This is our last accounting forum series, with one of the most prominent researchers in financial accounting. Prof. James Ohlson from Hong Kong Polytechnique University. Prof. Ohlson comprehensively discussed the false research findings in accounting research. Pujiharto, Ph.D. as the moderator and Prof. Ohlson started discussion at 13.00 and adjourned more than two hours afterwards.
Regarding the diagnostics of possible false research questions, Prof. Ohlson proposed some basic diagnostics. For example, he stated that critical independent variables were missing the table of descriptive statistics. He also argued that the increase in the R-Square is trivial when the interaction products were added to the regressions. Prof. Ohlson further argued other “clever steps” to run this diagnostics. For instance, we can rank the bivariate correlations and compared it with the t-statistics. Furthermore, the bias due to large observations used in the study is also problematic and additional test (t-stat/sqrt(N) > 0.03) to figure out whether the result is too modest or not.
Prof. Ohlson closed the session by arguing that there are many nicely crafted papers that are unlikely to be valid in substantive sense. He also stated that the use of hold out samples would most likely reject more than half of papers published.