Information Technology

After hearing from a colleague about inconsistencies she had experienced with not getting Qualtrics’ email notifications, I started investigating this issue. I looked through a huge number of Qualtrics knowledge base articles (XM Support), confirming that we were correctly setting up our notifications however couldn’t identify or explain the cause of their failures.

Not finding any concrete answers, I started exploring the XM Community, where other Qualtrics users share tips, tricks, or solutions. I came across an entry that was surprising but also an answer to my struggles with the issue at hand since the inconsistencies were so random and well inconsistent!

“… I’ve run into a few gliches with emails not sent using the email trigger. Qualtrics support last month said the most updated and reliable email trigger is the one set in the Actions Tab…”

I used the Action Email Task and tested the submission notification and it worked beautifully. After my test results supporting the functionality of the Action feature I switched the form’s notification from Email Trigger to Action Email Task.

So I wanted to share this experience since we all utilize Qualtrics for our various survey, forms, and feedback needs and a breakdown in notifications would impact the services we provide to our colleagues and constituents.

So, please use Email Tasks for your future notifications and update your existing forms and surveys if receiving submission notices are important to you:

• https://www.qualtrics.com/support/survey-platform/actions-module/email-task/
• https://www.qualtrics.com/support/survey-platform/actions-module/setting-up-actions/#CreatingConditions

If you ever need to get data from Twitter, and do not want to deal with coding, there is a neat application that you can use: Hydrator, a desktop application that takes in tweet IDs and returns the corresponding data from Twitter as JSON. Hydrator handles the Twitter API rate limits for you, and allows you to pause and continue the downloads if desired. Users have the option to convert the data to a CSV after the downloads are complete.

There’s just one catch: you do have to have tweet IDs to feed into the application. If you’re lucky and looking for a dataset for or from a common source, you may be able to find a set of IDs online. Otherwise, however, the Twitter API will still be your best bet to query and retrieve a large number of IDs.

Here at Ohio State, we can select between REDCap and Qualtrics for our survey collection needs. Are there important distinguishing factors to note between the two?

REDCap allows for more granular access rights to be granted for datasets, along with comprehensive logging and audit trails. Changes to data are more easily tracked using these features. Because of these features and on-premises storage location, REDCap is more suited to use with S4 data, and is often more suitable for research project. Qualtrics, on the other hand, offers an intuitive and easy-to-use UI. It’s often best suited for small-scale research projects and surveys.

A simple way to decide would be to decide if some of REDCap’s special functionality is necessary for your project. If not, Qualtrics may just do the job.

It’s true: R users have a mature, well-maintained development environment — RStudio. Whether you’re an R user looking into Python, or a Python user looking for options, there are good news for you.

The Jupyter Notebook is a popular, web-based development environment in which users can write, annotate, and run code not only in Python, but also R and many other programming languages. It’s interactive, in that you can run code and view the output incrementally as you write the code. Jupyter Notebooks also support displaying visualizations inline, which is an important feature for data science and related applications. The ability to interweave comments and visualizations in a visual environment where you can write code, run it, and view the output makes Jupyter Notebooks ideal for explaining, presenting, teaching, and collaborating on code.

Example of a Jupyter Notebook from DataQuest blog

And more good news: it’s free and open source. You can download and run your own instance today! If you’d like to first try it without installing, you have the option to do so from the official Jupyter website here.

Since the emergence of data science on the mainstream, a popular visualization has been circulating in an effort to categorize where it fits in among existing subject areas. Below is an example of this visual – the “Data Science Venn diagram”, which I found on datasciencecentral.com:

Data Science Venn diagram

This particular one has not commented on the intersections between the areas, but you will find other similar ones that do. Traditionally, it could be said that:

• Computer Science + Math / Stats = machine learning, scientific computing, and certain big data applications
• Computer Science + Domain Knowledge = “effective” software development
• Domain Knowledge + Math / Stats = quantitative research

If you have data science related questions, we would be happy to assist you with guidance at the Research Commons.

Have you ever wondered why web links use characters such as hyphens and underscores over spaces? There is a very valid and practical reason for this.

While humans use spaces in text to split parts of speech to make it more readable, this practice causes all sorts of issues for computers. For computers, spaces are instead used to separate parts of code. For example, on the command line, spaces separate commands and symbols from each other. If a command were two words separated by a space, the command line interpreter would not know to process that accordingly. 

In the old days of the internet, CGI scripts were used to run code on the command line to power the Web. Often, links on a website would correspond to a file path on a server that the user would access. Because of this, the same naming conventions carried over.

Despite the fact that the Web has seen significant changes since those days, the problem with spaces (as well as some other special characters) remains, and we continue to avoid these “nasty” characters to make our lives easier by using simpler, albeit less-readable URLs.

In this post, we will explore equivalents of data types between the two popular languages of data science: R and Python.

Despite the fact that R has traditionally been the language favored by statisticians for data analaysis, Python has emerged as a serious contender in the last few years. Being a flexible, general-purpose programming language, Python has drawn attention from the community especially with the development of two particular packages, Numpy and Pandas. These packages offer support for the R-like data types, a list of which you can see below:

Have you ever come across a coder working? Perhaps while getting morning coffee at a Starbucks, you may have seen a computer screen like this: white text, black background, and lots of words that probably make no sense to a bystander. But why is it that these people often work on Macs? To explain, I have to go into some history on programming.

Around 1970, the world received two technological developments that shaped programming of our modern day. These were the lingua franca programming language C, and the popular operating system UNIX which was written with it. Since then, C has become and remained one of the most popular languages. UNIX, however, was a proprietary system and was not accessible to the common user. For that reason, groups such as the Free Software Foundation (FSF) started rewriting free versions of UNIX software in the 1980s. These efforts were combined with the Linux operating system in the early 1990s, and became the popular open source GNU/Linux ecosystem of today, just in time for the dotcom boom.

When the World Wide Web spread, open source software popularized in tandem due to its accessibility and availability. Examples included the popular web server Apache, the web programming language PHP, and the open source database MySQL. This software, again due to availability, was used on GNU/Linux and together made up the LAMP stack. These technologies continue to power a large percentage of web applications today.

When developers work on a local copy of code on their computer, they try to mirror the components of a live system closely for compatibility. So in part, the preference is due to this legacy — both for compatibility, and to be able to run on a familiar system. Mac is also a UNIX system under the hood, and offers many of the same command-line utilities available on any UNIX that are useful to a programmer.

Other common reasons that can’t be overlooked, of course, hardware (speed, battery life, etc.), software compatibility (Apple makes both its own hardware and software), and simply community preference.