When John West and Cynthia Osborn assumed coeditorship of Counselor Education and Supervision (CES), they articulated their vision for the future of the journal in an editorial in the December 2006 issue. They spoke of a desire to publish data-based manuscripts "that describe research conducted in a disciplined, systematic, and rigorous manner (vs. research conducted simplistically, quickly, and conveniently)" (West & Osborne, 2006, p. 83). For manuscripts that contain results of quantitative research, the Publication Manual of the American Psychological Association (APA; 2001) outlines specific criteria to meet the standards of rigor and discipline. These include (among other criteria) a research design that "fully and unambiguously test[s] the hypothesis ... [and a sample that is] representative of the population to which generalizations are made" (p. 6).
Manuscripts that present the results of quantitative research make an important contribution, both to the journal and to the field. Quantitative research, in tandem with and often informed by, qualitative research, helps counselor educators, supervisors, counseling students, and practicing counselors make informed choices about what interventions they choose to use or avoid in their work. The impact of research published in CES can be significant, and it can serve as the foundation for much of the work in the field of counselor education. Thus, all potential contributors to the journal have a responsibility to ensure that the research contained in their manuscripts is of the highest quality and can be relied on to provide up-to-date and accurate information to the journal readership. What follow are some general guidelines and reminders about the foundations of quality quantitative research. The list of topics was generated at the request of the coeditors of CES and in consultation with them. The list is not intended to be exhaustive nor to inform actual statistical methodology, but to outline basic parameters that are used to guide editorial decisions about manuscripts submitted to the journal and to encourage and support the work of potential authors of quantitative manuscripts. This editorial is intended as both an overview of expectations for beginning scholars and a reminder for more seasoned researchers.
Editors' Note. Darcy Haag Granello is CES associate editor for quantitative manuscripts.
Darcy Haag Granello, Counselor Education Program, School of Physical Activity and Educational Services, The Ohio State University. Correspondence concerning this article should be addressed to Darcy Haag Granello, Counselor Education Program, School of Physical Activity and Educational Services, The Ohio State University, 448 PAES Building, 305 West 17th Avenue, Columbus, OH 43210 (e-mail: email@example.com).
Topics to Consider in the Development and Publication of Quantitative Studies
The abstract is an "accurate, succinct, quickly comprehensible, and informative" (APA, 2001, p. 15) summary of the manuscript. Authors sometimes mistakenly withhold results of their research from the abstract, using it instead as a "teaser" to encourage readers by telling them what the manuscript will contain, instead of simply summarizing all the contents. However, according to APA guidelines, abstracts should be self-contained, and readers who have access only to an abstract should have a basic understanding of all components of a manuscript. For quantitative manuscripts, this means that an abstract includes the problem that was investigated, the participants, the experimental methods, the results, and the conclusions and implications (APA, 2001). Admittedly, incorporating all of this information in 50 to 100 words is challenging, but it can be done with a succinct and well-written abstract that is typically written after the rest of the manuscript has been completed, thus allowing the author to identify and highlight the major points.
The Literature Review
The beginning of a quantitative manuscript introduces the specific problem and provides the reader with a background literature review. This review should be focused and include only research and writing that is relevant to the problem, not an exhaustive history of the issue with all of its tangential implications and findings. The literature review is intended to "demonstrate the logical continuity between previous and present work" and to "develop the problem with enough breadth and clarity to make it generally understood by as wide a professional audience as possible" (APA, 2001, p. 16). The primary audience for CES consists of persons who have been trained or are in training as counselor educators and supervisors and, thus, have some knowledge of the basic constructs of counselor education and supervision. When highly specialized subtopics within the field are introduced, they will require further elaboration and discussion. At the conclusion of the introduction, the purpose and rationale of the current research and the hypotheses of the research should be clearly articulated.
