At present, it is the golden time for graduate students to do dissertations. I can't help thinking of the relevant situation of reviewing master's dissertations in recent years, especially the common problems found. I think it is necessary to describe and comment on one or two in combination with practical examples, so as to avoid some improper phenomena. Now, based on a master of education thesis I reviewed, I present my comments and express my views. Of course, in order to avoid hurting the author of the paper, I did some technical treatment.

Taking middle school students' learning interest, academic self-concept, academic achievement and perceived teacher support as the research content, this paper investigates the mediating effect of academic self-concept between academic achievement and learning interest, the moderating effect of perceived teacher support between academic achievement and learning interest, and the topic selection meets the professional requirements, The statistical technology used has a certain cutting edge. However, there are some obvious deficiencies in this paper, including but not limited to:

First, the innovation of the research is limited. Firstly, there are few recent references in this paper; Secondly, the results obtained in this paper are mostly repetitive and belong to innovative research conclusions. For example, "this is consistent with the results of a previous comparative study on the dimension of learning ability on learning interest" and "this is consistent with the results of the positive relationship between mother tongue and the first foreign language field found by previous researchers".

Second, the measurement tools are questionable. The "adolescent learning interest scale" used in this paper is not involved in the previous "measurement of learning interest"; It is inappropriate to measure the dimension of "knowledge" with the title of "I did well in Chinese / Mathematics / English class"; The internal consistency reliability coefficients were. 968,. 971 and. 974 respectively, which were too high; In particular, there is a lack of references corresponding to this scale. There are the same problems in the measurement tools of academic self-concept. When introducing "academic self-concept", the scale of "mathematical self-concept" compiled by marsh et al. (2005) was not mentioned. It is problematic to simply replace "Mathematics" with "Chinese" and "English" respectively. The "teacher support scale" also has the same problem. In the previous "measurement of teacher support", this tool is not introduced at all.

Third, the research is suspected of being only statistical. For example, "3.2.2 mediating role of academic self-concept among disciplines", the analysis is complex, boring and close to chicken ribs. Figure 10, figure 11, Figure 14 and Figure 16 are so complex! However, it is difficult to get many meaningful results without thinking about the relationship between relevant variables from theory and practice, relying only on data statistical analysis and just trying to test or verify some theories. Therefore, it is impossible to put forward truly valuable educational countermeasures. The impression of this paper is to refer to relevant theories, complicate simple problems and test them with statistical analysis, but it fails to achieve the corresponding effect.

Fourth, the standardization of the writing is poor. On the one hand, the structure of this paper is inappropriate, some contents should not be presented in the paper (for example, research ideas), and there is a phenomenon of tiger head and snake tail. There are too many contents of "literature review" in the front and too few contents of "comprehensive discussion" in the back; On the other hand, all the research hypotheses involved in this paper appear twice (once in the second part and once in the third part), and the same contents in the research methods are presented repeatedly in "Study 1", "Study 2" or "Study 3" respectively. In addition, from the summary to the appendix, the content of each page needs to be improved.

For the above comments, the following interpretation and explanation are made:

First, the dissertation should be innovative. As a dissertation, we should make real contributions in the corresponding fields. Therefore, we need to be innovative, that is, the work involved in the dissertation can not simply repeat the existing research. Innovation, for academic master's thesis, should be mainly theoretical; For professional master's degree thesis, it should be mainly at the application level.

The author of a dissertation often demonstrates the novelty and frontier of his topic. However, if the dissertation does not have substantive references in the corresponding field, it is difficult to achieve this purpose. If the author does not find a real research problem, there is no basis for making innovative research. It is inevitable that the sentences of "this is consistent with the previous results of..." and "this verifies the research results of..." will appear in the expression of the research results. I pointed out in a blog post that this kind of expression means that the author of the paper has lost his subjectivity and become a footnote to other people's research.

Second, measurement tools should be guaranteed. The empirical paper of psychology analyzes the data obtained by measurement tools. If the measurement tools are not guaranteed, the data quality is difficult to be guaranteed, and the results of this analysis will become more and more unreliable and ineffective. For ready-made measurement tools, especially those prepared abroad, it is risky to use them after a little transformation.

Generally, the reliability of psychological measurement tools is not a big problem, mainly due to suspicious validity. For example, although the title of "knowledge" measured in the above paper is related to "knowledge", the title is ambiguous. "Good performance in class" may reflect that students master knowledge well, or that students abide by discipline and do not master knowledge well. In particular, the current research found that the internal consistency reliability coefficient of a measurement tool is too high, such as close to 0.98, which is problematic, that is, the corresponding topic tests a structure related to the original concept but different.

Another common problem related to measurement tools in the dissertation is that the measurement tools mentioned in the literature review are not used in the method part, which makes those reviews meaningless and the measurement tools used lose their foundation. If the literature review does not involve the measurement tools actually used by the author, it is likely to cause the problem that the concepts mentioned earlier by the author do not correspond to the measurement tools.

Third, statistical analysis should be meaningful. The current empirical research in psychology, especially the questionnaire survey, has an obvious tendency of only statistical methods. It is prone to intermediary analysis, regulatory analysis, chain intermediary analysis, regulatory analysis with mediation, regulatory analysis with mediation and latent variable growth model analysis, but the actual problems are not very important. There is a certain reason for academic master's thesis to do so, while professional master's thesis also does so, it is not necessary, because even if the corresponding model is tested, it is difficult or even impossible to apply in reality.

In fact, when conducting this kind of analysis, the author often does not really understand the principle of the method used, let alone the power of the method used, but simply imitates it, and even asks others to help with complex statistical analysis. This phenomenon is likely to encourage the impetuous style of study and let some graduate students make statistical results that they can't explain.

Fourth, the writing style should be standardized. Psychological empirical papers have a relatively fixed structure. It should be easy and reasonable to write with reference to that format. However, the authors of many dissertations often create another set and go their own way. Such dissertations were seen by other graduate students and imitated. As a result, some obviously inappropriate practices spread. For example, literature review, which was originally a part of the introduction, was written at the same level as the introduction. The corresponding contents were found in other papers, regardless of whether they had anything to do with the specific problems they studied, and were stacked in the so-called literature review part, so that 1 / 3 of a dissertation was literature review. Of course, there is another reason why many dissertations do this, that is, the length requirements of various training units for dissertations. For example, a master's thesis should not be less than 30000 words. The author's research content is not so much, so he has to use literature review to expand his space. This is how the "too much literature review and too little comprehensive discussion" appeared in the above papers.

At present, another common problem in the writing of master's degree thesis is that there are too many text expression errors. For example, typos, more words and less words are everywhere, which gives a very bad impression. It seems that the author has not read all his written materials at all. Of course, some graduate students may not have high requirements for themselves. Just like when sending messages or responses through QQ and wechat, they don't care whether they use punctuation at will or don't use punctuation at all and write words at will.

In short, there are often many problems in the current master's thesis, which requires graduate students and tutors to pay more attention to these problems, work together to improve the quality of the thesis, improve the training quality of graduate students, and transport more and better high-level talents for the country.