Recently a preprint was posted at ArXiv to explore the question “Can the Journal Impact Factor Be Used as a Criterion for the Selection of Junior Researchers?“. The abstract concludes as follows:
The results of the study indicate that the JIF (in its normalized variant) is able to discriminate between researchers who published papers later on with a citation impact above or below average in a field and publication year – not only in the short term, but also in the long term. However, the low to medium effect sizes of the results also indicate that the JIF (in its normalized variant) should not be used as the sole criterion for identifying later success: other criteria, such as the novelty and significance of the specific research, academic distinctions, and the reputation of previous institutions, should also be considered.
In this post, I aim to explain why this is wrong (and more, how following this recommendation may retard scientific progress) and I have a go at establishing a common sense framework for researcher selection that might work.
Continue reading “How to select junior (or other) researchers, and why not to use Impact Factors”
This post responds to a question by Matti Heino, partly phrased in this Facebook post, and partly in this presentation.
Wow, good question and points!!!
I’d say, in response to slide 28: yes, they are. A logic model is not a theory. I define a logic model in this context as a model that is built from theories and empirical evidence to try and explain one very specific, bounded scenario. I define a theory as a generic constellation of constructs and (e.g. causal) relationships between those constructs. (PN, e.g., is not a theory).
The goal of theory is to derive abstract laws about reality. Their level of abstraction grants them value; gravity works in general, not only in Padova. “Attitudes predict human behavior” is a theoretical statement. “Attitude predicts physical activity in my specific subgroup” is no longer a theoretical statement: whether it’s true or not tells us little about reality in general.
So, the logic model you construe for an intervention, which you base on theory (but where you deliberately omit variables that are irrelevant in your specific situation, even though you know they can be important predictors of behavior), and which you ‘fill in’ using empirical evidence regarding the beliefs (‘change objectives’ in Intervention Mapping lingo), is not a theory. It’s also not something to evaluate in your intervention evaluation.
It’s something to study BEFORE intervention development (step 2 of Intervention Mapping).
Then, once you have your logic model of change (as IM calls it), you move forward and start matching the relevant determinants to theory. If you don’t know in advance which determinants (and which sub-determinants or beliefs) you should target with your behavior change methods, your chances of success are already diminished before you even started.
So, this is not a matter of testing theory. Intervention evaluation is not fundamental/basic science. It’s application of science. You’re under no obligation to contribute to theory – in fact, you have the wrong design for contributing to theory. Your presentation clearly shows why this is the case.
If you want to test theory, design a study to test theory.
(Similarly, if you’re curious about mediation, design a study to test mediation – i.e. a factorial experiment with multiple measurement moments – and I haven’t checked that paper (“what’s the mechanism”) recently, you might need even more.)
People commonly respond to this by expressing exasperation that it all has to be so complicated. I sympathize, but believe that nobody’s served by conducting invalid science because that keeps things fun and easy.
Only learning one or two things from a study, even one with a huge dataset, is fine. Knowledge is valuable, so it’s ok to have to work for it 🙂
Based on a PsyArXiv preprint with the admittedly slightly provocative title “Why most experiments in psychology failed: sample sizes required for randomization to generate equivalent groups as a partial solution to the replication crisis” a modest debate erupted on Facebook (see here; you need to be in the PsychMAP group to access the link, though) and Twitter (see here, here, and here) regarding randomization.
John Myles White was nice enough to produce a blog post with an example of why Covariate-Based Diagnostics for Randomized Experiments are Often Misleading (check out his blog; he has other nice entries, e.g. about why you should always report confidence intervals over point estimates).
I completely agree with the example he provides (except that where he says ‘large, finite population of N people’ I assume he means ‘large, finite sample of N people drawn from an infinite population’). This is what puzzled me about the whole discussion. I agreed with (almost all) arguments provided; but only a minority of the arguments seemed to concern the paper. So either I’m still missing something, or, as Matt Moehr ventured, we’re talking about different things.
So, hoping to get to the bottom of this, I’ll also provide an example. It probably won’t be as fancy as John’s example, but I have to work with what I have 🙂
Continue reading “Why one randomization does not a successful experiment make”
In statistics, one of the first distributions that one learns about is usually the normal distribution. Not only because it’s pretty, also because it’s ubiquitous.
