Feedback PhD-Lab commission

Categories: Article 1, Reflections
Tags: No Tags
Comments: 1 Comment
Published on: 29/10/2011

About two weeks ago I handed in my first draft article. I received feedback on two aspects: 1) on the content of the article and 2) on my achievements as a PhD-student up till now and, based on that, some advice for the future. The feedback came from Paul de Beer, Judith Metz and Liesbet van Zoonen:

Content of the article

  1. Judith described 3 parts in which my article could be divided:
    1. Finding the organizational model to describe a local memory website with (based on the literature review and fine-tuning of the model by looking at cases from the field).
    2. Mapping the field by comparing the model to a set of local memory websites.
    3. Analyzing the on-line activity and its relations to aspects of the model (passive participation =  visitors and active participation = contributors).
  2. Both Judith and Paul: The parts 1-1 and 1-2 above (organizational model and mapping the field) is enough for an article. This gives the opportunity (demand?) to make both steps more transparent and thus convincing. This becomes a descriptive article which is a bit harder (less exiting) to ‘sell’.
  3. The results of part 1-2 makes 1-3 possible which could also be a independent article based on a question like: “How do organizational aspects or a combination thereof invite people to participate?”
  4. Paul (p. 3) : shouldn’t you use the word dimension instead of category? Yes. (Lesson learned: An earlier reviewer didn’t understand the word dimension, but I shouldn’t have followed here comments so relentlessly)
  5. Paul and Liesbet (p. 5): The research question should be clear much earlier in the article (introduction), at the least roughly. Two remarks on the present question:
    1. “What characterizes …” is too broad; it could become more exciting when you put in something like clusters or patterns.
    2. “…these fifty three cases …” is too narrow, we would like to get the idea that the conclusions (might) apply to all local memory websites.
  6. Paul (p. 5): With respect to inter-rater reliability: how did the two researcher coop with their differences? Explain.
  7. Judith and Paul (table 2 p. 5): For the part where the model is fine-tuned based on the cases (1-1 above), the dispersion of lesser importance. On the other hand, for describing the field (1-2 above) the dispersion is important, just as finding (snowballing) and selecting them is. For this I could use a set of Dutch cases first and then see whether a set of foreign cases make changes in the model. If not, we might say that the model is applicable internationally. If it does, though, then we might be able to claim that there are differences with respect to local memory websites in different nations.
  8. All reviewers (p. 8-9): put these results in tables and in only pay attention in the text to the most remarkable findings.
  9. Paul (p. 11 and further): It is not clear why I use two-step clustering within the dimensions and not across them. The two-step clustering based on log-likely hood produces 2 clusters within the ‘poor model’ area of the results. I could tell the two-step cluster to make for example 5 clusters, then the quality of the model goes up. But, with the hierarchical clustering algorithm it might be easier to decide myself where I make a cut in the dendogram and look at the clusters at that level.
  10. Both: Be careful with claiming that there are relations between characteristics, but also with claiming there is no relation.
  11. Both: percentages (with absolute numbers, because the set of cases is small).
  12. Paul (p. 15) questioned whether I could suggest the existence of three types based on my methodology. I acknowledged that, but now that I look at my text again I doubt whether he was right:
    1. If A seems to have a relation/ co occurs with B and B with C, then it must be allowed to suggest there seems to be a type characterized by A, B and C (?)
  13. Paul thought this peace was a bit disappointing:
    1. “However, it should be noted that a direct relation between involved parties and on-line activity was not identified. In other words: claims about the on-line activity related to the three types of initiatives cannot be made at present. This is one of the methodological consequences of the approach performed here. We applied strong clustering on the levels of all categories in order to make the number of variables manageable and reducing the complexity. At the same time this means a loss of detail about, for example, the involved parties in which libraries and universities were clustered into knowledge institutions.”
    2. When I look at this again, I don’t think there is anything wrong with the conclusion (in bold). The explanation, though, is not adequate and causes the disappointment.  I should have said something like: “This is an interesting finding, because it implies that on-line engagement according to our data relies mostly on the affordances. The hypothesis that involved professionals have an negative influence on the on-line engagement has to be further tested.”
    3. Mental note: I did not cluster the cases into three types to check the relation between type and on-line engagement.
  14. The last thing we talked about was the content. I should make clear that I did not really look at the content, but only at some characteristics of how the content is framed. Like I say in the text, the actual content might be of (social) interest in general, but also for the on-line engagement specifically.


  1. The advice will be send later by Paul, but it is something like “you have done a lot of work in your first year: continue your PhD-project!”. Some tips for me:
    1. There should be more focus in the article; this might mean that I should go for two texts with both their own message.
    2. It is very good to have an empirical start like this in an early stage, but it has taken a lot of energy/ time. The theoretical foundations (e.g how website characteristics invite people to participate) are lacking up till now. Time to put that in.
    3. I should specialize more in quantitative approaches (in general, but specifically in hierarchical clustering techniques).

Share this

1 Comment - Leave a comment
  1. […] Discussed the steps to take next with Liesbet related to  these two posts about the first draft of my article (my own reflections on my article and the commission’s feedback). […]

Leave a comment

Your email address will not be published. Required fields are marked *

Welcome , today is Saturday, 23/06/2018