Psychology and the New Media - Week 6 summary [UNIVERSITY OF AMSTERDAM]

It is possible that television viewing is helpful in the development of basic vocabulary but less helpful in the development of abstract vocabulary. Traditional reading appears to be related to the capacity for reflection. Online reading appears to require a different skillset.

The internet is easily accessible (1), has large amounts of information (2) and has interactive information (3). Experienced internet users show more activity in the prefrontal cortex during online activity (i.e. executive functioning and decision making).

The increased possibility for interaction can lead to more involvement and systematic processing but in the case of overload, there is less attentive processing. Superficial interactivity is related to the perception of interactivity and this is related to the appreciation of the website. It leads to heuristic processing. Real interactivity leads to deeper processing but also creates a cognitive load.           

Hypertext documents are not linear as they make use of jumps in the material. Each jump creates extra cognitive load. This is because it requires a choice and can lead to disorientation. The cognitive load in relation to hypertexts depends on the structure of links (1), the reader characteristics (2) and the text properties (3).

A greater working memory capacity makes a jump less disorienting and proficiency in self-structuring the reading experience makes reading hypertexts easier. Proximity of related matters (1), topical phrases (2), prior knowledge of the subject (3) and link previews (4) help form a situation model and this reduces cognitive load while reading hypertexts. Situation models refer to complex mental representations formed when readers integrate the statements in a text with their knowledge.

The flexibility of hypertexts may be beneficial for readers who are positioned to structure their learning activities (e.g. readers with prior knowledge). Low-knowledge readers may benefit from hypertext that transparently conveys the structure of the text.

Choice stress leads to spending less time per option (1), overlooking important features (2) and less of a test of its criteria (3). There is less satisfaction and less willingness to choose with choice stress due to post-choice dissonance and heavier weighing negative characteristics. A person with a maximizing strategy, a lot of choices and less selectivity lead to a worse choice than with fewer choices.

More information in the system does not increase the likelihood that people will encounter more diverse information. The choice strategy someone chooses influences choice stress (e.g. maximizing). A lot of texting during lectures is distracting but no texting during lectures is also distracting for the more recent generations. Directly reacting to text messages results in poorer memory performance during a lecture. A potential solution would be technology breaks.

Using Google has several effects:

• There is an increase in confidence in one’s cognitive abilities.
• There is an increase in confidence in future performance on similar questions.
• There is a misattribution of knowledge; people believe they already had the knowledge they were looking for.

However, these effects only hold with reliable access to a transactive memory system. The transactive memory system consists of the internal memory (i.e. memory storage of the individual) and external memory (i.e. memory storage of the collective about the memory of others). The transactive memory system plays a role in intimate relationships and teamwork. Using a transactive memory system (e.g. Google) leads people to be worse at remembering things (1), be better at knowing where to find the information (2) and having more capacity for creative problem solving (3).

Google shows search results based on popularity and relevance via user data. People tend to assume that popular results (1), preferred results (2) and fast results (3) are indicative of relevance. However, choosing between links may stimulate critical thinking. Distributive practice refers to repeated exposure to the same or meaningfully similar information over time. Repeated exposure to information improves the memory of that information.

Divergent and holistic processing is needed for creative problem-solving. Improving creative thinking may lead students to be better at finding information online because they approach the problem from different angles. New literacies refer to the competency to locate information, synthesize information and answer those questions and then communicate it to others.

Critical evaluation of knowledge online is promoted by a lack of distractions (1), high information value of false information (2), prior knowledge of a topic (3) and a sense of efficacy (4). The filter bubble refers to search engines making results more personally relevant which homogenizes the results. The personalization of search results shapes preferences.

The truth effect states that repeating the same information will make it more subjectively believable. The confidence people have in the answers people give to a variety of general knowledge questions is enhanced when those questions seem more familiar. Online communication can lead to more equality of influence among co-actors.

There is a difference between current academics and students. The academics see the knowledge as an intellectual landscape (i.e. more static) while students see knowledge as an ocean of information (i.e. fluid). Outsourcing the storage of knowledge (e.g. through Google) offers room for problem-solving skills.

The plausibility of misinformation (1), the credibility of the source (2) and the degree to which the information can be vividly imagined (3) are key factors which increase the possibility of creating false memories.

The large scale use of digital technology would enable a big data revolution through datafication (1) and the computational power of technology (2). Datafication refers to the phenomenon where more and more aspects of human behaviour can be analysed as data. This might make correlational research more informative due to big t, big v and big n.

Traditional psychological research usually focuses on some variables using highly structured measures with a small sample. The problems with this are the sample size, the poor generalizability and the static recording of dynamic structures. Big data allows for larger sample sizes (i.e. big n), measuring more variables (i.e. big v) and a more dynamic recording (i.e. big t).

There are several challenges regarding big data:

• Accessing data
  It is difficult to access the data sets and even more difficult to use a manipulation. A potential alternative is collecting own datasets.
• Processing data
  It is difficult to process large amounts of data and this might require more technological skills.
• Measurements
  It requires a more informative linking of measurements to constructs.
• Statistical approach
  It requires a different statistical approach than traditional null hypothesis testing.

One potential pitfall of big data predictions is model overfitting. This occurs due to tuning the model too closely to the training data or due to the changes in algorithm of Google and popular online platforms. Self-selection may be a pitfall of using Twitter data to predict.

It is important to have knowledge of demographic selection in big data due to the potential of Simpson’s paradox and other biases. Interpretation of big data is difficult as data is not objective. Furthermore, more data does not necessarily mean better data. There is a growing digital divide because of several factors:

1. Access to technology (i.e. access of technology of the users)
   This has consequences for the generalizability of research and the availability of participants.
2. Access to data (i.e. access by companies and researchers)
   The access to data improves competitiveness and therefore access to relatively more data.
3. Technological skills
   This has consequences for skill learning, independency and career perspective.

It is difficult to determine the impact of big data research on theory as it cannot be a-theoretical. A statistically significant result does not imply practical significance, even when the sample size is huge. As of now, big data is mostly correlational and thus does not answer why questions.

Articulated networks refer to networks that result from people specifying their contacts through technical mechanisms (e.g. followers). Behavioural networks refer to networks based on communication patterns (e.g. people who text each other). The ethical constraints of big data are important.