Confused about Copyright ? Assessing Researchers ’ Comprehension of Copyright Transfer Agreements

INTRODUCTION Academic authors’ confusion about copyright and publisher policy is often cited as a challenge to their effective sharing of their own published research, from having a chilling effect on selfarchiving in institutional and subject repositories, to leading to the posting of versions of articles on social networking sites in contravention of publisher policy and beyond. This study seeks to determine the extent to which authors understand the terms of these policies as expressed in publishers’ copyright transfer agreements (CTAs), taking into account such factors as the authors’ disciplines and publishing experience, as well as the wording and structure of these agreements. METHODS We distributed an online survey experiment to corresponding authors of academic research articles indexed in the Scopus database. Participants were randomly assigned to read one of two copyright transfer agreements and were subsequently asked to answer a series of questions about these agreements to determine their level of comprehension. The survey was sent to 3,154 participants, with 122 responding, representing a 4% response rate. Basic demographic information as well as information about participants’ previous publishing experience was also collected. We analyzed the survey data using Ordinary Least Squared (OLS) regressions and probit regressions. RESULTS AND DISCUSSION Participants demonstrated a low rate of understanding of the terms of the CTAs they were asked to read. Participants averaged a score of 33% on the survey, indicating a low comprehension level of author rights. This figure did not vary significantly, regardless of the respondents’ discipline, time in academia, level of experience with publishing, or whether or not they had published previously with the publisher whose CTA they were administered. Results also indicated that participants did equally poorly on the survey regardless of which of the two CTAs they received. However, academic authors do appear to have a greater chance of understanding a CTA when a specific activity is explicitly outlined in the text of the agreement.


INTRODUCTION
Publishers' conditions and restrictions on academic authors' own published work have a significant effect on academic activities pivotal to teaching, research, publication, and scholarly exchange with colleagues and collaborators.For example, funder and institutional mandates increasingly require researchers to make their journal articles open access within a given time frame from acceptance or publication (ROARMAP, n.d.;SHERPA Juliet, n.d.;Suber, 2012); depositing in institutional and subject repositories (green open access) is often the easiest and most affordable way to achieve this end.The complexity of the restrictions imposed by publishers means that it is no longer sufficient for authors to know if they retain copyright or even if the publisher allows self-archiving.Authors must now be aware of the minute details of the CTA, for example, which version of the paper may be self-archived and where, how the publisher should be acknowledged in any reproduction, under which conditions authors may make the article available to their students, and which methods of sharing the published version with colleagues are acceptable.
The growth of the open access movement and rise in institutional and funder open access mandates discussed above have resulted in more publishers explicitly permitting some variant of self-archiving (Archambault et al., 2014;Gadd & Troll Covey, 2016;Laakso, 2014).However, in tandem with this rise, the number of conditions under which self-archiving is allowed have proliferated at a staggering rate.According to one study, between 2004 and 2015, "the volume of restrictions around how, where and when self-archiving may take place has increased 119%, 190% and 1000% respectively" (Gadd & Troll Covey, 2016, p. 1).
By many accounts, academic authors experience a great deal of difficulty navigating these complexities.The literature consistently points to authors' self-reported confusion and anxiety around copyright, publisher policies, and CTAs as a major barrier to self-archiving (Cullen & Chawner, 2011;Frass, Cross, & Gardner, 2014;Kim, 2011).In addition, recent studies suggest that academic authors are active and frequent participants in the posting and distributing of their articles on academic social networking sites in contravention of CTAs they have signed.An analysis of a random sample of papers uploaded to Research-Gate found that just over half the non-open access papers analyzed were posted to the site in violation of publisher policy (Jamali, 2017).In the vast majority of these cases of infringement, the authors would have been permitted to deposit a postprint but had instead deposited the publisher's PDF (which was not permissible according to the publisher's policy).A recent survey of University of Rhode Island faculty corroborates this assumption, with over 60% of respondents unsure of the legality or compliance with publisher policy of posting their articles in ResearchGate (Lovett, Rathemacher, Boukari, & Lang, 2017).This practice, sometimes dubbed black open access (Björk, 2017), has increasingly attracted publishers' notice, resulting in their issuing takedown notices and filing lawsuits in an attempt to combat this practice (Chawla, 2017;Solon, 2013;Van Noorden, 2017).
This study seeks to move beyond academics' self-reported confusion and trepidation around CTAs to empirically test author comprehension of these agreements.We hypothesized that scholars with more publishing experience would demonstrate greater comprehension of CTAs.Furthermore, we hypothesized that authors who received a CTA from a publisher with whom they had previously published would similarly demonstrate greater comprehension than those who not worked with that publisher previously.We also expected that shorter agreements (calculated by word count) would have a positive effect on author comprehension, as this would improve the likelihood that the participants would read the agreement in full and circumscribe the amount of text that respondents would have to search to determine whether a condition of use is stated in the CTA.
By investigating the level and determinants of author comprehension, we hope to empiri-a major barrier to self-archiving in IRs (Cullen & Chawner, 2011;Frass et al., 2014;Kim, 2011).Creaser et al. (2010) report that scholarly authors' top two concerns when it comes to self-archiving are fear of committing copyright infringement and lack of certainty regarding embargo periods.Furthermore, 63% of respondents indicated a lack of understanding of publisher policy as an important or very important factor in failing to deposit an article in an IR (Frass et al., 2014, p. 14).Respondents in another international survey (Spezi, Fry, Creaser, Probets, & White, 2013) reported that determining their rights regarding self-archiving was the most challenging part of the process, with two-thirds reporting some degree of difficulty in doing so.Confusion concerning publisher policies is cited as a main contributory factor in studies on the massive gap between the proportion of scholarly journal articles that could be self-archived and those that are (Covey, 2009;Hansen, 2012;Laakso, 2014).For example, over half of respondents reported being unsure as to which version of an article, if any, they were permitted by their most recent publishing agreement to deposit in a repository or post online (Austin, Heffernan, & David, 2008, p.34).

