Effects of Splitting Text into Multiple Columns

Discussion

The goal of statistical analysis for any experiment is to prove or disprove a hypothesis.  In t-test analysis, two samples are compared with each other in order to determine whether there are statistically significant differences between the two.  Therefore the null-hypothesis for t-test is that there are no statistically significant differences between the samples.  The analysis proves or disproves the null-hypothesis.  On the other hand, single-factor ANOVA analysis looks at all the samples for a particular dependent variable and allows to analyze if there are no differences between all the samples, or whether some samples differ significantly from others.

In our experiment, we used both the t-tests and single-factor ANOVA to compare our results and determine whether splitting the text into multiple columns produces statistically significant differences in the final results.  T-tests were used to compare 1-column vs. 3-column samples for every window width and every dependent variable.  Single-factor ANOVA analysis was performed for all the samples for a given independent variable.

Time to Correct Completion

The analysis of t-tests for this dependent variable shows that computed t values are less then t-critical and the p-value is always greater then the significance factor (which is 0.05 for our experiment).  This fact shows that the null-hypothesis is valid and there are no significant differences between 1 and 3 column display for any window width.  The ANOVA analysis also shows that the computed F value is less then F-critical and p-value is greater then the significance factor.  This reinforces the fact that our experiment showed no significant differences in time it takes to read the articles and correctly answer the questions.

The results show that the splitting text into multiple columns does not improve the time it takes to read the text and complete the questions.  This finding does not agree with our hypothesis.

Number of Errors

The results for the statistical analysis of the number of errors (or incorrect responses to questions) are very similar to those for the time to correct completion.  Neither t-test analysis nor ANOVA analysis show any significant differences between the samples.  Once again, splitting the text into multiple columns does not improve comprehension of the text by the reader as the width of the window increases.

Subjective Rating

However, when we look at the results of statistical analysis on the subjective rating, a more interesting picture develops.  The t-test for 1 and 3 column samples for the 600-pixel window width shows no significant differences.  On the other hand, t-tests for both 800-pixel and 1000-pixel windows show statistically significant differences and display that for both of these window widths the subjective rating for the 3-column display is higher.  ANOVA analysis also confirms the fact that there are statistical differences between the samples.

The fact that for wider windows, the multi-columns display scores higher with the reader confirms a part of our original hypothesis.  Multi-column display is not necessarily preferred for narrower (600-pixel) window size, which can be attributed to the fact that the text line length is more manageable for the reader in such a window.

Department of Computer Science: Direct questions and comments to the student editorial team