Hello, I am a first year PhD Finance student and I?m working on my first academic paper. I have conducted a series of tests for weak form market efficiency for a range of individual stocks. I now need to check to see if my results have been distorted because of liquidity (or lack of). My plan is the run a series of univariate regressions with t stats from a weak form test (which were conducted on an individual basis for each stock) as the dependent variable and use a liquidity measure (e.g. bid-ask spread or Kyle?s Lambda) as the independent variable. My data set consists of over 400 stocks, so there are certainly enough observations, but my question is will this procedure yield robust and meaningful results?Many Thanks

i could see how you'd use standard deviations of parameters. but t stats? this sounds strange

why not just do a multivariate regression?