Examining People’s Sentiment towards Common Core State Standards (ccss) and Next Generation Science Standards (ngss)

Research Questions

Question1: How did people’s sentiment towards ccss and ngss change during the time when data were collected?

Question2: How do four dictionary-based approaches (i.e., Afinn, Bing, Loughran, Nrc) differ in analyzing people’s sentiment towards ccss and ngss?

Methods

This study used all the tweets data about ccss and ngss retrieved from the first five months of 2021. We adopted four different dictionary-based approaches to analyze people’s sentiment and compared the results.

Number of tweets over time

Overall, there were more comments about ccss than about ngss.

AFINN sentiment analysis

People showed more positive sentiment toward ngss than ccss. People gave more negative comments in the first two months.

Bing sentiment analysis

People showed more positive sentiment toward ngss than ccss. People showed more negative sentiment in the first two months.

Loughran sentiment analysis

People showed more positive sentiment toward ngss than ccss. People gave more negative comments during the first two months. There was a slight decrease in people’s positive sentiment toward ngss during the fourth month.

NRC sentiment analysis

This result is different from previous result: People showed more positive sentiment toward ccss than ngss.

Results

Research Question1: Overall, people showed more positive sentiment toward ngss than ccss.

Research Question2: AFINN, Bing and Loughran produced almost similar results. However, NRC generated totally different results. A closer examination revealed several issues. First, NRC generated scores for ten emotions and I just focused on the categories of “positive” and “negative”. Second, NRC gave high scores to positive and negative for tweets about ccss and comparatively low scores to positive and negative for tweets about ngss, making the difference in ccss much higher than the difference in ngss.

Discussion

Different lexicons may produce different results. When we choose lexicons to do sentiment analysis, we need to think more about our data features and the strengths and weaknesses of each lexicon. Human judgment is still an important part of using natural language processing techniques to process tasks.