The COVID-19 pandemic accelerated the use of online assessments, and they’ve remained popular because they offer flexibility. Lecturers are also able to incorporate multimedia into the assessment, and of course, online exams minimise paper waste.

Online exams also come with disadvantages. Students have reported increased anxiety around the potential for internet issues or technical problems, and the use of computers for exams may also add to cognitive load. But potentially the biggest downside of online exams now is the impact of generative AI on academic integrity. An estimated 83% of Australian university students are using generative AI platforms.  Concerns about cheating on assessments using generative AI are on the rise, with anecdotal reports that cheating in online exams is particularly easy.

In my research skills unit, which includes a focus on statistics, I replaced the online exam with a totally different type of online assessment: the digital decision index.

The digital decision index:

So, what is a digital decision index? My students didn’t know, and AI didn’t know either, when the students tried to ask!  (I’m sure it will learn soon).

For me, the digital decision index is an electronic tool that acts like a framework, or interactive decision tree, that guides a user through a decision making process, providing outcomes and additional relevant information.

Students were asked to design an index that would help them select the right statistical test for a given situation, and to interpret the results. They could build it in any electronic format that worked for them – software like Excel, Canva, Twine and Lucidchart were used. The aim was for students to create a resource they could use after completing the unit, to assist them when they began a new research project (perhaps in another unit).

For the assessment, students presented their index in an online class and discussed how they developed it. I then asked the student to use their index to help chose, run and interpret a statistical test to address a research question, and supplied a database. Students were each given different questions and did not see the questions beforehand.  Marks were given for the creation and effectiveness of the digital decision index, along with clarity and efficiency.

I was struck by the wide variety of digital decision indexes created by the students. Some used a flow chart style while others developed a linked database where answers to questions add up behind the scenes to produce a recommendation at the end. Students often made use of hyperlinks to link to additional information in an efficient way.  

Examples of different types of digital decision indices

How students found the assessment:

The preliminary feedback obtained via anonymous, optional Qualtrics survey (n=18, 2024-25) has been positive. The majority of students reported:

• They preferred the digital decision index to an online exam

“I liked using the tool, it was nice not to have another written assessment. I thought the process of putting together the tool was helpful in organising my thoughts, and felt like a more realistic application of stats than an exam would be.”

“It was much more interactive than it would have been if we had a test and we would have just read the text book to study. This made us apply it practically while creating the tool which was then us "studying" for the assessment”

• It helped them learn statistics

“I had very little understanding about stats initially. While the assignment seemed impossible to begin with, it really made me learn about and understand different tests, even just "simple" things like understanding different types of data which I could not get my head around at the beginning.”

• It helped them learn more about how to use software

“It was very stressful considering there was limited time to do the assessment, but once I upskilled with LucidChart and came up with the idea of putting SPSS commands into my hyperlinks, it was extremely quick, easy and efficient to use. Stress well worth it.”

• AI could only play a minor role in the assessment

About half the students did use AI – mainly for troubleshooting, selecting or learning new software, and clarifying statistical principles. Importantly, most felt AI couldn’t replace their own thinking in this task.

“I asked AI for ideas on the easiest software to use for this task”

• The tool would be useful for future research endeavours

“I had ZERO understanding before, so this really helped cement my learning. Also, now I have the tool to use in future.”

• Students also shared concerns around expectations and workload:

“Not having heard of the concept, nor any of my friends who work in programming/IT understanding what it was and the internet not being very useful in defining/explaining what it was either.”

“Structure and design was my biggest concern, but knowing it could be any format that makes sense to us helped with getting started.”

“It was an extremely time consuming assignment- when you weigh up the benefits x work in putted unfortunately I don't think the benefits outweighed the effort”

Overall satisfaction with this unit improved by 12.5% compared to the previous year, when the assessment was an online exam, as assessed by the university’s Unit Teaching and Evaluation Instrument (UTEI) survey.

Making it work in practice:

Based on feedback and reflection, there are a few key aspects to help make this assessment work well.

• Allow time in class to work on the index. I ran an activity where students broke up into small groups and investigated different types of software that could be used. They reported pros and cons back to the larger group and then decided as individuals what would best suit them and their ideas.
• Give practice questions in advance, along with answers, so that students can test their index and refine it.
• Be clear on expectations. Detail what is expected, how the assessment presentation will run, and what marks will be allocated for.

When I shared this work at a HERDSA conference roundtable in July 2025, delegates raised some additional helpful points:

• My students requested exemplars, but it’s likely that fully worked models might stifle creativity. A compromise could be showing partial demos without handing over a finished product.
• AI transparency could be improved by asking students to report how they used AI, and how they checked whether the information was reliable.
• The assessment should emphasise the thinking and iteration behind the index, not just the polished end result – ie, a focus on process, not just product.

These conversations reinforced that authentic assessment is about helping students engage meaningfully with the learning process.

Why this matters now:

Although concerns about academic integrity are pushing universities to move away from online exams, the value of online learning cannot be ignored.  There is a need for more authentic, engaging online assessment like the digital decision index that is process driven and resistant to academic misconduct.