What makes this tricky is that you don’t know what norm group you're being compared against; it could be everyone applying to that firm, or a broader population. You also don’t know how the firm is using the result. Some use it as a hard filter, others as just one factor among many. For now, I'd just focus on trying to get better and understand the type of reasoning demanded by each section.
From what I've seen the percentile will be relative to a reference group set by the provider rather than applicants to the firm. (It is possible to pay to request a personalised reference group, but this doesn't seem useful for law recruitment).
Getfeedback provide a "law graduates" reference group as well as "UK general population". (Along with other less relevant comparators.)
The firm can then determine its own cut-off, e.g, "70th percentile law graduate"
The issue with percentiles is that the number of questions is quite small - the standard WG test given by law firms is the less reliable 40-question 'short' test. This should reduce reliability by about √2 compared to a full 80-question test.
According to getfeedback's data, of all their available comparison groups, the group with the highest scores are law graduates, just ahead of senior management, which are in turn ahead of legal professionals.
Pearson claim referring to the WG III (2019) technical manual that the standard error of measurement is only 0.41. I.e., about 68% (1 sd from the mean in a normal distribution) of an individual candidate's scores would be +- 0.41 from the given raw test score
https://www.talentlens.com/content/...CTHub/Pearson_Efficacy_Watson-Glaser_2020.pdf
I find this implausible but don't have a copy of the 2019 manual to check.
I did find a copy of WG II (2009)'s manual, which says the SEM is 2.63, which seems more reasonable.
It also provides these statistics:
* standardisation group mean: 27, sd 6
* nursing students 27
* railroad dispatchers 25
* management position applicants 33.5-34.0, sd 4.2
* bachelor's degree holders 27.6
* small group of employees in a company rated by HR as critical thinkers 32 sd 3.9
* comparator group of those rated poorly by HR on same metrics 25.5. sd 6.7
It also says that a score of 35 for the Manager reference group is 86th percentile.
86th percentile is a z-score of 1.08. I estimate s.d of 5, so that would make the manager's group's mean ~30
It's not clear what the mean raw score is for law graduates, but it's evidently over 30.
If we said for example, law graduate mean = 31, sd 5, and UK population 27, 6 then:
25 - 12th/43rd percentile for law graduates/UK population, respectively.
26 - 16th/50th percentile (ditto)
27 - 21/57 (ditto)
28 - 27/63
29 - 34/69
30 - 42/75
31 - 50/80
32 - 58/84
33 - 66/88
34 - 73/91
35 - 79/93
36 - 84/95
37 - 88/97
38 - 92/98
39 - 95/98
40 - 96/98
It's also provided by Pearson that candidates rated below the 30th percentile are 'below average', 30th-70th percentile is 'moderately skilled', and 70+ is 'highly skilled'.
Thus it might be reasonable to reject those below the 30th percentile in comparison to law grads, even though this is fine for the UK population.
It's not entirely clear, further to my point above, what the standard error measurement actually is, but let's just treat our score as ± 2. Thus a candidate who scores what I'm guessing is average for law grads 31/40, could on a bad day score 28, which the firm should reasonably reject.
Similarly if we want to be sure that we score in the 'highly skilled law graduate' category, which I'm saying here is 34/40, then we'd need to consistently score 34+, so a safer target would be 36.