Plane Game is a metaphor for respondents completing surveys, with the two scenarios illustrating extreme ends of the cost vs speed spectrum. As a market researcher, you’ll probably be looking for a happy medium. Nfield achieves this via a formula which changes the number of “desks” as each quota target comes closer to completion. All you have to do is set your desired Max Overshoot number per quota target.

At the start

Number of desks = Quota target – Successful completes + Max Overshoot number

If we illustrate this according to the Plane Game, per quota it looks like:

Number of desks = 10 planes to be made – 0 completed planes + maximum 2 excess participants

This means we start with 12 desks per quota target.

One new participant is allowed to start after each new complete ONLY IF Successful Completes < the quota target. (Nfield doesn’t allow new interviews to be started once quota target for successful completes has reached its maximum.)

As completes start to accumulate, the number of desks gets reduced. This limits the number of active participants and, in turn, the number of excess participants at the time each quota target is reached.

Quota target = 10
Max Overshoot = 2

Number of successful completes	Number of desks available	Maximum number of active participants
0	12	12
1	11	11
2	10	10
|	|	|
8	4	4
9	3	3
10	2	2

What about setting the Max Overshoot to zero?

You can set the Max Overshoot to zero to keep costs to an absolute minimum. However, it will take longer to achieve the final complete as, by the end stage, only one respondent can be active at a time, which will amplify the effects of any drop outs.

You can make a judgement call on how seriously this might impact things by looking at your survey’s dropout data in Nfield.

• Drop-out rate: The higher the drop-out rate, the higher the chance of a longer wait for your final complete.
• Drop-out point: If drop-outs typically occur later in your survey, after a higher number of questions have been answered, the amount of time wasted due to drop-out is likely to be higher. Note that Nfield sessions automatically time out if there’s no activity for 15 minutes.

Ensuring value

Two important points to note when using Max Overshoot:

• Although Max Overshoot will limit the costs associated with going over quota, it still results in panellists being wasted when samples are flooded to fill quotas. Best practice is to use minimum and maximum targets in a quota structure that’s designed as simply as possible, in order to reduce or eliminate the need to flood the sample.
• Max Overshoot works in alignment with the maximum quota target. Its use is only beneficial if the target is reached.

Available on demand

Max Overshoot can be a very beneficial feature for controlling costs and provides flexibility to balance cost against speed. However, it does make quota evaluation more complex and may put a big load on Nfield processors. We are therefore rolling it out gradually and are currently only enabling it upon request. To enable Max Overshoot, please contact your account manager. At one point, we will enable this feature for all domains.

If you have any questions about Nfield’s Max Overshoot feature, don’t hesitate to Contact Us.

Quota management is important in fieldwork management. It ensures you have a representative demographic distribution and a good control on the cost (sample fee, respondents’ rewards, Nfield complete fee, etc). It starts with a simple concept – You set a limit of how many completes (5 males + 5 females = 10 total) should be done to meet your quota requirement. In this article we introduce the quota basics and variations and how you manage your quota well.

Let’s start by creating an example for the basics using Minimum Quota.

At the end of the fieldwork, when you are missing one final interview (the 10th respondent) needed for the quota target there’s a chance you may not get this straightaway (a female in this case, see the image below). It would be a waste to interview respondents and keep screening them out (3 males in this example) before the last female respondent is found.

Minimum quota

When your quota requirements get more complicated (e.g. a female of 20-25 year old living in a small city who purchased a specific wine in the last year), we would need to screen out even more respondents. To have more flexibility in the quota design, you can make use of minimum targets.

In this example, we fixed 80% of the required gender distribution and left 20% of the sample free in the choice of gender. Anyone can take the final two spots (20%) of the target. The principle behind this¹ is to fulfil the minimum gender target (so one female spot in grey is reserved) and limit the total target by setting a maximum. So once the minimum targets are fulfilled, it is more flexible for accepting the last ones instead of screening them out.

Benefits of this use of minimum targets are:

• It shortens the fieldwork period, no need to wait for the best-fit person.
• It does maintain a good demographic representation.
• It limits the amount, and possible cost, of sample that is used in the project, where members of this sample may be contacted only once in every three months.
• It saves on the sample cost where respondents who got screened-out get rewarded.