Design. Readers of quantitative manuscripts should be given sufficient information to understand the type of study that was conducted. Quantitative research encompasses a breadth of methodology, including case reports, controlled experiments, quasi-experiments, statistical simulations, surveys, observational studies, and meta-analyses (Wilkinson & the Task Force on Statistical Inference, 1999). Authors should include a statement describing the research design and statistics used (including a clear articulation of all of the independent and dependent variables and how they were measured) and a succinct explanation of what statistical methods were used to address each hypothesis. The appropriateness of the statistical methods used should be clearly demonstrated. Reviewers, readers, and other researchers appreciate statements such as, "To address the first research question, we used the AA statistical test. X was measured by XX instrument and used as the independent variable. Y was measured by YY and used as the dependent variable." A clear description of the statistical tests and all variables used is extraordinarily helpful for those unfamiliar with the specific research project. Because researchers are so close to their own research, they may make the assumption that others are able to follow their reasoning or infer from their listing of instruments which scores were used for each specific variable. Enlisting the assistance of another researcher who is outside the scope of the author(s)' current project is particularly beneficial in ensuring clarity. A researcher unfamiliar with the project should be able to replicate the design from the information given in the manuscript. For example, the person enlisted to help may be asked to read the Methods section of the manuscript and then be asked if he or she could describe the research design (independent and dependent variables, research questions, and specific methodology used) from the information given.
Population. It is crucial that the participants in any research investigation be recruited and selected from a well-defined population that is relevant to the research question(s). The interpretation of results of any research depends on the population intended for analysis, and authors should not overextend the population to whom they intend to apply their results. In most research, it is hazardous to draw inferences about populations that are considerably removed from the circumstances of the sample in the study (Popham & Sirotnik, 1992). For example, a study of doctoral supervisors at one university is not automatically applicable to all supervisors or even to all doctoral supervisors. The population from which the sample is drawn affects almost every conclusion in a manuscript (Wilkinson & the Task Force on Statistical Inference, 1999).
Sample. The degree to which the sample matches the population affects almost every conclusion that can be made in a manuscript. Inferential statistics are, at their core, only as strong as the ability to make inferences from the data collected. The ability to make these inferences is a function of the equivalence between the sample and the population to which the inference is to be applied (Popham & Sirotnik, 1992). In their foundational work on quantitative research design, Campbell and Stanley (1963) articulated six principles of quality research, two of which applied to sample. The first principle is that, whenever possible, the sample should be randomized. The second principle is that extraneous factors should be controlled for, to the highest degree possible, through selection, stratification, or, if feasible, statistical adjustment. In other words, Campbell and Stanley argued that the selection of the sample is the cornerstone of quality quantitative research. Thus, samples that are stratified or are randomly assigned are ideal. In practice, however, stratified sampling is not always possible, and convenience sampling is often used. Convenience samples should be identified as such in the manuscript, with the recognition that sampling based on accessibility or availability may significantly compromise the generalizability of the results and should be used only when other options are unavailable (Campbell & Stanley, 1963).
Response rates. Research must include sample response rates and a discussion of how the participants who were included in the research matched the characteristics of the intended sample. Low response rates are problematic because they imply that the typical respondent is somehow atypical of the sample as a whole. In other words, if only 25% of persons sampled through a survey actually responded, then it is the atypical person who is included in the research, one who is in the numerical minority of the sample and who is, in at least one significant way, different from the majority of the sample. There is no absolute cut-off for minimum response rates, and special circumstances must always be taken into consideration when determining the adequacy of a response rate for a particular study. However, in general, the lower the response rate, the more difficult it becomes for researchers to argue that their respondents are representative of the sample.
Many different strategies are available to increase survey response rates. Some of the most commonly used methods include multiple mailings, personalization of solicitations, and monetary or other incentives. A meta-analysis of 292 randomized controlled trials of response rates compared the odds of response for 75 different strategies designed to increase response rates (Edwards et al., 2002). The average number of participants per study included in the meta-analysis was 1,091, and the total number of participants in all the studies was 258,315. To be included in the meta-analysis, each study had to meet the inclusion criteria of random assignment of participants into two or more different strategies to improve response rate. Odds ratios compared each strategy against a baseline of no strategy, for example, monetary incentive versus no monetary incentive, or personalized letter versus no personalized letter. Several methods were found to more than double response rates (odds ratios of more than 2.0). The methods that were most effective for increasing response rates (in order of decreasing effectiveness) were (a) designing questionnaires of strong interest to participants (odds ratio: 2.44); (b) sending questionnaires via registered mail (odds ratio: 2.21); (c) providing surveys with monetary incentive (odds ratio: 2.02); (d) making questionnaires short (odds ratio: 1.86); (e) contacting participants by postal mail before sending questionnaires (odds ratio: 1.54); (f) using follow-up contact (odds ratio: 1.44); (g) providing nonrespondents with a second copy of the questionnaire (odds ratio: 1.41); (h) printing all or part of the questionnaire in colored ink (odds ratio: 1.39); (i) having questionnaires originate from universities, rather than commercial organizations (odds ratio: 1.31); (j) including a stamped return envelope (odds ratio: 1.26); (k) sending the survey via first class mail (odds ratio: 1.12); and (l) personalizing questionnaires and letters (odds ratio: 1.16). Researchers wishing to engage in methods to increase response rates are encouraged to consult the entire article by Edwards and his colleagues for more specific information on effectiveness of the various strategies. Whether relying on the Edwards et al. article, or other sources, or on the accumulated wisdom of the profession, it is clear that there are many strategies that researchers can use that have the potential to significantly increase survey response rates. One-shot postal mailings are seldom adequate to garner a sufficient response rate.
