Abstract
Despite scholars’ admonitions regarding the use of ratios in statistical analyses, the practice is common in management research. This is particularly true in the area of strategic management, where important variables of interest are operationalized as ratios. In this study, we employ simulations to demonstrate the implications of using ratios in statistical analyses. Our simulations illustrate that ratio variables produce inaccurate parameter estimates and can result in lower levels of statistical power (i.e., the ability to uncover hypothesized relationships). We also find that when an independent or a dependent variable is a ratio, the relationship between the independent and dependent variable fluctuates as the dispersion of the denominator changes. These fluctuations occur even when the correlations between the unscaled variables remain exactly the same. We also find that including ratios in models as control variables influences estimates of relationships between focal independent and dependent variables. This is true even when neither the independent or dependent variable is a ratio. We provide several recommendations for researchers who may be interested in avoiding the pitfalls of ratio variables.
Original language | English (US) |
---|---|
Journal | Organizational Research Methods |
DOIs | |
State | Accepted/In press - Jan 1 2018 |
Fingerprint
Keywords
- bootstrapping
- computer simulation procedures
- measurement design
- Monte Carlo
- multiple regression
- quantitative research
- quantitative research
- research design
ASJC Scopus subject areas
- Decision Sciences(all)
- Strategy and Management
- Management of Technology and Innovation
Cite this
Divided We Fall : How Ratios Undermine Research in Strategic Management. / Certo, Samuel; Busenbark, John R.; Kalm, Matias; LePine, Jeffery.
In: Organizational Research Methods, 01.01.2018.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - Divided We Fall
T2 - How Ratios Undermine Research in Strategic Management
AU - Certo, Samuel
AU - Busenbark, John R.
AU - Kalm, Matias
AU - LePine, Jeffery
PY - 2018/1/1
Y1 - 2018/1/1
N2 - Despite scholars’ admonitions regarding the use of ratios in statistical analyses, the practice is common in management research. This is particularly true in the area of strategic management, where important variables of interest are operationalized as ratios. In this study, we employ simulations to demonstrate the implications of using ratios in statistical analyses. Our simulations illustrate that ratio variables produce inaccurate parameter estimates and can result in lower levels of statistical power (i.e., the ability to uncover hypothesized relationships). We also find that when an independent or a dependent variable is a ratio, the relationship between the independent and dependent variable fluctuates as the dispersion of the denominator changes. These fluctuations occur even when the correlations between the unscaled variables remain exactly the same. We also find that including ratios in models as control variables influences estimates of relationships between focal independent and dependent variables. This is true even when neither the independent or dependent variable is a ratio. We provide several recommendations for researchers who may be interested in avoiding the pitfalls of ratio variables.
AB - Despite scholars’ admonitions regarding the use of ratios in statistical analyses, the practice is common in management research. This is particularly true in the area of strategic management, where important variables of interest are operationalized as ratios. In this study, we employ simulations to demonstrate the implications of using ratios in statistical analyses. Our simulations illustrate that ratio variables produce inaccurate parameter estimates and can result in lower levels of statistical power (i.e., the ability to uncover hypothesized relationships). We also find that when an independent or a dependent variable is a ratio, the relationship between the independent and dependent variable fluctuates as the dispersion of the denominator changes. These fluctuations occur even when the correlations between the unscaled variables remain exactly the same. We also find that including ratios in models as control variables influences estimates of relationships between focal independent and dependent variables. This is true even when neither the independent or dependent variable is a ratio. We provide several recommendations for researchers who may be interested in avoiding the pitfalls of ratio variables.
KW - bootstrapping
KW - computer simulation procedures
KW - measurement design
KW - Monte Carlo
KW - multiple regression
KW - quantitative research
KW - quantitative research
KW - research design
UR - http://www.scopus.com/inward/record.url?scp=85047924928&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85047924928&partnerID=8YFLogxK
U2 - 10.1177/1094428118773455
DO - 10.1177/1094428118773455
M3 - Article
AN - SCOPUS:85047924928
JO - Organizational Research Methods
JF - Organizational Research Methods
SN - 1094-4281
ER -