Authors' Perceptions of Rights Retained vs. Terms of CTAs
In addition to self-reported uncertainty about rights retained in CTAs, many studies find that there is a significant mismatch between authors' perceived and actual rights retained under the CTAs they sign.According to Spezi et al. (2013), authors believe they have fewer rights in general than they do in practice.A 2009 study that compared authors' perceived rights with publishers' policies found more specifically that authors overestimate their rights to self-archive the publishers' PDF, while underestimating the scope of rights publisher policies permit for the submitted and the accepted versions of their articles (Morris, 2009).Creaser et al.'s (2010) findings echo this point.
While much of the literature focuses on the self-archiving rights of authors, there are a variety of ways that authors may wish to disseminate their articles as dictated by the terms of their CTAs.This may include sharing the article with colleagues, incorporating it into other published works, or distributing it for teaching purposes.Morris's study indicated that authors consistently underestimate their rights in this regard (Morris, 2009).
The self-reported anxiety and confusion over copyright constraints paired with the discrepancy between perceived and actual rights accorded by CTAs raises the question of whether authors are unsure about copyright because they do not read their publishing agreements or because they do not understand them.Although there are no studies on this topic in the library or publishing literature, there is a broad consensus in contract law scholarship that very few consumers read standard form contracts (like CTAs) before accepting them (see Journal of Librarianship and Scholarly Communication Bakos, Marotta-Wurgler, & Trossen, 2014).Additionally, some publishers have incorporated the author's acceptance of the CTA into an automated manuscript submission system.This means that instead of receiving a hard copy or PDF of the document that the author can examine at leisure, authors are presented with an online click-through agreement with a checkbox to indicate consent.Recent legal and consumer behavior studies indicate that a small minority of users actually read contracts that are delivered in this manner (known as click-wrap agreements), often agreeing to onerous or one-sided terms and conditions (Helberger, Loos, Guibault, Mak, & Pessers, 2013).While the legal literature suggests that authors simply sign CTAs without reading them, when asked, the majority (with a range of 68 to 78%) of academic authors claim that they do indeed read and/or examine these agreements prior to signing them (Austin et al., 2008;Lawson & Lawson, 2011;Moore, 2011; University of California Office of Scholarly Communication et al., 2007).
As discussed above, previous studies have indicated that authors consistently state that they read and examine CTAs and that confusion and anxiety around copyright has a negative impact on green open access self-archiving.The literature also reveals a discordance between authors' perceptions of their rights and the actual scope of rights retained in CTAs.The question arises as to whether understanding the agreements themselves is a major barrier to self-archiving or if authors are not reporting their actual behavior.Instead of attempting to interrogate the veracity or accuracy of academics' self-reported behavior, our study aims to test how well academics understand the rights retained and relinquished in CTAs when they do read them.

