P values for null hypotheses of no association or no effect (null P values, which we denote by P _{0}) still thrive in most of the literature. Despite being data frequencies under a hypothetical sampling model, they are routinely misinterpreted as probabilities of a “chance finding” or of the null, which are Bayesian probabilities. To quote one biostatistics textbook: “The P value is the probability of being wrong when asserting that a true difference exists.”^{1} Similar misinterpretations remain in guides for researchers. For example, Cordova^{2} states the null P value “represents the likelihood that groups differ after a treatment due to chance.”
It is thus unsurprising that many call for improved education regarding P values.^{3–11} Others discourage specific uses of P values^{12–15} or, occasionally, all uses (as in the EPIDEMIOLOGY instructions for authors: “We strongly discourage the use of P values and language referring to statistical significance…”), regarding them as a confounded mix of estimate size and precision.^{16},^{17}
Many advocates of educational reform specifically decry use of P values in NeymanPearson hypothesis testing, in which results are placed in bins labeled “association” and “no association” based on whether the P value is below or above a prespecified alpha level (the maximum tolerable typeI or falsepositive error probability), usually 0.05. NeymanPearson testing has often been criticized as groundless and arbitrary^{9},^{17–20} and even mindless^{3},^{5}—aptly in our view. Unfortunately, NeymanPearson testing has also been called “significance testing,” and the alpha level has been called the “significance level” of the test. This secondary terminology has fed the oftlamented confusion of NeymanPearson testing with the significancetesting approach of Fisher.^{21},^{22} In the latter approach, significance level refers to the P value itself, which is treated as a continuous measure of evidence against the tested hypothesis, with lower P values corresponding to greater evidence.^{6},^{9},^{23–26}
This article is in the reform tradition. We first consider some arguments against P values. We then review ways in which P values and other ordinary frequentist statistics can be given correct Bayesian interpretations. We focus on how a onesided P value can provide a lower bound on the posterior probability that the point estimate is on the wrong side of the prior median (the prior odds that the true parameter is above vs. below this median is 50:50).^{27} In particular, when one sees a twosided null P value P0 for a difference or log ratio measure, one can immediately say that if one starts with a normal or uniform prior centered on the null, the posterior probability that the point estimate is in the wrong direction (ie, on the wrong side of the null) is no less than P0 / 2.
ARE P VALUES SUPERFLUOUS GIVEN CONFIDENCE INTERVALS?
It is sometimes said that confidence intervals render P values redundant because the two give “equivalent information,” with no indication of what “equivalent” means. True, for any parameter value, its P value is under 0.05 if and only if the parameter value is outside the 95% confidence interval computed by the same method. But this equivalence argument seems to restore 0.05level testing to the primacy it retains in most medical journals.^{28}
One can compute a P value from a confidence interval; therefore, the abstract (mathematical) information in a P value is indeed redundant given the confidence interval. But, unless the P value is 1, we can compute a confidence interval from the point estimate and P value, rendering the confidence interval redundant in the same technical sense. Furthermore, just as the estimate and P value do not convey a confidence interval well, a confidence interval does not convey desired P values well.^{10} For instance, we have found colleagues who believe that if a set of risk ratios (RRs) all have lower confidence limits above 1, the RR with the highest lower limit must have the smallest null P value P _{0}, especially if that RR has the largest point estimate. Others seem to think that if the lower limits are the same, the null P values are the same, or the one with the higher point estimate has the lower null P value. These beliefs are often the opposite of reality; for example, if the estimate
