P-curve and false positives
In this step, we’ll actually get to use P-curve to test some literature for publication bias.
We’ll use the P-Curve app: http://www.p-curve.com/ to test the evidentiary value of a series of papers on microfinance, as well as papers from psychology, including the power pose. The microfinance papers were all published in the January 2015 issue of the American Economic Journal: Applied Economics. You can find freely available versions of the final papers on the J-PAL website. We’ve also provided links below.
All six of the following microfinance papers describe randomized controlled trials of microfinance interventions. Table 2 in each paper describes two of the most basic outcomes — (i) whether people borrowed or not, or credit access, and (ii) how much more those in the treatment group borrowed from the microfinance institution (MFI) that was being studied, or loan amount.
Directions: Find the test statistics described below for each paper and separately test them using the P-curve tool and the following steps.
- Go to http://www.p-curve.com/. Click on “The online app 4.0”.
- For each of the six articles below, find the listed statistics.
- Calculate the Z statistic: In economics, reporting of coefficients and standard errors is the norm. But because the samples from these papers are relatively large and it’s difficult to back out degrees of freedom, we’ll use Z statistics instead of t statistics. To convert these data to a Z statistic, just divide the coefficient by the standard error. Each paper lists the standard error in parentheses below the coefficient.
- In P-curve, replace the default statistics in the text box with the Z statistics from all the six papers for both Credit access and Loan amount statistics. Click “Make the p-curve”. You’ll do this twice – once for each statistic. Keep these results handy as you’ll use them in the next step. An easy way to keep track is to make a table with 3 columns. In Column 1, put the paper numbers; in Column 2, put the Z statistics for all six Credit access coefficients; and in Column 3, put the Z statistics for the first five Loan amount coefficients.
“The Miracle of Microfinance? Evidence from a Randomized Evaluation” by Esther Duflo, Abhijit Banerjee, Rachel Glennerster, Cynthia G. Kinnan. Paper on the J-PAL website.
Coefficient from Table 2: Credit access from Spandana, Panel A. Endline 1
Coefficient from Table 2: Loan amount from Spandana, Panel A. Endline 1
“The Impacts of Microcredit: Evidence from Ethiopia” by Alessandro Tarozzi, Jaikishan Desai, and Kristin Johnson. Paper on the J-PAL website.
Coefficient from Table 2: Credit access from ACSI & OCSSC
Coefficient from Table 2: Loan amount from ACSI & OCSSC
“The Impacts of Microfinance: Evidence from Joint Liability Lending in Mongolia” by Orazio Attanasio, Britta Augsburg, Ralph De Haas, Emla Fitzsimons and Heike Harmgart. Paper on the J-PAL website.
Coefficient from Table 2: Credit access from XacBank
Coefficient from Table 2: Loan amount from XacBank
“Estimating the Impact of Microcredit on Those Who Take It Up: Evidence from a Randomized Experiment in Morocco” by by Bruno Crépon, Florencia Devoto, and Esther Duflo. Paper on the J-PAL website.
Coefficient from Table 2: Credit access from Al Amana, Admin data
Coefficient from Table 2: Loan amount from Al Amana, Admin data
“Microcredit Impacts: Evidence from a Randomized Microcredit Program Placement Experiment by Compartamos Banco” by Manuela Angelucci, Dean Karlan, and Jonathan Zinman. Paper on the J-PAL website.
Coefficient from Table 2A: Any loan from Compartamos: Admin data
Coefficient from Table 2B: Amount from Compartamos: Survey data
“The Impacts of Microcredit: Evidence from Bosnia and Herzegovina” by Britta Augsburg, Ralph De Haas, and Heike Harmgart. Paper on the J-PAL website.
Coefficient from Table 2: At least one loan outstanding from an MFI
© Center for Effective Global Action