METHODS
In order to determine the level of comprehension of CTAs by researchers we constructed a survey experiment including several testing questions (see Appendix A) to compare results from two different copyright transfer agreements.Specifically, the test questions were designed to gauge respondents' understanding of scholarly sharing in these agreements (e.g., self-archiving, presenting at a conference).We received research ethics board approval from our institution to conduct this survey.
The survey included basic demographic questions (i.e., academic discipline, academic publishing experience, how long ago participants published their first article, past publisher experience as corresponding author with a variety of publishers) alongside the anonymized text of one of two existing publisher copyright transfer agreements.In order to replicate natural conditions as much as possible, the CTAs were kept in their original formatting; however, any identifying information about the publisher (e.g., name, address) was removed.
In order to control for the effect of language complexity and readability on the scores, the SMOG readability formula (as detailed in Hedman, 2008) was applied to the agreements of four major publishers: Elsevier, Wiley, Springer, and Taylor & Francis.To include CTA length as a control variable, we selected one longer and one shorter CTA with proximate readability scores to be used in the experiment.

Participant Population
Participants were drawn from the article abstract and citation database Scopus (Scopus, 2018).Articles in Scopus typically contain the email address of the corresponding author, and this information was used to generate an email list for the sample population.To narrow our population in Scopus, we limited the search to English-language articles published in 2016.The following is the search string used: (DOCTYPE ( ar ) AND LANGUAGE ( english ) AND PUBDATETXT ( 2016) AND ( LIMIT-TO ( SRCTYPE , "j" ) ) AND ( LIMIT-TO ( PUBYEAR , 2016 ) ).The initial search provided a list of 1,656,414 articles.Due to Scopus's export limit of 2,000 items, in order to be able to extract randomly sorted data to Excel, the results were sorted newest-oldest then oldest-newest, then using the Scopus relevance ranking, in each case exporting the first 2,000 results to Excel (Elsevier, 2017).Using the resulting 6,000 items, articles with no e-mail address provided were filtered out and then duplicates removed.This provided us with a sample set of 3,245 articles from which to contact the corresponding authors as participants.
Participants were randomly assigned to one of two groups.The survey distributed to the participants was identical except for which CTA was included.One group received CTA 1, while the other received CTA 2. Participants were initially e-mailed on May 30, 2017.
A reminder e-mail was then sent on June 13, 2017.Of the 3,245 e-mails sent, 91 emails (45 and 46 for CTA 1 and 2, respectively) were undeliverable, leaving a set of 3,154 participants.In total, 122 full responses were received.This represented a 4% response rate overall.
To determine participants' comprehension of the agreements as presented via the survey, we co-created an answer key and then individually scored each respondent's answers.A review of the demographics of each population group (CTA 1 and CTA 2) demonstrated that our randomization process had been successful and that there were similar demographic makeups in both groups.Indeed, difference in means is never statistically significant between the two groups (see Appendix B for balance of controls and Table 1 for the demographic makeup of the two sample populations).

Statistical Analysis
To capture relationships between dependent variables and independent variables, we ran two main statistical analyses: one to capture the effect of the agreement on the overall score and another to capture how well respondents scored on individual questions depending on the agreement they received.

Pooled Analysis: Effect of Agreement on Overall Score
We investigated whether the overall score for each participant's survey was affected by which copyright transfer agreement they received (i.e., CTA 1 or CTA 2).We controlled for potential mitigating factors on the participants' scores by asking a series of demographic questions regarding academic discipline, academic publishing experience, how long ago they published their first article (called "Seniority" in our analysis), as well as past publisher experience as corresponding author with the agreement they received (called "Publisher 1 Before" and "Publisher 2 Before" in our analysis).
For academic discipline, we divided the respondents into two categories: those whose main academic focus was in the hard sciences (i.e., life sciences, engineering, medicine & health sciences, physical sciences & mathematics) and those in the humanities and social sciences (i.e., architecture & planning, arts & humanities, business, education, law, behavioral & social sciences).This division was made as the sample size for many disciplinary categories (e.g., law, humanities, etc.) was too small to complete analysis on its own.
To determine the effect of the independent variables (academic discipline, academic publishing experience as corresponding author, seniority, and past publisher experience) on our dependent variable (score), we estimated our models using OLS regressions.This regression was chosen because our dependent variable (score) is a ratio variable.

Single-Question Analysis: Effect of Agreement on Individual Questions
In addition to determining if one of the demographic variables had an effect on the participants' overall score, we conducted an analysis of each of the individual question's scores using probit regressions and controlling for the same independent variables as listed previously.We chose a probit model, as the dependent variables in this instance were nominal (i.e., a score of one for a correct answer, a score of zero for an incorrect answer).This analysis was done to ascertain if participants were more likely to answer certain questions correctly than others and whether or not participants had higher scores on individual questions according to which CTA they had received.

Effect of Agreement
Our analysis shows that those who received CTA 1 answered questions correctly overall 34% of the time, while those who received CTA 2 answered the questions correctly 33% of the time.The results between the two agreements are not significantly different (see Table 2), and thus the results indicate that which agreement the respondents received had no discernible effect on participants' overall scores.

Demographic Variables
Our analysis additionally showed that the independent variables (i.e.Seniority, Discipline, etc.) had no statistically significant effect on participants' scores (see Table 2) (p<.05).For example, a participant who had published more than seven articles was not more likely to provide correct answers than someone who had published fewer articles.

Individual Survey Questions
While overall respondents obtained the same scores regardless of which agreement they received, there were significant differences in the correct answers for specific questions between the two agreements.Table 3 outlines the probability that a respondent was more likely to answer a question correctly in the CTA 1 treatment than in CTA 2. For example, in Question 1 ("According to the agreement you just viewed you can post the article in a course management system"), participants who had received the CTA 1 treatment were almost 27% more likely to answer this question correctly.Conversely, participants who received the CTA 1 treatment were almost 44% less likely than those who received the CTA 2 treatment to correctly answer Question 7 ("According to the agreement you just viewed, you can post which version of the article to which of the following sites without seeking further approval or information: Personal website").As previously, all demographic control variables were included in the model but had no statistically significant effects.Questions for which there were statistically significant responses are marked with an asterisk below and their corresponding p values. 1 1 For instance: a positive and significant coefficient for Q1 implies that respondents receiving CTA 1 are 26.8% more likely to answer Q1 correctly compared to respondents treated with CTA 2 (the reverse would also be true, e.g. for Q1 respondents treated would CTA 2 would 26.8% less likely to answer correctly than respondents receiving CTA 1).
( In our analysis respondents completing the CTA 1 survey were more likely to answer these three questions correctly:

DISCUSSION
Contrary to what we expected, demographic variables had no effect on participants' overall scores on the survey.We anticipated that participants who had more publishing experiencing as a corresponding author would have more exposure to copyright transfer agreements and thus would have greater comprehension than those who had less experience; this turned out to not be the case.Similarly, we also anticipated that participants who had experience as a corresponding author with the publisher whose CTA they were administered would have higher scores than those who did not.For instance, participants who had had past experience publishing with Publisher 1 AND received CTA 1 had similar scores to those who had not published previously with Publisher 1 and received CTA 1.This lends credence to the perception that even experienced researchers lack understanding of copyright transfer agreements and that no particular population of researchers we identified in our survey (e.g.senior/junior, certain disciplines, past publishing experience, etc.) is more adept at reading and comprehending these agreements.Further, our results belie the assumption of a learning effect; while it might be expected that authors who had published previously with the publisher whose CTA they received would be more familiar with that agreement and therefore more likely to understand it, there was no evidence to support this hypothesis.

Explicit Wording
Although the overall number of correct answers was 33% for both surveys (or close to what might be expected if respondents had guessed at random) this does not appear to be due to random selection of answers by participants.The finding that between the agreements there was a significant difference on which questions they got right or wrong overall suggests that the respondents were indeed reading the agreement and attempting to answer correctly.In reviewing the answers for the five questions for which there was a significant difference between the two agreements, we hypothesize that explicitness was a contributing factor to the respondents' success rate on those questions.We define explicitness as the degree to which the question asked in the survey has an answer that is explicitly found in the text, with a prohibited or permitted activity and any attendant conditions detailed, as opposed to being inferred or not discussed at all.
For example, below are the three questions for which respondents were more likely to answer correctly if they received the CTA 1 treatment.Included under each question is the paragraph from the agreement that directly corresponds.• "Authors may self-archive the Author's accepted manuscript of their articles on their own websites."• Q8 "According to the agreement you just viewed, you can post which version of the article to which of the following sites without seeking further approval or information: Institutional repository."(Possible responses: Publisher version; Accepted version; Submitted/Prior Version; Not permitted for any version) • "Authors may also deposit this version of the article in any repository, provided it is only made publicly available 12 months after official publication or later." Although intuitively explicitness seems to aid in comprehension, and our results bear this intuition out to some extent, it does not appear to be a panacea.Both surveys included several explicit question/answer pairs that did not demonstrate statistically significant results.For example, respondents on the CTA 2 survey did very well in determining which version of the article was permitted for various uses, whereas respondents for the CTA 1 survey did poorly in spite of the answers being clearly stated in the text.While there is no readily apparent reason for this discrepancy, we speculate that this is one area in which CTA length was a factor; since CTA 2 was shorter and outlined fewer possible permitted uses, it may have been easier for respondents to isolate and identify those uses that were made explicit.
In addition, the statement of permitted uses in CTA 1 was organized by what was allowed for each version of the article, rather than by use, which may have confounded participants' efforts to identify the permitted activity.