Online data collection. As more and more researchers consider the use of electronic methods to enhance both the efficiency and the cost-effectiveness of collecting data, the question of how to maintain the scholarly rigor of such research has come to the attention of the CES editorial board. Survey research conducted over the Internet or via e-mail has many strengths that make it particularly appealing to researchers. Advantages include reduction in the time needed to receive results, reduced costs, ease of data entry, flexible formatting, and easy access to persons in populations who are not currently engaged in the mental health or educational system (e.g., persons with mental illness who are not currently receiving treatment), a strength that may be especially relevant to researchers in the field of counseling (Granello & Wheaton, 2004).
There are, however, limitations that must be overcome in any online research endeavor. Primary among these are sample selection bias and reduced response rates. It is unwise to assume equal comfort with technology and equal access to up-to-date computer hardware and software among individuals selected to participate in a study. Furthermore, little is known from a research perspective about participants' perceptions of an e-mail request to participate in a survey versus their perception of a survey they receive through postal mail. Research on sample selection bias for Internet- and e-mail-based surveys is in its infancy (Schonlau, 2004). It is incumbent upon researchers to indicate how selection bias has been addressed in their studies.
The second limitation of online research, reduced response rates, is more difficult to address within a research design. Some online data collection begins with a very specific e-mail or postal mail solicitation to an identified sample, often with a specific password required to complete the survey. Other studies have used postal mail to send hard copies of instruments with information about Web links that allow respondents to answer either in hard copy or by using an electronic version (Lusk, Delclos, Burau, Drawhorn, & Aday, 2007). With other surveys sent to listservs and large-scale samples, little effort is made to target specific respondents. Some surveys simply exist on the Web and are available to anyone surfing the Web who happens upon the Web site of such a survey. Although results vary from study to study, response rates have been reported that were 10% to 20% lower for Web-based surveys than for traditional mail surveys (Leece et al., 2004; Mavis & Brocato, 1998). Some researchers have even reported response rates more than 50% lower in Web-based surveys (Jones & Pitt, 1999). In a study of more than 5,000 health professionals, Lusk et al. (2007) found that, given a choice, participants in their study overwhelming preferred postal mail surveys over Web-based surveys, with 90% of the respondents in their study electing to complete the hard copy instrument. Web-based surveys that solicit participants from broad sectors of the population via e-mail often have even lower responses, and there is considerable debate about the appropriate calculation of response rates for this type of survey.
Accurate measurement of nonresponse to Internet surveys is confounded by technology such as spare blockers, which may prevent initial e-mail solicitations from ever being opened or read (Bowling et al., 2006). Methods to improve response rates are imperative in electronic survey design. Multiple solicitations may be required (Sills & Song, 2002). Introductory letters (which may be sent via e-mail) may enhance participation rates if they are personalized or if the letter is written by an individual who has what is perceived as a high level of power (Joinson & Reips, 2007). An important consideration is the development of strategies to bypass span blockers. Regardless of the tactic used, researchers engaging in online data collection may need to make an extra effort to enhance response rates. As with postal mail or in-person surveys, response rates should be included in all publications (Eysenbach, 2004). The most important message regarding online data collection is that the sane standards for quality research apply, regardless of what mechanism is used to identify, solicit, or study participants.