is 1.6 with lower limit 1.2, P0 is 0.001, whereas for
= 4.8 with lower limit 1.3, P0 is 0.020.
Equivalence and redundancy are thus not good arguments for suppressing P values as supplements to point and interval estimates. There are, however, good psychological arguments for that suppression. They are based on the observation that null P values are almost always interpreted badly, even by statisticians. For example, a small P0 is routinely taken as refuting the null and a large P0 is routinely taken as supporting the null; however, both interpretations are usually far from correct.^{8},^{11–13},^{25},^{26},^{28–37} We will explain how these problems could be avoided if P values were interpreted as distance measures and posterior probabilities rather than as tests of hypotheses.
Assumptions and Notation
Throughout, we assume that an analysis goal is inference about the true value θt of a parameter or some simple function of θt (eg, its antilog) in an analysis using standard models and approximations. Usually, θt is a coefficient in a linear, logistic, Poisson, or proportional hazards regression and analysis and thus represents a difference or log ratio; however, when is a productterm coefficient, it measures variation (heterogeneity) in a difference or log ratio.
will denote one of the usual efficient estimators of θt (eg, leastsquares, maximum likelihood), with
its estimated standard error. θ will denote the parameter name or a particular value for the parameter, possibly incorrect, but of special contextual interest. In almost all software and reports of P values, the null value 0 for θ is automatically taken to be of key interest, but other values may be as or more important to analyze. For example, in certain legal situations, a risk or rate ratio RR of 2 translates to a 50% lower bound for a causation probability,^{38} which leads to examining θ = ln(RR = ln(2).
CORRECT FREQUENCY INTERPRETATIONS OF P VALUES AND CONFIDENCE INTERVALS
We illustrate concepts using familiar Wald methods derived from the Z score
and its absolute value
which is the distance in standard errors from the observation
to the possibility θ. The twosided P value Pθ for θ is, then, the relative frequency (over study repetitions) that the absolute Z statistic
would exceed its observed value if indeed θt=θ and all other assumptions used to compute this frequency are correct. In particular, P _{0} is the usual twosided null P value for the hypothesis θt = 0, derived from
, and P _{ln(2)} is the twosided P value for θt = ln(2), derived from
. The usual 95% confidence interval with limits
can be derived by solving
for θ and thus can be defined as all θ for which P _{θ}. Conversely, given a 95% confidence interval confidence interval
, we can deduce
and thus compute a Z score and P value for any θ of interest.
P values such as Pθ are often described as measures of goodness of fit, distance, consistency, or compatibility between the observed data and the datagenerating model, or between a parameter estimate
and a hypothesized parameter constraint such as θt=θ or θt≤ θ.^{23},^{39} Pθ is the probability transform of the distance from θ to
; in particular, P0 is the transformed distance of
to the null. A small Pθ is taken to indicate a problem with the assumptions used for its computation (eg, perhaps, θt ≠ θ, or there is some uncontrolled validity problem, or both).^{21},^{22} This interpretation is popular among those seeking to avoid both hypothesis testing and Bayesian interpretations of P values.^{26}
To illustrate, consider a disease indicator Y = 1,0 and a logistic regression model for disease frequency given two binary factors X and Z,
Pr(Y = 1  X = x, Z = z) = expit(α + βx + γz)
This model asserts that only three parameters (α,β,γ) are needed to perfectly specify (or encode) the disease frequencies at every combination of X = 1,0 and Z = 1,0. There is rarely any justification for this assumption; however, it is routine and usually unmentioned, or else unquestioned if the P value for the test of model fit is “big enough” (usually meaning at least 0.05 or 0.10).
Consider the null P value P _{0} for θt = 0 when θ is the productterm coefficient in
Pr(Y = 1  X = x, Z = z) = expit(α + βx + γz + θxz)
θt = 0 yields model 1 and translates into constancy of the odds ratio relating X to Y as Z varies. P0 is a measure of
 (a). The distance from 0 to