¹ Remarks: When it comes to a choice (overshoot due to concurrency issue), it is more important to fulfil the minimum target (4 male and 4 female) and relax on the maximum total target (so it is overshoot).

Maximum quota

Quota requirements can be complicated. You may have different quota cells, interlinking with each other. Some quota cells are more important than the others. Together with minimum targets, you can have good grip of your quota control.

We revise the minimum quota example with smaller minimum targets.

Nfield ensures 3 minimum males and 3 minimum females and has 4 spots to be filled in by anyone. In extreme cases, it can be 7 males or 7 females.

Having 70% males may seem too much. To add a control on this, maximum targets can be added on the quota cells.

To understand this, we now add a border (maximum target) on how many circles can sit in one gender row. 3 males and 3 females are minimum targets. They must be filled. The total maximum target is 10 (ten circles in total), meaning four circles/spots are available. However, only three circles can sit in one row (in the male row or the female row). Then you can have 3 to 6 males or 3 to 6 females, totaling 10.

Benefits of a mix between minimum and maximum targets are:

• It keeps a minimum requirement on the demographic representation.
• It allows flexibility on the rest of the quota cells.
• The maximum targets can avoid extreme imbalances in the outcome.

Least-filled quota

This allows the survey to use routing based on the least-filled quota or offer selections based on a list ordered by least-filled quota. It helps lagging-behind quota to catch up by giving them advantages on higher position in the answer list or on routing to their sections. The following scenarios may sound familiar.

• A long questionnaire often is cut into sections and respondents are asked about only one of these sections to avoid boredom. To determine which section to show, a random test may be used. A random selection statistically makes an even distribution, especially when it is in a big-scale usage. However, in cases with imbalanced targets, you may need a better distribution method to fill all targets quicker based on least-filled quota instead of using a random selection.
• Sometimes, in the survey it is required to show a list on a question page to gear respondent selection towards the least-filled quota item.

The how-to is quite simple, with the use of command *GETLFQLIST (which literally means Get Least-Filled Quota List). It returns the quota list in its ascending relative filling percentage based on minimum target values. With the following example, the relative filling percentage is calculated like this: 3 completes on electric appliance kettle out of 5 targets, giving its relative filling percentage as 60%. The least-filled quota list will now show the order: Air conditioner (the least filled), refrigerator, kettle, and last television. You can then route your survey to show the section of the air conditioner.

The most needed quota item gets answered first, resulting in earlier completion of the quota and saving costs by using less sample.

Multi quotas

Least-filled quota can apply to multiple quotas too. You may ask “Which electric appliances have you used yesterday?” or “Which brands you have purchased in the last months?” Respondents can give multiple answers. The questionnaire can route to show one/more section(s) of the least-filled answers (appliances/brands) of his/her answers. Or it shows a follow-up question with his/her answers in an ascending relative quota filling. The most demanding quota is at the top, which hopes to gear the respondent’s selection to that item.

In Nfield, you can toggle the “Multi” button on in the Quota page.

And multi quotas should always be at the root (thus the lowest in the nesting quota frame). It can be standalone, or nested under another quota cell, but not having other quota cells under it. As one respondent gives multiple answers, the counts would not sum up nicely. In the following example, the minimum target is 10 male and 10 female with a total of 20. These numbers are controlled gender quota variable, not in the minimum targets set for each appliance. The minimum targets in each appliance are used as the calculation of the relative filling which is 100% x (minimum target – number of respondents in that answer). Another important note is that the multi quotas do not contribute to quota check itself.

Then this *GETLFQList command can be used to return its result.

When the quota status is like the following (same as the example in least-filled quota), the relative quota filling in ascending order is air conditioner (20%, the least filled), refrigerator (57%), kettle (60%), and last television (67%).

See this in action:

With the spread of coronavirus (COVID-19) disrupting daily life all over the world, we’ve noticed the changes in human activity being reflected in Nfield surveys. As regions have gone into lockdown and people have been discouraged, or even ordered, to avoid contact with others, CAPI interviewing has become all-but impossible in some places. Where this has been the case, there has been a significant increase in Online surveys to compensate. To illustrate, we’re sharing usage patterns for Nfield CAPI and Online in our China, South Korea, Spain and Vietnam deployments, so you can see how survey execution has changed along the coronavirus timeline.