Instrumentation. Instrumentation determines the type of data that can be obtained and how that data can be used to answer the research questions. Thus, the choices regarding instrumentation must be made with care and should be guided by the basic research questions that are being investigated. Whenever possible, researchers are strongly encouraged to use existing instruments with established validity and reliability rather than attempting to develop their own (Granello & Granello, 2001). Once variables in the study have been clearly defined, it is important to establish an explicit link between the variables and the measures used. The adequacy and appropriateness of instruments selected should be clearly articulated. This includes providing an overview of previously published psychometric data on validity and reliability for the total score, subscales, or factors. Even if the focus of the study is not on the psychometrics of the instrument, authors are encouraged to provide internal reliability coefficients for all instruments, using the data collected in the current study, because this contributes to the existing body of psychometric research about the instruments and is important information for future researchers (Wilkinson & the Task Force on Statistical Inference, 1999). Correlations among instruments (total scores and scores of subscales or factors) are also typically included, particularly when independence of the measures is a requirement of the statistical methods used.
It is not uncommon for researchers in counselor education to explore a topic that does not have an existing measure or that relies on instruments with very poor psychometric properties. If researcher-developed instruments are used for the first time, researchers should use review (or "expert") panels and pilot studies, whenever possible. Instruments that are poorly worded, ambiguous, not explicitly linked to the construct being studied, or contain researcher bias can prevent an otherwise well-researched study from being published.
Intervention. Experimental design research involves assignment of participants to one or more treatment conditions, the application of an intervention, and the comparison of results. In the Methods section of the manuscript, authors provide a detailed description of the intervention (e.g., specific teaching strategy, particular supervisory approach) and a description of the comparison group. According to the Publication Manual of the American Psychological Association (APA, 2001), "such a description enables the reader to evaluate the appropriateness of your methods and the reliability and validity of your results. It also permits experienced investigators to replicate the study if they so desire" (p. 17).
Statistical power Statistical power refers to "the ability of a statistical test to detect relationships between variables" (Newton & Rudestam 1999, p. 70). Statistical power measures the probability of finding a statistically significant result, if such a difference actually exists in the population being studied. All statistical tests carry with them the probability of either Type I (erroneously rejecting a true null hypothesis or concluding a statistical relationship or difference when there is none) or Type II (concluding that there is no statistical relationship or difference when there is one) errors. The lower the Type II (beta) error, the higher the statistical power of the test. In other words, if the power of a test is high and if the phenomenon exists in the population, then there is a high probability that the statistical test will detect that phenomenon. If the power of a test is low, then even if a phenomenon exists in the population, it is unlikely that it will emerge as statistically significant in the research. Clearly, the power of a statistical test is an essential component of interpreting research. If the power of a test is high, then researchers (and readers) can have much greater confidence in the results of a study. If the power is low, then it is unwise to place much confidence in the results.
Largely because of the work of Cohen (1988), there is more widespread recognition of the need to include a power analysis in research. Technically, a power analysis should be conducted before the research begins, using tables such as those provided by Cohen to determine appropriate sample size. Most statistical packages also provide post hoc analyses of power that can simply be added to the results generated. In general, a power of 0.8 (80 % of the time the researcher would find an effect that exists in the population) is the minimal level of power that researchers deem acceptable to conduct meaningful statistical analyses (Newton & Rudestam, 1999).
Power is a direct result of four variables: (a) alpha level, (b) sample size, (c) effect size, and (d) the type of statistical test being conducted. One way to increase power is to increase the alpha level, for example from .05 to. 10. However, most researchers would be hesitant to increase the alpha level above the .05 level that is traditionally used in social sciences research because the trade-off for the higher power would be a higher probability of a Type I error. The most commonly used method to increase power is to increase sample size. Because power is such an essential aspect of research, the APA Board of Scientific Affairs declared in 1999 that all research published in APA journals should include an analysis of power (Wilkinson & the Task Force on Statistical Inference, 1999). These guidelines are also appropriate for researchers who wish to publish in CES. The results of studies that include a power analysis are much more beneficial to readers. In addition, studies with a power analysis enhance the quality of generated research in the field, which is one of the prime responsibilities of the journal.
The Results section "summarizes the data collected and the statistical
or data analytic treatment used" (APA, 2001, p. 20). Results from all statistical tests are included in the Results section whether or not they support the original hypotheses or are statistically significant. In many cases, data can be presented with the aid of tables or figures. When reporting statistical information, authors should provide "sufficient information to help the reader fully understand the analyses conducted" (APA, 2001, p. 23). The Results section is not the place for commentary or discussion.
