The Fabulous Hitmontop’s Analysis of Artwork Popularity in the Met Collection

Preregistration of analyses

Analysis #1

• Research Question

  • Is the proportion of artworks completed before 1650 (pre-Renaissance) that are highlighted different from the proportion of artworks completed after 1650  (post-Renaissance) that are highlighted? 

• Description

  • According to our exploratory analysis, the majority of artwork has data for both object_end_date and is_highlight, indicating that there will be sufficient data to work with. 

• Hypothesis & Pre-registration

  \[ H_0 : p_{pre-1650} - p_{post-1650} = 0 \] \[ H_A :  p_{pre-1650} - p_{pre-1650} \neq 0 \]

  • Conduct a hypothesis test under the assumption that the null hypothesis is true and calculate a p-value (probability of observed or more extreme outcome given that the null hypothesis is true)

• What is the conclusion we want to draw from this analysis?

  • The analysis will reveal whether the difference between the proportions of highlighted artworks before and after 1650 is statistically significant. This shows whether the true proportion of highlighted artworks has significantly changed, based on statistical analysis. 

  • If so, we can make an inference about the public’s perception of older vs more modern artworks.

  • If the test results suggest that the data do not provide convincing evidence for the alternative hypothesis (p > 0.1), then we fail to reject the null hypothesis. If they do, then reject the null hypothesis in favor of the alternative hypothesis. 

Analysis #2

• Research Question

  • Is the proportion of highlighted artworks included on the Timeline of Art History website different than the proportion of highlighted artworks not included on the Timeline of Art History website.

• Description

  • By looking at the exploratory data analysis, we can see that there is sufficient data for each combination of isHighlighted and isTimelineWork. For each one of the combinations (True/False, True/True, etc.), there are >1000 artworks each. This will allow us to perform the analysis with a big enough population size.

• Hypothesis & Pre-registration (predict significant or not)

  \[ H_0: p_{timeline~highlight} - p_{non-timeline~highlight} = 0 \]

  \[ H_A: p_{timeline~highlight} - p_{non-timeline~highlight} ≠ 0 \]

  • Conduct hypothesis test under the assumption that the null hypothesis is true and calculate a p-value (probability of observed or more extreme outcome given that the null hypothesis is true)

• What is the conclusion we want to draw from this analysis?

  • The analysis will reveal whether the proportion of artworks included on the Timeline of Art History website when isHighlight == True is sufficiently different from the proportion of artworks included on the Timeline of Art History website when isHighlight == False.

  • If the test results suggest that the data do not provide convincing evidence for the alternative hypothesis (p > 0.10), stick with the null hypothesis. If they do, then reject the null hypothesis in favor of the alternative hypothesis.

  • If the null hypothesis is rejected, then it will be worth to discuss how the MET evaluates the value of an artwork in different contexts (by itself vs suitable for the timeline website).