This function extracts statistics from strings and returns the extracted values, reported p-values and recomputed p-values. The package relies on the program "pdftotext", see the paragraph "Note" for details on the installation.
statcheck(x, stat = c("t", "F", "cor", "chisq", "Z", "Q"), OneTailedTests = FALSE, alpha = 0.05, pEqualAlphaSig = TRUE, pZeroError = TRUE,
OneTailedTxt = FALSE, AllPValues = FALSE)
A data frame containing for each extracted statistic:
Name of the file of which the statistic is extracted
Character indicating the statistic that is extracted
First degree of freedom (if applicable)
Second degree of freedom
Reported comparison of the test statistic, when importing from pdf this will often not be converted properly
Reported value of the statistic
Reported comparison, when importing from pdf this might not be converted properly
The reported p-value, or NA if the reported value was NS
The recomputed p-value
Raw string of the statistical reference that is extracted
The computed p value is not congruent with the reported p value
The reported result is significant whereas the recomputed result is not, or vice versa.
Logical. Is it likely that the reported p value resulted from a correction for one-sided testing?
Logical. Does the text contain the string "sided", "tailed", and/or "directional"?
What proportion of all detected p-values was part of a fully APA reported result?
A vector of strings.
"t" to extract t-values, "F" to extract F-values, "cor" to extract correlations, "chisq"to extract chi-square values, "Z" to extract Z-values, and "Q" to extract Q-values (within, between, or in general).
Logical. Do we assume that all reported tests are one tailed (TRUE) or two tailed (FALSE, default)?
Assumed level of significance in the scanned texts. Defaults to .05.
Logical. If TRUE, statcheck counts p <= alpha as significant (default), if FALSE, statcheck counts p < alpha as significant
Logical. If TRUE, statcheck counts p=.000 as an error (because a p-value is never exactly zero, and should be reported as < .001), if FALSE, statcheck does not count p=.000 automatically as an error.
Logical. If TRUE, statcheck searches the text for "one-sided", "one-tailed", and "directional" to identify the possible use of one-sided tests. If one or more of these strings is found in the text AND the result would have been correct if it was a one-sided test, the result is assumed to be indeed one-sided and is counted as correct.
Logical. If TRUE, the output will consist of a dataframe with all detected p values, also the ones that were not part of the full results in APA format
Sacha Epskamp <mail@sachaepskamp.com> & Michele B. Nuijten
<m.b.nuijten@uvt.nl>
statcheck uses regular expressions to find statistical results in APA format. When a statistical result deviates from APA format, statcheck will not find it. The APA formats that statcheck uses are: t(df) = value, p = value; F(df1,df2) = value, p = value; r(df) = value, p = value; [chi]2 (df, N = value) = value, p = value (N is optional, delta G is also included); Z = value, p = value; Q(df) = value, p = value (including Qw, Qwithin, Qb, and Qbetween). All regular expressions take into account that test statistics and p values may be exactly (=) or inexactly (< or >) reported. Different spacing has also been taken into account.
This function can be used if the text of articles has already been imported in R. To import text from pdf files and automatically send the results to this function use checkPDFdir or checkPDF. To import text from HTML files use the similar functions checkHTMLdir or checkHTML. Finally, checkdir can be used to import text from both PDF and HTML files in a folder.
Note that the conversion from PDF (and sometimes also HTML) to plain text and extraction of statistics can result in errors. Some statistical values can be missed, especially if the notation is unconventional. It is recommended to manually check some of the results.
PDF files should automatically be converted to plain text files. However, if this does not work, it might help to manually install the program "pdftotext". You can obtain pdftotext from http://www.foolabs.com/xpdf/download.html. Download and unzip the precompiled binaries. Next, add the folder with the binaries to the PATH variables so that this program can be used from command line.
Also, note that a seemingly inconsistent p value can still be correct when we take into account that the test statistic might have been rounded after calculating the corresponding p value. For instance, a reported t value of 2.35 could correspond to an actual value of 2.345 to 2.354 with a range of p values that can slightly deviate from the recomputed p value. Statcheck will not count cases like this as errors.
checkPDF, checkHTMLdir, checkHTML, checkdir
txt <- "blablabla the effect was very significant (t(100)=1, p < 0.001)"
statcheck(txt)
Run the code above in your browser using DataLab