 ,
 (b). The goodness of fit of model 1 when taking the more general model 2 as the referent, and
 (c). The distance from the fitted model 1 to the fitted model 2 (where the fitted models are the above equations with the parameters replaced by their estimates).
For comparison, the nonnull value θt = ln(2) translates into a doubling of the odds ratio relating X to Y when we move from Z = 0 to Z = 1 and corresponds to the model:
Pr(Y = 1  X = x, Z = z) = expit(α + βz + γz + ln(2)xz)
P _{ln(2)} is a measure of
 (a).The distance from ln(2) to

 ,
 (b).The goodness of fit of model 3 when taking model 2 as the referent, and
 (c).The distance from the fitted model 3 to the fitted model 2.
Finally, the 95% confidence interval for θt is a measure of the spread of the fourdimensional likelihood function derived from the data and model 2 along the θ dimension. Consequently, confidence interval width is sometimes promoted as a measure of study precision in estimating θt.^{12} This interpretation, however, assumes model 2; using other models, we would expect all the confidence intervals and P values to change, demonstrating that statistical concepts of precision, fit, and distance are relative to a model rather than absolute properties of a study.
CORRECT BAYESIAN INTERPRETATIONS OF P VALUES AND CONFIDENCE INTERVALS
Typical misinterpretations of P values treat them as Bayesian posterior probabilities. For example, one common but extreme mistake interprets P _{θ} as the probability that θt = θ; however, these two probabilities are usually far apart.^{29},^{30} Such misinterpretations may be recognized and avoided by examining situations in which P _{θ} is indeed a probability (bet) about the true value θt. For a review of basic Bayesian ideas such as prior and posterior probabilities, see for example, Greenland.^{40–42}
ExtremePrior Interpretation
Under maximumlikelihood analyses, the most direct Bayesian interpretations of P values and confidence intervals arise when only two extreme kinds of prior distributions are allowed: point priors, which express 100% certainty that a parameter is a given value; and equalodds (uniform) priors on the usual normalizing (“natural parameter”) scale for setting confidence intervals (eg, typically the log scale for ratio parameters, and the coefficient scale in multiplicative models). If τ denotes the prior standard deviation for θ, point priors correspond to τ = 0 and equalodds priors correspond to τ = ∞.
A parameter θ with a point prior asserting it must be zero is usually inapparent (implicit) in the model because it is replaced in the model by zero and so disappears from sight. Model 1 is an example in which θ is invisible, yet the model is a special case of model 2 with the added constraint θt = 0, which in Bayesian terms corresponds to imposing the point prior Pr(θt = 0) = 1 for θ and placing equal odds on all possible combinations of α, β, and γ. Similarly, model 3 corresponds to imposing the point prior Pr{θt = 1n(2)} = 1 and placing equal odds on all possible combinations of α, β, and γ, and, but θ now remains visible in the model because it is nonzero.
If instead we assume only the more general model 2 and place equal prior odds on all combinations of α, β, γ, and θ, we obtain the following interpretations of frequentist statistics as approximate Bayesian posterior statistics:
 (a). Pθ is the posterior probability that

 is closer to θ than to the truth θt (ie, is Pθ the probability that

 .
 (b). Pθ / 2 is the posterior probability

 that is on the wrong side of θ relative to the truth θt; in particular, P _{0} / 2 is the probability that the observed association is in the wrong direction. Hence, if,

 > θ, Pθ / 2 is the posterior probability that θt < θ; if,

 < θ, Pθ / 2 is the probability that θt > θ; in particular, if

 is positive, P0 / 2 is the probability that θt is negative; if

 is negative, P0 / 2 is the probability that θt is positive.
 (c). The 95% confidence interval

 for θ becomes a 95% posterior probability interval for θ; hence, under model 2 and the prior, the probability that

 is 0.95 (parallel interpretations extend to other confidence levels).
 (d).

 is the posterior median (the odds of θt being above vs. below

 are equal).
Thus, the equalodds prior renders correct the usual misinterpretation of confidence intervals and provides Bayesian interpretations of P values.
As an example, suppose
= 1.40 and
= 0.60. Then, the following Bayesian posterior probability statements follow from model 2, the data, and an equalodds prior:
 (i).

 , giving P _{0} = 0.02 as the probability that 1.40 is closer to 0 than to θt and P _{0} / 2 = 0.01 as the probability that θt is negative or that e ^{θt} < 1.
 (ii).

 , giving P _{ln(2)} = 0.24 as the probability that 1.40 is closer to ln(2) than to θt, and P _{ln(2)}/2 = 0.12 as the probability that θt < ln(2) or that e ^{θt} < 2.
 (iii).

