Analyzing the Charts – should position count?
When analysing a CHART, i.e. a series of data that has been put in a particular order, should you take the ORDER into account, or simply the presence of the item in the list?
For Example:
I am hoping to assess the presence of rock / guitars in the charts over a period of time. Should I refer to / use the fact that a song with a guitar in it is at Number 12 in the charts, or simply that it was in the charts?
Surely it would be a more accurate representation of the impact of GUITARS / ROCK if I was to somehow use it’s chart position as a factor? If I was lookint at a Top 10 – Number 1 scores 10 points, number 2 9, etc)
If this was a formula to work out odds on a racehorse, insurance premiums for certain drivers, then placement within the table would be vital.
Likewise with this – the higher the album is in the charts, the more copies it sold, compared to the other items in the list.
(this does not take into account the NUMBERS sold, however – the Number 1 album may have sold a great amount more than #2). Without exact sales figures for an extended timescale, I am afraid this will be impossible to pursue.
Also – thanks to @bobbysparkes on Twitter for this message:
My sidestepping the issue of NUMBERS of sales we also then treat every position equally – and so ignore (for now) the gradual decline in album sales. 1970 will be compared on equal terms with 2010) as we focus on genre, not sales.
———-
Once I have secured a suitable data source (ideally end of year album charts from 1990 – 2010) then I can begin to categorize them.
I have just drawn up an Excel spreadsheet which requires minimal calculation/manual input:
| position | title / artist | acoustic guitar? | electric guitar? | guitar total | rock | pop | urban | dance | misc |
| A | B | C | D | E | F | G | H | I |
B – I – to be manually completed (I’m thinking of introducing another crowdsourcing experiment here)
Enter 1 for yes 0 for no
A separate part of the spreadsheet will calculate the score for each song and it’s categorization, and a third (not shown) will calculate the totals for each data set (ideally a YEAR)
| acoustic | electric | guitar total | rock | pop | urban | dance | misc |
| =B*(41-A) | =C*(41-A) | =D*(41-A) | =E*(41-A) | =F*(41-A) | =G*(41-A) | =H*(41-A) | =I*(41-A) |
HOWEVER …
When I actually analysed 2 sets of data (1 that took into account position, another that didn’t) these were the results
The Rating System shows – as expected – a lot more details – bringing in more rises and falls in the line.
Interstingly, we are no beginning to see a definite No Guitar Up when Guitar is down.
REALIZATION: But then , I WOULD – I am only basically comparing 2 elements – and out of 10, when 1 is up, the is going to be down. Definitely a case of being too close to the project at hand.
I need to try this theory with a larger data set – genre.
——————————————————————————–
Is rock dead? Support blog 
I think this could work and will certainly give you more specific results. You can always go less specific (i.e. simply “did it chart”) from the same data for an overview/summary to your study, but would be more difficult to go “more specific” if you decide you need to later, you may run out of time.
I would be fascinated to see the results incorporating chart position and I reckon the more detailed data you have, the more interesting things you could do with it.
Looking forward to the crowd-sourcing bit!
I’m thinking of gathering as much data as I can, you’re right – it’s harder to get it afterwards!!
Plus it will all be interesting for another project in the future perhaps!