Assuming Permission If Wording Is Unclear
Further analysis of the data revealed that in instances where the answer to the question was "unclear" (i.e., not explicitly discussed in the text nor implicit), typically 60-75% of respondents chose either "Yes, with conditions" or "Yes, absolutely."Given that both CTAs included as treatments in the survey constituted copyright transfers, authors should assume that they do not retain any rights that are not specifically licensed back to them in the text of the agreement.As such, this is either wishful thinking on the part of the respondents or a tendency to believe something is permitted unless it is stated otherwise.

Negative Response to Republishing in an Edited Book
Another interesting finding was that for Question 3 ("According to the agreement you just viewed you can republish the article in an edited book"), a majority of respondents for both surveys (CTA 1 and CTA 2) selected "No" as their response.This is notable as it was the only question for which more than 50% of respondents selected a negative answer.What is even more intriguing is that, in the case of CTA 1, this usage was explicitly allowed with listed conditions, and in the case of CTA 2, such permission could be inferred.It is unclear why respondents had such a strongly negative (and incorrect) response to this question; however, this finding is in keeping with Morris's (2009) study, which found that authors significantly underestimated the extent to which publisher policies permitted them to incorporate their published articles into subsequent publications.This misperception may be informed by scholarly custom, according to which republishing a piece of scholarly work would necessitate explicit permissions to be sought (Gadd, 2017).

Future Studies and Applications
There are several immediate applications of this research.Firstly, author understanding of these agreements is minimal.Respondents averaged a correct response rate of 33% demonstrating the importance of improving the comprehension of these agreements.Bearing this in mind, librarians who are involved in author rights education and advocacy should focus on providing academic authors with the tools to parse the language and phrasing of CTAs, particularly on how to isolate the conditions that are most important to the authors, such as the ability to self-archive to fulfill open access mandates or otherwise.Librarians should additionally focus on reaching out to related departments on their campuses such as research offices, promotion and tenure committees, faculty councils, graduate student associations, and legal offices to partner in coaching faculty on reading these agreements.Understanding legal documents related to publishing should be a campus-wide effort to engage and inform authors for which the library can take a leading role.
Second, we found that demographic variables and experience had no effect on the results.This finding overrides any assumptions librarians and publishers might have about CTA knowledge and understanding among faculty.For example, while one might assume that those who publish frequently or are senior faculty may be more adept at reading these agreements, this is not the case.Armed with this knowledge, librarians should cast a wide net in their author rights educational and advocacy efforts in order to reach all demographics.This could once again be aided by developing campus partnerships outside the library.
Third, explicitness has a potential effect to improve the readability of copyright transfer agreements.As one respondent wrote, "The wording [of the agreement] is very subjective and is not clear."Improving the clarity of wording in CTAs would have a twofold benefit: it would ensure that what is permitted/not permitted is clearly laid out, potentially mitigating the chilling effect of copyright confusion on author' self-archiving or other scholarly reuse behavior, and greater explicitness would avoid the "unclear" effect, whereby authors assume they have greater permission in the absence of a clear indication otherwise.
It would be interesting to test the explicitness hypothesis further, as a positive finding could be immediately utilized by publishers to their benefit.For example, there have been many conflicts recently between authors and publishers as publishers increasingly take steps to enforce their copyright policies by issuing takedown notices to social networking and university sites and, in some cases, take legal action (Chawla, 2017;Mika, 2017).Our research suggests that in many cases, the authors were not aware that they were breaching copyright by posting the publisher's final version to these sites.It has also been suggested that the language many publishers use on their websites portray their scholarly sharing policies as being more permissive than they actually are, which may lead to further confusion (Gadd, 2017).
In either case, a clearer statement in the CTA forbidding these activities may aid in mitigating such transgressions.Although many authors may still choose to ignore their agreements, it is also likely that some authors would adjust their behavior accordingly.
In addition, increasing explicitness in copyright transfers agreements would improve literacy, and librarians may find more researchers engaging with their author rights as they become more aware of what they are signing.With this in mind, librarians can advocate for more unambiguous language in CTAs wherever they have the influence to do so, be that as members of editorial boards, as providers of education and infrastructure through library publishing initiatives, or through scholarly communication education and outreach and engaging with faculty members and colleagues who sit on editorial boards.