 , 2.58 has 95% probability of containing θt, so that the probability of e ^{0.22} = 1.25 < e ^{θt} < e ^{2.58} = 13.2 is 95%.
 (iv). The odds of θt being above versus below 1.40 is 1, meaning equal probabilities that e ^{θt} < e ^{1.40} = 4.06 and e ^{θt} > 4.06.
Taking P _{0} / 2 as the posterior probability that the observed association is in the wrong direction is found among preFisherian writers, including Gossett, the inventor of the t test,^{4},^{34},^{43} and follows from Bayesian 2 × 2 table results published in 1877.^{44},^{45} More generally, (a)–(c) follow from taking the posterior distribution of the model parameters as proportional to the likelihood function, approximating the marginal posterior for θt with a normal (
,
) distribution (hence likelihoodratio statistics will render a better approximation than Z scores).
Objections to EqualOdds Priors and Responses
One objection to the equalodds prior underlying (a)–(c) is that it is improper (it has infinite instead of 100% total probability), which leads to undefined prior probabilities even though prior odds are defined. Although some Bayesians defend direct use of improper priors,^{46} one can instead replace the equalodds prior by proper priors that are weak enough to leave properties (a)–(c) approximately correct in practice, as is done in “objective” and “reference” Bayesian methods.^{47} We will call all these priors “weak priors,” whether proper or improper.
A more serious objection is that weak priors are usually contextually absurd.^{40},^{41} Suppose θt is a log deathrate ratio for an Food and Drug Administration–approved glioma chemotherapy versus radiation therapy. Then, an equalodds prior says ln(10^{−100}), 0, and ln(10^{100}) are all equally credible values for θt. But (10^{−100}) implies complete prevention and is disproved by any death with chemotherapy, whereas ln(10^{100}) implies complete causation and is disproved by any survivor of chemotherapy.
One response to these criticisms is to take interpretations (a)–(c) as Bayesian measures of the information content of the data relative to the model, before prior information about the model parameters is added. This approach provides a contrast against Bayesian results derived from informative priors and is supported by noting that (a)–(c) are precisely the Bayesian results one obtains when no prior data are added to the actual data.^{40},^{41} ReferenceBayes analyses produce similar interpretations in that they correspond to adding almost no data.
Bounding Results from More Informative Priors
Weak priors are most absurd when all the values of θ being debated (and thus all serious candidates for a prior median) are so close to the null that it is difficult to distinguish among them or distinguish them from the null.^{40},^{41},^{48},^{49} Many, if not most, modern controversies (eg, longterm nutrient effects and drug side effects) involve debates within ranges like ¼ < RR < 4. In these settings, confidence intervals (and thus posterior intervals under weak priors) typically cover a substantial part of the range under debate, and no one would take seriously claims of (say) RR > 100.
Even without controversy, the effort and discomfort of specifying detailed priors is daunting. Not only must we rationalize all choices to ourselves and the reader, but we may have to allow for the fact that readers may have different priors from ours. Anticipating the full spectrum of these differences with sensitivity analyses of various priors is even more demanding. Thus, in a world of sharply constrained time resources, demands for informative priors leads instead to avoidance of Bayesian analysis.
An alternative, however, is to offer bounds on posterior probabilities when the prior is restricted to a given class. These bounds show the range that a sensitivity analysis over that class would produce and thus may address concerns of those wary of detailed priors. In particular, suppose one would regard it reasonable to consider as a possible prior for θt a distribution that is symmetric around a single mode (maximum) at its median θm and weak for the remaining model parameters. This class includes the most common coefficient priors, including normal, t, logistic, and certain reference distributions with median θm, and implies that exp(θt) has prior median exp(θm). Of special interest, normal (Gaussian) priors arise when the prior information about θt is derived from a number of sources or studies, none of which contributes a dominant fraction of the information; this prior produces a lognormal prior for exp(θt).
Let Pm denote the twosided P value for the prior median θm. Then, Pm / 2 approximates the smallest possible posterior probability that the observed estimate
is on the wrong side of θt ^{27},^{50}; in addition,
