Well if you don’t know who is James Gaskin and you work with structural equations models with SPSS AMOS, you have definitely missed something! This guy is simply awesome and i love his work !
James holds a youtube channel wich he updates frequently about this topic. He offers for example Bootcamps (accelerated training sessions), and many recordings of his interventions on the subject for free! you can also find datasets to work with and enhance your usage of this method!
He also proposes a bunch ofsmart and handy plugins you can use to make your work easier with SPSS Amos! For example drawing your model directly from the SPSS PCA output! this is really a very cool work that deserves to be praised and encouraged!
Thanks James! you rule😉
The Statwiki of James: You can see here the outline of his courses, download the datasets and plugins
What should be the top 3 priorities for people working in analytics for marketing?
Attribution, Attribution, Attribution. Without attribution, what’s the point of measuring anything? You might be able to say traffic is up, or traffic is down, but you won’t have the slightest idea why, or what to do to change it.
But attribution isn’t easy, it takes a lot of knowledge and some clever tech to get it right. For those looking at improving their existing attribution methods, it is helpful to know what top marketers find the most effective way of achieving accurate attribution.
A new report from AdRoll can go some way to providing the answers. By asking a mix of client-side marketing professionals and agency marketers (75% in-company, 25 % agency) in the UK, France, and Germany, they discovered what types of attribution marketers found the most effective.
Algorithmic came top of the pack, with a whopping 96% of marketers reporting that it is very or somewhat effective.
The rise of gadgets in the digital age has led to a decline in human attention span, which is less than a goldfish. Photo courtesy of Shutterstock
In the digital age, where the news is limited to 140 characters and conversations take place in the form of emojis, our attention span has shortened. A recent study by Microsoft Corporation has found this digital lifestyle has made it difficult for us to stay focused, with the human attention span shortening from 12 seconds to eight seconds in more than a decade.
The rise of gadget use in the 21st century means our lives have become increasingly more digital at home, work, and school. In the U.S., nearly two-thirds of Americans are smartphone owners, with many users utilizing their phones for online access, according to a recent Pew Research Center poll. More than half of these smartphone users admit to using their phone in the past year to look up information about a health condition, but at what cost?
In the 54-page study, Microsoft sought to understand what impact technology and today’s digital lives are having on attention spans. The researchers collected data from surveys of more than 2,000 Canadians over the age of 18, who played games online to determine the impact of pocket-sized devices and the increased availability of digital media and information are having on everyday life. The researchers also monitored over 100 people’s brain activity with in-lab monitoring, using electroencephalograms (EEGs).
For the survey component, the researchers sought to gauge overall attention and gauge habits and perceptions by dividing the respondents into three equal sized groups based on performance — low, medium, and high attention — representing one-third of the sample. For the neurological component, participants’ brain activity was recorded and behavior was filmed while they interacted with different media and performed several activities across devices and in different environments. EEGs were used to measure their attention levels and activities were mapped against tasks and behaviors to view how attention varied by screen, task, content type, and structure.
Does it feel like every day you see a new article on marketing metrics? Lately, there has been a lot of buzz around the RIGHT metrics to focus on. Today it’s pipeline, yesterday it was MQLs, and last week it was customer acquisition cost (CAC).
In marketing, the one thing that is a constant is change, and it can feel overwhelming to finally agree on a set of metrics with your team, then go to a conference or see a tweet from an analyst and realize that you’re missing five more.
Today, I saw that pipeline is the #1 thing your business should be looking at if you’re in B2B. I can’t say I disagree with this, but I do really think it’s less about one metric like pipeline and more about what metric matters for when you’re measuring.
Here are a five things I’ve learned over the years about marketing metrics, and while they may not be the trendiest, they still hold true to this day:
1. It Starts With Understanding Your Business Objectives
Be clear on your business objectives first, then determine what metrics to measure. One way to do this is by mapping your business stages to the customer journey. In the example below, the customer journey is broken into stages, including: awareness, engage, convert, retain, and advocacy, but they may differ for your organization depending on your business model. These stages represent the different business outcomes your company is driving towards (e.g. retention or share of voice), and it’s critical to establish this first to determine the right set of metrics and timing that map to those objectives.
Thriving in today’s complex and ever-changing environment means companies need to gain rapid insights and translate those insights into actions. View this webinar to discover why predictive analytics has become an imperative for organizations as they strive to incorporate data-driven decision making into their processes by understanding potential future outcomes. Prescriptive analytics can then be applied to help determine the best solution or outcome among all the various choices, given the known parameters.When combined with predictive analytics, prescriptive analytics can suggest decision options for how to take advantage of future opportunities or mitigate future risks—and answering the eternal question: What should we do about this?
The following is a guest post by Steven Wastie, the CMO of Origami Logic. While it naturally aligns with his company’s solution, I think his Marketing Signals Framework — and this absolutely epic Periodic Table of Marketing Signals — transcends their own product and is a powerful concept for marketing technlogy management. CLICK ABOVE FOR A LARGER VERSION
As a modern marketer, you’re overwhelmed with information, all the time — and there’s no end in sight as complexity and the pace of change shows absolutely no sign of letting up.