Limitations
Our response rate, though sufficient for running our analyses, was a small percentage of the population that received our study invitation.Further studies could improve the response rate by offering incentives to respondents.Additionally, in attempting to recreate real-world settings by using the actual text and layout of existing CTAs, we acknowledge that even controlling for readability, there may have been elements of the layout or typography that affected respondents and for which we did not control.
Finally, the scope of our study was limited to testing author understanding of two CTAs.Subsequent research could examine the content, length, and syntax of a larger corpus of CTAs.
Where the current study was limited to CTAs that effected a copyright transfer, an expanded study could examine agreements which take the form of exclusive and nonexclusive licenses, including the Creative Commons license variants.

CONCLUSION
Our results indicate that academic authors have poor comprehension of the CTAs that govern the terms under which they may reuse and make available their own published journal articles.This lack of understanding does not vary significantly with seniority, academic discipline, experience publishing, or previous publishing experience with a given publisher.Nor does the length of the agreement seem to increase or decrease levels of understanding.The one factor that appears, in some cases, to have a positive effect on authors' levels of understanding is when permitted or prohibited activities are mentioned explicitly in the text of the CTA.

APPENDIX A Survey Instrument Background and purpose:
The aim of this survey is to investigate researchers' understanding of publisher copyright transfer agreements and the effect that word choice, syntax, and document length have on comprehension.

Study Procedures:
The survey should take approximately 5-10 minutes to complete.You may begin the survey by clicking on "Next" at the end of this information letter.Participants will be asked to read a sample agreement and then answer a series of questions.

Benefits and Risks:
Participating may improve your awareness and/or comprehension of copyright transfer agreements.There are no perceived risks with completing this survey.

Voluntary participation:
Your participation in this survey is completely voluntary.You may choose not to answer any questions, and you may withdraw from the survey at any time before submission of your results by closing your browser or selecting "Exit and clear."Incomplete surveys will not be saved.As the information being gathered is anonymous, it will not be possible to have answers withdrawn once the complete survey is submitted.

Confidentiality and Anonymity:
Your responses are completely anonymous and will not be traced back to you.Data collected via this survey may also be used in research articles or professional presentations.Any shared results will be in aggregate form only. Any free text responses that are shared will have all possible identifying information removed.This  Include the article in a dissertation or thesis.
Reuse tables/graphs from the article in a commissioned textbook.
Present the article at a conference.

2
survey has been approved by McGill University's Research Ethics Board.If you have questions about this survey, please contact either of the researchers of this project: Jessica Lange, McGill University Library & Archives or Alexandra Kohn, McGill University Library & Archives.If you have any ethical concerns or complaints about your participation in this study, and want to speak with someone not on the research team, please contact the McGill University Ethics Manager.By clicking Next, you consent that you are willing to answer the questions in this survey.Please save or print a copy of this document to keep for your own reference.1.Please select your area/discipline.ahow many articles have you been the corresponding author 2 in the past 5 years?With which publishers have you published an article as corresponding author?(Check all that apply.)Corresponding author is defined as the researcher who looks after the manuscript submission process and is the primary contact for the article.
many years ago was your first academic journal publication?(Choose one of the following answerssample publisher agreement.Please read this agreement.You will be asked questions on the content of this text afterwards.[TEXTOF CTA 1 OR CTA 2 HERE] 5.According to the agreement you just viewed you can: in a course management system.Use the article in a course pack / course reader / distribute print copies.Republish the article in an edited book.

Table 1 .
Descriptive Statistics for Variables Used in Analysis

Table 2 .
Share of Correct Answers Controlling for Demographic Variables: OLS Regressions with Standard Errors in Parentheses *p<0.05.

Table 3 .
Coefficient of CTA1 estimates the probability of respondents answering each question correctly when receiving CTA1 compared to when they receive CTA 2. Probit regression with marginal effects.Standard errors in parentheses *p<0.05.