is the furthest the posterior median could be from the prior median θm. These interpretations also apply using uniform priors on θt with median θm. Thus, Pm / 2 provides a statistic with a correct Bayesian interpretation, without demanding detailed specification of a prior or computation beyond Pm / 2. Again, suppose
= 1.40 and
= 0.60, with weak priors for all parameters but θ. Then, P _{0} / 2 is the smallest possible posterior probability that θt is negative when the θt prior is unimodal and symmetric (eg, normal) around 0; P _{ln(2)}/2 = 0.12 is the smallest possible posterior probability that θt < ln(2) when the θt prior is unimodal and symmetric around ln(2); and 1.40 is the furthest the posterior median could be from any prior median.
Some Limitations
Within the above class of priors, the discrepancy between the lower bound Pm / 2 and the posterior probability increases as the spread of the prior decreases, albeit the discrepancy is not large until the prior becomes fairly informative relative to the data and model. To illustrate, suppose we have a normal prior with mean 0 and standard deviation τ for a log rate ratio θt = ln( RR t). This is a symmetric, unimodal prior with median 0 for θt and median 1 for RRt = exp(θt). The Table shows the resulting posterior probabilities of for RRt < 1 various estimates
= exp(
), standard errors
for
, and prior standard deviations τ for θt expressed in multiples of
. The final column uses an infinite τ, which gives back P _{0} / 2.
The Table shows a general property of normal priors: in practical terms, when the prior mean is θm, Pm / 2 will notbe far below the posterior probability that
is on the wrong side of θm if τ is over twice (so that the weight 1 / τ^{2} for the prior mean θ is less than a quarter the weight 1/
^{2} for the ordinary estimate
). Of course, if one focuses on a particular normal prior, it can be easily used to calculate directly all desired posterior intervals and probabilities^{40},^{41},^{51}; Pm / 2 then remains a reference point showing what weaker normal priors with the same mean could yield.
The bounding interpretation also assumes weak priors on the rest of the parameters. Such priors are usually unrealistic. Nonetheless, provided care is taken in specifying the regression model (in particular, by centering the regressors^{51}), posterior probabilities for the target parameter tend to be much less sensitive to reasonably imprecise priors on the remaining parameters than they are to the target (θt) prior. Regardless, stronger priors on all parameters require much more specification effort than most are willing to provide; but again, if provided, one can use them directly in Bayesian procedures,^{51} and P values can serve as reference values for readers wary of more informative priors.
All the above approximations assume use of efficient methods such as maximum likelihood under standard asymptotics^{52} and that the data distribution satisfies certain conditions met by ordinary epidemiologic models.^{27} For less efficient methods (like inverseprobability weighting), the interpretations have to be qualified technically by stating they are approximate posterior probabilities if one is given only the estimating function, rather than the complete likelihood function.
P VALUES INTO BAYES FACTORS
There is a huge literature on Bayesian interpretation of P values, most of which demands nonlinear transforms and concepts unfamiliar in basic statistics. We describe the most common example,^{29},^{30},^{33},^{34},^{52–55} although others exist.^{56}
Your Bayes factor for a hypothesis is the ratio of your posterior odds on the hypothesis to your prior odds on the hypothesis; thus, it is how much your odds change (in multiplicative terms) in light of the data.^{49},^{57},^{58} Suppose your prior odds on the null versus all other possibilities combined was 1 and your posterior odds on the null was ¼ (you would bet only 20% on it after seeing the data); then, your Bayes factor for the null would be ¼/1 = ¼. When the odds are comparing simple point hypotheses, such as θt = 0 versus θt = ln(2), the Bayes factor simplifies to the likelihood ratio.
Consider the class of “spike and smear” priors for which there is a point probability q that θt = θm (ie, a spike or point mass of probability at θm of size q), and for the remaining possibilities θt ≠ θm, the prior follows a symmetric unimodal distribution around θm. This makes q / (1−q) the prior odds on θt = θ versus θt ≠ θ. Sellke et al^{34} show that, when Pm < 1 / e=0.37, −ePmln(Pm), is an approximate lower bound on the Bayes factor; in particular, for Pm = 0.10, 0.05, and 0.01, lower bounds on the Bayes factor are 0.63, 0.41, and 0.12, respectively. They further show that for normal priors on θt ≠ θm, these bounds are even higher: 0.70, 0.47, and 0.15. Thus, if P _{0}= 0.05 a Bayesian analysis with q = ½ = 50% prior probability on the null and meanzero normal otherwise leaves a posterior probability for the null of at least 0.47 / (1 + 0.47) = 32%; this maximum drop from 50% to 32% probability for θt = 0 is nowhere near as impressive as the usual “borderline significant” description of P _{0} = 0.05. We can also find the highest value of q for which P _{0} would deserve its common misinterpretation as the posterior probability of the null. For P _{0} = 0.05, this value is a prior probability of q = 10% for θt =0.