In addition, the normal distribution is often the reference that is used when discussion other distributions: right skewed is skewed to the right compared to the normal distribution; when looking at kurtosis, a leptokurtic distribution is relatively spiky compared to the normal distribution: and unimodality is considered the norm, too.
There exist quantitative representations of skewness, kurtosis, and modality (the dip test), and each of these can be tested against a null hypothesis, where the null hypothesis is (almost) always that the skewness, kurtosis, or dip test value of the distribution is equal to that of a normal distribution.
In addition, some statistical tests require that the sampling distribution of the relevant statistic is approximately normal (e.g. the t-test), and some require an even more elusive assumption called multivariate normality.
Perhaps all these bit of knowledge mesh together in people’s minds, or perhaps there’s another explanation: but for some reason, many researchers and almost all students operate on the assumption that their data have to be normally distributed. If they are not, they often resort to, for example, converting their data into categorical variables or transforming the data.
Continue reading “On the obsession with being normal”
[ primary audience: behavior change intervention developers ]
Threatening communication is a popular behavior change method used tobacco packaging, to promote seatbelt use and discourage substance use. However, much research also suggests that it is not the best weapon of choice when the goal is to really change behavior, or even when the goal is to raise awareness or educate people.
How is that paradox possible? This blog post will answer that question.
Continue reading “Fear is a bad counsellor”
This post is a response to a post by Daniel Lakens, “One-sided tests: Efficient and Underused“, whom I greatly respect and, apparently up until now, always vehemently agreed with. So this post is partly an opportunity for him and others to explain where I’m wrong, so dear reader, if you would take this time to point that out, I would be most grateful. Alternatively, telling me I’m right is also very much appreciated of course 🙂 In any case, if you haven’t done so yet, please read Daniel’s post first (also, see below this post for an update with more links and the origin of this discussion).
Continue reading “Why one-sided tests in psychology are practically indefensible”
In deze korte post wil ik uitleggen wat je moet doen op pakjes sigaretten. Ik leg kort uit waarom ik fel tegen angstaanjagende afbeeldingen en teksten ben; waarom ze zo populair zijn; en wat ik vind dat je wel op pakjes sigaretten moet zetten. (Haast? Ga gelijk naar de bottom line.)
Continue reading “Gezondheidscommunicatie op tabaksverpakking: angst is een slechte raadgever”
[ This is a Dutch post, as it concerns a “study” by a Dutch TV channel, BNN ]
Op dinsdag 22 september 2015 kwamen er verontrustende berichten de wereld in:
Schokkend feitje nummer 1: 35% van die jongeren zegt meer drugs te gebruiken door de ophoging van de alcoholgrens! Damn.
[citatie van Spuiten en Slikken]
Bijna een derde van de jongeren gebruikt elke week drugs, en een derde doet maandelijks aan drugsgebruik.
[citatie van nu.nl]
Dit lijken ernstige signalen. Gelukkig blijkt bij nadere inspectie dat het onderzoek waar deze conclusies op gebaseerd worden, ongeschikt is om dit soort conclusies te trekken. Er zijn zes serieuze problemen met dit onderzoek: Continue reading “Een niet-representatieve steekproef zegt ja tegen MDMA”
Earlier (ok, in the only previous, first post on this blog) I discussed the recent study of Zachary Horne et al. (2015), where they concluded that threatening communication may be an effective approach to counter anti-vaccination attitudes. One of the problems with this study was that the manipulation was not valid: the conditions differed on many variables, any of which may explain the results they found.
After I deliberated for a while whether to inform the authors of the blog post, I decided to do so in the spirit of academic debate, transparency, and learning from each other. He swiftly replied, and one of the things he dis was correct my assumption that they did not share their data. They did actually share their data! I think that’s very commendable – I strongly believe that all researchers should Fully Disclose. Zachary posted it at the excellent (and free) Open Science Framework repository, specifically at http://osf.io/nx364. After having downloaded the data, I decided to write a brief follow-up post about matching of conditions and validity of manipulations. Continue reading “The importance of matching: a case study”