Show, Then Do: Worked Examples Took Less Effort for Accounting Novices

It's tutorial week on adjusting entries. Half the room did accounting at school, so they breeze through the accruals. The other half stares at the trial balance as if it's written in hieroglyphics.

Abdel Halabi, Juhani Tuovinen and Alan Farley asked whether showing the first few solutions changes the cost of learning for that second group. For students with no prior knowledge, worked examples meant significantly lower reported effort and significantly higher efficiency, while test scores didn't differ significantly.

Statistical significance concerns how compatible the data are with a specified no-difference model. It does not measure the probability that chance caused the result or how important the difference is. A worked example shows the solution steps before students try similar problems.

Try this first: in your next adjustments tutorial, give full solutions to the first three items and ask students to do the rest. That's our suggestion, based on a single study, so treat it as worth testing.

What they found

Here are the numbers behind that. For these novices, reported effort averaged 3.50 with worked examples and 3.63 with problem-solving on the 1–5 rating scale. The authors report that 0.13-point gap as significant (p = 0.016).

Efficiency was −0.53 with worked examples and −1.18 with problem-solving (p = 0.006). "Efficiency" combined test score and effort into a single measure, using standardised scores. Standardising puts both on a common scale so they can be compared.

A negative value means the group's standardised test score was lower than its standardised effort score.

What about test scores? The novices' test means were 8.32 and 7.59, but test performance did not differ significantly by instruction type overall. The interaction on performance was not significant either, so prior knowledge didn't significantly change how instruction type affected scores.

So the authors say the efficiency gain came from lower effort, not higher performance.

The picture was different for students with prior knowledge. Their mean reported effort was identical, at 3.27 in both groups, and no statistically significant efficiency difference was detected. Although the authors describe the approach as beneficial for novices and not harmful for experienced students, a nonsignificant comparison does not establish equivalence or rule out harm.

Separately, prior knowledge itself had large effects on effort, scores and efficiency. An effect size describes the magnitude of a difference. Statistical significance answers a different question. The authors base that label on how much of the variation between students was explained by prior knowledge.

What they did

The study took place at Monash University in Australia with 93 students in Introductory Accounting A. Of these, 48 had prior accounting, meaning they had passed it at school, and 45 had none.

After a lecture on period-end adjustments, students were randomly assigned to one of two computer-based tutorials. That means chance alone decided who got which tutorial.

Both tutorials used the same trial balance and eight adjustments, with immediate feedback. The problem-solving group did all eight, while the worked-examples group got solutions to the first three and then did the remaining five.

Students rated their mental effort on a scale of 1 to 5. A 30-minute diagnostic test, marked out of 15, followed in the next week's lecture.

As with any single study, though, the setting matters.

The reality check: the study involved one university, one highly structured topic and computer-based material. So the authors say results should be generalised, at best, to other highly structured accounting tasks.

Effort was self-reported, meaning students rated it themselves. The test came about a week later, not at the end of semester. Finally, about 18% of initial volunteers dropped out, and students who said they had studied for the test were excluded.