1

I ran a fixed-effects regression and I am not quite sure about the interpretation category which it belongs to (i.e. level-level and so on), due to a paper interpreting it differently from what I think is correct.

The DV is a ratio: Number of sentences including a keyword / Total number of sentences

The IV of interest is a ratio as well: (aspired performance - potential performance) / stock price

The IV reads as a preformance pressure when it is positive. In other words, when IV is 0.01: the pressure to perform better is 1% of stock price.

From the regression I get a coefficient of 0.007.

I interpret it as a level-level case as neither variable is logarithmized: 1 unit increase in x c.p. leads to a 0.007 unit increase in y -> 1 percentage point increase in x c.p. leads to a 0.007 percentage point incrase in y

The paper I was speaking of suggests a different interpretation however: 1 percent increase in x leads to a 0.7% increase in y.

Can someone help me out here?

Mick
  • 23
  • 2

1 Answers1

0

I interpret it as a level-level case as neither variable is logarithmized: 1 unit increase in x c.p. leads to a 0.007 unit increase in y

Yes

1 percentage point increase in x c.p. leads to a 0.007 percentage point increase in y

No, we cannot interpret the results in terms of percentages unless we use logs.

The paper I was speaking of suggests a different interpretation however: 1 percent increase in x leads to a 0.7% increase in y.

This is incorrect for the same reason.

Robert Long
  • 5,722
  • 5
  • 29
  • 50