Effect size. There are many ways to measure effect size, and each of them represents an attempt to measure the size or strength of the relationship between two or more variables in the population. In other words, effect size is the magnitude of the effect or outcome measured. Many researchers refer to effect size as the practical significance of the research results. Studies with large sample sizes and high power can detect even very small effect sizes. In these instances, results can be statistically significant (with reported p values of < .001), but the practical results of these studies are less evident. What is the practical significance, for example, of studying two large samples and finding a difference of three IQ points between the two, even if those results are significant at the p < .05 level? Although the evidence is statistically significant, in practice, there is not much meaning in this difference. In 1999, the APA advised that all future publications of research should include effect sizes to prevent readers from making decisions based on statistical, rather than practical, findings (Wilkinson & the Task Force on Statistical Inference, 1999). In a 2005 review of the research published in the Journal of Counseling & Development between 1990 and 2001, Bangert and Baumberger found that fewer than one half of all published studies (43%) included any measures of effect size. They strongly recommended the inclusion of effect sizes in all future research studies published within the field of counseling, arguing that this would allow the practitioner who relies on such research "a more user-friendly way to evaluate the importance of research outcomes" (Bangert & Baumberger, 2005, p. 485).
Measures of effect size. Cohen (1988) suggested using an effect size index (referred to as Cohen's d) that is expressed in standard deviation (SD) units. Greater differences between the means of the two (or more) groups would be expressed by higher effect sizes. By using this measure, comparisons can be made across studies, across instruments, and across populations. Cohen argued that effect size differences between two groups that were at least .80 SD should be considered large, .50 SD should be considered medium, and .20 SD considered small. In analysis of variance, [eta.sup.2] is the effect size most commonly used, and it is used to represent the percentage of variability explained by the statistical phenomenon under study. In other words, it represents the percentage of the dependent variable that is accounted for by the independent variable(s). For [eta.sup.2], Cohen's determination of small,. 10; medium, .25; and large, .40 effect sizes is used. In correlations, the size of the correlation itself is a measure of effect size. Again, Cohen's determination of small, .10; medium, .30; and large, .50 remains the standard. Finally, in regression equations, effect sizes are typically reported as [f.sup.2], or the proportion of variance explained by the independent variables divided by the proportion of variance attributed to error ([R.sup.2] / 1 - [R.sup.2]). For [f.sup.2], effect sizes are small, .02; medium, .15; and large, .35. Regardless of which measure is used, effect sizes are an essential component of research and should be considered in the results. Because of the many different ways to report effect sizes, it is important for researchers to specifically state what measure they have included to report effect sizes, to include the determination of the strength of the effect (based on Cohen's levels or another respected reporting mechanism), and to discuss how this measure relates to the practical significance of their findings.
The Discussion section of a manuscript should focus on "credibility, generalizability, and robustness" (Wilkinson & the Task Force on Statistical Inference, 1999, p. 602), and in this section, authors "examine, interpret, and qualify the results [of the study, and] ... draw inferences from them" (APA, 2001, p. 26). This is the point at which authors compare their results with those from previous studies and theory. Authors are encouraged to include a discussion of the effect sizes from the current research and how these relate to results from previous studies. The Discussion section is not the place to introduce new theories or reviews of previous research, but links should be made to those items that were first introduced in the literature review. Researchers are cautioned not to overgeneralize results, nor to infer causality from nonrandomized designs. Rather, it is in the Discussion section that results of the study are interpreted, and the importance of the findings is articulated.
Careful and thoughtful recommendations or implications for future research are essential components of the Discussion section. Certainly every study could end with the phrase "more research must be conducted in this area," but this general statement is of little value to readers or to future researchers. Instead, an attentive analysis from one who has completed research in the area to help future researchers avoid pitfalls and build on strengths is much more useful. One of the essential components of this section of the manuscript is to discuss the implications of the findings for the reader. The Discussion section does not need to be long, but neither should it be a perfunctory recitation of previously made points. In CES, implications can emphasize the relevance of the research for those in the field, with emphasis on how the research contributes to the development of understanding of the phenomenon under study.
