I analyzed consumer reviews from IMDb and BestBuy in the context of Behavioral Economics and Information Economics specifically Loss Aversion, Negativity Bias and Search & Experience Goods Paradigm.
I found out that information sharing through consumer reviews is beneficial for potential buyers at least for Search Goods(Electronics). For Experience Goods(Movies) similar effect couldn't be observed.
Potential buyers find:
- Negatively and Positively rated reviews more helpful.
- Reviews with Negative sentiments more helpful.(Loss Aversion & Negativity Bias)
- Reviews with Positive sentiments less helpful.
- Moderately longer reviews more helpful.
Additionally when a review is negatively rated longer reviews are more helpful. Potential buyers may need more elaboration concerning why a certain good is bad. (Negative interaction effect between Review Rating and Review Length confirming Loss Aversion and Negativity Bias.)
- I scraped the data with Selenium Webdriver from BestBuy.com and IMDb.com.
- The scraper code can be found in this repository.
- Tobit Regression was used to test the hypotheses.
- NRC Word-Emotion Association Lexicon was used to get negative and positive emotion words from consumer reviews.
- All scripts which have
EDAin them were part of the experimental phase of my thesis. All Data Cleaning and Data Wrangling processes were initially done within these files. Report.Rmdis the first research report concerning the research question that I sent to my thesis advisor. Initial and the lesser version of my thesis's statistical model was conducted in this report as well.Paper.Rmdincludes both text and analysis code of the thesis.Paper.pdfis the rendered thesis.apa.cslconfigures the citation style.references.bibhas all the references.custom.texandnon-float-fig.texare the LaTeX configuration files that style my thesis into the appropriate format imposed by the Institute.