While we are, naturally, as concerned about the situation as everybody else, we are pleased to see that our customers have been switching between Nfield CAPI and Online without any problems. This is because we developed these two survey channels with the same scripting language and result format. Switching can therefore be done in just a few minutes, with minimal support needed from our helpdesk.

China

Nfield CAPI vs Online in China
At the time of writing, China remains the country most heavily impacted by coronavirus (COVID-19). This is reflected in a uniquely dramatic shift in survey channel usage. In normal times, CAPI very much dominates China’s survey activity. But with public spaces mostly deserted, and people being reluctant to interact with researchers, face-to-face interviews have almost completely ceased. Meanwhile, Online surveys have increased significantly to fill some, although not all, of the gap.

The correlation between Nfield usage in China and events on the coronavirus timeline clearly confirms how these are linked. A decrease in survey activity before long holidays such as Chinese New Year, which began on 25 January 2020, is common. Our graph shows an expected reduction in CAPI fieldwork leading up to this. Survey activity remained extremely low while the Chinese New Year holiday was extended to 2 February, due to the disease. As people gradually started returning to work in Beijing/Tianjin/Hubei/Sichuan, albeit from home, survey activity resumed on a very small scale. After the first ten days this increased to some extent, but almost exclusively via Online.

South Korea

Nfield CAPI vs Online South Korea
As of 5 February, there were fewer than 20 confirmed cases of coronavirus in South Korea, although the gradual increase in neighboring China was starting to cause alarm in other countries. By 7 February we were seeing a drastic decrease in CAPI face-to-face interviewing, while use of Nfield Online grew to twice its normal amount. As widespread infection took hold in South Korea, Online survey usage tailed off again to normal levels. Meanwhile, CAPI diminished greatly, but not completely.

Spain

Nfield CAPI vs Online in Spain
Spain’s Nfield usage pattern is very similar to that seen in South Korea, although the early February switch from CAPI to Online happened sooner and more drastically than in South Korea. In Spain, a 3-day Online spike suddenly dropped off again on 13 February, after which there was a reduction in both CAPI and Online. CAPI continued to play a diminished role in Spain’s survey landscape until the last two days of the month.

By 5 March (a week after CAPI all-but disappeared from use), the Spanish government advised companies to send workers home to reduce contact. On 6 March, Spain ranked 7th in the world for the number of confirmed cases. We expect to see the impact of these measures in March volume reports.

Vietnam

Nfield CAPI vs Online in Vietnam
Thanks to prompt and decisive governmental action, Vietnam did a very good job of containing the spread of coronavirus and preventing it from getting out of control. Like China, Vietnam had a relatively long new year holiday. However, the Vietnam government declared coronavirus to be an epidemic at a very early stage, on 1 February, when the number of confirmed cases stood at 6. As a result, Vietnam only had 16 reported cases, with the last one declared on 13 February. Usage patterns for both Nfield CAPI and Nfield Online very quickly returned to normal when new cases stopped being reported.

A WHO official, called Park, told Al Jazeera¹: “The country has activated its response system at the early stage of the outbreak, by intensifying surveillance, enhancing laboratory testing, ensuring infection prevention and control and case management in healthcare facilities, clear risk communication message, and multi-sectoral collaboration.”

¹ _{https://www.aljazeera.com/news/2020/02/infected-patients-vietnam-cured-coronavirus-miracle-200228035007608.html}

We proudly present our Nfield Top 15 Customers! We would like to take this chance to give them a round of applause and to recognize their project success with Nfield. Conducting projects in Nfield means they have also put security and data compliance to their top priority as we do. 

These top 15 customers are selected based on their usage in 2019. And the winners are (in alphabetical order):

Amárach Research	AmPuls Market Research	Awe Research
CTR Market Research	Elka Consulting	Hendal Market Research
Kantar (global)	Kantar BBSS SEE	Kantar Hoffmann
Kantar Info Research Austria	Kantar Worldpanel	Media Research Consultants
MERCAPLAN	TNS Ilres	Vital Research