Transforming P _{0} into a Bayes factor thus reveals a serious weakness of conventional interpretations of null P values. Bayes factors have several disadvantages, however. They cannot be read directly off tabulated estimates or P values; they require the conceptual steps of thinking in terms of Bayes factors; and they do not provide any posterior probabilities unless we specify q, an arbitrary analysis constant that represents commitment of an appreciable fraction of prior probability to a single specific value of θ. In contrast, the interpretations given earlier use only confidence intervals and P values, with no need for Bayes factors or probability spikes.
There are serious objections to the use of probability spikes in population research because of their contextual meaning.^{27},^{59} For example, a null spike represents an assertion that, with prior probability q, we have background data that prove θt=0 with absolute certainty q = ½; thus represents a 50–50 bet that there is decisive information literally proving the null. Without such information (such as a point prediction by a highly plausible physical law), a probability spike at the null is an example of “spinning knowledge out of ignorance.”^{31} This prejudice favoring the null characterizes common misinterpretations of frequentist tests^{37} and yet can lead to Bayesian conflicts with those tests.^{60} In contrast, weak priors have no spike.
Discussion
For more than 70 years, null P values have been the most common and yet most controversial inferential statistic, primarily because of their use for hypothesis testing against fixed alpha (significance) levels such as 0.05. A point often lost in statistical training, however, is that P values can be divorced from decision rules, testing, arbitrary cutoffs, and null hypotheses. In ordinary epidemiologic analyses, twosided P values can instead be used as probability measures (with both frequentist and Bayesian meaning) of the distance from (or fit of) the entire set assumptions (including model and validity assumptions and the explicit assumption of θt = θ) to the data. Onesided P values such as P _{0} / 2 can be used to measure support for or probability of their onesided hypotheses, given the remaining assumptions. Large P values are then as telling as small ones; for example, if P _{0} / 2 is large, then, considered in isolation, the study seems vague about the direction of an association when viewed through the same model lens used to compute the estimate
and the confidence interval. Furthermore, if one uses an informative prior to derive the posterior probability of the point estimate being in the wrong direction, P _{0} / 2 provides a reference point indicating how much the prior information influenced that posterior probability.
Confidence intervals provide valuable information about the precision implied by the data^{12},^{17} given the assumed model; hence, we would certainly not wish to see them replaced by P values. However, confidence intervals are rarely precise about the relation of observations to prespecified parameter values, which in part explains why null P values can be viewed as defensible supplements to confidence intervals when the null has legitimate claim to our attention (as in exploratory and many highly controversial settings).^{4},^{10} Furthermore, the frequency behavior and posterior interpretation of P _{0} are more robust to model misspecification than are confidence intervals and nonnull P values, which is worth noting given that we are always uncertain about the model form.^{5} They are also more robust to validity problems that produce bias toward the null, especially certain simple types of measurement error or misclassification of the exposure or disease,^{61} where, under a weak prior, P _{0} / 2 remains the posterior probability that the estimate is in the wrong direction,^{62} whereas confidence intervals become invalid.
Even when the null is an arbitrary hypothesis unworthy of special attention, nullrelated questions are encouraged by automatic software emission of P _{0}. Again, incorrect Bayesian interpretations of P _{0} remain the norm, encouraged by some instructional sources.^{1},^{2},^{63} It thus seems important to displace these misinterpretations with correct ones. With correct interpretation of P _{0} in mind, extensions of P values to θ of greater contextual interest is simple, and extensions to more realistic priors can also be obtained by translating the priors to data and adding them to the conventional analysis.^{41},^{51}