Your job looks nothing like it did five years ago — even one year ago. Yet despite this permanent state of change, you are, more than ever, accountable for knowing exactly what’s happening across your organization’s eclectic mix of marketing activations at all times. So you’d have no problem explaining how your campaigns are performing right now, today. Right?
It’s a simple question, but for most, it’s incredibly difficult — if not impossible — to answer. Mastering measurement has always been hard and, for many, even aspirational. As measurement vectors and approaches become increasingly diverse and sophisticated, many marketers are more than a little overwhelmed.
Visualizing relations between data instances can tell us a lot about our data. Let’s see how this works in Orange. We have a data set on machine learning and data mining conferences and journals, with the number of shared authors for each publication venue reported. We can estimate similarity between two conferences using the author profile of a conference: two conference would be similar if they attract the same authors. The data set is already 9 years old, but obviously, it’s about the principle. We’ve got two data files: one is a distance file with distance scores already calculated by Jaccard index and the other is a standard conferences.tab file.
Conferences.tab data file with the type of the publication venue (conference or journal) and average number of authors and published papers.
We load .tab file with the File widget (data set already comes with Orange) and .dst file with the Distance File widget (select ‘Browse documentation data sets’ and choose conferences.dst).
Le système de santé français connaît des crises structurelles récurrentes. Alors que l’inertie des politiques publiques risque d’en accentuer sa déliquescence, la e-santé apparaît comme une solution particulièrement adéquate.
Le mythe de l’Etat providence, grand ordonnateur de différents pans de la vie des citoyens, a définitivement volé en éclats. La révolution copernicienne induite par les nouvelles potentialités offertes par les TIC et le web social a fait advenir un nouveau paradigme qui redéfinit complètement les relations sociales. L’empowerment croissant des individus, dont les réseaux sociaux sont l’un des derniers avatars, a pavé la voie à l’établissement de nouveaux rapports de force entre l’individu et les structures holistiques.
Disruption, ubérisation et maintenant blockchainont contribué à questionner les piliers d’airain de systèmes, économiques, technologiques ou administratifs, que d’aucuns appréhendaient comme intangibles. La santé apparaît aujourd’hui comme le nouvel espace à conquérir et à disrupter.
Toutefois, contrairement à la disruption des services, tel que les transports de personnes, la livraison à domicile, le monde de la finance ou du droit, le secteur de la santé se caractérise par une forte barrière à l’entrée. Législation stricte en matière de santé, servitudes administratives, importance d’un bouleversement comportemental des parties prenantes ou encore sécurité des données constituent autant de blocages pouvant nuire au développement de la digitalisation de la santé.
Le secteur de la santé n’a jusque-là guère était touché par ce bouleversement paradigmatique et le développement exponentiel depuis la fin des années 1990 des forums santé (Doctissimo étant le plus le connu) ne saurait masquer la profonde inertie inhérente à cet écosystème.
Or la symbiose croissante entre les technologies de l’information et de la communication (TIC) et l’Internet des Objets (IoT) promet, en puissance, de modifier durablement le rapport des patients à la santé.
Voici l’une des méthodes les plus riches et intéressantes déjà mise en place à partir des années 1970 au MIT pour étudier l’effet des systèmes et des acteurs. Les travaux de Meadows (The limit of Growth) qui se révèlent par ailleurs très pertinents chaque jour un peu plus sont une brillante démonstration de ces méthodes et leur efficacité.
Cette modélisation permet de prévoir et intégrer plusieurs scénarios et voir les effets d’interaction possibles entre des acteurs et des dynamiques et les résultats potentiels. Certains journaux académiques en gestion comme le (Journal of business) ont déjà réservé un numéro spécial au sujet, mais ces méthodes restent assez peu connues et diffusées. Il y’a une courbe d’apprentissage et d’investissement à y consacrer, mais la méthode est assez riche et très peu mobilisée en gestion, notamment pour le marketing et la stratégie par exemple.
Ce genre d’approche se prête très bien à des modélisations d’organisations, ou d’acteurs, etc … et ce pour différents types de flux étudiés.
Il existe déjà des modèles par exemple de diffusion d’innovation et d’adoption, des modèles aussi pour le marketing qui sont mêmes brevetés. Par exemple, l’une des questions des cabinets de conseil actuels ainsi que des marketeurs c’est de mettre en place des réponses à ce qu’on appelle l’attribution marketing. C’est à dire quels sont les canaux, actions ou acteurs qui ont mené à la conversion? ceci requiert une vue systémique et pas une réflexion en silo comme c’est le cas aujourd’hui.
Ce billet est juste une première brève introduction sur un sujet qui est assez passionnant à explorer.
Pour plus d’informations voici quelques travaux qui ont mobilisé cette méthode:
Meadows, D. H., Meadows, D. L., Randers, J., & Behrens, W. W. (1972). The limits to growth. New York, 102.
Sterman, J. D. J. D. (2000). Business dynamics: systems thinking and modeling for a complex world (No. HD30. 2 S7835 2000).
