I'm going to focus on the first set of related questions about the German study, as per eNotes' policy. If you have more questions, you can submit them separately.
The study concerns the anchor effect, a heuristic where people rely too heavily on an initial piece of information when making a judgment. Specifically, the researchers showed that our judgments involving the correct value of a number—in this case, how long criminal sentences should be—are heavily influenced by exposure to other numbers, even random numbers.
In this study, researchers conducted four different experiments. In each one, researchers presented volunteers (judges and lawyers) with a packet of evidence regarding a fictional criminal case wherein the jury had issued a guilty verdict. They were also given relevant passages of the penal code (to tell them the minimum and maximum sentences stipulated by law).
After being allowed to review the materials for 15 minutes, these legal experts were asked to give their judgment as to what the defendant's sentence should be.
If the volunteers' judgments reflected purely fair and rational processes, they should have based their judgments on the information packets alone. The researchers wanted to know if these highly-trained individuals—who were accustomed to following a large body of procedural rules and deliberating in a rational way—would fall victim to the anchor effect. Before the volunteers delivered their sentences, they were given one additional piece of information—an outside suggestion about how long the sentence should be.
These outside suggestions were irrelevant. As you note in your question, one was framed as the suggestion of a journalist. In two other cases, the suggestions were explicitly acknowledged as random. Nonetheless, in every case, these legal experts were swayed by the suggestion. For example, in a rape case, half the volunteers were told to imagine that a journalist called them on the phone and simply asked:
"Do you think that the sentence for the defendant in this case will be higher or lower than one year?"
The other half were told to imagine that the journalist had asked if the sentence would be higher or lower than three years.
The judges were also told to imagine that they refused to answer the question and hung up. This simple, imagined phone call was enough to influence their sentences. Judges exposed to the "one year" suggestion issued, on average, a sentence of about 25 months. Judges exposed to the "three year" suggestion recommended an average of 33 months.
The same effect was observed in experiments where volunteers were given a suggestion they were told was totally random:
For experimental purposes, the following prosecutor's sentencing demand was randomly determined, therefore it does not reflect judicial expertise.
Researchers even observed the anchor effect when the only suggestion was a random number that the volunteers generated themselves by rolling the dice! Simply having that random number as a basis for comparison led to significant differences in sentencing times for a shoplifting offense—7.8 months versus 5.28 months.
Why were the recommendations so powerful? Why do anchor numbers have so much sway?
Evidence from previous studies suggests that selective accessibility plays a role, and researchers gathered more evidence for that explanation in this criminal sentencing study.
The idea behind "selective accessibility" is that thinking about the anchor number—as we inevitably do when we hear it—makes it easier for us to retrieve information that is consistent with that number. Thus, if somebody puts a higher number in your head, it helps make more accessible any information about the case that is consistent with getting a heavier sentence (e.g., the defendant used a weapon). Conversely, if someone puts a lower number in your head, it makes it easier for you to retrieve the information about the case that would be consistent with a lighter sentence.
In a fourth experiment, the researchers tested this idea by measuring how long it took people to classify different types of evidence after being exposed to an anchor number. They found that legal experts were quicker to identify exculpatory evidence when they had been exposed to a low anchor number. They were quicker to identify incriminating evidence when they had been exposed to a high anchor number.
Englich B, Mussweiler T, Strack F. 2006. Playing dice with criminal sentences: The influence of irrelevant anchores on experts' judicial decision making. Pers Soc Psychol Bull. 32(2):188-200.