After one separates P values from arbitrary testing cutoffs such as 0.05 and goes beyond the null, one may see confidence intervals as problematic for encouraging degraded binary inferences.^{28} Asking whether a value θ is in or out of the confidence interval turns the confidence interval into a 0.05level test of that value. If we try to be more precise, we hit failings of intuition. For example, if the 95% confidence limits for RR are 1.8 and 3.3, what is the onesided P value forRR ≤ 2? Few would recognize quickly that it is P _{ln(2)}/2 = 0.10, giving a 10% posterior probability that under a weak prior for all model parameters, and at least 10% if the prior for RR was more sharply specified as lognormal with median at 2.
Failure to appreciate the complementarity of confidence intervals and P values may stem more from historical accidents in how the two statistics entered common usage.^{22} The chief disaster is that for generations Bayesian perspectives were banished from introductory statistics. Much like what happened with other attempts at knowledge suppression, users have filled the gap with interpretations that miss crucial subtleties and are often simply false. A common false interpretation is that P _{0} is the probability that chance alone produced the observed association. Unfortunately, the premise that chance alone operated is logically the same as the null hypothesis (“chance alone” implies there is no effect and no bias), and thus, this misinterpretation is the same as claiming P _{0} is the probability of the null.^{11},^{37} Correct Bayesian interpretations can address this problem.
Given the extensive misinterpretations of frequentist statistics and the enormous (some would say impossible)^{64} demands made by fully Bayesian analyses, a serious argument can be made for deemphasizing (if not eliminating) inferential statistics in favor of more data presentation such as tables of raw numbers.^{65} After all, the only purpose of a single study is to contribute its description and data to a broad pool of relevant information.^{66} Without an inferential ban, however, an improvement of practice will require reeducation, not restriction.
REFERENCES
1. Glantz SA Primer of Biostatistics. 20015th ed New York, NY McGrawHill
2. Cordova ML. Giving clinicians more to work with: let’s incorporate confidence intervals into our data. J Athl Train. 2007;42:445
3. Ware JH, Mosteller F, Delgado F, Donnelly C, Ingelfinger JABailar JC, Mosteller F. P values. In: Medical Uses of Statistics. 1992 Boston, MA NEJM Books:181–200
4. Senn S. Two cheers for Pvalues?. J Epidemiol Biostat. 2001;6:193–204; discussion 205
5. Weinberg CR. It’s time to rehabilitate the Pvalue. Epidemiology. 2001;12:288–290
6. Sterne JA. Teaching hypothesis tests–time for significant change?. Stat Med. 2002;21:985–994; discussion 995
7. Whitley E, Ball J. Statistics review 3: hypothesis testing and P values. Crit Care. 2002;6:222–225
8. Goodman SN. Toward evidencebased medical statistics. 1: The P value fallacy. Ann Intern Med. 1999;130:995–1004
9. Pocock SJ, Ware JH. Translating statistical findings into plain English. Lancet. 2009;373:1926–1928
10. VanderWeele TJ. Re: ‘‘The ongoing tyranny of statistical significance testing in biomedical research.’’. Eur J Epidemiol. 2010;25:843–844
11. Greenland S, Poole C. Problems in common interpretations of statistics in scientific articles, expert reports, and testimony. Jurimetrics. 2011;51:113–129
12. Poole C. Low Pvalues or narrow confidence intervals: which are more durable?. Epidemiology. 2001;12:291–294
13. Goodman S. A dirty dozen: twelve pvalue misconceptions. Semin Hematol. 2008;45:135–140
14. Stang A, Poole C, Kuss O. The ongoing tyranny of statistical significance testing in biomedical research. Eur J Epidemiol. 2010;25:225–230
15. Poole C, Kuss O, Stang A. On a use of the null Pvalue (letter). Eur J Epidemiol. 2010;25:844–845, erratum 899–900.
16. Lang JM, Rothman KJ, Cann CI. That confounded Pvalue. Epidemiology. 1998;9:7–8
17. Rothman KJ. Curbing type I and type II errors. Eur J Epidemiol. 2010;25:223–224
18. Rothman KJ. A show of confidence. N Engl J Med. 1978;299:1362–1363
19. Rothman KJ. Significance questing. Ann Intern Med. 1986;105:445–447
20. Cohen J. The earth is round (p <.05). Amer Psychol. 1994;49:997–1003
21. Fisher RA Statistical Methods for Research Workers. 197014th ed New York, NY Oxford
22. Goodman SN. P values, hypothesis tests, and likelihood: implications for epidemiology of a neglected historical debate. . Am J Epidemiol. 1993;137:485–496
23. Cox DR. The role of significance tests (with discussion). Scand J Stat. 1977;4:49–70
24. Cox DR. Statistical significance tests. Br J Clin Pharmacol. 1982;14:325–331
25. Poole C. Beyond the confidence interval. Am J Public Health. 1987;77:195–199
