**Experimental procedure** The experiment will be conducted online in Qualtrix. During the experiment each participant will read one poem (in Dutch), presented in one certain font. After reading, they will answer some questions about that poem. The design is 2x2: *Factor Poem Fluency* (between: high, low) and *Factor Font Readability* (between: high, low). One between factor is poem fluency. Two poems differ in their fluency. One poem (‘Sterfbed’ by Jean Pierre Rawie) has a clear and predictable form (rhyme, sonnet), and is conceptually fluent in the sense that it is very easy to understand what the poem is about. The second poem (‘Alles gefilmd’ by Robert Anker) is of equal length, has a less clear structure (e.g. no rhyme) and is more opaque in terms of content. The another between factor is font, which differ in readability. We use 'calibri' as easy reading font, and 'mistral' as hard reading font. These choices are based on previous pretests in which participants scored the readability of fonts. ---------- **Hypotheses** We expect the interaction effect between font and poem fluency, specifically: The difficult poem will be appreciated more when font is hard to read than when font is easy to read; The easy poem will be appreciated more when the font is easy to read than when the font is hard to read. ---------- **Dependent variables** The dependent variables are 6 questions asking about appreciation, structural fluency, and conceptual fluency (see material). ---------- **Sample size** 400 participants is the maximum amount we are willing to run. ---------- **Sequential Analyses** Since we don't have specific previous research to estimate reasonably accurate effect size, we calculate the effect size based on the willingness to collect at most 400 participants for this study, in order to achieve 80% power and 5% of alpha level. G power software is used to calculate the effect size. The input and output of G power are in the following frame: F tests - ANOVA: Fixed effects, special, main effects and interactions Type of power analysis - Sensitivity: Compute required effect size - given α, power, and sample size Input: α err prob = 0.05 Power (1-β err prob) = 0.8 Total sample size = 400 Numerator df = 1 Number of groups = 4 Output: Noncentrality parameter λ = 7.8871042 Critical F = 3.8650483 Denominator df = 396 Effect size f = 0.1404199 We set this effect size (f = 0.14) as the smallest effect size of interest (SESOI) to conduct sequential analyses. We decide to perform two-side interim analyses after collecting 200 and 400 participants. The alpha boundaries for the two analyses are calculated by WinLD software (for download: https://www.biostat.wisc.edu/content/lan-demets-method-statistical-programs-clinical-trials). The input and output of WinLD software are in the following frame: Comupte -> Bounds Input: Interim Analyses (k) = 2 Information times(t) = Equally Spaced Test Boundaries = Two-Sided Symmetric Overall Alpha = 0.05 Function = Power Family Phi = 1 Truncate bounds? = No Observed Z? = No Output: Nominal Upr Alpha 1 = 0.01250 2 = 0.01679 Upper Bound 1 = 2.2414 2 = 2.1251 alpha level for the first two-sided ANOVA performed after 200 participants is 0.0125 * 2 = 0.025, and alpha level for the second (last) analysis is 0.01679 * 2 = 0.03358. Z boundaries of 2.2414 for the first interim analysis, and Z boundaries of 2.1251 for the second analysis. The data collection will be either terminated or continued based on interim analysis (after 200 participants): if: then: p-value & f data collection <0.025 & > 0.14 stop <0.025 & < 0.14 stop >0.025 & > 0.14 continue >0.025 & < 0.14 stop In this study, we mainly focus on the interaction effect between readability and poem fluency. Thus, the criterion of alpha level and SESOI for sequential analysis will be applied to the interaction effect. The data analyses will be conducted in JASP. The effect size reported in the result (eta square) will be converted to f in order to compare it to SESOI. For whole sequential analyses procedure, please check: https://osf.io/qti32/ and: Lakens, D. (2014). Performing high‐powered studies efficiently with sequential analyses. European Journal of Social Psychology, 44(7), 701-710.