Dans cette conférence, Pablo Servigne, Raphaël Stevens, co-auteurs du livre « Comment tout peut s’effondrer », retrace la littérature scientifique sur l’effondrement systémique à venir et la psychologie de l’effondrement. Ils décortiquent les ressorts d’un possible effondrement et proposent un tour d’horizon interdisciplinaire de ce sujet – fort inconfortable – qu’ils nomment la « collapsologie ». Enfin, Geneviève Azam (porte-parole du mouvement Attac et économiste) intervient pour exprimer son analyse sur le livre et la problématique de l’effondrement.
» Il a l’argent, il a le pouvoir, il a une Audi….. Il aura la femme », voilà la célèbre formule d’une publicité pour une voiture datant de 1993, formule qui en dit déjà long sur les représentations et mythes véhiculés par la publicité.
Dans cet épisode, Gull enquête sur la mythologie contemporaine, plus particulièrement l’automobile dans la publicité. Car, oui, nous n’avons jamais été aussi imprégnés par les mythes qu’à notre époque.
Très souvent toutes les méthodes de tests d’hypothèse et d’expérimentation classiques se basent sur des tailles d’échantillon fixées au départ. Aujourd’hui, on présente une méthode assez peu mobilisée et pourtant très intéressante qui permet d’utiliser des échantillons dont la taille n’est pas connue à l’avance. Ainsi, les données sont évaluées au fur et à mesure de leur disponibilité et le test s’arrête selon une règle prédéfinie une fois qu’on a obtenu des effets ou résultats significatifs. Assez souvent, ce type de méthodes permet donc d’avoir des résultats statistiquement fiables avec des tailles d’échantillon réduites au strict minimum.
Il existe plusieurs familles de tests statistiques conçus spécifiquement pour ce type d’approches, comme par exemple: (Test du rapport de vraisemblance, Test triangulaire, Test restreint, Test répété, Taux de signification à la sortie, Facteurs pronostiques).
Le test séquentiel a 3 hypothèses et non 2 comme pour les cas classiques:
1- H0 Hypothèse nulle
2- H1 Hypothèse alternative
3- H2 Besoin de plus de données, information insuffisante
Ce genre de méthodes peut être utilisé pour de l’ A/B Testing par exemple, ainsi que toutes les dispositions expérimentales ou on peut avoir besoin de prendre une décision chemin faisant. Certains travaux remontent au début du siècle dernier ou la méthode était déjà utilisée pour des essais cliniques, mais c’est une approche originale qui mérite d’être redécouverte notamment dans les recherches en management!
Mukhopadhyay, Nitis, and Basil M. De Silva. Sequential methods and their applications. CRC Press, 2008.
Doucet, Arnaud, Nando De Freitas, and Neil Gordon. « An introduction to sequential Monte Carlo methods. » Sequential Monte Carlo methods in practice. Springer New York, 2001. 3-14.
This is not a joke nor a teasing title! Medical studies, humanities, business research, psychology, etc … everybody is concerned! Actually, it’s a very serious issue about the flaws of P-value used as a scientific objective mean for rejecting the null hypothesis. This kind of statistical validation is being held in all scientific fields as the mainstream way to establish a rule or a scientific law or make a result accepted and published. The way we report our experiments via P-value gives as the false illusion of precision and validity!
In a famous paper published by Loannidis in 2005 (Why Most Published Research Findings Are False) the author conducted simulations proving that most scientific results are more likely to be false than true. The chase of statistical significance is not a warranty that the results are true. According to the author the bias could be coming from many elements such as : weak effects, experimental design flexbility, sample sizes, etc… Obtaining a significant P-value can be the result of chance and not the evidence of an effect (the opposite is also true, where research can be unable to observe a true hidden effect).
Pr. Geoff Cumming defends the idea of a « New statistics » way of working to ensure that we report in research the confidence intervals instead of the misleading P-value. The above video is a simulation explaining how P-value is also in someway a kind of -random metric- depending on other parameters! and that the only way for science to advance is to express results as an estimation or likelihood confidence interval to compare.
Researchers should conduct many experiments and encourage metanalysis that aim to make those confidence intervals smaller (more precise) by replication and comparing if those effects can validate (or invalidate) our expectations or research conjectures.
This approach will reset the way we do research by creating incentives to search for the cummulative truth, instead of looking for an artificially probably biaised P-value. It’s also a more humble, honnest and precise way to share scientific research findings and to boost the cummulative aspect of science instead of seeking unreliable novelty or working with a silo research mindset. Our P-values gives us very poor information and exposes heavily research to the danger of false inferences.
Ressources to consult:
www.thenewstatistics.com : The excellent website of Pr. Cumming, you can download here ESCI software (Excel files) that allow you to run experiments and simulations manually & discover how P-value can be tricky and misinform research.
Buy or rent the book : Cumming, Geoff. Understanding the new statistics: Effect sizes, confidence intervals, and meta-analysis. Routledge, 2013. (A new version will be published in August 2016)