26. Rothman KJ, Greenland S, Lash TLRothman KJ, Greenland S, Lash TL. Precision and statistics in epidemiologic studies.. . In: Modern Epidemiology.. 20083rd ed Philadelphia, PA LippincottWoltersKluwer:148–167
27. Casella G, Berger RL. Reconciling Bayesian and frequentist evidence in the onesided testing problem. J Am Stat Assoc. 1987;82:106–111
28. Poole C. Confidence intervals exclude nothing. Am J Public Health. 1987;77:492–493
29. Berger JO, Delampady M. Testing precise hypotheses (with discussion). Stat Sci. 1987;2:317–352
30. Berger JO, Sellke T. Testing a point null hypothesis: the irreconcilability of p values and evidence (with discussion). J Am Stat Assoc. 1987;82:112–139
31. Poole C. Causal values. Epidemiology. 2001;12:139–141
32. Schervish M. Pvalues: what they are and what they are not. Am Stat. 1996;50:203–206
33. Goodman SN. Of Pvalues and Bayes: a modest proposal. Epidemiology. 2001;12:295–297
34. Sellke T, Bayarri MJ, Berger JO. Calibration of p values for testing precise null hypotheses. Am Stat. 2001;55:62–71
35. Senn S. A comment on replication, pvalues and evidence, S.N.Goodman, Statistics in Medicine 1992; 11:875879. Stat Med. 2002;21:2437–2444; author reply 2445
36. Hubbard R, Lindsay RL. Why P values are not a useful measure of evidence in statistical significance testing. Theory Psychol.. 2008;18:69–88
37. Greenland S. Null misinterpretation in statistical testing and its impact on health risk assessment. Prev Med. 2011;53:225–228
38. Greenland S, Robins JM. Epidemiology, justice, and the probability of causation. Jurimetrics.. 2000;40:321–340
39. Pearson K. On the criterion that a given system of deviations from the probable in the case of a correlated cystem of variables is such that it can be reasonably supposed to have arisen from random sampling. Philosophical Magazine Series 5. 1900;50:157–175
40. Greenland S. Bayesian perspectives for epidemiological research: I. Foundations and basic methods. Int J Epidemiol. 2006;35:765–775
41. Greenland SRothman KJ, Greenland S, Lash TL. Introduction to Bayesian statistics. In: Modern Epidemiology. 20083rd ed Philadelphia, PA LippincottWoltersKluwer:328–344
42. Held LAhrens W, Pigeot I. An introduction to Bayesian methods with applications in epidemiology. In: Handbook of Epidemiology. 2013; New York, NY Springer in press
43. Gosset WS. The probable error of a mean. Biometrika. 1908;VI:1–25
44. Seneta E. Carl Liebermeister’s hypergeometric tails. Historia Mathematica.. 1994;21:453–462
45. Nurminen M, Mutanen P. Exact Bayesian analysis of two proportions. Scand J Statistics. 1987;14:67–77
46. Taraldsen G, Lindqvist BH. Improper priors are not improper. Am Stat.. 2010;64:154–158
47. Berger JO. The case for objective Bayesian analysis (with discussion). Bayesian Anal.. 2006;1:385–472
48. DeGroot MH. Comment on Shafer. J Am Stat Assoc.. 1982;77:336–339
49. Greenland S. Probability logic and probabilistic induction. Epidemiology. 1998;9:322–332
50. Greenland S. Re: “P values, hypothesis tests, and likelihood: implications for epidemiology of a neglected historical debate” (letter). Am J Epidemiol. 1994;140:116–117
51. Greenland S. Bayesian perspectives for epidemiologic research. II. Regression analysis. Int J Epidemiol. 2007;36:195–202
52. Lindley DV Introduction to Probability and Statistics from a Bayesian Viewpoint; Part 2, Inference.. 1965 London, UK Cambridge University Press;
53. Edwards W, Lindman H, Savage L. Bayesian statistical inference for psychological research. Psychol Rev.. 1963;70:193–242
54. Pratt JW. Bayesian interpretation of standard inference statements. J Royal Stat Soc B. 1965;27:169–203
55. DeGroot MH. Doing what comes naturally: interpreting a tail area as a posterior probability or as a likelihood ratio. J Am Stat Assoc. 1973;68:966–969
56. Hodges JBernardo JM, Berger JO, Dawid AP, Smith AFM. Who knows what alternative lurks in the heart of significance tests?. In: Bayesian Statistics 4. 1992 New York, NY Oxford University Press;:247–266
57. Good IJ Good Thinking. 1983 Minneapolis, MN U Minn Press:140–143
58. Goodman SN. Toward evidencebased medical statistics. 2: The Bayes factor. Ann Intern Med. 1999;130:1005–1013
59. Casella G, Berger RL. Comment. Stat Sci. 1987;2:344–417
60. Lindley DV.. A statistical paradox. Biometrika.. 1957;44:187–192
61. Carroll RJ, Ruppert D, Stefanski LA, Crainiceanu C Measurement Error in Nonlinear Models. 2006 Boca Raton, FL Chapman and Hall
62. Greenland S, Gustafson P. Accounting for independent nondifferential misclassification does not increase certainty that an observed association is in the correct direction. Am J Epidemiol. 2006;164:63–68
64. Senn SJ. You may believe you are a Bayesian, but you are probably wrong. RMM.. 2011;2:48–66.
65. Greenland S, GagoDominguez M, Castelao JE. The value of riskfactor (“blackbox”) epidemiology. Epidemiology. 2004;15:529–535
66. Poole C, Peters U, Il’yasova D, Arab L. Commentary: this study failed?. Int J Epidemiol. 2003;32:534–535