Quantitative Research and the Generative Conversation
Quantitative research is an essential component of the ever developing understandings in the profession of counseling, and publication of that research through peer-reviewed professional journals is one of the most commonly accepted methods for the dissemination of those professional perspectives. The primary purpose of CES is to publish quality manuscripts that expand and enhance those professional perspectives, and the goal of the entire editorial board is to assist in that process. With that goal in mind, the recommendations and expectations contained in this editorial are provided to assist potential authors and researchers with their research design and implementation as well as with their manuscript preparation. We hope to continue to publish the very highest quality research in the field, maintaining the journal's legacy of excellence. Clearly, this effort requires a shared responsibility. Contributing original research is one of the most important contributions a scholar can make to his or her field, and it is only through the dedicated efforts of potential authors and researchers that the excellence of CES will be maintained and strengthened in the years to come.
American Psychological Association. (2001). Publication manual of the American Psychological Association (5th ed.). Washington, DC: Author.
Bangert, A. W., & Baumberger, J. P. (2005). Research and statistical techniques used in the Journal of Counseling & Development. 1990-2001. Journal of Counseling & Development, 83, 480-487.
Bowling, J. M., Rimer, B. K., Lyons, E. J., Golin, C. E., Frydman, G., & Ribisl, K. M. (2006). Methodologic challenges of e-health research. Evaluation and Program Planning, 29, 390-396.
Campbell, D. T., & Stanley, J. C. (1963). Experimental and quasi-experimental designs for research. Chicago: Rand McNally.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). New York: Academic Press.
Edwards, P., Roberts, I., Clarke, M., DiGuiseppi, C., Pratap, S., Wentz, R., & Kwan, I. (2002). Increasing response rates to postal questionnaires: Systematic review. British Medical Journal, 324, 1183-1191.
Eysenhach, G. (2004). Tackling publication bias and selective reporting in health informatics research: Register your eHealth trials in the International eHealth Studies Registry. Journal of Medical and Internet Research, 6, e35.
Granello, D. H., & Granello, P. F. (2001). Counseling outcome research: Making practical choices for real-world applications. In G. R. Walz & J. C. Bleuer (Eds.), Assessment: Issues and challenges for the millennium (pp. 163-172). Greensboro, NC: ERIC/CASS.
Granello, D. H., & Wheaton, J. E. (2004). On-line data collection: Strategies for research. Journal of Counseling & Development, 82, 387-393.
Joinson, A. N., & Reips, U.-D. (2007). Personalized salutation, power of sender and response rates to Web-based surveys. Computers in Human Behavior, 23, 1372-1383.
Jones, R., & Pitt, N. (1999). Health surveys in the workplace: Comparison of postal email and World Wide Web methods. Occupational Medicine, 49, 556-558.
Leece, P., Bhandari, M., Sprague, S., Swiontkowski, M. F., Schemitsch, E. H., Tornetta, P., et al. (2004). Internet versus mailed questionnaires: A randomized comparison (2). Journal of Medical Internet Research, 6. Retrieved June 19, 2007, from http://www.jmir.org/2004/3/e30
Lusk, C., Delclos, G. L., Burau, K., Drawhorn, D. D., & Aday, L. A. (2007). Mail versus Internet surveys: Determinants of method of response preferences among health professionals. Evaluation and the Health Professions, 30, 186-201.
Mavis, B., & Brocato, J. (1998). Postal surveys versus electronic mail surveys: The tortoise and the hare revisited. Evaluation and the Health Professions, 21, 395-408.
Newton, R. R., & Rudestam, K. E. (1999). Your statistical consultant: Answers to your data analysis questions. Thousand Oaks, CA: Sage.
Popham, W. J., & Sirotnik, K.A. (1992). Understanding statistics in education. Itasca, IL: Peacock.
Schonlau, M. (2004). Will Web surveys ever become part of mainstream research? Journal of Medical Internet Research, 6. Retrieved June 19, 2007, from http:// www.jmir.org/2004/3/e31
Sills, S. J., & Song, C. (2002). Innovations in survey research: An application of Web-based surveys. Social Science Computer Review, 20, 22-30.
West, J. D., & Osborn, C. J. (2006). Enhancing the profession through published scholarship: The purpose of Counselor Education and Supervision. Counselor Education and Supervision, 46, 82-87.
Wilkinson, L., & the Task Force on Statistical Inference. (1999). Statistical methods in psychology journals. American Psychologist, 54, 594-604.…