CHOICE membership

Scoring in Reviews

I noted in the recent tea bag reviews that the decision was made to issue scores.

My question is: Does every Choice review need to include numerical scores?

Take this tea review. Tea is down to a very personal taste, and flavour will be impacted by how it is brewed and served. So these scores will only be accurately reflected if people drink it in the same way as the testers. And even then there was only a 10% score difference between the best and worst teas! Although all foods are subjective, there’s no way to score health, consistency, texture etc. so tea stood out as very odd

My suggestion: There are definitely people who will benefit from a review of these products. But perhaps Choice should consider that a numerical score is not always useful to the reader (one of my favourite teas scored near the bottom in fact). Some products it may be better to simply offer review of the qualities, price and facts of each and suggest a best for certain tastes.

I’d like to know everyone’s thoughts on this.


You are right that the assessment is fairly subjective. The panel were volunteers who were not trained tea tasters. At least 62 tasted each tea and rated on a 7 point scale. How that scale from terrible to excellent was converted to a percentage score is not stated. The narrow range of scores between 70% and 80% is a consequence of three things. The underlying variation in quality, the perception of the panel and the score conversion formula. Included in the perception component is the willingness of people to award high and low scores. Often untrained testers have no way to relate to the scale given and become conservative avoiding extreme ratings.

We are relying on the number of tasters to even out the subjective nature of their taste and any idiosyncratic preparation methods.

If the ratings are not converted to a single score what is the alternative? Publish the spread of ratings perhaps. So instead 0f 75% you might get a frequency table from terrible to excellent that looks like this:
0, 3, 7, 15, 34, 25, 3

How does the average punter compare that to another set of seven numbers?


May I rate that a ten out of ten. :wink:

Any objective taste test should at the very least explain what factors are being valued and what are negatives. The taste of a cup of tea is an excellent example to consider. There are so many variables before the cup is tasted and even more with the actual consumer.

Our number two household tea drinker favours Dilmah extra strength. The key attribute? It produces a satisfactory cup in 10-20seconds of brief agitation of the string. Not in the Choice test method?

By comparison the less important number one tea drinker lets his tea (a cheaper leaf) brew too long and drinks it too cold.

We do agree on skim not full cream, although straight is best for some teas.

Would anyone like extra sugar with that? Some teas are best taken with full cream milk and a good two or three heaped spoon fulls. The tea is cheap, but perhaps not so once enhanced. :joy:


I’m not a tea drinker
This kind of feedback helps me decide what I buy for guests
Maybe Choice could ask us what tea we buy and include this in the review. This could extend to other food and beverages tested by choice.
I find the numbers useful - noting they alone should not determine the purchase.


The thing is though ultimately this review indicated that on an score basis there’s little difference in anything tested

1 Like

Or that the testing and scoring methodology used was not able to detect much difference, the two are not the same.


Thanks for the feedback @tpeter267 (and sorry for my slow reply). The challenge from our point of view is to give people a quick indicator of our findings. However, I think you make some good points and for anything subjective like taste, it always comes down to personal preferences. I’ll will pass this on to our staff that run these tests and see if there’s a way we can incorporate this in the future.

@jonat186 - Keep an eye out on the food and drinks section of this forum, we’re running lots of polls and surveys to help inform our work :slight_smile: